PIP-4048MS and PIP-5048MS inverters

[weber]

Thanks for the info on 24 V MOSFET drivers, Coulomb.

It was nice to have visits today, at different times, from my friends Trevor Berrill and Greg Breslin (Bladecar). Both interested in the Black Monolith, and both helpful. Thanks guys.

The 5 watt AC/DC module finally arrived (the orange rectangular thing on the lower right in the photo below). So I installed it, and the energy meter, and the breaker for the genset inlet, as you can see below.





Now that the AC switchboard is complete, I have labelled it. Note the gaps beside the contactors for cooling.

[offgridQLD]

Looking good,
             It would have to be very close to installation now on location?


Kurt

[weber]

Looking good,
             It would have to be very close to installation now on location?

Yes. It's booked for Thursday, and I am very grateful that my friend and former co-teacher (of grid-PV design and installation), David Chaplin, has agreed to be the attending electrician.

But it's a bit of a worry that I'm still making changes to the battery management software at this late stage. I'm currently running the battery down, with both coulomb counting (whole battery) and estimated OCV (individual cells) to see how well the estimated OCV detects the 20% SoC point under varying load conditions. OCV = Open Circuit Voltage.

[weber]

I've noticed that very few people know the difference between a relay and a contactor. (We're talking electromechanical here, not solid state). If you google it, you'll find forums with all sorts of nonsense being confidently asserted in response to the question, "What is the difference between a relay and a contactor?". Wikipedia doesn't even get it right.

They will go on about how it has to do with the power rating, or the voltage rating, or the presence of arc suppression or blowouts, or whether it is used for power switching versus control, or whether it is used with a motor, or relays go click while contactors go clunk, or "contactor" is just a trade name for a type of relay. While some of these statements may be rescued from complete falsehood by beginning them with the word "Typically", none of them actually tell you what the difference really is. In some forums, there may be one guy giving the right answer, but he will be drowned in the shower of bullsh*t.

It's dead simple.
In a relay, the moving contacts are hinged.
In a contactor, the moving contacts are floating.
That's it.

Not hinged like a door hinge, but flexibly attached at one end so they only make or break at the other end.

Not floating as in mid-air -- they are still mechanically connected to a piece of iron that is moved by the magnetic field of the coil -- but electrically floating -- not permanently connected to any terminal. Because they are floating, they make or break at both ends at once.

Normally open contacts

    /
---/  o---  relay

   ____
---o  o---  contactor

Changeover contacts

    /o---
---/  o---  relay

---o  o---
---o  o---  contactor

Here's a link to the only page I found that gets it right. It has some good diagrams showing how the two types work.
http://machinedesign.com/engineering-es ... contactors

[T1 Terry]

Thankyou Weber, I'll commit that one into the memory banks, and maybe cut and paste it some where as that is likely to be more reliable form these days I knew there had to be a fundamental difference as technical people speak of them as two different terms within the one sentence.
Learning every day till I die.... not quite sure what happens with all that knowledge after that

T1 Terry

[weber]

Thankyou Weber, ...

Thanks T1 Terry, for all your contributions to this thread. And thanks Johny, for those papers with much more precise and readable charts of open circuit voltage vs SoC for LiFePO4.

Learning every day till I die.... not quite sure what happens with all that knowledge after that

Well, with good enough backups, whatever knowledge you post here may outlive you.

And I'm in need of that knowledge right now. I admit I haven't been very interested in your methods before now, because I was planning to do something better, wasn't I , controlling the PIP-4048MS on an individual cell basis, with Coulomb's and my BMS software. So why am I now eating a big serve of humble pie?

Because the system is booked in to be installed on Thursday, and although last night I went to bed happy (mind you that was at 3:30 am) because the PI-control was working well and had been tuned to perfection (while charging from the grid), this morning I got up and tried it charging from the PV array and it failed utterly.

The "thinking" of the MPPT device in the PIP seems to go something like this:
Oh, you want me to do something.
OK I see the battery voltage is less than the set absorb voltage.
I'll connect the PV array to myself. Relay clicks on.
I'll slowly reduce the operating voltage of the array until I get enough power to put enough current into the battery to bring it up to the voltage you want.
What's that you say? You have a new absorb voltage for me.
Never mind that it's the same as the previous voltage. I'd better drop what I'm doing and start over. Relay clicks off.

So because it gets a new voltage setting every 2 seconds, it spends all its time just getting started and never puts any current into the cells. Or only puts it in for about 2 seconds. It seems to ignore the voltage commands until it starts actually putting a little current into the battery, otherwise the relay would be clicking every 2 seconds.

When I disconnect the optic fibre to the PIP and manually set the ideal voltage for the present state of balancing it works beautifully. By watching the array voltage with a multimeter I can see it operating between max power voltage and open circuit voltage. When I have no AC loads I see it working near the OC voltage. When I put loads on, I see it working near the MP voltage.

So the only problem is this stupid business where it thinks it has to disconnect the array and start over every time its absorb or float voltage is updated (even if it doesn't actually change in value).

I have just 24 hours to get this thing completely working, with the customer's cells, and loaded in the back of the Prius with all the needed tools. David Chaplin (electrician) is arriving at my place 8am day after tomorrow. Therefore it would be madness to start trying to implement some new control scheme.

So it's time to forget about PI control and it's either find a Plan B, or find a warm bath and a razor blade.

And since a couple of people say they like having me around, it had better be Plan B (no, not you Kris). So I think I'm ready to place enormous and ridiculous faith in the inherent balanceability of LiFePO4 cells (after an initial manual balance).

So Plan B is:

Set the CMUs for a 95% SoC charge. Forget this 75% nonsense for now. I'll have the IMU send fixed bulk/absorb and float voltages to the PIP on startup and not do any PI control. Just turn the contactor off at stress 12 and back on at stress 7 to save an individual cell if necessary. And hope it never gets to stress 15 which will drop out the battery contactor and require a human to cycle the stop button to get it going again. I'll set the bypass voltage to just below the average cell voltage of the PIP's setting and hope it gets enough balancing to keep them all balanced.

Then I'll build Monolith #2 and work on more refined software at a more leisurely pace.

What I need from you, T1 Terry, are the magic numbers you suggest for the fixed bulk and float voltage setting for the PIP, and the bypass and alarm (contactor dropout) voltage settings for the CMUs. The CMUs can only bypass 0.8 A. It's a 16 cell system, nominally 48 volts with a maximum charge current from the array of 60 amps. The cells are new 180 Ah grey CALBs.

[offgridQLD]

Do you have any control over the absorb time? Or is it back to the stock PIP rules. Time to reach absorb V = time at absorb V (min 15min max 8hr) ?

55.5V absorb and 53.5v float is my vote. It's what am setting with as my initial starting point numbers for float and absorb on my charge controllers and (400AH calb grey cells)

The contactors on my BCU pull the pin at 4v pr cell high and 2.5v pr cell on the low side.

Hope it all works out in the end.

Kurt

[weber]

Do you have any control over the absorb time? Or is it back to the stock PIP rules. Time to reach absorb V = time at absorb V (min 15min max 8hr) ?

That's right. We've been over the protocol manual with a fine toothed comb and there is no way to even find out if you are in float mode or not, let alone control which mode you are in. Can't even find a clear statement of what makes it return to bulk, from float.

My manual, which I need a microscope to read, says
"T1 = 10* T0, minimum 10mins, maximum 8hrs"

I read that as "absorb time is 10 times bulk time", which is insane. Where do you get that they are equal (which is sane).

55.5V absorb and 53.5v float is my vote. It's what am setting with as my initial starting point numbers for float and absorb on my charge controllers and (400AH calb grey cells)

The contactors on my BCU pull the pin at 4v pr cell high and 2.5v pr cell on the low side.

Hope it all works out in the end.

Kurt

Thanks for that. I will weigh up those numbers with whatever others may suggest. and with what I know already.

[offgridQLD]

" Where do you get that they are equal (which is sane)."

Well I just took the liberty to assume the (*) wasn't referring to multiply and it was the more commonly used in basic charge controllers..... time to reach absorb = time at absorb with some min / max limits as listed.

Like you say how it reads in the pocket size manual is crazy.

Kurt

[T1 Terry]

From what I remember, the absorption time is a minimum ?? mins, I thought it was 20 mins but it could be 10 mins in your version, absorption time was equal to the time charge begins till absorption voltage is reached, absorption voltage and bulk voltage are the same. I set absorption voltage at 3.45v per cell and cut the solar/240vac if a cell hit 3.6v for X mins, depends on the load as to how long it takes to drag the high cell back to 3.3v, this is the high voltage for a full cell under load, then resume charging and reset the timer circuit. Each time the charge stops and then resumes the clock is reset for the absorption mode, so it progressively gets shorter till the float stage is reached before a cell reaches 3.6v. The whole game starts again the next time the sun comes up. Float voltage of 3.4v or 3.35v per cell is sufficient to hold the battery close enough to fully charged for the remainder of the solar day.
I know it sounds far too simple, but it does work

T1 Terry

[weber]

The cavalry arrived a few hours ago, in the form of Coulomb, bearing many headlight bulbs in cups with clip leads and a dual lab power supply. Bless his heart. He has the new battery close to balanced now.

Thanks heaps for those numbers, Terry. Can you tell me what the maximum charge current is, compared to battery capacity, in systems where you're using those voltages?

I note Kurt's numbers agree fairly well with yours, at 3.47 V absorb and 3.34 V float.

Terry, I was going to ask you what voltage your cell balancers turn on at, but I seem to remember you don't use them. Is that right? And yet you must have individual cell voltage monitoring since you say you shut off charge when any cell goes over 3.6 V. Is that right? What do you do about any cell going undervoltage? What voltage setting do you use for that?

[weber]

Kurt, what voltage do your balancers turn on at?

When T1 Terry confirmed that float_time = Min(Max(10min,absorb_time),8h)
Coulomb suggested the apparent "10*" might actually be "1.0*" but you'd need an electron microscope to see the decimal point.

[offgridQLD]

They start to shunt from 3.6v (up to a max of two amps) though I am treating balancing as one would a EQ charge in the old lead acid days and scheduling that periodically (yet to be determined time intaval days-weeks and for how long each time will be determined After a few trial runs ( just long enough to bring the cells back in line)

Yes I know not as easy with the PIP as it doesn't have a EQ feature.

Kurt

[T1 Terry]

If/when we encounter a cell constantly going high, we charge the lowest cell to the same 3.6v. We are trialling an automated system but it takes mths to collect any data as these conservative charging rates seem to over come the cell run away issue. Charging current can be 1 amp to 100 amps, just depends on the weather conditions at the time. we monitor 4 cells in each group and control the charging for that group, 4 cells in series take mths if not yrs to go out of balance by more than 0.05v, we ignore any thing smaller than that as it's never the same cell in a group that goes high much under that so 0.05v became the "take notice" point.
I have chosen to do mostly 12v systems as a lot of my interest is in the RV game, some 24v systems for those with 24v charging systems in their vehicles.
we just use the trusty cell loggers for cell monitoring, a number of companies producing them now, 2.8v is absolute min cell voltage, but we use battery SOC monitoring to activate alarms and secondary charging systems if available at preset depth of discharge, we have even successfully set up a system that controls the 240v charging between set SOC levels, eliminating the need to cell voltage control the mains charger, but we still run the cell over voltage cut off as a secondary back up system, belt and braces logic, but that's what I worked with throughout my original trade so it stuck.

T1 Terry

[offgridQLD]

What I find interesting is how out of balance your cells (measure) is usually in proportion to how how much you push the upper and lower limits of SOC.

Is it not the case that the big flat area of SOC in between the two knees upper & lower can mask imbalance thats present and perhaps growing?

That said, Why are imbalance discrepancies the evil villain? My thinking is as the end result would be ever diminishing usable capacity scope as each end of the usable upper and lower SOC range within the cutoff voltage limits narrowed.

I guess this only becomes a issue if the discrepancy's start to creep into and become measurable within the range that you are cycling the battery.Though perhaps by the time they do things could be very out of whack. Think one cell 65% soc the other 40%...? as your using voltage and its a very flat line.

4 cell 12v packs i guess are more forgiving with balance then a 16 cell 48v bank.

My 16 x 400ah Calbs will be online this long weekend so will see how it pans out.

Kurt

[weber]

Thanks for all that, Terry. At present we're going with your 3.45 Vpc (55.2 V) for absorb and 3.375 Vpc (54.0 V) for float. All praise and no blame.

We're setting our 0.8 A balancers to turn on at 3.325 V. Disconnecting charge sources if any cell goes over 3.625 V. Back on again when all cells drop below 3.375 V. System shut-down if any cell goes over 3.775 V.

Still working out low voltage settings.

Charging current can be 1 amp to 100 amps, just depends on the weather conditions at the time.

Sure. But how many amp-hours is the battery?

[weber]

What I find interesting is how out of balance your cells (measure) is usually in proportion to how how much you push the upper and lower limits of SOC.

Is it not the case that the big flat area of SOC in between the two knees upper & lower can mask imbalance thats present and perhaps growing?

Quite correct. Except that we've seen there are two flat areas (40% to 65% SoC and 78% to 90% SoC) with a ramp between them, according to figures 2 and 3 of this paper that Johny found.
http://www-personal.umich.edu/~hpeng/pu ... 3_Weng.pdf
So there is a small possibility of detecting imbalance on that 68% to 75% ramp, as well as above 95% and below 35%.

That said, Why are imbalance discrepancies the evil villain? My thinking is as the end result would be ever diminishing usable capacity scope as each end of the usable upper and lower SOC range within the cutoff voltage limits narrowed.

I guess this only becomes a issue if the discrepancy's start to creep into and become measurable within the range that you are cycling the battery.Though perhaps by the time they do things could be very out of whack. Think one cell 65% soc the other 40%...? as your using voltage and its a very flat line.

Yes. But loss of available battery capacity is not the only problem caused by imbalance. The more serious issue comes when your end-of-charge and end-of-discharge voltages are determined only by looking at the overall battery voltage. When in absorb mode at 55.2 V (3.45 Vpc) at 60 amps into 16 cells having a 1 milliohm internal resistance, if one cell had 10% more charge than the others, it would go to 3.69 V while the other 15 cells are still at 3.43 V. This will do some (minor) damage to the high cell.

4 cell 12v packs i guess are more forgiving with balance then a 16 cell 48v bank.

Absolutely right -- and something many people fail to appreciate until it's pointed out. The same 10% imbalance scenario with 4 cells charging at 13.8 V (3.45 Vpc) would result in 3 cells at 3.43 V and one cell at 3.51 V. No big deal.

Then consider 218 cells in series, as we have in MeXy. With 217 cells at 3.43 V and one, a violently-venting molten mess at 7.79 V, dissipating 467 watts. Still only averaging 3.45 V per cell, still only 10% imbalance. So with large numbers of cells, there is no way you can base end of charge on overall cell voltage. With 4 cells you can get away with it. With 16 cells it's marginal, and you must have some means of regular balancing to prevent it getting to 10% SoC imbalance.

One correction of my previous message. I plan to turn on the 0.8 A balancers at 3.375 V which is the average cell voltage in float mode.

[weber]

The installation has been postponed to Tuesday.

I made one tiny but necessary change to the BMS bootstrap loader code yesterday morning, tried to upload it, and the whole thing fell in a screaming heap. You can get some pretty subtle bugs when you use a bootstrap loader to upload its own replacement. But Coulomb found it after many hours of investigation. Had I written some code differently about 6 months ago, or had I set it to do an immediate update rather than wait for me to type a command to do so, it would not have happened.

Unfortunately this meant we didn't get to fix another known problem that would have caused contactors to chatter instead of turning off cleanly, and I didn't get to finish making the cover for the monolith, which I probably would never have got finished in any case.

[T1 Terry]

Terminal voltage is only for use on the primary control circuit, cell voltage is for the secondary cell protection circuit. as long as the cell remain reasonably within balance, the system runs on the primary control only, once it's out of balance it runs on the primary until the upper stages of charge are reached, then the secondary system takes control until a balance is again reached.
No matter how many cells there are in series, broken up into sets of 4 and each individually charge controlled, by either passing the current through the cell group or bypassing the cell group, the pack stays very close in balance.
Sure. But how many amp-hours is the battery?


Works with 100Ah 12v batteries through to 1200Ah 24v battery packs and 100Ah 48v battery packs, though the set up of the 48v pack failed due to incorrect placement of the SSR's and the voltage took them out.
Life is a never ending learning process, you know you've learnt something when you recognise a repeated mistake before it fails

T1 Terry

[weber]

Charging current can be 1 amp to 100 amps, just depends on the weather conditions at the time.

Sure. But how many amp-hours is the battery?

Works with 100Ah 12v batteries through to 1200Ah 24v battery packs and 100Ah 48v battery packs...

So you're putting 100 amps into 100 Ah batteries in some cases?

What I'm trying to find out here is what is the maximum C-rate, from a PV array, at which you have successfully used that 3.45 Vpc absorb voltage.

It matters because charging current will causes a voltage rise due to internal resistance, particularly in winter, that may prevent the battery from taking the full current available from the array, at 3.45 Vpc, at any state of charge above 40%.

The system I'm working on has max PV charging current of 60 amps and uses 180 Ah cells, so that's 0.33C. If you're successfully using that 3.45 Vpc absorb voltage through winter, with 100 A into 100 Ah cells (1C) then I needn't worry.

You're using mostly Winston, is that right?

[offgridQLD]

"broken up into sets of 4 and each individually charge controlled "

So Terry, your saying you have two separate 12v chargers on your 24v 8 cell banks?


Your secondary system being cell level motoring with balancing ? what are the voltage limits befor the secondary system kicks in?


The key point I guess is how good is the manufacturing process / consistency of the Calb/Winston cells. That's the one thing we cant change and as long as your interconnections are up to scratch . Most likely the only reason they go out of balance. How much and how long it takes is on average with our kind of assuage is the key bit of info.

There was a saying...

Can you measure it and prove it's true? yes it's true and measurable. Will it matter and effect me in the real word? probably not.

Could this be the case with cell balancing. We all use shock tactic examples of what could happen (as we know cell imbalance to be true and measurable)But the shock tactic examples of one cell at 60% and another at 40% SOC. Are these examples to extreme and would this really happen and how many years would that take with healthy cells.

Just pondering really.

Kurt

[weber]

The key point I guess is how good is the manufacturing process / consistency of the Calb/Winston cells. That's the one thing we cant change and as long as your interconnections are up to scratch . Most likely the only reason they go out of balance. How much and how long it takes is on average with our kind of assuage is the key bit of info.

I eventually figured out that was meant to be "our kind of usage" but only after I rejected "our kind of sausage".

Series interconnections can't have any effect on cell SoC balance, no matter how bad they are. In a series circuit, whatever charge goes through one part goes through all parts. SoC imbalance is purely due to differences in internal self-discharge (affected by temperature and history), and differences in the current consumption of cell monitoring units. Both of which are typically 1 to 5 milliamps.

There was a saying...

Can you measure it and prove it's true? yes it's true and measurable. Will it matter and effect me in the real word? probably not.

Could this be the case with cell balancing.

With a 16 cell or more battery? Good luck with that.

We all use shock tactic examples of what could happen (as we know cell imbalance to be true and measurable)But the shock tactic examples of one cell at 60% and another at 40% SOC. Are these examples to extreme and would this really happen and how many years would that take with healthy cells.

Who is "we" white man? My example only involved a 10% SoC imbalance, and showed that this would begin to cause damage in a 16 cell pack, although not a problem in a 4 cell pack, and utterly disastrous in a 218 cell pack.

It doesn't really matter how many years it takes. If I'm building systems with more than 4 cells, for other people, some of whom wouldn't know an amp-hour from a aardvark, I won't be around when it happens, so I have to build in balancing.

[offgridQLD]

Sorry yes I used the 20% SOC imbalance example and as you said your example was 10%. I wasn't trying to say cell balancing wasn't unnecessary ever and agree a automated balancing setup is a good move for customers system and will have the same on my system. But just questioning how much inballance we would really be dealing with when the cells are being cycled as they would in a typical offgrid system. How long would it take before say a 5% inballance was created (9ah) on a 180ah cell.

" SoC imbalance is purely due to differences in internal self-discharge (affected by temperature and history), and differences in the current consumption of cell monitoring units. Both of which are typically 1 to 5 milliamps"

When you say 1-5ma I would assume then if the cells where being charged or held at float for 12hrs you could have 5ma x 12 = 60mah a day or 420mah a week gap growing. That's about 20 weeks to see 5% inballance.

Or am I looking at that totally wrong?


I'm thinking about the option between a short balance every day and for me anyhow with fixed V ballancing units its a excursion to a higher voltage each day) or a gap between of X number of days then potentially a longer time balancing time. In the end I guess it could work out to be the same time spent at the higher voltage anyhow.

I will find out starting from tomorrow when I install my cells on the house.(wish it would stop raining) I think I will set up a dayly balance to start with end then extend it out and see how much time it takes to even things out after a day or two gap between balancing and work up or down from there.

Kurt

[weber]

Your calculation of 20 weeks for a 5% imbalance would only be true if there was a difference of 5 mA between cells. That's unlikely. If most cells have self-discharge plus CMU consumption = 5 mA and one has only 4 mA, it's only the difference of 1 mA that matters.

And you seem to be assuming that self-discharge and CMU consumption only happen on charge. But they happen 24 hours a day.

1 mA for a year is 9 Ah. And I expect you could get away without balancing for the first year with a new pack, all from the same batch. But then again, you just might be unlucky.

The difference in consumption between CMUs tends to be a lot greater than the difference in self-discharge between new cells, which is one reason why some people say you're better off without CMUs. But I'd prefer to have CMUs that balance on every full charge. Then the difference in their consumption will be irrelevant. It's not like there is anywhere else you can use the solar energy the resistors burn off, when the cells are already fully charged.

In your case, when your CMUs can only balance at 3.6 V, I think balancing once a month would be fine. But it may take a few consecutive days to get the initial balance.

[T1 Terry]

So you're putting 100 amps into 100 Ah batteries in some cases?

What I'm trying to find out here is what is the maximum C-rate, from a PV array, at which you have successfully used that 3.45 Vpc absorb voltage.

It matters because charging current will causes a voltage rise due to internal resistance, particularly in winter, that may prevent the battery from taking the full current available from the array, at 3.45 Vpc, at any state of charge above 40%.

The system I'm working on has max PV charging current of 60 amps and uses 180 Ah cells, so that's 0.33C. If you're successfully using that 3.45 Vpc absorb voltage through winter, with 100 A into 100 Ah cells (1C) then I needn't worry.

You're using mostly Winston, is that right?

Mostly Winston cells, I did fit a few systems with Sinopoly cells, but the majority are Winston cells.
The system of stopping the charging for a set length of time if a cell reaches 3.6v eliminates any issues with high CA charge rates, it sort of becomes a very slow pulse charge if the acceptance rate is not up to the charge rate. One of the things we found was the faster the charge, the better the cells remained in balance, it's the slow trickle charge at the very top end of charging that seems to cause the most run aways.

T1 Terry