Problem: Current source has spikes at the output with capacitive load

[elektronchika]

Dear colleagues,

Recently I designed a circuit what should generate bipolar, constant current (programmable) pulses that should be DC free. I attach the circuit with name "current-source".

The load is attached at CN600 and for test purposes I use 1kohm resistor. C600 and C601 are used to ensure DC free pulses.

So far the circuit is workings. But because of the capacitors in series with the output, there is a spike at the first pulse (attachments "waveform-uncompensated" and "waveform-uncompensated-spike". I added a RC snubber in parallel with R603. Experimentally I have better results with C = 10 ohms and R = 1nF. The waveform is named "waveform-compensated". But this leads to slower second and third slopes and still the first slope has a big spike. Of course when I simulate the circuit everything looks better.

Could you give me some guidance how to solve the problem. I tried to compensate the feedback of IC604 but without any better results.

Best regards,
Lazar

[StillTrying]

24 hours, 48 reads and no replies.

there is a spike at the first pulse

My first thoughts would be that C603 opamp saturates the constant current generator's B-E while trying to produce a current into an open circuit. So when the switch first closes the set current is too high for a few uS, while it's base charge is removed.
A 27V zenner between it's collector and the +30V could allow it to be at or near it's set current before the switches close and divert the current through the load.
Or leave all 4 switches closed and just open the appropriate 2 to allow the already set current to go through the load.
Or something else..

[David Hess]

I am sure you identified the problem; the current sink is out of compliance before the first pulse and the operational amplifier is trying to source the entire current through the base-emitter junction.  Fixing it may just involve clamping the collector voltage.

[Powermax]

That's an interesting design, I am not really sure why you would be seeing those spikes like that, as it only occurs when switching to the negative half cycle pulling current, the schematic shows that the H bridge stuff looks pretty symmetrical.

Also, it appears that the high-side differential amplifier (high side current sense amp) is redundant, as you could just use the effective 500 ohm resistor between the ground and the emitter of T600 (and also, transformers are normally labeled with 'T', while transistors are labeled with 'Q'. I'd call it Q100 or something.).

My guess as to the problem is software. It appears you are using a microcontroller for the switching of the 4 H bridge devices, and it may be a software issue causing shoot through when switching, say, A and C on, but leaving the load temporarily disconnected from the current source when switching to say, B and D.

[David Hess]

Try changing the drive to the switches so that all 4 switches start closed.  This will pull the output of the current source to 30 volts and keep it out of saturation before the switching starts.

I am assuming that the transistor can handle the power dissipation.