0

I am using this circuit to interconnect a pairs of crystal

enter image description here

Now the resulting signal has a ringing problem. I am not sure if it is due to ground loops or to my probe ( normal 6" probe). But I noticed if I use a 50 Ohm Resistor between the probe and the output it's getting better (still present.). Now, I would like to feed this signal to a PLL and I am afraid of possible false triggering. Is there a way to render it as a normal square wave? Should I use a S.Trigger (the inverter already has a S.Trigger btw) or something else?

enter image description here

enter image description here

Ken90
  • 17
  • 4
  • The added series resistor probably adds a little dissipation is all. The ringing seems like a few hundred nanoHenries and maybe 15-20 pF. Not sure if that's the circuit layout, the probing, or both. – periblepsis Jun 22 '23 at 02:12
  • Show us a photo of the actual circuit you constructed as well as how you're probing it (both tip and ground of the scope probe). – brhans Jun 22 '23 at 03:02
  • Different crystal driving circuits exist, why do you need to use this one? The inverter gate logic family and exact type is also left undefined so the type or resistor values might be OK or wrong, and there is no AC coupling between inverters. – Justme Jun 22 '23 at 05:16
  • @periblepsis yes that can be the cause. Adding a 4013 and diving the frequency totally remove the ringing but still I would be curious to know how to flat that signal without dividing – Ken90 Jun 22 '23 at 05:17
  • @Justme right, I used a 74HCT14N hex inverter for this. Maybe changing the resistor values will do it, will give a try. AC coupling what do You mean? adding a non-polarized capacitor between the two inverters (1-2)? – Ken90 Jun 22 '23 at 05:21
  • @Ken90 There's a discussion [here](https://www.sitime.com/support/resource-library/application-notes/an10028-probing-oscillator-output). Take a look at Figure 3.6, for example. Also consider using an active probe, as that's kind of what they are designed for. – periblepsis Jun 22 '23 at 05:31
  • 1
    Why did you use a HCT14? That's a buffered CMOS inverter with Schmitt trigger inputs and TTL compatible voltage levels. Extremely unusual selection for a crystal oscillator. – Justme Jun 22 '23 at 05:35
  • @periblepsis beautiful article, thanks a lot – Ken90 Jun 22 '23 at 08:57
  • @Justme was the only HEX inverter I had here :) – Ken90 Jun 22 '23 at 08:58
  • @periblepsis looks like the exact problem, thanks – Ken90 Jun 22 '23 at 09:09
  • If you can interpolate an answer with the relative article, I will mark it as correct @periblepsis – Ken90 Jun 22 '23 at 09:15

1 Answers1

1

Thanks to @periblepsis for posting this interesting article.

enter image description here

It is clear to me ringing effect in this case is due to the ground loop of the inductance of the probe.

Ken90
  • 17
  • 4
  • It might be. Or it might be due to 1x probes, as you did not mention if you use 1x or 10x probes. Or the fact that the output is driven with one inverter and the other inverter via 1 kohm load resistor to the opposite direction. That's not how a crystal drive waveform should look like anyway, because if you want square wave logic levels out, only the final buffer should be Schmitt trigger. – Justme Jun 22 '23 at 14:26
  • 1
    Nope using 10X. Indeed if You use a 1X probe You will not have a square wave from those small crystals check out this answer https://electronics.stackexchange.com/questions/58223/output-of-oscillator-expected-to-be-square-wave-but-looks-sinusoidal – Ken90 Jun 23 '23 at 00:56
  • 1
    @Ken90 You can select your own answer. Or wait. But yes, I think you've got ground loop inductance and probe capacitance on the order I mentioned (well, I didn't just 'think so' but also tested it with LTspice and the curves match nicely and with the same damped sine output with same number and duration and relative peak magnitudes. So I have some support for those numbers. +1 on your answer. You can greatly shorten that ground lead, by the way. More custom work to do it. But it's not hard, either. – periblepsis Jun 23 '23 at 16:11