Convert 0-3.3v PWM signal to +/-12v Please Help

[ronsimpson]

show periodic oscillation

What frequency? How much? "this one"??

 

[milolouis]

To be honest I'm sure it's something I did but with a 40KHz square wave on V1 at 50-99% duty cycle It will randomly drop from 12v to -12v and then back up every millisecond or so.
Could you explain how it's supposed to work?

 

[ronsimpson]

Could you explain how it's supposed to work?

U2 is a low pass filter. The frequency sweep says that at 100hz it will pass the signal and give it gain. At 40khz it will only pass -60db of the signal. So it should remove the PWM signal but pass the "average" of the signal.

U1 is an example of a simple RC filter. (with a buffer) How is it doing?

It will randomly drop from 12v to -12v and then back up every millisecond or so.

You need to set V4 to 3.3 volts. (supply for your PWM source) "Set to 3.3 for DC. Set to 0 for AC study "

 

[ronsimpson]

You are right! It is oscillating. I have never pushed the gain beyond 2 or maybe 4. OK let me see why.

 

[ronsimpson]

I changed C2 to 0.0047u and it is stable.

 

[ronsimpson]

Stepping from 0 to 50%. Takes 6mS to get there. Seems stable.

 

[milolouis]

Thanks Ron this, is fantastic thanks for the explanation as well. I get about 10mV noise with just one op-amp. Cut-off is a bit low however so just fiddling around with it a bit. Hoping to get 1000Hz pass through rail to rail. Or at least more than 500hz.

 

[alec_t]

Here' a 3V to +-12V switcher which should give a pretty clean output at 50kHz :

 

[ronsimpson]

Alec_t,
What does the collector current look like? Q3, Q4

 

[alec_t]

Not very nice. There's considerable shoot-through .

 

[milolouis]

Still alec_t that's exactly what I was looking for originally. Spot-on Thanks so much.

The target goals are:
- 40-50KHz 0-3.3v PWM input
-Smooth Analog Voltage output very low noise with effectively zero-drop off at 1KHz (PWM duty modulation)
-Lowest part-count/cost
-Must be able to reach less than 0.1v

As there will be 8 of these on one board.

So I think now with all the advice that probably the best way is two op-amps with RC filter before the voltage translator and then second op-amp as Sallen Key Filter.
So 3-poles and translation with two-opamps.

If you or anyone disagrees or thinks there is a better way better way then I'd love to hear it. I'll post a picture of the schematic when it's finalised

 

[ronsimpson]

Alec_t,
The shoot-through comes from two places.
The first two transistors were really slow at turn off. So I made them "grounded base" and keep them out of saturation.
The last two transistors need a "turn off" resistor.
I have not gone back and checked each resistor. This is just a first try. Shoot through 10 to 15 mA.

 

[alec_t]

That's a big improvement, Ron.