Need help designing a circuit to adjust charge level of a cap bank

[terramir]

I got this circuit to precharge my LiFEPO4's and well it works but it's a bit dirty in it's cut-off values maybe because the circuit is not ver insulated etc.
https://www.electro-tech-online.com...ging-circuit-will-this-work.92969/#post740212
The circuit I need to is for a capacitive spotwelder, I need to adjust it so it will charge to various voltage levels (11,11.5,12........15V).
The bank is 2 caps 16V with a total capacitance of 1.22F-1.58F exact data is not availible on that point the value was extracted experimentally, 1RC method gave me the high value discharge over a known resistor gave me the low value. it's somewhere inbetween. Now I have several possiblities when it comes to charging this cap the most likely method in the long run is some sort of battery that will have a voltage of around either 16V nominal or 19.2V nominal the output for most of the discharge curve would be either 18 or 15 respecitively.
So I'd like to have a circuit that does not limit the current but turns the current abruptly off when the cap reaches a pre-selected voltage level (which also determines the availible Ws (joule) for the weld. Hysteresis should be kept to a minimum (0.1V) a relay will disrupt the current flow when a weld is in progress.
I'm asking the following two questions
A. can part of this circuit be reused to cut-off the current flow when the pre-determined voltage has been reached?
B. would using something like a LM 339 be better in this situation?
terramir

 

[user_88]

You might consider an integrative approach. The idea is to use a low value shunt resistor to generate a voltage that is proportional to the current going into the target capacitor.
Then use the current proportional voltage as the input to an op-amp integrator.
Possible details might include constructing a reasonable precise integrator .... getting it to function properly. Choosing the correct capacitor for the op-amp feedback ... electrolytic, non-polar ....
You should end up with a circuit that would achieve reproducible numbers relating to the current*time quantity .... essentially the total charge .... that would pass into the capacitors. The number might not correct in an absolute sense, but it would be the same for a given setting, and would serve the purpose.

 

[terramir]

:S english plox, I'm a beginner LOL I know enough about Li-ion chemistries and well learning about capacitive welding transistors are already places were I make errors but what you wrote I understood 0
p[lox translate
terramir

 

[user_88]

This is not an extremely complicated project .... I suggest starting a new forum thread .... maybe something about op-amp integrator .... Then set up an op-amp comparator to switch off the current at a desired set point, using the op-amp magnitude as one input, and the set point as the other.

This approach might get you some suggestions for specific parts .... Also, it might be a good idea to have a breadboard, so that you can build the actual circuit and check it out ... find any problems.

 

[crutschow]

Don't see any reason to go to the complexity of an integrator. The capacitor voltage should be a good enough indication of the amount of charge. Just use an LM339 comparator, as suggested, with a switch to select the desired trip points from a resistive voltage divider string, to shut off the charge at the desired capacitor voltage. The LM339 could drive a relay in series with the charge circuit. You want to add a small amount of hysteresis to the LM339 trip point to avoid chattering of the relay.

 

[terramir]

how about using a mosfet to switch the thing instead of a relay, as I understand it a mosfet when it's not (rapidly) switching doesn't have all that much in losses going on. I'm already using a relay for the fireing of the weld circuit, that's already one part that will need to be replaced every so often I'd like to keep the rest solid state.

One question I'd like to ask how exactly does a 339 work? I put a referance voltage on one pin, and the second pin is the measured voltage right? So what does the other pin output? I read something about the supply voltage determining if the output pin pulls high or low?
terramir

 

[crutschow]

You certainly can use a MOSFET of the correct size and voltage in place of a relay. If you need isolation of the control voltage from the relay output, you could use a solid-state relay.

On the 339 you put a reference voltage on one pin and the voltage to detect on the other pin. It "compares" the two voltages and gives an output accordingly. When the (+) input input is higher the (-) input pin, the output goes high, otherwise the output is low.

The output is an open collector NPN transistor connected to negative supply (typically ground) so you need to add an output resistor connected the the plus supply voltage (it can sink 6mA maximum). Read the data sheet to learn more https://www.electro-tech-online.com/custompdfs/2009/06/LM339.pdf.