Need help reviewing design

[nesse]

I'm designing a battery monitoring system for large strings of serially connected lithium cells.
It will continuously measure individual cell voltages by connecting one cell at a time through opto-couplers, AQS221.
I would be very interested in hearing any thought about this design from you brilliant guys, will this work or will it explode

 

[kubeek]

The schematic si unnecesarily large so it is very hard to read. Fist thing is that you could loose all those or gates and use diodes instead.
Second, I am not sure what kind of optocouplers you are planning to use, but you would probably need something that has no Vf like transitor optocoupler would have. Also the opto needs to be able to withstand the full battery voltage on the off transistor.

 

[nesse]

Thanks for your reply kubeek
20pcs of quad OR gates are cheaper, easier to mount and uses less board space than 160 diods
The optos are AQS221 and they will not see the full pack voltage. The idea behind this design is that only one pair of optos are open at a time, connecting one cell to the OP/AD part.
The 4051 are a safety, making sure only one of the 4514's are enabled.

 

[kubeek]

The optos are AQS221 and they will not see the full pack voltage.

I don´t think that is true, say you turn on one opto at the start of the pack, so the sensing rail is close to ground. The opto at the far end of the pack will see the whole pack voltage across the switch - between the terminal of the cell and the sensing rail.
You need to find better optos, or divide the circuit into banks so that the voltage across one bank isnt so high.

Also, I didn´t look much through the sensing circuit, but it seems rather odd. Could you somehow redraw that circuit into sections? Also, what does the -HV terminal mean, is it connected to the low side of the battery? Better would be to just call it ground and show the battery as floating.

 

[kubeek]

Also, what is the purpose of the top half of the U28 opamp?

 

[nesse]

You are correct, I need to use something like the CPC1025 instead of the AQS221.
Every other cell measured will reverse polarity so I need to rectify the cell voltage.
The top half of the OP with the diode forms an ideal diode.
For a positive-going input signal, the bottom opamp can only function as a unity gain buffer, since both inputs are driven positive.
For a negative-going input signal, The ideal diode prevents the non-inverting input from being pulled below zero volts. The bottom opamp will then functions as a unity gain inverting buffer, with the inverting input maintained at zero volts by the feedback loop.
Input impedance is non-linear, having an almost infinite impedance for positive half-cycles, and some 50K input impedance for negative half-cycles, so I need U71 to drive it.

 

[kubeek]

I think I would use a standard differential amplifier plus a precision fullwave rectifier, or feed it straight into the ADC and correct in software.

 

[ronsimpson]

You have 80 batteries?
The OR gates are clever. It looks like you are passing the + and - voltage for the selected battery to a ADC.
At first I thought you have a differential amplifier but now I see the ADC is isolated and climes up the ladder of batteries.

I would use a PLD to (combine the 4514 and the OR gates) into one device. It should be an easy design.

I don't understand the amplifiers in front of the ADC. I think you can remove the amplifiers.

Where is the +5HV supply coming from?

 

[nesse]

Yes, up to 80 cells. The +5HV and its ground -HV are from an isolated SMPS
PLD is a very good idea, but I have no experience with it, not sense the old PAL circuits in the early 90:th :-o
I wouldn't know where to begin :-(

 

[ronsimpson]

on the logic side; I also thought about using a long shift register and loading it with 01100000 and shifting the 11 around.
Care must be taken not to have too many switches on!!! Maybe there is a SR with tristate outputs that can be enabled (tri-state) until good data is shifted in.

 

[nesse]

I think I will replace the multiplexers and OR's with a ATmeg640, it has 86 GPIO and is only around $5.
That will simplify allot and thus reducing cost

 

[ronsimpson]

Your switches are isolated so I don't see the need to also isolate the ADC. The micro can measure the voltage with out the external ADC.

 

[ChrisP58]

The topmost cell could be ~340 Volts above ground, so the ADC does need to be isolated.

Neese, it looks like R10 is inline with the ADC input. I would change that single resistor to one at each cell tap point. That way, if something goes wrong and more than two adjacent switches get closed, the resistors will keep things from blowing up.

 

[canadaelk]

Asides from finding the circuit far too complex, I would really like to know of the charging scheme of such a battery arrangement. E