Other Parts Discussed in Thread: TPSM5601R5
Tool/software:
I am having trouble with a REF2033 ringing badly in my application.
I am using a shunt and instrumentation amplifier to measure current in the high side FET of a soft starting 600v DC switch (think buck converter, but with 100% duty cycle in normal operation).
The output of the in-amp goes to a comparator, and the comparator threshold is set using a microcontroller DAC output (heavily filtered). The in-amp output is also filtered and fed into the micro ADC input to measure current once steady state is reached.
The comparator drives some logic that latches the FET off when current passes the overcurrent threshold.
The REF2033 generates a 1.65v reference for the in-amp reference input and a 3.3v reference for the micro DAC and ADC, as shown in the REF2033 datasheet - with one exception, the shunt, amplifier, reference, comparator and microcontroller are all floating at line potential. The 0v of all of this circuitry moves together with the source of the FET when the FET switches on or off. The circuitry is on a 6 layer board with good ground planes, lots of decoupling capacitors, and I've worked to minimize coupling between the PCB and earth.
The board main supply is +/- 12v from a monolithic isolated converter. The +/- 12v (with filtering) feeds the in-amp supply rails. A 5V LDO (TPS7A2550) from the +12v feeds the REF2033 and the comparator. A separate switching supply (TPSM5601R5) creates 3.3v for the digital circuitry.
Whenever I switch the FET on, I see a large amount of ringing on the REF2033 1.65v output at about 580kHz, lasting for multiple microseconds.
I'm using some fancy optically isolated probes on my scope to measure the analog voltages - Lecroy HVFO108 - and I see some small glitches (<100mV) for about 100ns on the 5v supply when the FET switches on but +/-500mV ringing on the reference lasting for multiple microseconds. When the FET switches off, it's even worse - the transient on the 5v is larger and the reference rings for 10s of microseconds. I suspect this is because the FET drive is configured for a harder turn-off than turn-on.
I have tried adding capacitance to the reference output (started out with 3x 100nF X7R, added 3x 1uF X7R), extra filtering on the 5v line, and loading the reference with a 680 ohm resistor. I have not seen substantial differences in the reference output ringing with those changes. I can't pin down anything on the PCB that would be resonating at that frequency, since it's pretty small.
Persistent ringing for 10s of microseconds does not make sense to me, it looks like instability in the reference itself somehow.
Any help or suggestions on how to control this would be very beneficial.
Thanks
Ben Edwards
