Smartgauge ponderings

[nicknorman]

First of all I want to make it quite clear that I think the Smartgauge is a great bit of kit. I have one and don't regret buying it. Its beauty is its cost and in particular its simplicity of installation and autonomous operation compared to other SoC gauges. But its not perfect, and I decided to try to understand how it works, so as to better understand its limitations.

 

I feel slightly guilty because I suspect Gibbo might be banned again, thus denying him the opportunity to respond straight away. I am no doubt saving up for a mighty beasting when he is back, but hey ho you have to take a risk in life... And as I said, I am not trying in any way to dis the product, in fact the opposite because the better it is understood the better it is to use and hence the more will be sold!

 

So, I have taken my SG home. I have always suspected that it works solely by measuring voltage and then carrying out some very clever processing on that data to give SoC, probably starting out by calculating dV/dt and d²V/dt². Other theories about it measuring battery impedance etc just can't be true IMHO.

 

So, I connected it to a regulated power supply set to 12.6 volts. It powers up showing the default 75% SoC. I then wind the voltage up to 14.4 and go away for a while. When I come back, its showing 78% SoC. Then I turn the voltage back slowly to 12.8 (simulating or no little load), and then down towards 11.8v to simulate a heavy load, notching the voltage slowly down over a few minutes. Result is that by the time I have finished typing this the SG now shows 55% SoC. Therefore, since I can drive the SoC up and down simply by changing the supply voltage from a regulated power supply, it seems reasonable to assume that at least the primary thing it is measuring is voltage.

 

My current observation is that our batteries struggle in winter. Of course this is due in part to early darkness, too much telly, and being physically cold (they sit on top of a thin bit of wood on the swim, battery temperature has been around 4 deg C recently). The reason I say "I think they struggle" is that the SG shows a much lower SoC (more rapidly dropping off from 100%) than does the Mastershunt Amphour counter. Of course the batteries do have lower capacity in cold weather (or at least the ability to produce the energy reasonably quickly) but I also think that the rested open circuit voltage of a Lead Acid bat decreases with decreasing temperature whilst its SoC remains constant. This is slightly counter-intuitive since we know that charging voltage should increase with decreasing temperature.

 

So if SG works primarily on voltage and the voltage of a cold battery, versus a warm one in exactly the same SoC, is less, that implies that the SoC reading will be less for a cold battery, than for a warm one in exactly the same SoC.

The SG doesn't know the temperature of the batteries and, if its in the cabin when the bats are elsewhere, its own temperature is irrelevant.

Put this together with probable reduced actual capacity at low temperature, and increased demand, and they all help to result in an unexpectedly low SoC on the SG.

 

Yesterday, having done an equalise and got the bats to 100% I turned off the charger and ran various loads all day. The SoC on the SG was dropping much more quickly than the MS. I ended up with 50% on the MS and 35% on the SG. Eeek! I then turned everything off and left the bats to settle. They came back up to 12.2v quite quickly, which suggests to me that the real SoC was in fact nearer the 50%. It was perhaps because the bats were cold and the voltage was dipping more quickly due to slow chemical reactions, that had caused the SG to under-read. If true, that's a useful pieced of info because it means I can afford to run down below 50% on the SG in cold weather without worrying.

 

I intend to prove this by getting a small battery and measuring its voltage having soaked it warm, cold and then warm again without changing the charge. I think the SG does react to static voltage over long periods of time, it will be interesting to see if it gives a different SoC for the same battery, same charge state but different temperatures.

 

I'll let anyone who has managed to wade through this tome without falling asleep know the outcome!

Edited January 23, 2013 by nicknorman

[wiccan]

I will be interested in your findings.

[Nickhlx]

Likewise, I will too...

 

I am slightly swayed to believing that it is likely to be "correct" by the enormous and deep knowledge of the designer though, but will maintain an open mind

 

 

Nick

[Mac of Cygnet]

Is Gibbo banned? If so I missed it. He wasn't banned the first time anyway, but left of his own accord (and came back ditto).

[nicknorman]

Is Gibbo banned? If so I missed it. He wasn't banned the first time anyway, but left of his own accord (and came back ditto).

Same difference in effect!

[matty40s]

what you are doing with your homegrown electrical regulated supply is almost exactly what your smartgauge would do when first linked to a battery bank - giving a straight forward voltage reading. Effectively tricking the smartgauge. Gibbo always suggested that the smartgauge only started to work properly after around 3 full cycles - you havn't done this.

[nicknorman]

Likewise, I will too...

 

I am slightly swayed to believing that it is likely to be "correct" by the enormous and deep knowledge of the designer though, but will maintain an open mind

 

 

Nick

I totally agree with your sentiments about the designer, but great engineering is all about compromise. In the case of the SG it is a compromise between cost, installation difficult (doing away with the need for shunts and temperature sensors) versus accuracy, SoC is really not an exact science so the compromise struck is a good one for most purposes. As I said, I have total respect for the design, but it is not and can't be perfect. I want to know its limitations!

 

what you are doing with your homegrown electrical regulated supply is almost exactly what your smartgauge would do when first linked to a battery bank - giving a straight forward voltage reading. Effectively tricking the smartgauge. Gibbo always suggested that the smartgauge only started to work properly after around 3 full cycles - you havn't done this.

It learns the behaviour of the battery by looking at the relationships between voltage, dV/dt and d²V/dt² and perhaps long term rolling averages of these, so as to build a model of internal resistance and all the other complex behaviours of the individual battery, so as to refine its readings. It is very complicated and clever, but ultimately only looking at voltage and its derivatives. What else do you suggest that its looking at? Much as we respect its designer, it doesn't work by divine intervention!

Edited January 23, 2013 by nicknorman

[nb Innisfree]

Using lots of agorithms to compare various voltage fluctuations synoymous with charging/discharging characteristics of whatever battery type is selected at set up thereby making increasingly educated guesses as to SoC by narrowing down the options.

 

IMO.

Edited January 23, 2013 by nb Innisfree

[nicknorman]

Using lots of agorithms to compare various voltage fluctuations with whatever battery type is selected at set up thereby making increasingly educated guesses as to SoC by narrowing down the options.

 

IMO.

 

Exactly! Although I would have put it more kindly!

[nb Innisfree]

Exactly! Although I would have put it more kindly!

 

Read my edit, an improved description.

[Bod]

what you are doing with your homegrown electrical regulated supply is almost exactly what your smartgauge would do when first linked to a battery bank - giving a straight forward voltage reading. Effectively tricking the smartgauge. Gibbo always suggested that the smartgauge only started to work properly after around 3 full cycles - you havn't done this.

I too think there must be a bit of averaging going on, to only get reliable results after several cycles, shows to me a form of learning!

 

Do your basic tests, once a day, for a week, any changes in results would be interesting.

 

Bod

[by'eck]

Although I don't have a Smartguage, I have on many occasions pondered how it works and agree with your conclusions, as its only input is voltage (plus time of course). So to put it in laymans terms, it fine tunes its results by measuring battery voltage change with time and the rate of change of such against time.

 

Doubtless untold amounts of testing went into developing the algorithms, but given the diversity of battery types and states of same, it can at best only provide a compromise on accuracy albeit a good one I'm sure.

 

I'm surprised that a battery sensor is not included with the Smartgauge. Would this not allow further input to the algorithms to increase accuracy for no great extra cost? Having said that I remember reading a recent response from Gibbo himself confirming that battery SOC doesn't change with temperature but that is does modify the Peukert component as a result of rate of chemical change with temperature.

 

To sum up, for me there is no substitute for regularly monitoring battery voltage and charge/discharge currents of your own batteries under whatever regime you use them, and understanding the results. Any sudden change of expected results may well pre-empt something sinister on the horizon before chance of damage.

[nicknorman]

To sum up, for me there is no substitute for regularly monitoring battery voltage and charge/discharge currents of your own batteries under whatever regime you use them, and understanding the results. Any sudden change of expected results may well pre-empt something sinister on the horizon before chance of damage.

 

That's fine if you are a battery enthusiast, but for joe public the SG is a fit and forget, instantly read, reasonably accurate indication of SoC.

[nb Innisfree]

It has to be remembered that the SG isn't a substitute for skilled reading and translating of a voltmeter, ammeter, hydrometer and ah counter, it's designed as a simple to read battery SoC, aimed at the technically challenged or for those who are skilled enough to translate readings but can't be bothered to.

[by'eck]

That's fine if you are a battery enthusiast, but for joe public the SG is a fit and forget, instantly read, reasonably accurate indication of SoC.

 

True - but that's why I included "for me"

[Paul C]

Post #42 on this forum :

 

http://www.ybw.com/forums/showthread.php?73924-Smartbank-battery-management&p=716133#post716133

 

more or less admits that it works by AC impedence spectrography.

 

2 wires doesn't necessarily mean its only a voltmeter. I believe the 'clever stuff' is the algorithms developed from empirical data, to relate its readings to a simple % state of charge. I suspect it will also use measurement of the battery voltage, as a 'sanity check' and in the initial 3 discharge/recharge cycles. It probably has an algorithm that weights whether it uses AC impedence and/or volts depending on the time connected, discharge/recharge cycles, and/or closeness of fit between the two methods, so that it is accurate more often that not.

 

There are other battery monitors which use the same technology (ie it is not unique) although the next one (in price) is about £250 or so.

[smileypete]

INTERESTING If you have access to a scope try putting both the scope (set to AC) and the SmartGauge on a lead acid battery, see if it uses AC as well.

 

http://www.canalworld.net/forums/index.php?showtopic=52921&view=findpost&p=994993

 

Yesterday, having done an equalise and got the bats to 100% I turned off the charger and ran various loads all day.

Just an observation, it often seems to me that '100%' isn't always quite the same as 'fully charged' as described in The Battery FAQ.

 

cheers, Pete.

~smpt~

[dmr]

Post #42 on this forum :

 

http://www.ybw.com/forums/showthread.php?73924-Smartbank-battery-management&p=716133#post716133

 

more or less admits that it works by AC impedence spectrography.

 

2 wires doesn't necessarily mean its only a voltmeter. I believe the 'clever stuff' is the algorithms developed from empirical data, to relate its readings to a simple % state of charge. I suspect it will also use measurement of the battery voltage, as a 'sanity check' and in the initial 3 discharge/recharge cycles. It probably has an algorithm that weights whether it uses AC impedence and/or volts depending on the time connected, discharge/recharge cycles, and/or closeness of fit between the two methods, so that it is accurate more often that not.

 

There are other battery monitors which use the same technology (ie it is not unique) although the next one (in price) is about £250 or so.

 

I have read the link and it appears to be a response to a previous critical post (that I could not see). However reading it carefully I do not think Gibbo is saying that Smartgage is using AC spectroscopy, or even measuring impedance, he is just citing these as examples of things that could be done with a two wire system, probably in response to something said in the critical post. I have access to a high frequency current probe and full data acquisition equipment so I could put this on a Smartgage wire and see if Gibbo is using current pulsing. But this is not really fair is it? Its getting a bit too close to reverse engineering????

 

.........Dave

[Paul C]

Since Gibbo has pretty much said that he can't (probably for commercial confidentiality reasons) reveal how it works, there's 2 options

 

1) persuade him to tell you

2) reverse engineer it

 

Personally I think its a bit daft that the underlying technology can't be revealed. Its like buying a car, and being told it has a 200bhp engine, then when asking how many cylinders the engine has, being told they can't tell you.

[nb Innisfree]

Since Gibbo has pretty much said that he can't (probably for commercial confidentiality reasons) reveal how it works, there's 2 options

 

1) persuade him to tell you

2) reverse engineer it

 

Personally I think its a bit daft that the underlying technology can't be revealed. Its like buying a car, and being told it has a 200bhp engine, then when asking how many cylinders the engine has, being told they can't tell you.

 

Understandable really as revealing it will lose their commercial exclusivity,

[Timleech]

Since Gibbo has pretty much said that he can't (probably for commercial confidentiality reasons) reveal how it works, there's 2 options

 

1) persuade him to tell you

2) reverse engineer it

 

Personally I think its a bit daft that the underlying technology can't be revealed. Its like buying a car, and being told it has a 200bhp engine, then when asking how many cylinders the engine has, being told they can't tell you.

 

Option 3 -

don't worry, just use it (or not, your choice)

 

Tim

[by'eck]

I see here a challenge to the OP. Do you have a 'scope handy Nick?

[GUMPY]

Im just sitting here smiling wondering why you are bothering unless its a case of tryi g to prove how clever you are.

I have known Chris for a long time and had one of the prototypes to test also have a fair idea how it works having been involved and you are not even close.

[Paul C]

Understandable really as revealing it will lose their commercial exclusivity,

 

I don't understand what you mean. Are you saying that somehow its more "exclusive" because its technology is unknown?

[TNC]

Im just sitting here smiling wondering why you are bothering unless its a case of tryi g to prove how clever you are.

I have known Chris for a long time and had one of the prototypes to test also have a fair idea how it works having been involved and you are not even close.

 

Mrs TNC thinks NB Rusty Dustbin's Smartgauge works by magic!