Older DSO for audio?

[martyh]

Will an HP 54645A with the storage/math option be good enough for audio work? (100MHz 8bit 200 MSa/s) I’ve been wanting an upgrade from my even more ancient Tek T 922 for a while and this one is cheap enough. I’m hoping that I can use the FFT function to measure distortion and use the extra bandwidth to sniff out HF instability. I’m sure there are better options these days but will something like this do what I want?
Thanks,
Marty

 

[Mark Tillotson]

8 bits in a scope means they do not have the dynamic range for audio distortion measurement.


BTW if its 200 MSPS, the bandwidth cannot be 100MHz in practice, more like 40MHz useful bandwidth.

 

[1audio]

It's really 20 MHz single shot. However its more than adequate for audio. To look at distortion add a notch filter.

 

[martyh]

Thanks guys.

 

[nigelwright7557]

8 bits in a scope means they do not have the dynamic range for audio distortion measurement.


BTW if its 200 MSPS, the bandwidth cannot be 100MHz in practice, more like 40MHz useful bandwidth.

The problem with bits is the number of bits relative to number of screen pixels.
I designed a 12 bit USB scope and then found only 8 bits would fit on the pc screen !

Also digital MHz needs to be divided by about 50 to get actual sine wave bandwidth on the screen.
Anything less than 50 bits per sine wave starts to give stepped or uneven waveform.

 

[Mark Tillotson]

The problem with bits is the number of bits relative to number of screen pixels.
I designed a 12 bit USB scope and then found only 8 bits would fit on the pc screen !

But what about vertical zooming? Or even anti-aliasing? Those 12 bits are great for measurement, and the FFT use of course.

Also digital MHz needs to be divided by about 50 to get actual sine wave bandwidth on the screen.
Anything less than 50 bits per sine wave starts to give stepped or uneven waveform.

A decent scope will sin(x)/x interpolate upto close to the Nyquist limit giving very nice displayed waveform for a sine wave. Data capture and data visualization are two distinct things. You do need about 10 samples to get a convincing rectangular or pulse waveshape however, ie 10MHz squarewave needs at least 100MSPS.

 

[nigelwright7557]

But what about vertical zooming? Or even anti-aliasing? Those 12 bits are great for measurement, and the FFT use of course.


A decent scope will sin(x)/x interpolate upto close to the Nyquist limit giving very nice displayed waveform for a sine wave. Data capture and data visualization are two distinct things. You do need about 10 samples to get a convincing rectangular or pulse waveshape however, ie 10MHz squarewave needs at least 100MSPS.

You can get away with a low sampling frequency for square waves but not sine waves. While sampling will average out the sine wave then it wont pick up anomalies in the sine wave which is what we are often looking for.

I found by the time I get scope display, FFT display and menu's on the screen there wasn't enough room for 512/1024 samples height. Just using top 8 bits loses a lot of information at the low end of voltage.

 

[phofman]

A decent scope will sin(x)/x interpolate upto close to the Nyquist limit giving very nice displayed waveform for a sine wave.

That is what every waveform visualization software should do, instead of connecting the sample values with lines.

 

[nigelwright7557]

That is what every waveform visualization software should do, instead of connecting the sample values with lines.

If you take a sine wave which fills the screen vertically which uses an 8 bit sample. That is fro mzero 64 steps up and 128 steps down then 64 steps up again. So 256 steps per sine wave cycle.
So in this case you would need 256MHz sampling rate to capture 1MHz sine wave fully.

 

[phofman]

Standard procedure is correct resampling/interpolating to pixel density.

 

[Mark Tillotson]

If you take a sine wave which fills the screen vertically which uses an 8 bit sample. That is fro mzero 64 steps up and 128 steps down then 64 steps up again. So 256 steps per sine wave cycle.
So in this case you would need 256MHz sampling rate to capture 1MHz sine wave fully.

If its only a sine wave there's no information above 2f. None at all. sin(x)/x interpolation is all you need.

Looking for distortion is a completely different matter, you need to sample above twice the frequency of the highest harmonic of interest, and with enough dynamic-range / SNR. Scope's are not able to do that for quality audio because 8 bits is way too little.

My scope has a resolution enhancement mode that can get upto 3 extra bits equivalent, and that's not enough(*) (though perhaps after a good twin-T notch filter it could be used).

(*) Except for valve amps

 

[H713]

Will an HP 54645A with the storage/math option be good enough for audio work? (100MHz 8bit 200 MSa/s) I’ve been wanting an upgrade from my even more ancient Tek T 922 for a while and this one is cheap enough. I’m hoping that I can use the FFT function to measure distortion and use the extra bandwidth to sniff out HF instability. I’m sure there are better options these days but will something like this do what I want?
Thanks,
Marty

Given your location I can only assume you're looking at the scopes on UW Swap.

I've used the 500MHz version of that scope at one point and it's fine. The requirements for audio work are pretty minimal.

Not sure about the 100MHz version, but the 500MHz version will let you play Tetris on it. Believe it or not it was one of my professors who showed me that.