I have 8 cells in series and I want to check them, whether all are connected. I came up with the following schematic:


simulate this circuit – Schematic created using CircuitLab

My question is about R and how viable this solution is.

My reasons for calculating R is the following: I saw that a CMOS gate leaks 1uA. While I don't want to waste energy, I want to be as closely as possible. Not too close, due to noise. So I came up with 470k. Is this too high? Does this work? (Sorry, I don't have 2V Zeners at hand, so I can't check it myself)

Let's say I have a RC element (R=100, C=0.1uF) between the cell and R. Does this help and make those 470k a safer solution?

Do you see any other problems?

  • \$\begingroup\$ To avoid an AB-problem, why do you want to see if all are connected? Especially with them in series like that, one way to see if they're all connected is if your load sees any power at all! \$\endgroup\$ – Orotavia Apr 29 at 1:24
  • \$\begingroup\$ I want to make sure the cable from the cell to R is ok. They are used for measuring the cell voltages. Long story short: Each cell has an IC for checking the voltage, but it's powered by each cell. So there is no way to see whether the IC is working and the cell is ok, or the IC has no power, due to broken cable \$\endgroup\$ – duedl0r Apr 29 at 1:29
  • \$\begingroup\$ Perhaps some kind of differential voltage measurement across each cell to to check if cell is good or not good. \$\endgroup\$ – Leoman12 Apr 29 at 4:16

Zeners, particularly low voltage ones, leak far more than the CMOS input (1uA is a very loose specification at high temperature). For example, 100uA at 1V for a 2.2V zener.

So I suggest you lose the Zeners and use a different kind of clamp. For example, a diode clamp to a 2.5V shunt reference for each input. You only need one 2.5V shunt reference, but each input needs the series resistor and diode.

The other problem is that if the input is left open (not grounded), there is no reason for it to go to logic low, so you really need a resistor to ground on each input. You could use 470K resistors to ground and varying values for different series resistors, but the first one would call for a relatively low value of resistance, perhaps 100K.


Clamping scheme:


simulate this circuit – Schematic created using CircuitLab

(not shown the Vref to cathode connection on the TL431 to set the voltage to 2.495V nominal).

The TL431 can clamp significant current (up to 100mA total). Input voltage will be one diode drop plus 2.5V.

Using the supply voltage runs the risk of causing it to rise out of regulation as most regulators cannot sink current and also you would have to use Schottky diodes, which are relatively leaky compared to signal diodes such as the 1N4148/LL4148.

| improve this answer | |
  • \$\begingroup\$ All the inputs are connected. If there aren't as many cells as the board, all the open ones just get connected to the highest cell. So we can safely assume, all is connected. If not, the output should be logic high. \$\endgroup\$ – duedl0r Apr 29 at 10:33
  • \$\begingroup\$ Could you elaborate on the diode clamping with a 2.5V shunt reference? Could you describe the schematic in a bit more detail? Is it something like the LM431? I looked at this one earlier and thought it is almost the same as a zener? \$\endgroup\$ – duedl0r Apr 29 at 10:36
  • \$\begingroup\$ Btw I have a 3.3V supply on the board for the NAND gate, maybe we can use this. \$\endgroup\$ – duedl0r Apr 29 at 10:37
  • \$\begingroup\$ Thanks for the detailed explanation. This looks promising! \$\endgroup\$ – duedl0r Apr 30 at 0:31

Save yourself all the trouble and go with a chip designed to do the job like the LTC6813: -

enter image description here

The zener idea is flawed and realistically, you need to measure each one to ensure they are balanced so, just go for the chip solution (or a variant of it).

| improve this answer | |
  • \$\begingroup\$ Thanks for your input. I know those chips. It's not that trivial to build something with those.. :) \$\endgroup\$ – duedl0r Apr 29 at 10:38
  • \$\begingroup\$ Well, it comes down to whether you want something that works reliably and is proven or do you want something that doesn't work (the zener diode and NAND gate idea in your question). \$\endgroup\$ – Andy aka Apr 29 at 10:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.