Glitches or ringing on the pulse output of the AS3340 and CEM3340

[antoine.pasde2]

While trying to calibrate/tune my #1222 VCO I used my oscilloscope to measure the frequency as I don’t have a good enough ear for tuning, and I got some weird readings from the scope. The frequency reading would sometimes jump up when it should be going down.

After some poking around, I noticed that there was some high frequency oscillation on the falling edge of the pulse output.

The lower the VCO frequency, the more ringing/oscillation on the falling edge, and it can get quite bad.

This starts happening when the VCO frequency falls below about 200Hz and because this unwanted oscillation is quite high frequency (on the order of 15kHz) it is not heard and it doesn’t show on an oscilloscope when the time base is set to see at least one full cycle of the oscillator.

The ringing is such a high frequency that it get slew limited by the opamp on the #1222’s output, but it is still clearly present.

I only have AS3340s so I haven’t experienced this myself with the CEM3340, but after observing it, I found this web site which mentions it also happening with the CEM3340.

I tried connecting an additional 0.1µF decoupling capacitor directly between the VEE and GND pins of the AS3340 as I thought the one in the #1222 VCO might not do its job properly because it is on the wrong side of the R6 resistor, but it did not make any noticeable difference.

So just a heads up, this is probably not a big issue if the output of the VCO is just connected to the input of a filter, but if it used as a clock for a responsive enough circuit (e.g. a fast clock divider) it might have unexpected results.

Has anyone else seen this issue?

Between this and the frequency drift caused by the PWM input, it seems like the pulse output of the AS3340 and CEM3340 should not be used and a pulse signal should be derived from the triangle output using a comparator instead.

[fredrik]

@Caustic mentions it briefly here (later updates):

AS3340 VCO /w hard/soft sync, FM modulation

Hey yall, I got some time today to take a look at my old VCO build. Couldn’t figure out what the hell was wrong with it, so i just built it again. Rather than just be boring, i also attempted to add the other features that were not included in the original stripboard design. Things like soft / hard sync and FM modulation. I dont really have the capacity to check that this works right as is, i did my best given the diagram in the datasheet and some example schematics on Electric Druid. All i ca…

And yeah, the raw pulse output works for a handful of audio octaves, but seems somewhat broken by design beyond that (hence things like the AS3340 hybrid that I haven’t used).

(decoupling at VEE instead of −12V seems wrong, btw, pretty sure you want a low impedance path to the capacitor)

[fredrik]

Btw, what pulldown value are you using? (the resistor between pin 4 and ground, whatever that is in the #1222).

[antoine.pasde2]

Thanks for the pointer @fredrik.

For the 3340 decoupling, I am not sure if you are agreeing with me or not.
You have to think of the internal zener diode as the voltage regulator for this chip.
VEE is then the negative supply for the rest of the circuit, you want decoupling capacitors between this negative supply and ground and between the positive supply at pin16 and ground.

This is what the #1222 currently has:

And this is what it should be:

@fredrik

Btw, what pulldown value are you using?

I am using 47k for R9, one of the values recommended by Sam in the #1222 VCO documentation for the AS3340.
I see the AS3340 data sheet shows 51k, but their example design is running at +15V, so maybe a value closer to 12/15 x 51k = 40.8k would be more appropriate? It doesn’t seem like it should make much of a difference as that output drives the high input impedance of an opamp, but maybe a lower resistance could reduce the ringing?

EDIT I see that @Caustic is using a 10k resistor in AS3340 VCO /w hard/soft sync, FM modulation and is still seeing some ringing so a smaller resistor doesn’t seem to help.

[KardashevTypeII]

Antoine,

I’ve been looking what everyone had thought to use for value of that resistor on this Performance VCO and I now realize it was right here.

-Fumu / Esopus

[Rusty]

Try this:

[antoine.pasde2]

I saw that mentioned elsewhere, basically adding hysteresis through positive feedback on the PWM comparator, but haven’t got around to trying it.
Has it helped for you?

[Rusty]

I never had this issue as I put that resistor there from the start. But I checked what happens if I remove it and it is as you described.

[antoine.pasde2]

Cool, good to know.
I’m deep into digital stuff right now with my Synth in an FPGA project, but as soon as I get time off from that, I’ll modify my #1222 oscillators with that fix. Thanks!

[analogoutput]

In another topic we had @trumac22 chasing down the PWM/frequency interaction and reporting:

@Trumac22

I also added a 1M resistor across legs 4 and 5 (something about hysteresis, means nothing to me) and it helped, but also reduced the pulse width pot to doing its full sweep in about 10 degrees of rotation. I cba to rectify that and it didn’t seem to make a huge difference, so away with it.

but I don’t think he was looking at the trailing edge glitches.

The Kassutronics 3340 design as of rev. 2.2 has this 10M resistor.:

On Rev 2.1 PCBs there may be some high-frequency oscillations around the pulse waveform transitions. While these oscillations are not directly audible, they can have an effect when using the pulse output as sync or clock source for other modules. To fix this issue, it is recommended to solder a 10M resistor directly between pins 4 and 5 of U2…

The T. Henry VCO Maximus has a resistor there but it’s 1M.

[Rusty]

I will try to experiment some with different values. But my PWM sweep looks fine.

Oh and there is another issue that I have seen floating around.
Lot’s of designs use -12V for scale and power. This is far from ideal.
Using -5V gives better tracking, more stability, less heating.

In general the less current you drive thru the IC the more stable it is.

Same goes with 10K/51K pulldown on pulse out. You won’t see any noticeable difference, but again the same applies here! Draw less current!

[analogoutput]

The Kassutronics design uses a -5V supply, and reports are this improves the frequency/PWM problem.

The 10k resistor is specified by the CEM3340 datasheet. The AS3340 specifies 51k. But anecdotal reports say either will work with either.