Need some advice for 3340 pulse output

I have a typical setup for AS3340 IC.
This is my pulse wave with 51K pulldown, as per AS3340 specification:

This is the same output with 4K pulldown - just played around with a trimmer pot.

Generally the lower I go the sharper falling edge gets, but it drops the peak voltage as well.

It looks great up to around 2kHz, the pictures shows the most extreme example at ~18kHz.

This is measured without any additional load on the output.

I have seen that in general 3340 has pretty bad pulse wave shape, with somewhat slanted falling edge, end corners are not crisp at all. But this seems way out of line. I will probably add some additional shaping to make it crisper, but still would like to know what is wrong with it.

While I have your attention - has anyone implemented this solution ?
As far as I understand it stabilises the PWM, but will it clean up the wave shape as well?

4 Likes

What circuit are you using? Hard to say what the problem might be without knowing that. By the way, you can copy and paste pictures directly here, which is easier for readers than an imgur link.

1 Like

Doing my research it doesn’t seem anything that I have done. Pulse output on 3340 is just crappy. Anyway it looks this bad only at really high frequency > 10kHz. I added a Schmitt trigger to reshape the wave and now it is practically perfect. Slant at rising and falling edge is just 1-2% from pulse width at 8,44kHz, so I’m good with that. And thanks for the img tip, didn’t know that.

2 Likes

Nice, I imagine this is a pretty simple mod but if you have a diagram you could share that would be great.

On the other frequency end, have you seen this problem?

Between that and your problem and the well known PWM/frequency interaction it seems like the 3340 pulse wave is quite problematic. I think @lookmumnocomputer was talking a while back about looking into not using it and instead making a pulse wave using a comparator on the ramp or triangle output but I haven’t heard anything more about that.

2 Likes

I will definitely share the fix once I adjust the outputs. Now I have -10v to +10v so that needs to be cleaned up. But for the inpatient ones it is a simple comparator:

I take square wave as input and compare it to Vcc / 2 = that is ~ 6V. You would probably get even more precise the closer you get to the top of the square, especially in really high frequency.

And for that ringing glitch, just put 10M resistor between PULSE_OUT and PWM pin.

image

6 Likes

This cleans up my pulse perfectly. If you don’t have +5V reference then just put a voltage divider, anything from +5 to almost the upper limit of your pulse output is fine. If you want to be OCD about it then the closer you get to top the more precise your pulse will be.

My pulse is around 11V high coming out of the first op-amp, so 91K/82K divider takes that down to around 5V. You can replace one of them with a trimmer pot. to get perfect -5V to +5V.

4 Likes

Hot damn, I knew I set up my power supply’s optional 5v for a reason. Great idea!

3 Likes

I have received advice from smarter people than me that it is always better to have voltage regulator on the module. Thou I don’t think it matters much in this case, as it is just for a reference. And even if it fluctuated it wouldn’t mess with the actual tone of the pulse.

I guess I need to rename this post to - “Here’s some advice for 3340 pulse output” :smiley:

4 Likes