Home Uniquely NZ Travel Howto Pauline Small Firms Search
Upgrading Corinna's Electrical System
With Solar Power, Battery Monitoring and a Sine Wave Inverter

Introduction

We have been using the same power system on our narrowboat Corinna since she was new. A major component is an independent 24 volt system used purely for an inverter providing 230 volt power enabling us to use a mains fridge. It worked well when we were traveling for long distances every day but we have always had problems with rapid battery degradation which has become more acute over the years. This is partly because modern basic batteries have a lower specification in terms of life cycles and partly because we suspect as the fridge has become less efficient as it has aged. The result is that we have increasingly needed to run the engine for a period of half to one hour in the evenings, especially if we have had a short day, to avoid running the batteries down to an unacceptable level overnight. Even then we are lucky to get a couple of years life out of the batteries for the 24V system. Just after another set of batteries started to fail I saw that Sterling were selling Lithium batteries at a more sensible price of ~£550 per 100 amp/hr battery. Lithium batteries are good for over 5000 cycles to over 80% depth of discharge and offer a life of up to 20 years. That provoked me to look at upgrading the electrical system on Corinna. Initially I looked at switching to a 12V system using Lithium batteries with a common bank for the housekeeping and inverter allowing me to have three Lithium batteries which would more than double my existing usable capacity and outlast Corinna but that was nnot to be.

There is little doubt in my mind that Lithium batteries are the correct solution if one is starting a boat design from scratch but, to cut a long story short, they pose considerable problems in location for us as they have a much more restricted temperature range compared to Lead acid batteries. We can not meet the temperature constraints with our battery location in a tiny engine room with keel cooling tanks which means the batteries can reach over 50 deg C in the summer yet have little protection against low winter temperatures. Lithium batteries are also much less tolerant when it comes to charging etc. and despite most having a battery management system built in to protect them trying to charge a Lithium battery directly is a good way to destroy a standard alternator. Despite rejecting an immediate switch to Lithium it started a search for other alternatives to solve my problems and the information and ideas I gathered had a big impact on my thinking and the options I should leave open for the future. For example, the Sterling Lithium batteries need to be charged via a battery to battery converter to both protect the alternator and charge the Lithium batteries in safe way - without such a converter their guarantees are void. These battery to battery converters can interface between 12 and 24 volt 'domestic' systems and provide a way of using a 24v system with a 12 volt alternator using the engine start battery as the 'buffer'. One of my main worries with my existing system is that the 24 volt alternator is probably no longer replaceable and a battery to battery could be a simple way to replace the alternator and provide an optimised charging environment at the same time.

Initial Objectives of Upgrade

It is difficult to know the best order to document the way the new system design has developed as there has been much iteration. However laying down ones objectives is always a good way to start! The requirements were comparatively soft (ie not quantitative initially) and could be satisfied in several ways. For a starting point, the main objectives were to:

Some of the main ways the first two objectives can be addressed are:

Background - Lead Acid Batteries

Before we can go further I have to give some background on Lead Acid batteries to cover the first two points above. The specified capacity of a Lead Acid battery is not fully available in practice for a number of reasons. Firstly the specification assumes a slow discharge rate, normally C/20 where C is the nominal capacity and even a C/10 rate reduces the capacity. When used with an inverter the demand rate will be much higher when, for example one is running a microwave or boiling a kettle. Secondly the life in terms of cycles is very dependent on the depth of discharge and discharging to 50% may only provide 70 cycles with a basic leisure battery and perhaps 300-800 with the much more expensive sealed AGM (Advanced Glass Mat) batteries. (NCC Verified Leisure Battery Scheme Information).

Charging also plays a major part. An alternator initially gives virtually its full rated charge rate then slows as it approaches the desired voltage then holds it. The final voltage should be chosen to be such that the current taken by the battery tails off to a few percent over time, any higher and the excess charge breaks down the electrolyte and produces hydrogen and oxygen, an explosive combination and this gassing also means the battery needs frequent topping up. The stage at which the fixed voltage charge is first reached corresponds to only ~80% of full charge if the battery has been significantly discharged and it may take several hours to complete the charge with a steadily decreasing current. Stopping the charging too early is not good for the battery if repeated often and we are only left with greatly reduced usable capacity if we stop charging as soon as voltage limit is reached. Overcharging breaks down the electrolyte into hydrogen and oxygen not only drying out the battery but producing an explosive mixture. Standard Alternators can not be adjusted to suit different types of batteries.

The best chargers will charge at their highest rate initially (Bulk charge), then maintain the fixed voltage (Adsorption charge ) for a period calculated based on the discharge history to reach full charge and/or by monitoring the reducing charge current. Then one can change to a much lower voltage level to maintain the battery at full charge which can be continued for ever without gassing (Float charge). The voltages vary a little with different types of Lead Acid battery construction but typically '12V' batteries use an 'Absorption' level of 14.4V and a 'Float' of 13.8V. In addition the levels are temperature sensitive and better chargers will have temperature sensor on the batteries and a voltage measurement on the terminals to compensate for temperature and voltage drops in the cables.

There are alternator controllers which can provide the ideal behavior for Lead Acid batteries but they mostly require modifications to be made to the alternator. My 24V alternator has already been modified to have lorry controller which allows several voltages but it still takes a long time to add the final 20% of charge - fine if one is running most of the day but no good if you are stationary and want to minimise running time. An alternative to an alternator controller is a Battery to Battery charger. They usually sit between an engine start battery (charged by an alternator) and the Leisure batteries. Good battery to battery chargers are expensive but provide an optimised charging profile and can convert from 12 to 24 volt at the same time. They are almost mandatory for Lithium batteries and can remove the need for an extra alternator on a 24V system.

It is actually surprisingly difficult to find Good background and reference information on management of Lead Acid batteries with sufficient depth to be useful. One site which I have found useful is the Car and Deep Cycle Battery FAQ 2021 by Bill Darden which I have used for some of my reference information.

Adding Solar Power

I chose an approach for the first phase of my upgrades which helped several of the existing problems, namely to add some solar cells with a sophisticated controller which charges batteries with a multistage adaptive algorithm with temperature and voltage monitors on the main battery pack. The idea is that during much of the day the solar can keep the batteries charged in an optimum way and, if required, bring the charge up the last few percent before the end of the day. Running my alternator on its lower setting avoids overcharging. This involved a certain amount of experimenting to check if a reasonable area solar array would provide useful power during our main cruising season of Late April to early August, say 8 weeks either side of the longest day. The only non-discretionary 24V power demand is the fridge and hot days with maximum demand are like to have plentiful solar! The initial phase of upgrades was therefore to add some Solar Power and some battery monitoring to inform the next phases.

Solar Panels

I had heard good reports on Renogy Solar panels (available through Amazon) and they produced panels which were ideally sized to fit the two areas available. My initial order was for two 100 watt Mono-cystalline panels to fit between the Pigeon box and first mushroom vent leaving space to walk down one side of the roof. They are connected in series to match a 24V system. After my initial assessment I added two more 80 watt panels. The roof is not flat and is 'planked' which ruled out flexible panels and my flat panels are spaced from the roof by 4 rubber doorstops bolted with wide washers to some of the many fixing locations. At present they can be moved round. The panels came with 50cm cables terminating in the universally used MC4 Solar connectors. I bought a kit with ten connectors, strippers and the special crimping tool from Amazon (Prostar Solar Crimping Tool, Stripper with 10 male and Female Connectors and two assembly spanners currently £27.99) I bought 10 m of 6mm2 solar cable as part of my initial order from OnBoardEnergy. Strip Fuses and Fuse Holders came from Midland Chandlers as we passed. The complete cost of the four panels, connectors, special tools and solar wiring was under £400.

Solar Controllers - The Victron SmartSolar MPPT 100/20 charge controller

It is impossible to over emphasise the importance of the solar controller whose job is to match the completely different supply characteristics of solar panels with the demands of charging batteries with the highest efficiency. This requirement has lead to the creation of controllers which contain DC to DC converters so the maximum solar voltage times solar current point can be chosen at any solar input and converted to the required battery voltage required for the charge point reached. Compared to earlier chargers this gives about 30% more useful power than when any excess voltage was wasted. This can be seen when you are using a Maximum Power Point Tracker (MPPT) controller as the output current is typically about 30% higher than the current input from the solar array despite the losses in the DC to DC converter. Multiple arrays can now also be connected in series as well as parallel and the controllers can often handle a number of output (battery) voltages.

Wikipedia has a reasonable explanation of Maximum Power Point Tracking if you want to understand more of the details. Another good source of information I found recently is at Clean Energy Reviews and their explanation of MPPT controllers and their advantages is as easy to understand for a layperson as any I have found. They have reviews which seem relatively free of bias and include Victron and Renogy Devices but note they are Australia based so conditions are very different to UK!

I looked at two of the main manufacturers of marine inverters, solar controllers, battery to battery chargers etc: Sterling and Victron both have good reputations. Sterling have the most [only] affordable Lithium batteries and seem to have an edge on battery to battery converters whilst Victron excel on Solar Controllers and Battery Monitoring as well as having excellent Sine Wave Inverters. Victron's more recent offerings are 'Smart' with built in Bluetooth for configuration and monitoring via an App. Suitable Victron Smart devices can be networked via bluetooth so a temperature and voltage monitor fitted on the battery pack can be used by a Solar controllers with no additional wiring. Many devices automatically sense the battery voltages and can be used in 12, 24 and sometimes 48 volt systems. Unfortunately, but not surprisingly, that does not extend to inverters and battery to battery chargers but does include the Victron Solar Controllers, Voltage and Temperature Monitors and their Smart Shunt for Battery measurements. This meant that I did not have to initially make a number of major choices and could make measurements to inform those choices. I have chosen to use Victron devices wherever possible.

My initial Victron order was from OnBoardEnergy who are based at Springwood Haven Marina near Nuneton. We had looked in whilst passing on Corinna and I gained a lot of useful information from Kevin as well as few initial purchases. OnBoardEnergy offer on their web site to match an genuine internet offers and Kevin Mascall has done very well by us on our initial and subsequent orders which have been delivered by a premium delivery service - DPD. During the design phase I fortunately worked up in solar controller size to the Victron MMPT 100/20 Smart tracking controller. This device automatically determines the system voltage on installation then continually adjusts the working point to optimise performance. The combination of Renogy panels and Victron controller has proved very satisfactory and has produced up to 336 Watts from my horizontally mounted 360 watts of panels in July which exceeds the output I expected from horizontal mounted panels.

The Victron controller is small, lightweight and easy to mount and wire and can be extensively monitored and configured via Bluetooth even from the other end of the boat. It is networked via Bluetooth to a Victron Smart Battery Sense voltage-and-temperature sensor which is mounted directly on the batteries and came with an inline fuse and pre-crimped eyelets. The Solar Charger uses these measurements to optimize its charge parameters. The controller is capable of providing up to 480 Watts at 24 volts and larger arrays can be used to maintain performance in lower light conditions including series connection up to 100 volts. The existing panels have regularly provided 1000Wh/day and peaked at 1200Wh/day ie the equivalent of 2 leisure batteries discharged to 50% without even being in full sun on a clear day in late July. Since fitting the initial 200 Watts I have never needed to run the engine in the evening so the first of the objectives has easily been achieved and the full 360 Watts of panel are still giving adequate power well into September. See the Appendix for actual measurements. A good solar controller with temperature measurement will set one back almost £200. In our case for £600 we expect to reduce engine hours by 60-90 per year, extend battery life and have quiet evenings!

Monitoring - The Victron SmartShunt and SmartConnect App.

The final component in my first phase of upgrading was to add a Victron Smart Shunt. This has proved very enlightening as it not only provides current, voltage and power for the batteries but also the Consumed Amp/hours and an estimate of State of Charge which is automatically kept synchronised. It also has an addition voltage input which, in my case, monitors the 12v leisure battery voltage but could also monitor the battery mid point to indicate if equalisation is required. The shunt goes in the negative (ground) lead to the battery and there should be no other connection to the battery, it handles up to 500 Amps at which it drops 50 millivolts so the losses are negligible provided the wiring is kept compact. The only input required to the device (once via the App as it is stored) is the battery pack capacity. The Bluetooth range seems much less than my other Smart devices possibly because the Shunt is currently mounted right above the batteries to keep leads short. The short range is a pity as the outputs are arguably some of the most useful for controlling the system. Other users have also suffered from short range and have used the option to connect an external Bluetooth transmitter, an expensive but pragmatic solution.

I needed to modify and in some cases beef up my high current wiring for the SmartShunt which involves making crimp connections onto heavy duty cables. Wiring requirements and ratings are covered in an Appendix. I found that an impact crimping tool by Draper worked very well with a mooring hammer - they are available from Amazon for ~£20 as is battery/welding cable and the crimp connectors. Adding monitoring will cost from £125 to £200 depending on rewiring - heavy duty Copper cable is very expensive.

The Victron SmartConnect App

Smart Victron devices are all accessed from the same Victron SmartConnect App over Bluetooth. Most of the devices have default configurations which allow them to be initially used but further configuration and monitoring is done via Bluetooth. Overall I am very impressed with the Victron App, most device have a main Status display page, a History page and a page titled Trends where you can plot in real time almost any of the data available from the device. History shows information stored within the device for up to 30 days whilst Trends shows plots of the real time data from the device which are lost when you disconnect or the screen goes off both of which are irritating shortfalls. Fortunately there is the option in the settings menu to keep the screen on while SmartConnect is active, causing SmartConnect to stay connected but quickly flattening the android device batteries even with the screen turned to low brightness - I end up keeping them in the dark to save power and connected to a power brick!

In addition to accessing devices you can also run demos of every device so you can investigate all the configuration options available before you purchase the device or set it up. The App is also used to update the firmware in the devices, once the App has been updated with new firmware you can not change any settings until the Device has been updated which ensures they are always kept in step. The App also allows you to set up a Bluetooth network which is, for example, used to connect the Battery Monitor to the SmartSolar Controller and any other devices needing the information.

Its main shortfalls are that you can only connect to a single Smart device at a time from each App and can not currently leave the App running in the background so it is easy to lose all the real time data collected for the Trend graphs although the more important History is stored internally in the devices. This has meant I have rescued several old phones/pads from storage for my investigations so I can monitor several devices simultaneously. Although the instructions make it clear the devices have to be paired within the App rather than within Android it is worth repeating. If you have to remove the connection for any reason that is done within Android and I have found you must also take the phone or whatever well out of Bluetooth range. I have only had to do it once after a device became inaccessible because it seemed to be permanently locked to a particular 'phone' whilst they are normally shown on all the Android systems with some 'grayed out' if already in use on a different system. More recently I have had an even more comprehensive lock out of the Bluetooth system possibly provoked by the Android device shutting down through flat batteries but this was fixed by removing power for 30 seconds from the Smart Shunt by pulling the inline fuse for ten seconds then reconnecting.

Stop Press: From Victron Energy BV August 4, 2021 "Hi, We are working on new firmware that stores the trends in the product so no need to keep VictronConnect open. If you want to try it you can manually install the current beta release: https://www.victronenergy.com/live/victronconnect:beta" I have not tried this as it changes the device firmware on most of my devices and I do not see a route back but it should be a major improvement when it is issued. It appears that the stored data is at a lower time resolution than the current live data (ie 2m at 1s interval (RAM only), 2h at 30s interval, 1d at 5m interval, 45d at 30m interval).

Actual System Power Usage Measurements

Once the Solar and the SmartShunt monitoring were connected I was able to get far more information on what was actually going on than I had ever had before with graphical displays (Trend graphs) available for all of the important parameters versus time. This confirmed what I suspected, the current Matchpower inverter with Power-saver drew negligible power until a load was applied but the idle (running but without load) power was huge at 73 watts and with fridge started at ~ 200 watts falling a little after about 5 mins as it 'warms up' consistent with fridge itself being ~100 watts. The Fridge labeling says 90 watt so the Matchpower inverter efficiency is under 90% plus the considerable no-load overhead. The cycle time is obviously very dependent on ambient temperature but seems to have an on off cycle of approx 20 to 35 minutes (~40%) overnight giving a daily usage of 190x24x0.4 = ~1800 Wh which alone requires an engine run in the evening to avoid taking the batteries below 50%. There are measurements for most of my 'appliances' in the Appendix below with power inputs measured using the SmartShunt and outputs from the Victron Phoenix 24/1600 Smart Inverter once it was available,

Input power measurements. The alternator does exactly what is on the tin and provides 25 Amps initially tailing off to stabilise at 28.8 volts on the high setting but taking several hours to complete the process. The solar controller does an excellent job and the output current is typically 30% higher than the input current from the cells (which are at a higher voltage) due to its MMPT feature. You can see on the Trend Graphs that it does an optimisation every ten minutes or so. The Solar controller keeps a lot of valuable information in the daily summary (History tab) including the maximum power, total Wh and times in bulk, adsorption and float for the last 30 days which can also be output as a data file.

Solar Power Measurements

This section covers covers both my general experiences and specific measurements on the solar panels. There are a number of constraints for me when making measurements which are generally applicable, both in the installation and the site. Firstly I have flat panels which are sitting near horizontal on the roof of the boat so they will never produce full power as they are always at an angle to sun so their effective area is reduced and there will be increased reflections at low incidence angles. The other constraint is that there is a certain amount of shading after 1200 on my home mooring which will get very much worse in the winter. This will hopefully not be a problem as over the winter as Corinna is, by default, on a power hook-up so the batteries are just being maintained fully charged with very little solar input needed for the 24V system as the inverter is off. The major power requirement on shore-power is for the 12V domestic system powering water pump, lighting, hifi and internet.

So there were several regimes which needed to be understood:

The shore power hook-up makes test measurements of the maximum solar power available more challenging as the 24V batteries never have any significant load and kept fully charged - one has to deliberately change to battery power and discharge the batteries but remembering keeping significantly discharging the batteries will shorten their life. I use the kettle for an initial ~100 Wh discharge and then use the fridge via an inverter. The old inverter has an idle-power of 70 Watts which provides a useful background usage. Even with these tricks it is easy to end up with the batteries full and no idea of the solar input which could be available although keeping the SmartConnect App running and connected to the SmartSolar device provides good insights.

There is also a lot of very useful information in the History stored by the SmartSolar where the times in each phase are accessible along with a graphical display of the power used in each and the maximum solar power during the day. The History information data can also be stored in a file which can be emailed or otherwise transferred in a .csv format (comma separated) which can be input to any spreadsheet and build up a picture over time. It is helpful to know the regime for each day but it is usually fairly obvious when you look at the data. Sample periods of selected data from these files is shown in an appendix.

Perhaps the easiest regime to understand is maintenance on a shore hook-up. At the start of the day the charger become active once the voltage of the solar panel is 5 volts above the batteries and this occurs much earlier than I expected with solar availability often 15 hours or more in mid summer. The battery voltage when the charger starts is used to estimate the state of discharge and adjust how long an Absorption phase will be needed to fully charge the battery once the bulk level has been reached. The initial power will be low and it takes a while in Bulk charging to reach the voltage of ~29V and but the Absorption time when just maintaining the batteries seems to be only a few minutes before going to Float for the remainder of the day. The total solar power provided during a 'maintenance' day is only 20 to 30 watts which is unlikely to be overcharging the batteries with an initial time in 'Bulk' of about an hour at the start if the day, a short time in 'Absorption' and the remainder in 'Float'.

In the other two regimes the batteries will be at least partially discharged when the charger turns on in the morning and the Absorption phase timing depending on the state of discharge (estimated from the battery voltage at the time solar become available) - that could be up to 6 hours although I have never seen it that long, one hour seems more typical and I have once seen 3 hours. Note: the absorption phase is also terminated if the current falls to a low and programmable level and that seems to be the norm. I have rarely been able to keep sufficient load to know exactly what the solar array was capable of yielding but have certainly seen 1200 Wh on days which only had broken sunlight. In August it easily keeps up with the old fridges demands up till 1900 when I finished checking and had recharged the batteries and  entered 'float' by 1300. As I said earlier the SmartSolar controller records the maximum power during the day and I will be able to plot an envelope during the year of time solar is available and peak power. The figures I initially made indicated that the panels are capable of producing close to their specified power in July even though they are horizontal rather than oriented towards the sun. Sample measurements were:

Solar with the initial 200 Watt array when I first got the solar running up and running ~ 500 Wh/day (400, 560, 288, 410 in mid July with peak of 870 on clear day) which would scale to an average power of 900 Wh/day with a 360 watt array.

Solar with 360 Watt array on 4 broken days and one mostly sunny day on the garden mooring (with some shading during afternoons) at end July 950, 1030, 1110 and 1270 with max powers recorded of 309, 310, 305 and 325. Average power ~ 1087 Wh/day. When making these measurements I had to keep discharging the batteries using the techniques above otherwise the solar spends most of the time throttled back in 'Float'.

Solar with shaded panels and cloud cover: This is going to be typical situation in the winter due to shading by my trees. The results have been surprising in that the panels give power when the sky is overcast or the panels are completely shaded just from reflected light. On a day with broken cloud it is possible to get far more at the times when it is cloudy than clear over the panels and up to 60 watts has been seen in October with no direct sunlight as seen in the snaps taken with the phone below which also shows the arrangement of the panels on the roof in two pairs. Also note how the charging current is 30% higher than the Solar current showing how well the MPPT controller works and the comprehensive display.

 

Tables with 3 months of measurements are available in the Appendix

Victron Phoenix 24/1600 Sine Wave Inverter

The initial tests and measurements were done with the inverter lying flat on the engine room floor and when I was satisfied it was mounted on the bulkhead. The inverter is heavy ~ 11 kgs so needs to be secured well to the bulkhead. The inverter is hung from a wide 'bracket' which has has three 6mm holes to fix to a wall. My bulkhead is only 16mm ply so I used 3 SS No 10 by 16mm pan head screws with a small flange and also drilled 3 holes for screws below the bracket overlapping the edge slightly to take additional load. The inverter was then lifted into place and hung on the bracket. There are then two additional mounting holes under the bottom cover to secure the bottom and prevent it lifting off. There may be two more holes at the top, they are shown on the drawing but are only accessible it the whole cover is removed which I have not done [yet] and only five screws were provided indicating they were not needed.

I have a heavy duty fuse of 125 Amp rating as recommended between the battery switch and Phoenix. The positive cable all the way to the main battery is 25mm2 which is what is recommended for wiring up to 6 metres in the handbook although the internet version recommends 35mm2 up to 5 metres. My wiring will be closer to 3 metres (two way) so I am intending to keep to the 25mm2 for the positive. During initial testing the negative was 1.2m of 16mm and although that gave no problems it has been changed to 35mm and connected directly to the Smartshunt to minimise loses.

The App does not offer any history or trend pages for the Phoenix which is disappointing, it is mainly used for configuration of the eco mode. The status page only displays load in Volt Amps and input voltage are useful information but the voltage allows one to see voltage drops in the cables compared to voltages from the SmartShunt.

Measurements on Phoenix Inverter

Idle: The idle is the 9 watts as in the specification. The eco mode works by turning on power briefly every few seconds. The levels for turn on and off can be set in the app as can the frequency. The power in eco mode is specified to be 1.3 watts but is changing so fast it is difficult to measure but looks about correct at the default 3 secs and .16 secs. It should be possible to change these to much longer numbers and look at the trend graphs to get a more accurate estimate for other settings.

Efficiency: The efficiency in converting power from the battery to 230V AC input to a device can be thought of as being influenced by three components. The losses in wiring etc increasing roughly with the square of the power input, the fixed component of power to run the inverter which is independent of power output and the efficiency the power conversion which we do not know the shape of but Victron claim an overall efficiency of up to 94%. We can get an overall idea of the shape from measurements on a series of resistive loads as we have in the Appendix and the data points have been fitted to a model proposed by Victron which is covered in the following Appendix.

In summary: The end-to-end efficiency seems to rise to be over 94% from 180 to 480 Watts output and then reduces to ~90% at 1000 Watts and ~83% at 1600 Watts .

General impressions/conclusions on Phoenix inverter.

The change from the Matchpower Square Wave inverter to the Phoenix Sine Wave inverter has been a dramatic saving in power when running the old fridge. Solar power now easily keeps up with fridge consumption on a partially sunny day in early August and enabled 24 hrs from solar alone fully re-charging the batteries by 1300 in early August. I get the impression the fridge motor is slightly more efficient/powerful on a sine wave and runs with a lower duty cycle. It is now early October and I have not had to run the engine to charge batteries yet.

Eco Mode: The mode of operation in Eco is very efficient but has idiosyncrasies which can leads to some strange effects. For example the pulse of power is sufficient to flash the fridge light every 3 secs but not hold the inverter on. In contrast my Bosch palm sander draws enough start up power to activate the inverter but once running it does not draw enough power to keep it the inverter on so it stops after a few seconds then restarts a few seconds latter. However for most applications it works fine provided you can tolerate the short and variable delay in starting, for example, power tools.

Some appliances reduce their power by alternately switching on and off every 10/20 seconds to vary the mark space, these include many microwave ovens and Induction hobs and it is also best to switch out of Eco when running them.

Upgrades to the 12V Domestic system

It is looking possible that the combination of solar, an improved inverter and a lower consumption fridge will allow us to go several days without needing to run the engine to charge the 24 volt batteries whilst away from home and should keep them charged in a an optimum manner throughout the year whilst on a shore line. That still leaves a problem with the 12V 'Domestic' supply which provides for Lighting, Water Pump, Mobile Wifi, HiFi, and charging for Laptops, Pads, Phones etc. Currently the only charging for the 'Domestic' leisure battery is from the engine's alternator - this never proved a problem whilst traveling and running the engine daily but care has had to be taken whilst on a shore hook-up not to use much lighting or water. Now we also have continuous Mobile Broadband provided through an old phone and a GiffGaff 100 Gbyte data package and far more use of phones and pads to monitor and control the new systems so we need to look at ways to keep the 12V systems charged.

The obvious way is to use solar but a separate 12V system would be expensive and we do not have an easily accessible area to fit even a small addition solar panel. However for much of the day there is excess power available from the 24V solar system and the batteries end up on Float charge. So what we need is a means to charge the 12V Domestic battery from the 24 Volt batteries whenever they are well charged and the solar is active, ie at the float voltage or higher.

The best way would be with a battery to battery charger with phased charging and temperature compensation. They tend to be higher power than we require and only the Sterling version has temperature compensation.They are also expensive (~£280) Victron however make a simpler device which is intended more as a power source but they say can be used as a trickle charger. The smallest version provides 110 watts (approximately 9 Amps) which is far more than any current requirements for 12V (apart from a few seconds surge when the water pump starts and possibly the horn). It provides a fixed voltage which can be adjusted by a potentiometer to a suitable battery float level to keep it trickle charged when not in use or recharged after discharge by internal lights etc when solar power is available. Bulk/Absorption charging to full capacity will occur on days with the engine running. The device in question is the Victron Orion-Tr 24/12-9 isolated 110 watt DC-DC converter

Using the Orion-Tr 24/12 isolated DC-DC Converter to trickle charge a leisure battery.

A Victron Orion -Tr 24/12 isolated DC-DC 9 Amp Converter (110 Watt) has been added to our system and is powered from the 'switched' load from the Victron SmartSolar Charger to trickle charge the 12 volt domestic battery at ~13.7V. The SmartSolar MPPT 100/20 has been programmed to turn on the Load output once the 24 volt battery has been charged to the close to the Absorption level and disconnect when it drops significantly below Float level. This can be achieved using the App and there is a convenient delay time built into the load switching algorithms to avoid switching on surges. The Load switching algorithms can be overridden in the App to allow continuous or no charging.

The settings may need to take into account a battery temperature range and it is not clear if the temperature compensation changes the levels for the SmartSolar load switching. The Orion has connections for a remote switch and one could be added at the unit to manually top up the 12v when required. In the meantime I have set up the SmartSolar using the SmartConnect App load switching to User Defined Mode 1 switching a little below the bulk/absorption level and off below the float level. It is a bit difficult to test as it does not work exactly as expected as it switches on a transition so you can have to think of the effects if you manually switch the solar off and on. The delay is also stated to be 2 minutes but it seems to be quite a lot longer and initially I thought it was not working.

Initial set-up and measurements on joint Smart Solar MPPT 100/20 and Orion System for 12V

The Orion is not programmable so the output voltage had to first be set using the potentiometer accessible from outside the bottom of the box beside the terminals to a suitable voltage to 'Trickle charge' the battery. This has been initially chosen to be 13.7V ie just below a normal 'Float' level of 13.8V at 25 degrees C and we will watch the current under no load over time to get a compromise between keeping the battery charged over long periods without engine running and overcharging. When the Orion was first connected the leisure battery partially charged and reading about 12.4V. There is no direct measurement of current but the Load Output of the SmartSolar is available as both a current and power which means we have a good estimate of the '12V' current if we assume a figure for the Orion's efficiency. The SmartSolar load current started at about 2 amps and fell to 1.4 amps within minutes then continued to fall to 0.4 amp after a few hours.

The 12V system was the put under load by progressively turning all the lights on and the Load Output power from the SmartSolar limited at circa 123 Watts. This corresponds to what one would expect as the Orion is supposed to limit at 110 Watts/9 Amps and has been set to 13.7 volts. By full load the voltage had dropped to ~13.2V at the battery (measured by the auxiliary input to the SmartShunt), again roughly what one would expect as the limiting will not be abrupt. The voltage rose to 14.4V when the engine was running again as expected and the power into the Orion dropped to zero. In other words it works as expected although longer term measurements are required to determine an optimum voltage to maximise the battery life.

Some further checks were also made using the History tab in the SmartSolar App where the daily total of the solar power and load power are available. The invertor was turned off so all the power used was through the switched load output from the SmartSolar controller. All the lights were turned on for 2 hours giving a power usage of 250 Wh which was matched to a few percent by the Solar input power after time was allowed for the 12V battery to be fully recharged, the difference being consistent with the residual 'trickle' and 'float' charge currents over the period which had fallen back to under 0.4 amps after 3 hours for both 12 and 24V supplies. The solar power was not steady and high enough to provide all the power so so it was often being drawn from the 24V battery which needed to be recharged at the end making the check more realistic but less accurate.

There is a table of my measured domestic (12V) power requirements in an Appendix

Thoughts on Solar Safety, Switching and Fusing

Firstly Lead Acid batteries are Dangerous, they produce hydrogen and can explode if not ventilated, short circuited even momentarily, or otherwise abused. Wear safety goggles - a frightening number of people are blinded by acid from Lead Acid battery explosions in the USA and probably elsewhere.

Systems with Solar Input need special care to work on safely. It is no longer sufficient to disconnect the batteries physically or by switches when working on the system, for example, changing batteries. In a conventional Solar Power system there are seldom physical switches to turn off the Solar Power feeding the batteries [which may be considerable] and control may only be possible via the controller, by remote control switches or by Bluetooth to enable work to be carried out on the system. Worse still Solar Controllers have direct connections to the batteries and many have outputs designed to provide considerable power even when there is no input from the Solar Arrays probably conditional on battery voltage. Even without Solar modern systems often include battery to battery converters so disconnecting the 12V domestic battery can still leave the 12V domestic system powered from the engine start battery, or a 24V source itself powered by battery or solar input!

Fusing remains important but most high quality controllers and converters have, in theory, a lot of inbuilt short circuit, open circuit and overload protection and adapt to different system voltages so it may not be necessary to have fuses between every device. Fuses can also be a useful way to disconnect parts of a system but many high power fuses are not a plug in type making disconnecting by 'pulling a fuse' difficult or risky

When working on a system with a Solar Controller providing input it should be turned off as should be as many of the battery to battery converters/chargers but even so one should always assume the system is live. Just checking with a meter is insufficient as one must assume the electronic controllers could switch on at any time. This means that great care must be taken when changing batteries - even a brief short with a spanner or a spark can cause major damage, burn out wiring and even cause explosions from hydrogen gas. Insulate your spanners with self amalgamating tape! There should be warning notices for any engineers brought in to repair or service the engine or change batteries about all multiple power sources.

Bluetooth systems should be properly protected and passwords changed from the default ones to prevent accidental or malicious changes by other people and nearby systems. It may be necessary to have dedicated bluetooth device available for service personnel and very full instructions on how to disable as much as possible of the system.

Another problem is that there will be far more wiring directly to the batteries, much will be low power for sensors to measure battery voltages and temperatures or for shunts to measure currents. These still need to be protected by local fuses - Victron sensors have fine wires protected by local fuses but a diagram is required to ensure they are connected properly after a battery is changed and they are placed in the correct place. Photographs are a useful tool and I use the mobile phone to take pictures before disconnecting or modifying any wiring.

This section nay be extended as we gain experience with our system.

Overall Summary and Conclusions

The changes we have made have exceeded even our most optimistic expectations. We have run the engine for less than an hour total for charging batteries in the 4 months since we fitted the solar and have simultaneously greatly reduced the depth of discharge of our batteries which, along with much improved charging profiles should greatly extend the battery life. The improved monitoring has played a major part in understanding our problems and guiding our usage in the future. The new sine wave inverter has almost halved the battery consumption of existing fridge so we have left fitting the new fridge till the spring. We can now use a microwave hob for cooking during the summer months to complement the squirrel stove during spring and autumn when solar is reducing.

We hoped that solar and other changes would have an impact during our main touring season but it seems it will cover most of the time from Easter to the start of the stoppage season in November.

Appendices

There is a lot of information in the appendices and they grow with time. The ordering and indexing is Work in Progress.

Appendix A - Wire Gauges, Current ratings, Resistance and example voltage drops.

Conductor area in mm2 AWG equivalent Normal Rating in Amps IEE Rating in Amps Resistance in mOhms/metre
70
2/0
485
234
0.27
50
1/0
345
182
0.39
35
2
240
141
0.55
25
4
170
114
0.78
16
6
110
87
1.21
10
8
70
65
1.91
6
10
50
47
3.30
4
12
40
37
4.95
2.5
14
30
27
7.98

The Table is for Hi-Flex multi strand battery cable ratings and ~ equivalent AWG (American Wire Gauge). First rating is from normal sales literature for battery and welding cables ie free air and spaced cable, second is the more conservative IEE ratings from table-4d1a column 6 and resistance from https://www.cse-distributors.co.uk/cable/technical-tables-useful-info/stranding-chart/

Calculations for typical inverter cable lengths 2 x 1.25 m of 25mm2 = 2 milli-ohms

Nominal inverter output of 1600 Watts at 24 volts = 67 Amps == .13v drop == .54% voltage drop (8 watts spread over 2.5 metres of cable ~warm)

Max 2000 watts at 25 volts = 80 Amps == .16v drop == .64% drop

Appendix B - Measurements on Matchpower 1 KW Square Wave versus Victron 1.6 KW Sine Wave inverters

The Matchpower Square Wave inverter does not have a stabilised output and the output voltage and hence power drops considerably due to internal losses and wiring losses. It also has a very large no-load power consumption but does have a power saver which only turns it on if a load is applied which works extremely well. The Victron Phoenix 24/1600 is a fully stabilised sine wave inverter so there are no output reductions when a high load is applied.

The input power from the batteries measurements are from the Victron Smart Shunt and the Solar input has been turned off temporarily in the Victron Smart Solar MPPT 100/20 settings. The Phoenix Smart 24/1600 also has a Volt Amp display available in the App which differs for non resistive loads such as the microwave and the motor in the fridge compressor because the phase of the current and voltage differ for capacitive and inductive loads. The Smart Shunt figures should be best for power drawn from the batteries whilst the VA is more appropriate for inverter load and wiring loss calculations. The figures for the Fridge have a VA figure which is 20% higher.

The quirks of the Matchpower make some of the comparisons seem odd, for example, the Matchpower looks to use less power than the Victron boiling a kettle but that is because the output voltage and power drop under high loads and the resulting power is less than from from the stabilised Victron. The difference between the SmartShunt measurements and the Victron VA readings gives an indication of overall efficiency but for Resistive loads only . The losses will come from the idle power, losses in wiring and actual efficiency above and beyond the idle power.

Device: Measured by Matchpower: SmartShunt (input) Phoenix: SmartShunt (input) Phoenix: Phoenix VA (output) Comments
No-Load, Idle
72
9
0
Power used when no load connected and power saving modes off.
750 Watt Kettle
750
754
696
Resistive Load
Cooker (Top only)
700
700
660
Resistive Load
Cooker (Top and Oven)
1020
1084
970
Resistive Load (1050 watts nominal at 230V)
Microwave
640
1312
1155
Microwaves have a seriously reduced output power when driven with square waves
Remoska
443
418
Resistive Load (500 Watts nominal at 230 volts)
Slow Pot High
103
90
Resistive Load
Slow Pot Low
84
72
Resistive Load
Light Bulb (40Watt)
118
46
33
Resistive Load
Light Bulb (100Watt)
102
87
Resistive Load

Induction Hob (1200 watts)

-

1351

1151
Sterling Induction hob can not be run on square wave
Induction Hob (1000 watts)
-
1078
923
Sterling Induction hob can not be run on square wave
Induction Hob (800 watts)
-
830
715
Sterling Induction hob can not be run on square wave
Old Zanussi Fridge
202 falling to 194
106 falling to 102
130 falling to 126
Power reduces slightly as compressor warms up. *

The fridge is, of course, the most important as it is on 24/7 and is non-discretionary and the change to the Victron Inverter comes close to halving the power used every day.

* Some latter measurements show an even more dramatic fall in power input with time on the old fridge of 112 W falling to ~91 W after 15 mins.

Appendix C- Inverter Power loss and Efficiency (based on Victron paper)

Victron do not provide efficiency curves for most of their inverters but if you search you can eventually find https://www.victronenergy.com/upload/documents/Output-rating-operating-temperature-and-efficiency.pdf. It has some tables of measurements on similar but higher power inverters (3000 Watt) and the following (rather basic) justification for their calculated curves which gives a very reasonable starting point and a logical way to size an inverter.

Victron Model for Inverter Efficiency

Electric current generates heat in the conductor through which it flows. The basic formula to calculate the rate of heat generation, or power dissipation, is:

P = R x I2 (1)

where P stands for power (measured in Watts), R (Ohm) is the resistance of the conductor and I (Amps) the current.

In power electronic circuits the situation is much more complicated: one has to deal with DC losses and switching losses, losses in semiconductors and in high frequency transformers, etc. Very often, however, formula (1) appears to be a fairly good approximation of the overall losses in the circuit. The power dissipation P of the circuit can then be calculated by defining R as the overall resistance between input and output of the circuit and I as the output current.

An even better approximation is obtained if a factor is added to (1) to account for the no-load power consumption, the power dissipated by the circuit when it is switched on without any load connected. It is an important specification especially of inverters since in the long run it can drain a battery and dominates the efficiency at very low power outputs

Taking into account the no-load power consumption results in the following formula:

Ploss = Po + Rinv x Iout2 (2)

Ploss is the total power dissipation in the product; Po is the no-load power dissipation (and by definition also the no-load power consumption); R is the “resistance” between input and output; and Iout the output current.

With formula (2) efficiency, an important specification of inverters, can be calculated:

η = 100 x Pout / (Pout + Ploss ) (3)

where η is the efficiency in % and Pout the output power (Pout = Vout x Iout ).

Tests and extrapolations on Victron Phoenix Smart 24/1600 Inverter Data

The above analysis by Victron has a some physical basis but can also be regarded as a power series fit and it would be logical to try a quadratic fit to the losses from experimental data which would show if there is any significant linear term. I found an App https://play.google.com/store/apps/details?id=com.ashermobile.math.curvefitterfree&gl=GB which I used to do a basic quadratic fit to some of my data and it does look as if the Victron suppositions are true up to the 1000 Watt level (table-top oven), the maximum I had inputs for, and at that point the efficiency was ~89.2% and was 92% at the 750 watt 'kettle' level.

The peak efficiency is at just under 25% of maximum output and is ~94.7% which is very much what is specified by Victron (max 94%), in fact it stayed at over 94% from 185 to 480 Watts output. Finally extrapolating to the maximum inverter load of 1600 watts the efficiency would be ~ 83%.

It is worth noting that these are end to end figures from the SmartShunt measurements at the batteries to the 230V output VA measurement by the Phoenix and included the losses in the wiring, contacts, switch and fuse. These figures have been derived for resistive loads but should be indicative of the performance on loads such as motors.

Appendix D - Estimates of typical 230V weekly power usage.

Note these are nominal powers for planning purposes and do not use measured powers although most of these have now been confirmed by measurements.

TOTAL per day averaged over a week = ~650 watt hours plus the existing Zanussi 240V fridge @ 1000 Wh/day ~= 1650 Wh/day which requires 3 hours engine average without solar of which 2 hours essential for fridge. This ignores Solar contribution which should cover over half during mid summer. Recent actual measurements on the Zanussi fridge in September show an average overnight consumption of ~1.3 amps equivalent to ~820 Wh/day. That brings down the daily requirement to ~ 1500 Wh/day with the lower average temperatures in September

Appendix E Estimating Solar Power

This appendix will bring together various sources of information in the literature on measurements and ways of estimating the solar power available through the year for various configuration of panel installations. What immediately stands out to me is the variability of available solar power with wide variations on all time frames. There is, for example, a wide variation in sun hours per day and month from year to year by factors of close to two fold, for example, see https://www.statista.com/statistics/322602/monthly-average-daily-sun-hours-in-the-united-kingdom-uk/

Despite the wide variations between years it is possible to see a pattern emerging with May typically having the highest monthly hours of sunshine. The actual hours of sunshine is far from the whole story as the solar arrays can give reasonable output under cloudy conditions and the figures also vary considerably depending on location within the country.

Victron MPPT Tool: Victron supply an interactive tool for their MPPT Solar Controllers at https://mppt.victronenergy.com/ which includes a monthly forecast of yield per day at a chosen location in the world. This is useful for planning but does not contain information on how it has been derived or the panel configuration in use - I believe it represents the configuration used by most installations which are fixed and giving peak output when pointed South and angled at an angle determined by the latitude whilst ours is horizontal. Even so I wish I had found that page earlier on. It is also the only tool where you enter the detailed differences between solar panel technologies and manufacturers although I have not exploited that fully.

What the curve shows for Oxford is that the solar power available is almost flat from late April to the begriming of August but then falls more rapidly to approximately 80% at the end of August and 60% by the start of October. In other words it is flatter and less symmetric about the longest day than one might have expected without taking into account average weather patterns. It means that there could be almost constant power during during our main cruising period, late April to the start of the School holidays but considerably less in the period starting September after the Bank holiday when we start again to cruise on the Thames.

The EU PVGIS Tool: The EU produce a much more comprehensive/complicated system, the PVGIS, for estimating the energy from a solar array. PVGIS is a web application that allows the user to get data on solar radiation and photovoltaic (PV) system energy production, at any place in most parts of the world. It is completely free to use, with no restrictions on what the results can be used for, and no registration is necessary. A good starting point is the manual at https://ec.europa.eu/jrc/en/PVGIS/docs/usermanual - it looks very complex but is actually quite easy to use provided you have a doctorate in Atmospheric Physics. You must also remember to select a location before anything else. I chose to enter "Oxford, UK" It enables you to specify the azimuth and elevation of a fixed array and will even calculate the optimum elevation for your location. Most of the inputs have sensible options and you probably only need to input the specified peak power of your array before hitting the Visualise results button to get a graphical output. The following shows the difference between optimised position array and a horizontal array, the summer is little changed whilst the power is much reduced by the low incidence angles in winter. I have included a compromise angle of 20 deg as 39 deg is not ideal on a boat even when moored!


Horizontal 360 Watt Polycrystalline Array


Array at compromise angle of 20 degrees elevation.


Array at optimum angle of 39 degrees elevation.

Most of the solar radiation data used by the EU PVGIS have been calculated from satellite images. There exist a number of different methods to do this, based on which satellites are used. The default choice is the PVGIS-SARAH data set. This data set of hourly solar radiation estimates has been calculated by CM SAF and the PVGIS team with 11 years of hourly data from 2005 to 2016 on a 0.05° x 0.05° (~ 5 km) grid and covers Europe, Africa, most of Asia, and parts of South America. The data sets are available for download but are tens of Gbytes in size indicating the work that the EU put into this tool. It would be interesting to see the effects of more recent data now global warming is starting to change the weather patterns. The PV technology 'presets' should also separate mono-cystalline and polycrystalline silicon which is an increasingly important choice.

The full print out (pdf) from PVGIS for our configuration is here. Peak yield is calculated by PVGIS to be ~1500 Wh/day and an average of >1200 Wh/day during our main cruising period. Half power points seem to be mid March and end of September which should reduce or avoid use of the engine to top up batteries in the evening during that period.

Appendix F - SmartSolar History from the SolarConnect App.

The following table uses some of the output which is available from the SmartSolar History output starting with my original tests with 200 Watts of solar panels as well as the 360 watt panels. Yield is the total power provided by the Solar Array during the day to the Battery. Load is the power from the switched output from the battery in addition to the power to the inverter which is taken directly from the battery. The switched load is only used to top up the 12V battery when solar power is available.

Date         Yield (Wh) Load (Wh) Max Solar
(W)
Max Solar
(V)
Min Bat'
(V)
Max Bat'
(V)
Bulk
(mins)
Absorb
(mins)
Float
(mins)
9/30/21 100 60 30.00 42.60 25.68 29.27 204 1 476
9/29/21 140 80 119.00 43.64 25.80 29.23 162 5 529
9/28/21 170 20 95.00 42.89 25.42 29.19 174 42 459
9/27/21 430 380 135.00 42.24 25.29 29.12 620 23 51
9/26/21 200 240 103.00 42.67 25.31 29.18 483 2 222
9/25/21 240 30 52.00 39.77 25.60 29.09 629 1 69
9/24/21 450 40 188.00 39.91 24.79 26.92 718 0 0
9/23/21 550 0 182.00 40.96 25.03 29.17 720 1 14
9/22/21 20 0 17.00 42.69 25.86 29.21 49 1 682
9/21/21 30 0 17.00 42.70 25.90 29.16 70 1 671
9/20/21 30 10 15.00 42.60 25.88 29.12 50 1 691
9/19/21 20 0 16.00 42.03 25.86 29.24 123 1 607
9/18/21 30 0 35.00 43.02 25.76 29.16 72 3 664
9/17/21 30 10 31.00 42.96 25.80 29.17 59 1 697
9/16/21 110 0 58.00 43.12 26.06 29.35 225 1 520
9/15/21 310 50 194.00 42.60 24.16 28.28 246 40 473
9/14/21 340 20 124.00 39.98 23.73 28.57 611 60 0
9/13/21 500 70 190.00 42.33 24.04 28.60 430 115 182
9/12/21 590 120 262.00 42.38 24.22 28.43 546 21 196
9/11/21 570 90 259.00 43.22 24.17 28.32 498 21 258
9/10/21 520 140 275.00 43.24 24.20 28.53 374 20 342
9/9/21 460 100 255.00 42.83 24.19 28.33 578 30 151
9/8/21 1000 70 231.00 42.64 24.06 28.34 529 20 233
9/7/21 310 60 169.00 41.41 24.15 28.24 526 39 217
9/6/21 180 80 199.00 42.04 25.01 29.14 337 19 437
9/5/21 120 110 48.00 42.39 0.01 29.16 129 2 631
9/4/21 180 30 106.00 42.28 0.01 29.62 193 17 527
9/3/21 270 160 133.00 42.12 25.53 29.11 189 19 593
9/2/21 390 300 121.00 42.71 24.13 29.94 423 7 360
9/1/21 30 20 65.00 42.42 0.01 29.18 90 2 523
8/31/21 260 0 162.00 43.14 25.06 29.13 416 15 357
8/30/21 20 30 10.00 42.16 25.80 29.12 99 1 708
8/29/21 600 0 317.00 42.55 24.12 29.15 502 34 279
8/28/21 310 10 246.00 43.51 24.23 29.18 225 21 571
8/27/21 390 0 180.00 42.68 24.46 29.05 362 18 431
8/26/21 330 10 280.00 43.39 24.59 29.95 295 17 492
8/25/21 630 10 307.00 43.55 24.23 29.12 372 29 432
8/24/21 550 10 285.00 42.85 24.20 29.18 312 20 509
8/23/21 30 30 25.00 42.88 25.86 29.16 107 20 708
8/22/21 640 0 275.00 43.21 24.15 29.11 662 20 164
8/21/21 30 40 15.00 42.79 25.77 29.12 82 20 731
8/20/21 30 40 13.00 41.43 25.80 29.11 66 20 758
8/19/21 40 50 17.00 42.67 25.83 29.13 93 20 747
8/18/21 530 0 251.00 43.00 24.15 29.15 485 20 328
8/17/21 40 50 13.00 42.81 25.86 29.14 46 20 773
8/16/21 30 50 14.00 43.32 25.94 29.10 73 20 787
8/15/21 420 10 286.00 43.03 24.03 29.14 215 20 588
8/14/21 30 70 11.00 43.14 26.00 29.12 48 20 804
8/13/21 120 40 88.00 43.12 24.22 29.19 147 20 701
8/12/21 30 10 38.00 43.10 0.01 29.11 126 20 675
8/11/21 30 10 69.00 43.16 26.08 29.12 72 20 759
8/10/21 780 0 256.00 42.49 0.01 29.06 464 58 309
8/9/21 460 10 225.00 43.18 23.84 29.12 499 20 324
8/8/21 280 0 202.00 44.17 0.01 29.14 237 20 609
8/7/21 200 10 228.00 44.12 0.01 29.16 162 61 466
8/6/21 50 50 12.00 43.83 26.08 29.15 91 20 796
8/5/21 60 40 16.00 43.30 26.12 29.14 61 20 809
8/4/21 70 30 85.00 42.78 26.11 29.09 77 40 677
8/3/21 50 10 45.00 43.18 26.13 29.13 99 40 281
8/2/21 70 10 33.00 42.84 26.17 29.04 210 21 249
8/1/21 890 30 300.00 39.91 24.30 28.77 661 257 0
7/31/21 440 0 266.00 42.64 24.42 28.50 774 2 129
7/30/21 640 0 336.00 42.12 24.67 28.80 877 50 0
7/29/21 1270 0 305.00 41.15 24.07 29.01 908 22 0
7/28/21 950 0 305.00 42.57 24.57 29.01 842 46 44
7/27/21 1060 0 310.00 42.45 24.49 28.93 652 32 238
7/26/21 1110 0 309.00 42.05 24.47 29.01 796 20 118
7/25/21 270 0 149.00 42.73 25.57 28.98 225 4 682
7/24/21 150 0 60.00 42.20 25.81 28.97 331 15 602
7/23/21 270 0 174.00 42.57 25.60 28.88 386 1 485
7/22/21 90 0 14.00 41.97 26.17 28.89 139 1 802
7/21/21 90 0 14.00 42.03 26.17 28.88 145 1 810
7/20/21 110 0 14.00 41.96 26.19 28.88 179 1 769
7/19/21 350 0 133.00 41.31 25.93 28.93 475 60 418
7/18/21 810 0 144.00 42.17 25.52 28.97 840 64 40
7/17/21 410 0 119.00 42.01 0.01 28.92 435 62 293
7/16/21 540 0 131.00 41.73 24.13 28.47 406 196 351
7/15/21 440 0 190.00 39.83 24.09 28.11 935 0 0
7/14/21 230 0 159.00 43.49 24.45 29.03 600 17 346
7/13/21 290 0 158.00 43.96 25.67 29.07 496 24 403
7/12/21 110 0 28.00 43.62 26.25 28.87 234 1 701
7/11/21 170 0 117.00 43.53 23.60 28.87 241 64 646
7/10/21 360 0 122.00 42.42 23.81 28.21 905 34 20
7/9/21 600 0 164.00 43.31 23.72 28.86 711 72 193
7/8/21 530 0 194.00 42.19 23.72 28.18 747 62 151
7/7/21 410 0 232.00 44.38 23.83 28.19 714 60 180
7/6/21 280 0 134.00 45.26 24.11 29.58 514 65 367
7/5/21 560 0 142.00 41.21 24.81 26.56 970 0 0
7/4/21 400 0 141.00 42.98 25.02 29.77 788 3 172
7/3/21 110 40 68.00 43.45 24.90 29.01 399 5 558
7/2/21 120 20 120.00 40.31 23.97 28.83 49 87 206

Note 1: Patterns of use. Very few of the readings in table are from 'Normal' operation ie when traveling with a pattern of use involving use of the fridge overnight, a kettle early morning then the engine running for a few hours before depending on the solar keeping the batteries fully charged until sunset. Most days were on a shore hook-up so the solar was just 'Maintaining' the batteries in a fully charged state. A number of 'Testing' days involved switching to battery use during the day to make measurements of the solar and used fridge, kettle or other loads to discharge the batteries enough to activate full solar recharging to obtain peak power and solar yield measurements.

Note 2: Effects of Maximum Adsorption Time and Tail Current settings. There were some configuration changes part way through the data set - the use of Tail Current to terminate Adsorption was turned off and the maximum Absorption time reduced from 6 to 2 hours, then the use of Tail Current was restored after a few weeks. This resulted in a number of days with an absorption time of 20 mins. The reason was that the manual implied that when the switched load was in use it was used for tail current measurements whilst my configuration would only use the switched load for supplying power to the 12V system and the high currents for the inverter would be directly from the batteries as recommended. This did not turn out to be the case. When Tail Current is in use and the batteries are not being discharged overnight (ie when on shore supply) the current is very low already at the end of the Bulk phase initiated at sunrise and adsorption is terminated very quickly and a large number of short readings of 1 minute appear.

Appendix G - Setting up Charging and Measuring systems [WIP]

Both the SmartSolar Charge Controller and SmartShunt need some information to give optimum performance. I will first look at the SmartShunt which has a large number of parameters which can be adjusted to optimise its estimate of State of Charge (SOC). They fall into two classes: Information to detect and reset the State of Charge estimates so they do not 'drift and information to translate the measurements of current into and out of the battery to a figure for Consumed Ah and then to State of Charge.

Estimating the Battery Capacity [WIP]

The most basic input you really must make is the Capacity of the Batteries which sounds very easy as it is printed on the side, however that assumes they are new and properly formed. In practice the capacity is reduced quite rapidly with use and the degredation depends very much on the depth of discharge they are taken to. Any discharges below 50% will cause rapid degredation and, depending on type, you will get 150 to 850 cycles when discharging to 50% before the capacity is halved according to the manufacturers. So you really need to know what the actual capacity is to estimate how deeply you are discharging them as they age or feed in very conservative figures.

This problem does not seem to be addressed in any places I could find but there are ways of estimating the State of Charge itself so I am looking at ways of working backwards from depth of discharge figures combined with the Consumed Ah to estimate the Capacity. There are two main ways to measure the State of Charge - for unsealed flooded batteries you can measure the Specific Gravity of the electrolyte, for sealed batteries you can use the voltage with no load. The use of the Open Circuit Voltage (OCV) depends on knowledge of the type of chemistry of the battery, the temperature of the battery and the battery must be completely free of any load for ideally several hours. You really need information from the manufacturers to make a good estimate but if the battery manufacturer's State-of-Charge (SoC) specifications are not available for the battery there are tables available on http://www.batteryfaq.org for various generic types of batteries and I thank them for allowing use of their information and spreadsheets such as that below.

These tables show that this is not going to be easy as there are significant differences due to temperature and type of chemistry which we will also need to address in the setting up the Solar Charge Controller. You will note that the 100% charge voltages are significantly higher for the various sealed batteries and the voltage change between 0 and 100% SOC is 32% greater which means that the chemistry plays a large part. In contrast the temperature change offsets seem to be the same for both. Note: The temperature coeficients for variations in Open Circuit Voltage in the table above are not the same as those recommended for setting up the Absorption and Float voltages for charging. If we plot Open Circuit Voltage versus Consumed Amp hours it will give us the Chemistry from the the 0% consumption if we fully charge the battery and leave it for a few hours and the SOC from the slope for the appropriate battery type. At least it will be better than my current guess of 67% of new capacity. One problem will be the need to allow several hours between each reading, another is there are many other factors that can affect the voltage readings so every battery manufacturer will have a slightly different voltage performance curve under load.

Another problem is surface charge. All the figures above indicate that at 100% the batteries should read between 12.64 and 12.78 times 2 for a 24V system at 20.1 degrees whilst a quick look at the output from the SmartConnect shows a minimum battery voltage which is much higher at ~25.84 (equivalent to 12.92V) on days on a shore supply when the load is small! This memory effect after a very full charge is well known and is due to surface charge. It occurs because Lead Acid batteries cannot convert lead sulfate to lead and lead dioxide quickly during charge so some of the charge activities to occur on the plate surfaces, resulting in an elevated state-of-charge on the outside. A battery with surface charge has a slightly elevated voltage and gives a false voltage-based SoC reading. It dissipates quite slowly especially at temperatures below 25 deg C. - it means we will either have to wait for weeks or carry out a small discharge then wait before starting.

The following graph shows a typical surface charge effect on an almost new 105 Ah Flooded Lead Acid battery discharged at 5.25A ( ie C/20 ) over 20 hrs after an initial absorbtion level charge followed by a period under no-load. It shows a reading well over 13 volts at the start which did not fall to the expected voltage until over 1Ah had been taken out.

It is not only the State of Charge that depends on an accurate value for the Battery Capacity, the accuracy of available Ah will also be affected as the Peukert Exponent corrections uses it as is the efficiency (I have left them as default values) and we need to sure the battery is fully charged when starting the calibration. At least we have temperature measurements available! Overall I am not convinced I will improve on my guesswork by much but keeping an eye on battery voltage will still be useful and allow some subjective corrections to my guess.

Keeping the batteries above 24.5V after a short recovery time with no load at normal temperatures seems to be a useful way of ensuring that most Flooded Lead Acid batteries are not going to seriously damaged and was, in retrospect the point at which my batteries used to degrade rapidly if regularly reached in the morning before I added Solar!

Despite my reservations above I decided to try to develop a method to improve my estimates of battery capacity and hence state of charge optimised towards the 50% level where damage rapidly increases. Some of the ground rules for making measurements derived from above are:

So I have taken a number of sets of readings overnight when only the fridge is running with a duty cycle of ~40% and a cycle time of ~ 1 hour allowing 40 mins for the Open Circuit Voltage to stabilise. The average discharge rate is 40 watts giving a relatively low discharge rate of ~ C/75 so the Open Circuit Voltages have stabilised within .01 V by the end of the off cycle from inspection of the voltage versus time plots in the SmartShunt trend screen. Readings are ignored until ~ 5% of capacity has been drawn to avoid surface charge effects. After much lose of sleep I have ended up with 45 points in 8 data sets some of which are 'extended' by the additional consumption of boiling a kettle (4 Ahr) before solar kicked in. I have plotted the results which show scatter as one would expect but are reasonably consistent even so. Each series has been plotted individually with a different colour but a single best fit trend line is shown on the first plot shown below. The intercept with the 0 Ahr discharge axis is at 25.88 volts which indicates the batteries are behaving more like sealed cells and the intercept with the the 24.5 volt line (assumed to be ~50% discharge) is at 52 Ahr indicating a remaining battery capacity of only ~104 AHr. This is below the figure I currently assume and indicates my batteries are already at under 50% of new capacity. So further investigation of this data is required.


Data from 8 nights combined for a single trend line

Individual trend plots for selected short and long series and a second order polynomial fit to the whole data set

The second plot shown above has trend lines for some individual series which show a spread with the longer series in general having a higher intercept. If a second order fit is done there are further indications that the capacity would be higher - more readings at higher total discharges are required to confirm that.

The second set of readings and and curve above comes from an independent set of measurements on a pair of Powermax batteries which were replaced after just under 2 years prior to our new system. They still show green on their built in indicators and are still matched. The plot hower shows that their capacity is down to about 50% (27 Ah discharge to 24.5 Volts ~= 54 Ah Capacity instead of 110 Ah claimed when new). The intercept at 0% discharge is at 25.46 is almost the classic figure for a flooded lead acid battery which can be topped up, which they have been every 200 hrs. The new batteries in contrast behave like sealed batteries with an intercept at 25.88V. This shows the potential problems of mixing batteries from different manufactures and ages and use with simple alternators without adjustments or charge controllers.

There has also been no corrections for battery temperature which would bring down the initial readings in each set slightly as the batteries were still cooling again tilting the initial curves slightly to increase the capacity. I do not have the temperature data to correct the curves properly but the correction curve below is plotted from a table allegedly originating from The Battery Council and will be of use in the future.

Note: This has much larger corrections to those in the table above for SOC from batteryfaq.org

Appendix H - What is this Peukert’s law and the Peukert exponent?

I have refereed to Peukert's law above as important when Coulomb counting is used to calculate State of Charge as used by the Smart Shunt. There is a nice simple explanation of a very complex phenomenon on the Victron Web site at https://www.victronenergy.com/media/pg/SmartShunt/en/battery-capacity-and-peukert-exponent.html which I have simplified even more here.

Peukert’s law and the Peukert exponent

The apparent capacity of a battery depends on the rate of discharge. Basically the faster the rate of discharge, the less capacity will be available from a battery. The relation between slow or fast discharge can be calculated by Peukert’s law and is expressed by the Peukert exponent. Some battery chemistries suffer more from this phenomenon than others. Lead acid are more affected by this than lithium batteries are and a Lead acid battery discharged completely in two hours will only give about half the Amp hours (capacity) available if discharged over 20 hours. The battery monitor takes this phenomenon into account with Peukert exponent. If the efficiency and Peukert's law were ignored counting Amp hours in and out (Coulomb Counting in technospeak) would not work.

The formula that states the Law in a usable format is as follows:

H is the rated discharge time, in (hours).
C is the rated capacity at that discharge rate, in (Ampere-hours).
I is the actual discharge current, in (Amps).
k is the Peukert constant (Victron recommends using a value of 1.25)
t is the actual time to discharge the battery, in (hours).

The formula can be reorganised and written in a form to enter a calculator as:

I*t = C*(C/(I*H))^(k-1)

I*t is the new AH rating (the discharge rate at the time to discharge).

which in my case gives a reduction from 160Ah to 160*(160/(20*40))^(1.25-1) = 107Ah for a 1 kW load.

Where does the charge go: It is worth noting that that is not the end of the story. It looks as if the charge has “disappeared” but that is an over-simplification. What happens is that the chemical process which generates the current is progressing too slowly to allow diffusion through the plates and only takes place at the surface, so the battery looks as if it has a smaller capacity. A 200Ah battery discharged with 200 A and “flat” in 30 minutes will therefore also be (nearly) full again after recharging with 100 Ah at a slow rate, while the same battery which is discharged with 10 A and is flat in 20 hours will be nearly fully charged after recharging with 200 Ah.

In fact a battery which has been discharged at a very high rate will recover over time and most of the remaining capacity can be retrieved after the battery has been left at rest for several hours or a day.

There is lot more about all of this at https://www.victronenergy.com/upload/documents/Book-Energy-Unlimited-EN.pdf - the book is well worth reading in full.

Appendix I - Settings in Use for the SmartSolar Controller, SmartShunt and Phoenix Inverter

The easiest way to show the current settings of the various devices (as of 14 October 2021) is to show screen dumps of their settings screens in the SmartConnect App. Only the SmartSolar Controller, SmartShunt, and Phoenix Invertor need to be configured via the App. The only other settings is on the Orion Tr which supplies power to the 12V domestic battery and that has to be set manually with a screwdriver and is set to 12.71V to trickle charge it.

 
SmartSolar Settings Screens

 
SmartShunt and Phoenix Settings Screens

Appendix J - Power measurements in Domestic 12V system

The SmartSolar measures the power provided to the Load circuit, namely the Orion Tr 24/12 and knowing its efficiency is specified to be 85% allows us to calculate the power used by the various system components it feeds. Some of them turned out to be somewhat surprising to me: a 10Watt halogen light was exactly what was expected but that was not the case for an LED replacement which had so low a power consumption it could not be measured reliably (under 2 watts). Our main lighting is by fluorescent lights which each contain three tubes each of a nominal 8 watts but only drew half that expected but a single 13 watt tube drew significantly more than the nominal power. The Hifi drew a surprisingly high 12 watts playing a CD at low volume. The powers are that from the 24 volt supply as that is what is measured and actually matters in this case. Potentially the largest steady demand on the 12V system is when running or charging our laptops. The two most likely to be present have 45Wh batteries and run for ~4 hours on batteries so will consume about 10 watts when charged. When in use and charging this could be up to 40 watts which are their nominal charger ratings.

Item Nominal Power (Watts) 24V Load Power (Watts) Hours per day Wh/day (24V) Comments
Water Pump
-
45
0.2
9
Shurflow
Tunnel Light (LED)
~9
8
--
-

620 Lumen LED - Occasional use only when engine running

LabCraft Lights (5)
24
13
3 x 4
156
3 x 8 Watt tubes
HLC Light 20"
13
18
0.5
9
Corridor Single Tube
HLC Light 11"
8
10
0.5
5
Bathroom and Bedhead Single Tube
Wall Light Halogen
10
12
-
-
Hardly used
Wall Light LED
2
2
2
4
Often used
Hifi playing CD
-
12
3
36
Medium Volume
S3 Mini Phone (Tethered for Internet)
24
14
Provides Wifi 24 x 7 (Tethered 3G) 6Wh battery
Charging Phones
3 x 2
14
2 Samsung A6 phones 12Wh batteries discharged to ~50% daily
Samsung Tab S6
5 (Estimate)
4
4
16
Browsing or Streaming TV 27Wh battery discharged to ~50%
Multipurpose 12V PS (No Load)
-
2
24
12
Idle keeping lap tops charged
Laptop using 12V PS
10 (Estimate)
14
2
28
Chillblast Helios Laptop ( charged and in use) 40Wh battery provides ~4 hours use
TOTAL
303

The 12V power consumption is dominated by lighting during winter cruising - most are Labcraft which seem to be quite low power consumption despite having up to 3 fluorescent tubes. We tend to operate daylight hours but we normally have two lights on in the living area if reading and one or two in the kitchen whilst cooking plus occasional use on demand elsewhere. Wifi provision, charging of phones and tablets and laptop use is not greatly dependent on season. The total use would allow ~2 days use to 60% charge on a new 100Ahr battery out of season hence the need to link to a solar power source especially if on a shore supply to avoid running an engine just for a 12V supply

Appendix K - LEC R5017W 230V fridge.

Purchased but not yet fitted

The fridge is our main (only) non-discretionary power consumer on the 24/230V system. We initially intended to change to a low voltage 12/24V fridge which thought would have a major impact when we looked at the power consumption of the modern 12/24 volt fridges available from chandlers. It turned out they were mostly modifications of standard fridges with a low voltage Danfoss compressor and had the same power consumption as their parent fridges. The were obviously still some savings as in did away with the losses in the inverter but in the end we decided to get a 230V version of the LEC fridge which forms the basis of the Inlander Rir405W which is well proven in marine applications.

The LEC R5017W 230v is a 50cm wide fridge with 4* freezer and A+ rating which the manufacturers allege only uses 167 units year which equates to 458 Wh/day. Assuming the Victron Phoenix 24/1600 inverter is 94% efficient plus idle power, in Eco mode and the fridge runs for 6 hours a day -> 458÷.94+6×9+18×1.3 = 564 Wh/day which would be ~1/3 of existing fridge/inverter combination. More recent measurements on the Victron Inverter Zanussi fridge combination give a consumption during a warm September of 820 Wh/day giving ~250 Wh/day or 30% calculated saving from changing the fridge which will still be useful.

In summary the reasons for choosing the 230V version via an inverter were:

Measurements on LEC R5017W in 'ideal' conditions before installation

Measurements on LEC R5017W set with Fridge ~6 deg and freezer -18 deg C with room temperature 20 deg in free air has 9.5/42 on ratio = 22.5%. The compressor is a D53CY1 using R600a coolant which I found on the web to have a power input of 65 Watts (and a 80 watt cooling capacity). 65 x 24 x .225 = 351 Wh/day which looks low compared to the LEC data. It is also described as a 1/10 hp unit which gives 765 / 10 x 24 x .225 = 413 Wh/days. I wonder if this mean that manufacturers Energy Efficiency figures include factors for opening/closing the door and freezing/cooling of food or assume a different ambient temperature of say 25 deg C ?

In normal operation the sides of the cabinet quickly rise about 10 deg C above ambient and the compressor rises much more slowly towards 18 degrees above ambient. These measurements were taken with the fridge sited well away from any obstructions and sitting on tight weave kitchen tiles from Heuga. The measurements were taken with an IR remote thermometer I bought from Screwfix several years ago to check domestic radiators and also use for 'Covid' fever checks on a forehead.

[Plans for] Installation of LEC R5017W on Corinna [Work in Progress]

The manufacturers specify a clearance of 2.5 cm on the sides and top and 5 cm at the back when installed under a counter. The existing fridge has cooling coils on the back so there is much tighter fit at the sides with circa 3 cm total clearance available. At the back there is a gap at the bottom allowing cold air to be drawn from the bilges and from below the water line. The actual gap is however a bare 5 cms until the top where there is an exit gap upwards for hot air.

The cupboard beside the fridge is shallow front to back (30cm) so does not overlap the whole side and can and will be moved back and sideways to increase the clearance by 2.5cm. The fridge may need to be slightly forwards of the work surface to increase clearance at the back but will still be inset from the pillar on the worksurface end. I have a USB powered variable speed fan providing up to 50 cubic ft/min available which can be mounted to provide cool (river temperature air) over the compressor and aiding flow up the sides if required. I also have a small 40 deg thermostat which can be bonded to the compressor but it may be too slow to respond to be useful.

Appendix L - My Original 'Day-Book' for Reference

Whilst I was planting and implementing the upgrade I kept a 'day-book' with all my thoughts, references, measurements, planning etc. Much has ended up in the in a polished version above but there is always useful information and dead ends recorded, which may be useful - one should never destroy a day-book and original measurements.

You can Click to Expand or hide my original working document ( effectively a day-book kept as a text file in the cloud so accessible from the nearest machine). I have done a little formatting from the original text file to display better in HTML and new sections may, on occasion be transferred from the working document.


Before you leave

I would be very pleased if visitors could spare a little time to give us some feedback - it is the only way we know who has visited the site, if it is useful and how we should develop it's content and the techniques used. I would be delighted if you could send comments or just let me know you have visited by sending a quick Message.

Link to W3C HTML5 Validator Copyright © Peter & Pauline Curtis
Content revised: 24th October, 2021