Low Amplitude and DC Offset on the Crystal

[drad1]

Hello, 

We use STM32H755 with a 25MHz crystal oscillator connected to OSC32_IN and OSC32_OUT OSC_IN and OSC_OUT pins. Probed on the bench, I see a sine wave with a DC offset and low amplitude. Can you comment if the attached waveform looks okay? My expectation was that the sine wave would swing from 0V to around 3.3V, and that there should be no added DC offset. 

 Thanks!

[AScha.3]

No, that's ok.

The amplitude depends on the drive strength, so if you set high drive level, amplitude is bigger, but consuming more energy from the driver.

[TDK]

This is not sufficient as-is.

However, the probe adds capacitance. It's possible that the signal is fine without the probe attached. Your probe probably has a 10x mode which should be used to reduce (but not eliminate) the capacitive load.

Additionally, and most critically:

> STM32H755 with a 25MHz crystal oscillator connected to OSC32_IN and OSC32_OUT pins

The OSC32 pins expect a ~32 kHz signal. A 24 MHz signal won't do.

[drad1]

Just to address the typo on my end: It is connected to OSC_IN and OSC_OUT pins. 

[drad1]

How do you determine that it is not sufficient as-is? Is the amplitude too low, or is the relatively large DC offset worrisome? Frequency does look good.

Are there any techniques that can be used to get higher level of confidence that the oscillator works correctly? My only idea is to get a low-capacitance differential scope probes. 

[AScha.3]

Is clock / CPU working fine?

If so, why you worry about it's signal level?

Or do it the way, I choose for 100% reliable clock: use a clock oscillator. Not much more expensive than a crystal and caps, but connect and forget. And less risk of coupling in any spikes from ambient.

Settings for HSE bypass then.

[drad1]

The MCU works fine - no issues that can be traced back to the crystal. 

Not too worried - I was just curios as I didn't see such low amplitude/high DC offset before. I wanted to make sure I am not accidentally designing the circuit close to the failure borderline.

Thanks for the clock oscillator suggestion. I prefer them for the simplicity as well. For this particular question, it is too late to make a change. 

[AScha.3]

So just accept the way the CPU is driving the crystal, as long as it is giving perfect results.

These crystal oscillators are built to be low power, so the drive level is low. And about mid scale of supply, that's normal.

If you don't want to worry about driving a crystal, take an oscillator next time.