[Solved] Need help debugging my first PCB project

[Cory Parsnipson]

Don't attach the scope across the crystal. Attach the ground to to the ground and probe either of the pins. Ideally XOUT, as it has lower chance to cause disruptions.

Usually with crustal oscillators when things start to work if you touch them means that you don't have correct load capacitors. Usually you need to have load capacitors to the ground from both XIN and XOUT.  Some value may be already integrated into the IC, but the actual value you need depends on the crystal parameters. The datasheet here does not have any information, so all you can do is guess.

Ah ok, oops... 

Here's the probe on XOUT with the ground on the J1 connector ground. Looks like it's got a period close to 80ns, though there seems to be some interference happening.

If it keeps being unreliable, add 5-15 pF capacitors to the ground from the crystal pins. In your case it would be a bit tricky, since you have 5V pour on the soldering side. This is another case why you want to have mostly ground pours.

Are there guidelines for ground/power pours? I just read some very brief tips about using KiCAD about wanting one ground and one 5V on either side of the PCB.

[ataradov]

The clock looks strange. Capture it with DC coupling. The amplitude is really high, this has to overdrive the crystal, but it is possible that it is by design.

On 2 layer boards use ground pour on both layers and everything else with traces or local polygons. For 4 layer boards, two internal layers are typically power and ground while outside are for the signals. And if signal density on the outside layers is not too high, add ground pour there too.

[Cory Parsnipson]

Ok, what about this one?

[ataradov]

This is very strange. DC coupling yet there is no DC component. They must use some very strange oscillator design. But I guess if it looks clean and works, then I would not mess with it.

[Cory Parsnipson]

Maybe I'll be able to come back to this after I get a few more projects under my belt and understand the design more deeply. That's the hope anyway.

Thanks again for your help!

[pcprogrammer]

This is very strange. DC coupling yet there is no DC component. They must use some very strange oscillator design. But I guess if it looks clean and works, then I would not mess with it.

The shown menu is for the trigger settings. I don't know how this scope works, but could it be that the channel it self is still in AC coupling?

[Cory Parsnipson]

The shown menu is for the trigger settings. I don't know how this scope works, but could it be that the channel it self is still in AC coupling?

Oh, good catch. I dipped into the TDS 210 manual and it says that I need to press the "menu" button under the channel. When I did that, it was on DC coupling the entire time.



I switched it to AC coupling, but... it's somehow still the same? I tried this a couple of times. However, as you can probably tell, I'm not too familiar with the oscilloscope either.



EDIT:

Actually I had an idea. So far I've been using probes that were two alligator clips. I pulled out the "needle-y" one and used that one instead. What I found was that I got a weird DC signal if I connected the probe to the XOUT and ground to ground. But if I reversed it--connecting ground to XOUT and the probe to ground, I get this waveform:



And here's the same one with AC coupling:



Probe pic:

[pcprogrammer]

With the actual probe, did you check if it is set to 1x or 10x, or is a fixed 10x probe?

On your scope it is set to 1x, and if the probe is 10x the voltage readings are of.

Also if the probe is 10x, the input attenuation of the scope needs to be changed to get a proper reading. With the scope ground connected to the USB device ground, while the device is connected to the computer can cause problems when the scope ground is not isolated and the computer is grounded too.

These are things you should be aware of when doing measurements. Connecting a probe ground to something that is not ground could damage both your scope and your device under test.

[Cory Parsnipson]

The probe is set to 1x.

These are things you should be aware of when doing measurements. Connecting a probe ground to something that is not ground could damage both your scope and your device under test.

Oh damn I didn't know that..

[Zero999]

If it keeps being unreliable, add 5-15 pF capacitors to the ground from the crystal pins. In your case it would be a bit tricky, since you have 5V pour on the soldering side. This is another case why you want to have mostly ground pours.

Note, it should work with the capacitors connected to 5V. As far as AC is concerned, the ground and supply rails are interchangeable.

[ataradov]

The probe is set to 1x.

It needs to be set to 10x. Forget that 1x mode exists for general measurements.

[tooki]

The probe is set to 1x.

These are things you should be aware of when doing measurements. Connecting a probe ground to something that is not ground could damage both your scope and your device under test.

Oh damn I didn't know that..

Yep, oscilloscope ground really is ground, not a floating negative like on a multimeter. Dave has a video named something to the effect of “how to not blow up your oscilloscope” which I highly recommend watching.