Build 1760 2-10-2014

[boB]

Zone,   I think you're really going to like the next rendition...

Averaging is about smooth as melted  butter now.

boB

[zoneblue]

Hi Bob,

You are a miracle worker! Playing with my dataset it looks quite tricky to do.

1. The first graph is for today. You can see at about 8.30am the noise goes up by an a order of 10. I figure thats probably caused by our sound system.

2. the second graph is an early morning detail. It shows, absent load noise, a pretty clean WB signal in general.

3. in the third graph i try out different moving averges. 5-10s seems about 'optimum' for this data.

4. the final graph is the first graph put through a 10s moving average filter. No negative loads, and the EA is more discernable but not perfect. I really dont know how you could get it bang on.

Today WB EA occured at about 2.5A, when set to 5A. 

[boB]

Well, I'm glad it's working OK but this one is SO much better !  Revamped the whole averaging and
rounding code.

The MNGP code is still build 1760 from 2-10-2014.

Classic is  1769  2-13-2014.  You will probably only need to update the Classic itself.

http://fusion.midnitesolar.com/MidNiteSolarUpdate_1769_2-13-2014_v4.22.exe

The main WB Jr. current is of course rounded to +- 0.1 amp at modbus address 4370.
(register 4371)     Looks like this is what you have been reading all along though.

boB

[zoneblue]

its around 9am here, 1769 is in, will see how EA works today.

Yes, using reg 4371. Do you have other registers holding the averaged data, how does this all work now?  If you have anything scrawlled on the back of an envelope for the new WBJr/SOC/etc registers, thatd be great.

Re my "Iload" datapoint, did you manage to get the Iwb and Icc figures averaged in the same time span? If not, ill have to rely on my proposed post process filtering i guess. (6s moving avg, then resampled at 60s intervals).

[boB]

its around 9am here, 1769 is in, will see how EA works today.

Yes, using reg 4371. Do you have other registers holding the averaged data, how does this all work now?  If you have anything scrawlled on the back of an envelope for the new WBJr/SOC/etc registers, thatd be great.

Re my "Iload" datapoint, did you manage to get the Iwb and Icc figures averaged in the same time span? If not, ill have to rely on my proposed post process filtering i guess. (6s moving avg, then resampled at 60s intervals).

No, I did not adjust any time constants for the Classic amps and the WB amps.  Maybe some time...

Now, you may be interested in a couple of other registers regarding this.

For instance,  Register  4391  (address 4390)  will give you averaged but Pre-rounded WB Jr. amps that
show 10s of milliamps.  May be accurate to 50 mA or better.

Pre-averaged non-rounded WB Jr. amps is in register 4390  (address 4389), also at 10s of milliamps resolution.

As for Classic amps, if you want "raw" unprocessed amps,  register 4272  (Ibatt) holds this value BUT it is offest by around 4.0 amps (value = 40) that is subtracted out before being used.  That offset is not available on the list of modbus registers though.  It's there if you want to use it for time constant verification though.

boB

[zoneblue]

Bob,

4371 certainly is now a LOT smoother. One thing is strange though. In places it reads higher than 4117. See graph.

Will post float transition in a minute.

[zoneblue]

Float transition was only 8 minutes off.

[zoneblue]

I just went and measured the  house load current (fridged cycled off) with a clamp meter (4.4A). Thats the background load here all morning. So there should be a 4.4A gap between the two curves... one of those readings is off by about 2 amps.

[boB]

I just went and measured the  house load current (fridged cycled off) with a clamp meter (4.4A). Thats the background load here all morning. So there should be a 4.4A gap between the two curves... one of those readings is off by about 2 amps.

Not sure what you are measuring exactly but reg. 4117 is the classic's shunt which will read different than the 4371 WB Jr. shunt.  I would not expect them to always read the same.

For Ending Amps, you should be using the WB Jr.   Of course, Ending Amps may be a short lived way of ending Absorb because of new light coming to the world of charging strategies and battery life awareness... (Do I sound like a politician or visionary now ?)

Using E.A. only all of the time it seems will possibly make batteries sulfate if a nice long absorb time or EQ is not done every week or two.

I would trust the WB Jr. current measurement into/out of the battery.

Please keep us posted on your findings.   This is good stuff !

boB

[zoneblue]

THe way im thinking is this. If there is no load on the battery at all then 4117 and 4371 should agree 100%, right?. However with loads: Iload=Icc-Iwb, ie the difference between the two curves. Iwb can never be greater then Icc, at least not in the same polarity. So something amiss. Will check my math again to make sure.

BTW this only started with this last firmware.

[boB]

THe way im thinking is this. If there is no load on the battery at all then 4117 and 4371 should agree 100%, right?. However with loads: Iload=Icc-Iwb, ie the difference between the two curves. Iwb can never be greater then Icc, at least not in the same polarity. So something amiss. Will check my math again to make sure.

BTW this only started with this last firmware.

Those two numbers should agree except for the idle current of the Classic.  When the PV is live and it
is charging, the idle is powered by the PV side and both of those numbers ~Should~ agree very closely.

But even with the idle, they shouldn't be "amps" apart in reading.  Did you say you say something
like fourteen (14.0) amps of difference ?  That's gotta be a math problem ?  This just does not
make "sense".

So, from the main status menu, hold down the left arrow key and tap the enter key.  When the
Classic is Resting, those current + and -  numbers should be sitting at around 40 which means
4.0 amps of offset from zero.  If there is some kind of problem with the current sense, those
numbers will be quite different that 40.  35 to 45 is normal.  But that is offset.  If there is
a gain problem, that's different.  But just a change of software shouldn't change that.
I will look closer on this end and have others look to make sure things are OK.

I would really like to know what's going on with your system.  Which one is wrong ?

And, DGD, you wired your WB. Jr. backwards and had to re-wire it ?  There IS a bit that reverses the
polarity...  Not a menu item though but a modbus enable bit (a Murphy bit).

boB

[zoneblue]

No it was only a couple amps i couldnt account for on feb 14th. Problem is i havent been able to reproduce it to that extent. Today (Sunny day) it looks perfectly reasonable, see pic 3 below..

Re idle/tare. Ahh, yes, i had forgotten about that, was sort of cognisant that the the classic amps didnt go negative at night. But... that would explain iwb being higher than icc at rest, by a predictable enough 0.1-02A.

But during bulk/absorb?. During my testing here, i have a couple times during the day shut down all loads except the blackbox. When i do that i get Iwb being 0.1 - 0.2A higher than Icc.
This may just be an accuracy issue on Icc that was previously undetectable inside the prev WB noise floor. Or might be an artifact of the new averaging filter?

It seems to me that even if the the classic was drawing tare from the bank but not accounting for it in Icc, the net result would be a higher Icc, not a lower Icc as is experienced here.  Its impossible to get more amps at Iwb than Icc. Thats free energy type stuff.

Sample raw data pic 1, another instance: pic 2. Both in bulk.  0.1-0.4A out. Given that blackbox draws 0.1A, inverter tare 0.4A, and sundry dc converter tare loads, computers in standby etc puts my absolute min load at around 0.7A, also not accounted for.

It would be nice if there was a good relationship betw Icc and Iwb, becasue the Iload datapoint is an incredibly useful one,  probably number 2 in terms of importance, after Vbat/SOC. It takes the reliance on inverter data completely out of the equation and avoids resorting to another shunt, and asociated ADCs. However its not super critical, we can probably live with it. Work up some fudge factors etc.

In terms of end amps, its all looking pretty darn tight to me. People need to remember that EA waits 90s for the peak noise layer to clear EA, thus if the noise layer is about 0.5A, then set your EA 0.5A higher than your desired EA. By 4pm charge acceptance here was down to 0.3A (AGM), and setting EA as low as 1A, looks like it will work with this new firmware. Nice job.

[boB]

OK, zone.  Looks good.

A couple of tenths of an amp difference is probably OK...   It's when you see  13.1A and  15.6A  that I do not understand.

At night, or when Resting, the Classic will show up as -0.2 or  -0.3 amps on the WB Jr..  And can be higher IF you catch
the Classic running, but sitting at Voc as it may do from time to time when either starting up or recovering from a sweep
or something like that.  If the Classic's input is sitting at a high voltage, say, +130 volts and the relay is on, it can draw
quite a few watts but what I think I am seeing in your tables is the WB Jr. showing higher charge current than the
Classic Ibatt 4117 register...  A out 0.2 amps more in a lot of spots.  That's OK I think.  The Classic could have probably
stood to be calibrated somewhat  better maybe.

The Classic's current sense can be very accurate... But from what I have been seeing from WB Juniors, I would tend to
believe the WB Jr. over a calibrateable Classic current sense.  The WB Jr. doesn't have any adjustments.  It should be
repeatable without any adjustment at all and I am very happy with what it reports back as far as accuracy.
This, even over a large temperature swing.  It's just not very fast at sampling current.

boB

[zoneblue]

A couple of tenths of an amp difference is probably OK...   It's when you see  13.1A and  15.6A  that I do not understand.

From the first table at 8:43am? Isnt when Icc is higher than Iwb ok, becasue some of Icc is going to load? I would have thought its the reverse where you end up with a situation where something doesnt add up. Like say the two previous minutes, at 8:41 and 8:42am where Iwb is a full half amp higher than Icc. How can more come into the batterys than is being generated? And like i said its not just a half amp, because there are also loads running on top of that. The discrepency could be as high as 1.5 or even 2amps. Hard to say without better testing. I might be not understanding what you are saying about the tare, as i want to say this might account for it, but not sure.

Sure makes my head hurt thinking about negative negatives

At night, or when Resting, the Classic will show up as -0.2 or  -0.3 amps on the WB Jr..  And can be higher IF you catch the Classic running, but sitting at Voc as it may do from time to time when either starting up or recovering from a sweep or something like that.  If the Classic's input is sitting at a high voltage, say, +130 volts and the relay is on, it can draw quite a few watts

I think im following you.  Kinda. If the classic did decide to draw some tare off the battery during bulk, but not account for it in Icc (4117),  what happens...   Iwb reduces, right?

It seems to show up in real changey sun conditions.

[Cniemand]

boB,

Do you have another link to the latest firmware 1769 or up. I'm running 1758. Wanted to update to see a smoothing of EA.

The link posted a few replies up brings back a 404.

Thanks.

Cloud