Battery Monitors and Capacity

[nicknorman]

9 minutes ago, dmr said:

I've never done it myself as my interest was always fast data acquisition, but I think its the method used in digital voltmeters; charge the capacitor with the applied voltage and count up, switch to a good quality current source and discharge back to zero, again counting. As long as the current source is accurate this gives a measure of voltage independent of capacitor value.

..............Dave

That’s dual slope integration. I’m pretty sure that’s not what gibbo used since if you look at the current waveform on the supply terminals (measured by scoping a series resistor in the supply) and with the display off to lose the led display multiplexing clutter, you get a series of pulses as the capacitor is quickly discharged each cycle. I can’t exactly remember if I looked at the capacitor waveform but pretty sure I did and it was a sawtooth, not triangular.

[nicknorman]

19 minutes ago, Dr Bob said:

The bit in bold is the key to how the SG operates. This is not an easy thing to do and is accomplised by taking a set of input data viz dV/dt, V over time (not quite the same as dV/dt) etc etc and 'guessing' output data ie SG. The 'guessing' is done by a software model - likely a linear regression type (or multi linear regression) unless Gibbo was using more advanced techniques like topology or sparse data matching (which is unlikely as he would have said that to help sell it). There is always going to be some error in that model and particularly if a 'system' being monitored has different characteristics to the systems used to build the models (ie is outside the training set of the model). I know from what is said here that 'thousands' of systems were tested so the model should be quite robust BUT it will not be a 100% global model capable of being 100% correct with EVERY system. It obviously works extremely well on Nick's system. It may have difficulty tracking some 'outlier' systems. Nick you have on a few occations said that if the Voltage is correct then so will the SoC. That will only hold if the model is a 100% global model. I dont think anyone knows the answer to that (I certainly dont) apart from Gibbo. I dont think I have seen a 100% global model for inputs of the complexity we  have here.

Yes I think this is all very true but how much variation is there between the behaviours of different lead acid batteries (and bearing in mind there are different models for gel, AGM etc)? The installation of the batteries shouldn’t really come into it as the device is connected directly to the terminals. Also the SG is able to learn by comparing the calculated SoC during discharge with the actual SoC when there is subsequently no discharge (zero dV/dt) and back-calculating to tweak the parameters used so that next time, the calculation during discharge is closer. In other words, in between the guesstimating, there are periods when the SoC is known pretty precisely. This is why there aren’t the typical cumulative integration errors found in AH counting gauges.

[Dr Bob]

8 minutes ago, nicknorman said:

Yes I think this is all very true but how much variation is there between the behaviours of different lead acid batteries (and bearing in mind there are different models for gel, AGM etc)? The installation of the batteries shouldn’t really come into it as the device is connected directly to the terminals. Also the SG is able to learn by comparing the calculated SoC during discharge with the actual SoC when there is subsequently no discharge (zero dV/dt) and back-calculating to tweak the parameters used so that next time, the calculation during discharge is closer. In other words, in between the guesstimating, there are periods when the SoC is known pretty precisely. This is why there aren’t the typical cumulative integration errors found in AH counting gauges.

....and I agree with all you have said. The SG is a clever machine given its programming but it could predict incorrectly on an outlier system.

Your question on how much variation is a good one. I would think there is a lot - ie all the things that cause SoC to change, from things on the boat, like all the different potential loads and chargers (alternator/solar/genny), to the batteries and how sulphated they are, temperature, state of plates, electolyte level etc etc and then the instalation itself, resistance on wires, conections etc. How many of these did Giboo address when building the model? By address, I mean include in the training set. A variable that has a big effect is temperature. A 10 deg C swing IIRC gives a difference of 0.05V (long time ago since I looked at this) which is circa a 5% difference in SoC if temperature effect is not considered. I dont know what people do on NBs but on the lumpy water boats, very few of my mates had their meters connected up to measure battery temp. Maybe everyone here does?

[cuthound]

5 hours ago, dmr said:

I reckon there are broadly 4 "types" of lead acid battery.

Starter/car/leisure batteries    Thin plates, calcium, smallest shortest cells.

Semi Traction "Trojans"  Thicker plates, Antimony, bigger and taller cells (more reserve space for shedded crap). Usually 6 volt.

Traction. Even thicker plates? Antimony, very big and very tall cells. Always 2 volts

AGM, a few variants but the acid is not a free liquid but trapped in an absorbent material.

There are then various sub-types to these types, some with significant differences, and some "hybrids" like the 12 volt Trojans.

Each type has various brands.

..............Dave

Don't forget lead acid batteries are also available with spiral and tubular plates as well. These pack more power into a smaller battery by increasing the surface area of the plate and in the case of the spiral one increasing plate strength as well.

[Dr Bob]

2 hours ago, mross said:

All resources are finite!  The cost of extracting and incorporating Li in batteries will likely fall, even if its commodity price rises.

To contradict the FT article, this one from Science Today says the future for the important metals is good - just short term bottlenecks.

https://www.sciencedaily.com/releases/2017/10/171011131704.htm?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+sciencedaily%2Fmatter_energy%2Fbatteries+(Batteries+News+--+ScienceDaily)

There is enough info on the tinternet to make an argument either way. Just who do you believe?

[dmr]

4 minutes ago, cuthound said:

Don't forget lead acid batteries are also available with spiral and tubular plates as well. These pack more power into a smaller battery by increasing the surface area of the plate and in the case of the spiral one increasing plate strength as well.

Yes, thats exactly what I was thinking off when I mentioned "sub types". There are also some reports of carbon nano-tubes. These things are probably "details of construction" rather than "types" but all classification schemes are somewhat arbitrary. If I was trendy I would be calling them "forth generation batteries" or even 4.5G

What I was trying to say is that "fundamental construction" and Brands are two very different attributes.

.............Dave

[nicknorman]

15 minutes ago, Dr Bob said:

....and I agree with all you have said. The SG is a clever machine given its programming but it could predict incorrectly on an outlier system.

Your question on how much variation is a good one. I would think there is a lot - ie all the things that cause SoC to change, from things on the boat, like all the different potential loads and chargers (alternator/solar/genny), to the batteries and how sulphated they are, temperature, state of plates, electolyte level etc etc and then the instalation itself, resistance on wires, conections etc. How many of these did Giboo address when building the model? By address, I mean include in the training set. A variable that has a big effect is temperature. A 10 deg C swing IIRC gives a difference of 0.05V (long time ago since I looked at this) which is circa a 5% difference in SoC if temperature effect is not considered. I dont know what people do on NBs but on the lumpy water boats, very few of my mates had their meters connected up to measure battery temp. Maybe everyone here does?

The installation itself (resistance on wires, connectors etc) is not relevant. Different loads and chargers just corresponds to different net current in or out. Very easy to simulate and gather data on.

How sulphated they are probably does have some impact but in the main, a sulphate battery tends to behave like a good battery but just with a lower capacity, and since the SG has no concept of battery capacity (it is purely a relative SoC calc). Gibbo did say that the model broke down once the actual capacity was below 50%.Temperature does seem to make the SG read a bit higher, but then temperature also increases the capacity. Depending on how you measure it!. So maybe that is right. Once again, it all down to the difficulty in defining SoC, let alone measuring it! Ditto I’ve noticed that when the batteries need topping up (acid concentration higher) the SG tends to read a few % higher. But again, perhaps the capacity as measured in the normal way, is higher with stronger acid?

[alan_fincher]

2 hours ago, nicknorman said:

 

 

Obviously calibration is important, but can they work well when they are  that much out of focus?

[nicknorman]

9 minutes ago, alan_fincher said:

 

Obviously calibration is important, but can they work well when they are  that much out of focus?

Hmmm well that is a small chunk of a much larger photo. It didn’t look blurred until I cropped it severely!

[MtB]

14 minutes ago, alan_fincher said:

 

Obviously calibration is important, but can they work well when they are  that much out of focus?

And in a similar vein, the type of capacitor is obvious to me from the photo, even though blurred. It's one of those big black buggers.  

[nicknorman]

12 minutes ago, Mike the Boilerman said:

And in a similar vein, the type of capacitor is obvious to me from the photo, even though blurred. It's one of those big black buggers.  

Naughty - I think you mean “a capacitor of colour”!

[WotEver]

2 hours ago, Dr Bob said:

... unless Gibbo was using more advanced techniques like topology or sparse data matching (which is unlikely as he would have said that to help sell it)

No he wouldn’t

He never spoke at all about how it works other than to say it contained “a complex, self-correcting algorithm”

’Son of SmartGauge’ otherwise known as Datacell is much more complex to install (as it uses multiple shunts) and can tell you an awful lot more about your bank but at its heart it still contains two SmartGauges. It also costs a grand. 

[mross]

3 hours ago, Jon57 said:

Diesel

If diesel fuel has gone up in price that is due to tax!

[WotEver]

3 minutes ago, mross said:

If diesel fuel has gone up in price that is due to tax!

Hmmm... in 2006 the duty on fuel was 61.35p per litre. Since 2012 the duty has remained constant at 57.95p per litre. So not only has duty reduced in real terms but it hasn’t even kept pace with inflation so... no, you’re wrong on that point. 

[Robbo]

4 hours ago, mross said:

Give me some examples..........

Computer memory regularly has had its shortages over the years.   Prices rise when it does.   However memory prices is only a small part in a computer, lithium prices are around 35% the cost of the battery.

Edited October 23, 2017 by Robbo

[mross]

13 hours ago, WotEver said:

Hmmm... in 2006 the duty on fuel was 61.35p per litre. Since 2012 the duty has remained constant at 57.95p per litre. So not only has duty reduced in real terms but it hasn’t even kept pace with inflation so... no, you’re wrong on that point. 

Go back further; the price of crude, which determines the price of diesel (ex tax) is about the same as in the 80's despite a 50% increase in demand.

[Robbo]

30 minutes ago, mross said:

Go back further; the price of crude, which determines the price of diesel (ex tax) is about the same as in the 80's despite a 50% increase in demand.

Oil prices have been going up and down for decades for one reason or another ...

 

http://www.macrotrends.net/1369/crude-oil-price-history-chart

[Dr Bob]

13 hours ago, WotEver said:

He never spoke at all about how it works other than to say it contained “a complex, self-correcting algorithm”

So he isn't using one of the advanced techniques then.

 

7 minutes ago, Robbo said:

Oil prices have been going up and down for decades for one reason or another ...

 

http://www.macrotrends.net/1369/crude-oil-price-history-chart

And mostly due to the actions of the Cartel (OPEC). There is no cartel for Lithium yet. What about other metals?

[Robbo]

8 minutes ago, Dr Bob said:

So he isn't using one of the advanced techniques then.

 

And mostly due to the actions of the Cartel (OPEC). There is no cartel for Lithium yet. What about other metals?

Well lithium prices have significantly increased recently looking at the charts.

[Dr Bob]

7 minutes ago, Robbo said:

Well lithium prices have significantly increased recently looking at the charts.

Defo supply and demand at the moment but look at the post I put up earlier on long term availability. No one knows how it will go. The FT said one thing. My link says no problem. Long term it may not be Lithium......and there are other metals involved. We just do not know how prices will look in 5 or 10 years time. 

[Robbo]

26 minutes ago, Dr Bob said:

Defo supply and demand at the moment but look at the post I put up earlier on long term availability. No one knows how it will go. The FT said one thing. My link says no problem. Long term it may not be Lithium......and there are other metals involved. We just do not know how prices will look in 5 or 10 years time. 

Well supply and demand is what we are referring to!   If the demand increases (like for electric cars) and the supply can't increase with it the prices will go up.   

They do say it won't be lithium that will be the issue tho, for electric cars it will be the Cobalt part, which is rare!

[BruceinSanity]

1 hour ago, Robbo said:

Well supply and demand is what we are referring to!   If the demand increases (like for electric cars) and the supply can't increase with it the prices will go up.   

They do say it won't be lithium that will be the issue tho, for electric cars it will be the Cobalt part, which is rare!

There's a piece in the Indy today about Chinese recyclers gearing up to deal with Li batteries from electric cars in due course. At the moment it costs 8,540 yuan to produce a tonne of metal worth 8,110 yuan, but those numbers will cross over as demand rises, I guess.

[Dr Bob]

5 hours ago, Robbo said:

Well supply and demand is what we are referring to!   If the demand increases (like for electric cars) and the supply can't increase with it the prices will go up.   

They do say it won't be lithium that will be the issue tho, for electric cars it will be the Cobalt part, which is rare!

This link I posted earlier does not seem to think supply in the future will be a problem.

https://www.sciencedaily.com/releases/2017/10/171011131704.htm?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+sciencedaily%2Fmatter_energy%2Fbatteries+(Batteries+News+--+ScienceDaily)

Who knows?