P-channel FET switch producting oscillation

[Aeternam]

I wired up a pretty straightforward p-channel switch. At a basic level, the circuit works. I'm using an IRF4905 for the FET, the BJT is a BC547 (designations for M1 and Q1 are not correct on the schematic, but resistor values are, sorry for that.)

I'm feeding it a +3V square wave with approx 10 ns rise time. At frequencies above 100 Hz I start to see artifacts that look suspiciously like oscillations (pink plot, this is measured across the load). I'd like to get rid of that, I want to see a clean output up to 250 Hz.

Can someone tell me where I should start looking for the cause of these ripples? Is this Miller effect? Is it too much gate charge in this specific FET (the datasheet specifies 180 nC for the IRF4905)?

Thanks

[Zero999]

Move R3 to the emitter of Q1 and reduce its value to 270R.

[Aeternam]

Thanks for replying Hero. There is no R3 in the schematic, I suppose you mean R4? Moving it to the emitter won't do any good, the transistor is wired as switch.

I tinkered a bit more with the circuit and the ripples do look like exponential charge and discharge curves of an RC. I suppose that excessive gate charge is the culprit here. I could lower R4 to get more oomph at the gate but that's not really an option since I'd like to switch 12V in the final circuit and anything below 1kish will smoke the resistor.

Guess I'll have to order some FET with less Qc and see how that goes.

[Benta]

How "stiff" is your VCC? I suspect decoupling issues on the supply. You could try tracing VCC as well on the scope.

[danadak]

You changed your time base from 5 mS to 1 mS between screen shots, I suspect
scope probe may be issue and its grounding. Is there a long ground lead on scope
probe ?

The earlier point about bypassing Vcc at the device, use a .1 uF ceramic as well
as a 10 uF tantalum, preferably a polymer tantalum. Probe that an look at noise
as suggested earlier.


http://teledynelecroy.com/doc/passive-probe-ground-lead-effects



http://www.linear.com/solutions/4450


Frequency of oscillation appears to be ~ 500 Hz, so seems to confirm
maybe supply bypassing / regulator issue.


Regards, Dana.

[T3sl4co1l]

The time constant is some orders of magnitude too long to be anything in the circuit shown.

That looks more like step response from a poorly compensated switching supply.

Let me guess, a little 5V plug pack or charger?

If you like, you can fix it by opening the supply, locating the compensation capacitor, and making C smaller, and adding R in series with it, until the response is usefully stable.

Tim

[Aeternam]

How "stiff" is your VCC? I suspect decoupling issues on the supply. You could try tracing VCC as well on the scope.

Bingo!

A hefty electrolytic across VCC solved the issue. The output is now clean as a whistle. 

Thanks, all !!

[Zero999]

Thanks for replying Hero. There is no R3 in the schematic, I suppose you mean R4? Moving it to the emitter won't do any good, the transistor is wired as switch.

Sory it was a typo. I meant R2. Try changing R2 to 270R and moving it to the emitter of Q1.

[Benta]

Thanks for replying Hero. There is no R3 in the schematic, I suppose you mean R4? Moving it to the emitter won't do any good, the transistor is wired as switch.

Sory it was a typo. I meant R2. Try changing R2 to 270R and moving it to the emitter of Q1.

Please expand on this idea. As far as I can see, the only effect would be reducing the gate voltage to the FET. Not really desirable.

[Zero999]

Thanks for replying Hero. There is no R3 in the schematic, I suppose you mean R4? Moving it to the emitter won't do any good, the transistor is wired as switch.

Sory it was a typo. I meant R2. Try changing R2 to 270R and moving it to the emitter of Q1.

Please expand on this idea. As far as I can see, the only effect would be reducing the gate voltage to the FET. Not really desirable.

Adding emitter degradation will increase the stability of Q1, making it less likely to oscillate.

The reduction in gate voltage is a non-issue, since the MOSFET is more than capable of switching 1.2A with a gate voltage of -8.5V. In fact a lower gate voltage may be desirable, since it will switch off more quickly.