The blackbox project

[Gregy]

Zoneblue

Thanks for response - good news to know that access to the BBP data will be "easy"

You started me reading up on rpi vs cubic vs beaglebone black
Wow I didn't realize there was such a difference
Certainly the last 2 are a lot more capable ... Price points similar (frankly not a consideration, there
Are other more important considerations for the task at hand)
Both have much greater IO as c/w Pi ... Hence making "on board" IO a definite possibility if using
Beaglebone or cubie
Eg beaglebone black has 7 ch 12 bit ADC (but only 1.8V max .. Noted by some as RPITA)

Sounds like you have already settled on cubie (perhaps that was a timing issue)

So now I'm rethinking my plan - seems cubie or beaglebone black are the current better solutions
For hosting a BBP

[Gregy]

DGD

Do you have a link to a post where you describe details of how you are doing your panel isolation
And switchover to heating loads (I'm not thinking of heating)
But I'm speculating this may be a better way to achieve some other things I'm contemplating

Not clear if you are switching a complete controller (assuming you may have more than one) away from main bank (I'm assuming that's not possible because the CC. Are linked)
So if that's not the case, you are switching over at the PV panel high voltage - into ? Another CC?
Or somehow direct across the load ? (But then no MPPT operating )

Just curious ..... DC switching at 150V or higher .. That can get exciting I suspect - but if there is minimal load I guess it's not as bad

[Westbranch]

FYI, I came across the short video part way down this page, there may be some info that applies here

http://www.adafruit.com/products/1278

[RossW]

Just popped in for a coffee and thought I'd check on reference voltages and check more carefully on the later 08M2 parts analog.

Seems the picaxe now has access to an internal (quote) "Accurate 2.048 volt reference" for the ADC.
Using 0V/2.048V internal reference gives us immediately a 2mV precision.
If I use the opamps to reduce cell drain and purely as a level-shifter, I can get away with a dual rather than quad opamp (smaller, cheaper, even lower power) and maintain a 2V measuring span for each cell.
My only concern is how I accurately level-shift without a reference....

*ding*  Blinding flash of light....  use the "top" of the previous cell! Ie, differential amps. Each one references its output on the negative of the cell under test!!  0-2V (nominal) out. I will still need a small bias or divider to be able to measure over 2.048V (really need to get to about 2.5V). Simple 4:3 divider with 1% resistors still leaves me around 3mV/bit resolution. If we're aiming for 10mV or better, we're comfortably in!

As for electronics in the battery box: they could be potted.

As for calibration: it could be done digitally and stored in the pics NV scratch. (I currently do this with another device I make for light aircraft)

[Gregy]

Ross
I think we need to aim for per cell measurement of at least 2.7V (for the 2 V cell scenario)

for flooded lead acid cells eg surette rolls that they recommend
A EQ charge in range "61.92-64.08" (48V string)
Hence if all cells equal during  EQ,  This could already give 2.67Vpc
(Not sure if this temp compensates up or down)

And I'm thinking that during EQ is exactly one of the very useful times to have accurate individual
Cell measuring (To supplement SG testing of course)

Not sure about other battery chemistries .. Is the flooded lead acid battery our "worst case"? Design scenario
If we level shift to measure against the 2.048Vref. We can discard the lower end
Eg if cell less than 1.0V it's got much bigger issues, hence we can get 2V measuring range centred around 2V
Eg  1-3 Volt measuring range

[dgd]

Ross
I think we need to aim for per cell measurement of at least 2.7V (for the 2 V cell scenario)

Now it would be nice if the per cell measurement could range high enough to cover 3.2v Lithium Iron cells   

dgd

[dgd]

DGD

Do you have a link to a post where you describe details of how you are doing your panel isolation
And switchover to heating loads (I'm not thinking of heating)
But I'm speculating this may be a better way to achieve some other things I'm contemplating

Not clear if you are switching a complete controller (assuming you may have more than one) away from main bank (I'm assuming that's not possible because the CC. Are linked)
So if that's not the case, you are switching over at the PV panel high voltage - into ? Another CC?
Or somehow direct across the load ? (But then no MPPT operating )

Just curious ..... DC switching at 150V or higher .. That can get exciting I suspect - but if there is minimal load I guess it's not as bad

I tried switching out two strings of panels (from 4) but it made no difference to the Classic. Its a 150 with 2.8Kw PV in 4strings by 5 panels.
24v bank, so too much PV input but designed to give some useful power input in non ideal conditions.
I just insert the heating element on the output of two strings and protect the Classic with a diode to stop risk of Classic feeding the heater.
Auctually use a 3phase rectifier so get 3 diodes. Also have two 50,000uf  100vcaps on pv output- a leftover from when I used classic pwm wastenot as relay control. A 20A 200v ssr inserts the heating element in circuit enabled when PV in rises about 7v over the mppv of the arrays.
I have not seen the Classic drop out of float (yet!) when the heater ssr enables. its only a 800w heater so the two atrings, 1400W, seem to easily cope
When experimenting with this setup I was concerned about the Classic and its possible reaction to seeing PV voltage pulled down as heater came online. I used the AUX1 float output on another ssr to disconnect the PV bank from the Classic. In normal mode the float high is used to keep the ssr closed and current flowing from PV strings to Classic. When float V reached this AUX1 goes low disabling the ssr and
disconnecting the two PV strings from the Classic. Shortly after as the PV voltage of the disconnected strings rises fairly quickly the ssr to the heater activates.
I gave up on the ssr disconnectig scheme and for simplicity just went with the one ssr inserting the heater. The Classic's mppt logic is smart enough to always select the higher voltage input from the two strings not connected to the water heater, as least so far when I have seen it scanning
This is why I am interested in a potential black box that could more efficiently use the available power. I'm simply now inserting an 800w parasitic load on the output of a couple of PV strings. The real situation is that when in float on a good day there must be over 2Kw available so what the heater does not use is wasted.

dgd

[RossW]

OK. Severe case of "Feature-Creep" here.
This was going to use an 08M2, the tiny little 8-pin device.
I'm thinking I might have to go up to the 14M2.

The chip as an ADC output. It's not very precise, but it's reasonably accurate.
It has 32 steps.
I wonder if I could make the differential amp a summing differential amp, and add my chosen offset (from the ADC output). This would let the PIC ramp the reference voltage up or down to monitor a 2V span from 0-2.048V up to 2.048-4.096V (theoretically), with the 2mV resolution.

Or, do I simply set the differential amp gain at 0.5, thus giving me 0-4.096V with 4mV resolution?

Do we feel the need to measure the temperature, and if so, do we need one sensor per cell? Per group?

When do we stop? Do I add three chunky FETs so we can try to automatically balance the string?

Do we try to crawl first, then see what else we can do?

[Gregy]

Feature creep always an issue
It's one of the challenges to determine how many different scenarios to cover
Aside from 3? Or so lead acid battery chemistries .. Then there is LIXx ..... And that covers
A few more permutations

Ross - there is now a separate thread for battery ADC
Zone blue has started a couple new threads - which well help us better organize

.. I've already wrong posted today also !

[Gregy]

I agree - crawling first is good
Just need to define "crawling" .....

IMHO
Keep temp sensing separate - it doesn't have the common mode isolation issues,
It's simple and already quite easy to do, and add x number sensors as users requires
Eg daisy chain 1wire DS18xx sensors along the cells cases

[Gregy]

DGD
Thanks for detailed explanation - I can see the "limitations" in your current setup (no criticism intended) .. I'm impressed by your creativity ... And the approach to the "high voltage switching "
Challenges ... 150 VDC and big currents ... Not for the faint hearted (or inexperienced)

I see why/how the BBP. Could benefit your scenario
- once we have more intelligence available to us/BPP on opportunity energy .. Then we can
Chase the holy grail ... Use every drop available!!

[Gregy]

Thinking aloud .... yes I know I have too much thinking time .... It's only a temporary aberration)

The BBP "vision" statement could include;
chasing the PV RE holy grail of "use every available watt of available energy at all times"

[Gregy]

 LiOn/LiPo etc requires voltage sensing up to 4.2Volt (the max allowable for these type of cells)
I sense ross has one foot already over the balcony rail

[Westbranch]

Ok guys, you are getting way ahead of a lot of us interested bystanders..fast, real fast.... 

so is it possible to start new BBP sub topics for a flow chart?

and (not sure of the right term here)

a schematic of all the different hardware that is being proposed

ie, rPi, BBL, Arduino, Cubie, etc et al.?

These in addition to the "Vison Statement"

[Gregy]

Not sure about "flow chart"
But have a look at my thread "BPP sensing modules"
My attempt at some "structure" (at least for my thinking)