Digital scopes - a beginner question.

[ginetto61]

Hi !
I have a question about digital scopes.
I am using the pc and a soundcard with a software, Arta, providing a spectrum analyzer function.
When i select the max sampling rate available (i.e. 192kHz) i get a spectrum extending up to 96kHz (i guess half the Fs).
Then i see digital scopes with bandwidths of GHz ... i am confused.
Are not they working on the same principle ? i mean using an adc ?
How can they have a bandwidth this wide ?
I am very confused
Thanks for any kind explanation.
Kind regards, gino

 

[abraxalito]

They're using a much, much faster ADC. Or perhaps an interleaved array of ADCs to achieve those very high bandwidths into GHz.

 

[ginetto61]

They're using a much, much faster ADC.
Or perhaps an interleaved array of ADCs to achieve those very high bandwidths into GHz

Hi and thanks a lot for the very kind and interesting reply.
So the principle is similar. Always a very high freq sampling.
Thanks a lot again, gino

 

[easp]

Hi !
I have a question about digital scopes.
I am using the pc and a soundcard with a software, Arta, providing a spectrum analyzer function.
When i select the max sampling rate available (i.e. 192kHz) i get a spectrum extending up to 96kHz (i guess half the Fs).
Then i see digital scopes with bandwidths of GHz ... i am confused.
Are not they working on the same principle ? i mean using an adc ?
How can they have a bandwidth this wide ?
I am very confused
Thanks for any kind explanation.
Kind regards, gino

For what it is worth, bandwidth and sample rates are two different things.

With digital scopes, bandwidth generally refers to the frequency response of the analog front end and is defined as the highest frequency it can handle without attenuating the signal by more than -3dB. Digital filtering techniques can also impact the achievable bandwidth of the instrument, as can software limitations imposed by product-line-engineering on a common platform.

The sample rate is more familiar. It refers to the rate at which the analog voltage is sampled (after passing through the analog front end). The sample rate should be at least least 2x the target bandwidth, and probably more like 3-4x.

You may see scopes with bandwidths of 1GHz and up, but they aren't exactly commonplace. A "budget" 1GHz scope probably starts at $5,000 or so, and most people who actually need such a scope are probably going to spent 2-5x as much.

Much more common are scopes with 1 billion/s sample rates and bandwidths of 100-200MHz.

They're using a much, much faster ADC. Or perhaps an interleaved array of ADCs to achieve those very high bandwidths into GHz.

Plus, most digital scopes have 8-bit sample resolution. They aren't precision instruments, at least not on the Y axis.

There are a variety of approaches to analog to digital conversion, and they generally tradeoff between precision and sampling speed (and in some cases, repeatability/accuracy).

 

[ginetto61]

For what it is worth, bandwidth and sample rates are two different things.
With digital scopes, bandwidth generally refers to the frequency response of the analog front end and is defined as the highest frequency it can handle without attenuating the signal by more than -3dB. Digital filtering techniques can also impact the achievable bandwidth of the instrument, as can software limitations imposed by product-line-engineering on a common platform.
The sample rate is more familiar.
It refers to the rate at which the analog voltage is sampled (after passing through the analog front end). The sample rate should be at least least 2x the target bandwidth, and probably more like 3-4x.
You may see scopes with bandwidths of 1GHz and up, but they aren't exactly commonplace.
A "budget" 1GHz scope probably starts at $5,000 or so, and most people who actually need such a scope are probably going to spent 2-5x as much.
Much more common are scopes with 1 billion/s sample rates and bandwidths of 100-200MHz.
Plus, most digital scopes have 8-bit sample resolution.
They aren't precision instruments, at least not on the Y axis.
There are a variety of approaches to analog to digital conversion, and they generally tradeoff between precision and sampling speed (and in some cases, repeatability/accuracy)

Hi ! thank you very much indeed for your very kind and valuable explanation.
My main interest, for now, are noise measurements on usb soundcard (it is the only thing that I can do more or less easily ).
I am using Arta software (very nice indeed).
The usb soundcard I am testing has sampling frequencies up to 192kHz
and so i get the spectrum of the soundcard noise floor up to 96kHz.
It would be perfectly fine if I did not read some discussion about how much is important higher Hz noise for digital equipment.
In particular they are referring to the noise coming from power supplies.
I am using mostly smps now (but I am about to test some linear ones diy).

In particular i would be very interested to know:
1) how much detrimental can be the psu noise > 100 kHz for a AD/DA card/converter ?
2) how much noise > 100kHz has a common smps

All the range above 96kHz is a complete mistery to me
this makes me very worried.
Maybe what I do not see it is what really matters ?
because up to 96kHz even a cheap but decent quality smps for laptop can be very fine, in terms of resulting noise floor of the soundcard I mean.
For now i am experimenting different psu solutions with the same usb soundcard in order to see the lowest and flattest noise floor I can get, as a start.
Thanks a lot again, gino

 

[easp]

For ~$100 you can get a DAC with low THD, which, when fed to a suitable amp + speakers has a noise floor below hearing thresholds, and peaks near the pain threshold.

Or, you can spend much more than that, and end up with equipment that measures phantoms because your test lash-up is flawed in ways you don't even know to worry about

If you are going to go down the rabbit hole, I'd suggest forgoing the DSO and getting an old analog scope capable of running in differential mode. Use that to verify "noise" deliberately injected into the power supply of your DUT (device under test). Use your sound card to measure the output of the DUT under various test conditions. Can you measure a difference between its output in normal operation from a clean linear supply or battery and that under various test conditions?

FWIW, Power Designs precision lab power supplies, when operating properly, have incredibly low noise levels, less than 100uV, and often much less than that.

 

[ginetto61]

For ~$100 you can get a DAC with low THD, which, when fed to a suitable amp + speakers has a noise floor below hearing thresholds, and peaks near the pain threshold.

Hi ! thanks again but with the dac I am done.
I am using a usb soundcard that is used also by recording engineers.
Some reviews have spotted a little issue with the smps. So they have slipped a little on the psu.
Replacement is made difficult because the device is very little an available space a big constraint (see picture attached).
I hate this approach of building small units.
I like rack size units.
So AD and DA converters are very good. Only this small issue with the stock smps.

Or, you can spend much more than that, and end up with equipment that measures phantoms because your test lash-up is flawed in ways you don't even know to worry about
If you are going to go down the rabbit hole, I'd suggest forgoing the DSO and getting an old analog scope capable of running in differential mode.
Use that to verify "noise" deliberately injected into the power supply of your DUT (device under test). Use your sound card to measure the output of the DUT under various test conditions.
Can you measure a difference between its output in normal operation from a clean linear supply or battery and that under various test conditions?
FWIW, Power Designs precision lab power supplies, when operating properly, have incredibly low noise levels, less than 100uV, and often much less than that

I have the feeling that measuring directly the noise coming from a psu is beyond my reach.
I can only see on my pc the noise floor spectrum of the usb soundcard connected to the pc itself. In the end this is my final goal.
For sure if I power the card with a precision extremely low noise lab supply this would be extremely useful to reach the lowest possible noise floor for the soundcard. I could see the limits of the soundcard, at least on terms of noise.
However from what I have gathered its limits are very good.
It is very well designed and built, smps aside
Power Designs is not very common here in Europe where I live.
Do you have other brands maybe they could look for ?
Thanks a lot again, gino