Measuring Distortions with the Scope:What you see is not what you really have..

[nctnico]

Think of it this way: if you can make extra information appear magically from an 8 bit ADC, then why aren't we all using scopes with 1 bit ADCs? Even cheaper to make!

There is no "magic" in these concepts and in fact many of the slower Delta Sigma types are simply 1 bit core ADC types (1 bit comparator) with massive Oversampling and Multi-Order Modulators followed by high order Decimation Filters and easily achieve well beyond 20 ENOB, these are the common 24 bit Delta Sigma ADC chips that only cost a few $!!

Many of the higher speed ADC chips employ techniques similar to the 1 bit core Delta Sigma ADC, except they utilize more than 1 bit (usually 3 ~ 4) in the "Comparator" and DAC "Feedback" path to speed the overall conversions up, and digitize the Signal - Feedback difference instead of integrating such as the DS ADC do (if they utilize integration then some of the benefits of Modulator induced "Noise Pushing" can be employed, but this slows things down).

Yes, but these ADC cores are designed to meet the specifications of the end goal which is to have and ADC with X bits and not X+'magic number' bits.

Anything beyond is just luck. If you buy a whole bunch of 3.5 digit DMMs and stick 3 extra digits to them, a few may seem to be accurate beyond 3.5 digits but that is more due to luck and cirumstances (right temperature for example) rather than solid engineering. In the end measuring is about having a certain confidence level in what is being shown on screen.

The same goes for measuring signal levels using an oscilloscope. The numbers Performa is listing in a posting above, have a 0.5dB error margin. IOW: a different unit may show different results within the specified error margin.

[nctnico]

@nctnico, do you now of a signal that can be generated easily with an AWG where artifacts would appear within the effective dynamic range. Would be nice to do a side by side comparison, and also useful to see what to watch out for...

The two tone test mentioned earlier on is a good one (and also industry 'standard').

[switchabl]

From a DSP perspective, if you are only interested in a low frequency (relative to f_s/2) sub-band, you could either do a full FFT and throw part of it away or decimate first and do a shorter FFT. The latter is of course more efficient and this is basically what happens with ERES on. So you can get more frequency resolution (and processing gain) with the same resources.

Of course, caveats mentioned by gf apply (only works if ERES implementation actually does decimation, ERES filter less than perfect).

[mawyatt]

Yes, but these ADC cores are designed to meet the specifications of the end goal which is to have and ADC with X bits and not X+'magic number' bits.

Anything beyond is just luck. If you buy a whole bunch of 3.5 digit DMMs and stick 3 extra digits to them, a few may seem to be accurate beyond 3.5 digits but that is more due to luck and cirumstances (right temperature for example) rather than solid engineering. In the end measuring is about having a certain confidence level in what is being shown on screen.

The same goes for measuring signal levels using an oscilloscope. The numbers Performa is listing in a posting above, have a 0.5dB error margin. IOW: a different unit may show different results within the specified error margin.

This IS the way things work, whether you want to believe it's due to "magic" or "luck" is of course your prerogative.

The mentioned 24 bit DS ADC just uses a 1 bit comparator (ADC if you will), yet achieves better than 20 Effective Bits (ENOB). This is achieved, even specified in the data sheets, using Oversampling and Decimation, basic Signal Processing 101 methods.

The idea of "dithering" to achieve higher resolution and improve linearity is not new, this was utilized in the early 70s in the Ring Laser Gyro (RLG) to "unlock" the counter rotating photon (laser) beams at low rotation rates (dead zone), and likely utilized well before this time. Don't think folks back then would rely on "luck" or "magic" to guide/navigate an airplane, ship, submarine, satellite, rocket or missile.

So the techniques being discussed are real and repeatable, not conjured up non-sense as we often see. As much as many like the Siglent DSOs (we have a couple), the results shown are not unique to these instruments, but achievable and repeatable by any quality DSO with the necessary features to achieve such, and would expect the Rigol, GW, Keysight, LeCroy, Tek and so on, would respond similarly to these techniques. 

Anyway, hope you'll realize this isn't "magic" nor "luck", just good old Engineering/Science Signal Processing being put to use for our (and others) benefits.

Best,

[2N3055]

...The same goes for measuring signal levels using an oscilloscope. The numbers Performa is listing in a posting above, have a 0.5dB error margin. IOW: a different unit may show different results within the specified error margin.

Every measurement ever will have result within it's certainty interval. On any instrument.
0.5dB error margin for a scope is not bad at all...
And scope Performa used has specified BW flatness..
So I don't see any problems...

[mawyatt]

No doubt that the real deal – a true higher resolution ADC – is the superior solution. Yet there is nothing wrong with a proper ERES/HiRes implementation as long as the user thinks a little. And that should not be too much asked, since there are so many situations in T&M where we cannot trust blindingly and just copy some values from an instrument to the lab protocol without a second thought.

This is the basic issue with folks utilizing an instrument beyond their understanding, as many knowledgable individuals have said before, "Know Thy Instrument", or KTI!!

We've all been there, done that, at least us older folks!! Where we grab a little utilized instrument from storage to make a measurement or troubleshoot without taking the time to revisit what/how the instrument works/behaves, only to make a useless measurement that confounds the problem at hand!!

Today many of these little utilized instruments/functions are now contained within the modern DSO, especially true with the ERES/HiRes, FRA/Bode, and SA(FFT) Function/Mode/Instruments, and there seems to be little understanding of how the Signal Processing, Input Amps/Attenuators and core ADCs behave (especially WRT each other) when utilized outside the usual Time Domain Scope operation, as with these mentioned functions.

Anyway, thanks to the demonstrated truly knowledgable individuals like yourself and very few others, some of the less seasoned folks can sift thru all the rubbish and glean a sparkle of understanding regarding these increasingly complex data acquisition systems we now call DSO/MSOs

Best,

[Martin72]

Now I only have the SDG2122X, the SDG1062X is gone....
But not far away, it landed on my work(sold to it).....
This has the nice effect that I can continue to make measurements with it, but then with the HDO6034A...Not the worst option.

[Martin72]

Today just before closing time quickly measured what, must still adjust the scaling, for better reading.
But nothing new, the second harmonic dominates the spectrum, even with the HDO.
Then again quickly our last analog generator, a Wavetek, connected.
Then I experimented with and without ground, but these were only marginal changes.

[2N3055]

Short measure today, must set the axis scaling more proper..
But nothing new, the 2nd harmonic is domenating on the HDO too.
Today just before closing time quickly measured what, must still adjust the scaling, for better reading.
But nothing new, the second harmonic dominates the spectrum, even with the HDO.
Then again quickly our last analog generator, a Wavetek, connected.
Then I experimented with and without ground, but these were only marginal changes.

Wise choice of level and V/div. Try similar with SDS2000X HD. Also try using fine vertical and watch for the sweet spot...

[Martin72]

Try similar with SDS2000X HD

Can do but now with the SDG2122X only(or I take the HD to work...Hmmm..)
Meanwhile I got an idea, we have several high precision power analyzers from ZES here.
They can show THD and SHD, when their input sensivity is low enough, I would connect the SDG1032X to one of them, could be interesting.

[elecdonia]

@nctnico, do you now of a signal that can be generated easily with an AWG where artifacts would appear within the effective dynamic range. Would be nice to do a side by side comparison, and also useful to see what to watch out for...

The two tone test mentioned earlier on is a good one (and also industry 'standard').

I recall Prof. W. Marshall Leach and others (Walt Jung?) using a two-tone test comprised of equal levels of 10kHz and 11kHz to demonstrate TIM (transient intermodulation distortion), aka “slew-rate induced distortion.” For systems with a very wide bandwidth the test tones may be increased in frequency up to 19kHz and 20kHz. The frequency difference between the two tones is kept at 1kHz. The resulting intermodulation distortion product always appears at 1kHz. Such a large frequency displacement makes this distortion product relatively easy to hear and measure.

Prior to the discovery of TIM, the two tones in the SMPTE/DIN standard IM distortion test were 60Hz along with 7kHz at a relative level of -12dB compared to the 60Hz signal. Non-linearity will generate sideband signals at 60Hz intervals immediately above and below 7kHz. Another similar IM test signal substitutes 4kHz instead of 7kHz for applications with lower bandwidth. In this case the distortion products appear in side bands near 4kHz.

[EA1DDO]

If you need to measure a "Direct injection box" (aka transformer) audio characteristics, eg. attenuation, THD distorsion, etc.
What piece of equipment do you recommend?

Thanks

[Martin72]

Hi...
I had made a small setup today to measure the amplifier tomorrow using bode plot.
Before that I had sent a 1khz sinus into the amplifier and plotted the FFT of it.
Following the discussion here, the result can not be trusted...or is it?
Or in other words, how could you make sure that what is displayed is coming from the amplifier and not from the scope itself...

[pope]

If you need to measure a "Direct injection box" (aka transformer) audio characteristics, eg. attenuation, THD distorsion, etc.
What piece of equipment do you recommend?

Thanks

An audio interface with the appropriate software. Or an audio analyser if you can afford one.

[BillyO]

Or in other words, how could you make sure that what is displayed is coming from the amplifier and not from the scope itself...

See if you can get a hold of a sine wave reference oscillator.

[Martin72]

Well, finally it´s time to assemble this kit here..

[mawyatt]

Is that an oscillator based upon the Wien Bridge, kinda like what LT Jim Williams did long ago with the Opto-Coupler replacing the incandescent bulb in the famous HP oscillator!! Recall Jim used this oscillator to evaluate 20 bit ADCs, so should be good enough for our DSOs 

BTW looks like fun ahead

Best,

[Martin72]

Hi,

It´s this one :

Ultra low distortion oscillator (below -140dB)

And yes, a classical Wien-Bridge...

[BillyO]

Well, finally it´s time to assemble this kit here..

Where can you get those boards?  Looks like a nice ting to have.

[Martin72]

Here:

http://www.janascard.cz/aj_Vyrobky.html

You can get it completely or only the pcbs.

[Sighound36]

I have only just read this thread our workload currently is quite manic so as we have a quality audio analyser UPV66 have set up the unit to measure its own internal high precision genny @ 1 khz ref voltage is 1V 20Hz-20Khz FFT 256K 4 averages, Van hann window


On the Wavepro HD I used max bandwidth 4Ghz, NO filters ( I have ERES and digitial filter package) as close as I can get to Martin's set up here, ground is on 600Ohm output on the genny selected. SA set to Von han / power spectrum / RBW20Hz. 1Mohm  input 1m long RG59U 22 AWG. with an XLR<>BNC adapter, I didn't de embed cable impedence tonight.

Next image same setting etc just with 10 averges

The following image now spots 200Ks and 1Ms sampling same HTB then the 10 avergae of that as well.

Then 5MS memory points and 25MS/s sampling rate & that average

Last pairing 200Meg points of memory & 1GS/s rate.

Finial image the noise sat on the genny output with it off!

[Martin72]

Hmm..
So the first pictures show the "truth", because actually there should be no harmonics in the amplitude range...

[Sighound36]

Hi Martin

I think it with either NTICO or Sinisa? who made the observation about what's really buried in the noise.

Its Sunday evening maybe I have not been as accurate as I could have been? The analogue gennerator is the precision version so it should be pretty clean. Again we are back to how good is the set up/opertor and actual ability of the equipment to deliver a result that is as free of measurment uncertanty as possible.

If you have specfic set you wished checked happy to do this.

Sighound

[Martin72]

Hi,
I'll try to finish the oscillator as soon as possible or use the generator from my neutrik A1 to recreate your measurements.
From my (probably too simplistic) thinking, if a signal with say -100dB thd is fed in, there should be nothing to see in the displayed FFT when it goes "down" to -80dB.

[Martin72]

I'll try to finish the oscillator as soon as possible

The missing parts ordered today at Mouser, the sum has already become three digits.