New Rigol DS1054Z oscilloscope

[David Hess]

I remember similar parts in the past costing at least an order of magnitude more for less performance and functionality.

Its not that long ago you couldn't buy such functionality as a chip, even at exotic niche product prices. Times change, and devices like this do have volume applications now. Hittite describes (or is it described now they are part of ADI) itself as a microwave company. That tells you what their core market is.

Tektronix was making their own custom high speed ADCs when they sold their silicon fab to Maxim and then they bought from Maxim and others got access to those high speed ADCs as well.

[coppice]

Tektronix was making their own custom high speed ADCs when they sold their silicon fab to Maxim and then they bought from Maxim and others got access to those high speed ADCs as well.

That was 20 years ago. Back then Maxim would not have been able to sell a high performance converter to a Chinese company. Even now Chinese companies wanting state of the art converters end up sending delegations to Washington to plead for them.

[David Hess]

Yes and what we have seen has nothing to do with the impurities of the applied signal. Yes impurities above Nyquist will fold back and show up, but even with my DSA815's 'crappy' TG this would be far down in the noise (i.e invisible).

I agree because I was looking for two different things and I think I got the math wrong and saw the image instead of the distortion but now I have a better idea.

And (but maybe I misunderstood David) reconstruction will not restore the original 24 MHz waveform.

I cannot find it now but you or someone else said the distortion would not have been high enough to create the AM modulation depth we saw so I was thinking about what could cause that.  The image added to the fundamental would do it if the image was lower amplitude and the reconstruction filter is attenuating the image would cause that.  Does that explain it?

If the reconstruction filter has a finite attenuation at the mirror frequency, then it should add back to the fundamental and cause large but not 100% AM modulation.  Does that make this test a way to measure what order and type of reconstruction filter Rigol used?

This fits with the earlier post marmad made about leakage.  I know I have run across that before but did not recognize it because the terminology was different.  A sharper reconstruction filter would allow one to get closer to the Nyquist frequency.

[David Hess]

Tektronix was making their own custom high speed ADCs when they sold their silicon fab to Maxim and then they bought from Maxim and others got access to those high speed ADCs as well.

That was 20 years ago. Back then Maxim would not have been able to sell a high performance converter to a Chinese company. Even now Chinese companies wanting state of the art converters end up sending delegations to Washington to plead for them.

I had not considered that but I am sure it was the case because of export restrictions.  They continued to make custom ICs for Tektronix as well for at least some time.  I have one Maxim marked part with Tektronix part number with a date code of the 50th week of 1994 which fits with the other Maxim parts showing up in the late 24xx series oscilloscopes which were made between 1989 and 1996.  Wikipedia says the date was 1994.

[coppice]

I had not considered that but I am sure it was the case because of export restrictions.  They continued to make custom ICs for Tektronix as well for at least some time.  I have one Maxim marked part with Tektronix part number with a date code of the 50th week of 1994 which fits with the other Maxim parts showing up in the late 24xx series oscilloscopes which were made between 1989 and 1996.  Wikipedia says the date was 1994.

I thought all the 24xx scopes used CCDs for sampling, and had very mediocre ADCs. I think Tek made those CCDs, and they were pretty neat technology for the time.

[marmad]

This fits with the earlier post marmad made about leakage.  I know I have run across that before but did not recognize it because the terminology was different. A sharper reconstruction filter would allow one to get closer to the Nyquist frequency.

No, this wouldn't help. Reconstruction filters are, of course, low pass filters designed to remove spurious high-frequency content (above Nyquist) in the sampled data. As stated in the physics post I made (and visible in the examples), leakage "...introduces spurious low frequency components in the sampled data." A look at the power spectra graph from that post confirms that the "power of the fundamental frequency" leaks into other lower frequencies.

As I noted before, there has been a continual development of adaptive sampling-frequency algorithms (to minimize the mismatch between input frequency and sampling rate) over the last few years to combat leakage (specifically for use with FFTs). There are a number of various papers on it if you're interested.

[David Hess]

I had not considered that but I am sure it was the case because of export restrictions.  They continued to make custom ICs for Tektronix as well for at least some time.  I have one Maxim marked part with Tektronix part number with a date code of the 50th week of 1994 which fits with the other Maxim parts showing up in the late 24xx series oscilloscopes which were made between 1989 and 1996.  Wikipedia says the date was 1994.

I thought all the 24xx scopes used CCDs for sampling, and had very mediocre ADCs. I think Tek made those CCDs, and they were pretty neat technology for the time.

I mean the analog 24xx series oscilloscopes.

On the digital 24xx series DSOs, I think the CCDs were what made them mediocre (but fast) rather than the slow ADCs.  For many years CCD based digitizers were many times faster than ADC/memory digitizers for a given price.

Another problem with the CCD digitizers is that they have terrible overload recovery; the CCDs may be saturated just like a photographic CCD suffers from blooming.

[David Hess]

No, this wouldn't help. Reconstruction filters are, of course, low pass filters designed to remove spurious high-frequency content (above Nyquist) in the sampled data. As stated in the physics post I made (and visible in the examples), leakage "...introduces spurious low frequency components in the sampled data." A look at the power spectra graph from that post confirms that the "power of the fundamental frequency" leaks into other lower frequencies.
interested.

Now I am confused.  Are you referring to the analog antialiasing filter before the digitizer or the reconstruction filter after the digitizer?  If aliasing occurs then the later cannot do anything about it.

As I noted before, there has been a continual development of adaptive sampling-frequency algorithms (to minimize the mismatch between input frequency and sampling rate) over the last few years to combat leakage (specifically for use with FFTs). There are a number of various papers on it if you're interested.

I like the paper but I am surprised someone wrote one about this and I do not understand how it applies in this case.

Phase locking the sampling frequency to the source to remove spectral leakage in FFTs is obvious especially if you are only concerned about one frequency or multiple phase coherent frequencies.  Some GPSDOs do the same thing to remove jitter in the time domain or spectral leakage in the frequency domain when they have a DDS synthesized output.  DDS driven PLL multipliers may do this to avoid dual modulus PLL multiplication.

The idea also reminds me of synchronous demodulation and orthogonal frequency-division multiplexing.  In the later case, don't they phase lock the receiver to the transmitted carrier so the FFTs do not suffer from spectral leakage?

Don't some FFT based spectrum analyzers sweep their sample rate just a little bit to take leakage free measurements on multiple signals?

[marmad]

Now I am confused.  Are you referring to the analog antialiasing filter before the digitizer or the reconstruction filter after the digitizer?  If aliasing occurs then the later cannot do anything about it.

Both are low-pass filters. Leakage is a phenomenon that is not associated with aliasing (since it happens below Nyquist) and is not preventable by antialiasing or reconstruction filters.

[pa3bca]

Now I am confused.  Are you referring to the analog antialiasing filter before the digitizer or the reconstruction filter after the digitizer?  If aliasing occurs then the later cannot do anything about it.

Both are low-pass filters. Leakage is a phenomenon that is not associated with aliasing (since it happens below Nyquist) and is not preventable by antialiasing or reconstruction filters.

Yes, and we established that in the screenshots I made there were no frequency components above Nyquist, so aliasing was not occurring.
But when reconstructing after the digitizer the scope is unable to reliably reconstruct the real 120 MHz signal, as this is too near the Nyquist frequency.
note: what results is not really an AM signal, but a DSB signal (or AM with suppressed carrier, the sine does not ride the wave) -equivalent to the summation of the original "real" 120 MHz and the leaked "mirror" at 130 MHz. In the FFT's a few pages ago and attached here you can see that the amplitude of the mirror grows as the sampled frequency approaches Fnyquist. The result is shown as a double-sideband signal, not really an AM signal as there is no power in the central frequency of 125 MHz.
All this is after the digitizer and no aliasing occurring.
So as Marmad noticed this leakage is no issue till say Fsample/2.5 (my 100 MHz). There the leakage is so small is does not show any more.
As I understand it (now) up until Fnyquist you can reconstruct the frequency, but the amplitude information gets lost above Fsample/2.5. Just too few samples, and these samples "shift" along the wave resulting in the AM like waveform
Looking at the samples near Fnyquist themselves I am not even sure that a reliable reconstruction of the amplitude is even possible, and that it has nothing to do with errors in the reconstruction algorithms used by Rigol? At least that is what the documentation Marmad supplied is suggesting.
Interesting learning experience. This goes to show that one should be really really careful when interpreting displayed waveforms even long before FNyquist.  So the DS1000Z's are 100 MHz scopes? Yes, but beware..

[pascal_sweden]

If the source in my electronic design has a max. frequency of 100 MHz, then where can I expect to get the higher frequencies from? Interference, noise, higher harmonics?

Everybody is discussing here that the low-pass filter in the analog front-end of the Rigol scope is no good, as it is not a higher order filter and does not have a steep slope behind the cut-off frequency.

But who cares if the signals in the electronic design are below or at max. 100 MHz?
I have a hard time to understand where the higher frequencies can come from, and if there are any, that their impact is noticeable.

Would like to have a better understanding on unexpected higher frequencies that can show up, their origin (interference, noise, harmonics) and their impact?

Aren't the higher harmonics far enough away, even if the analog filter isn't perfect?
Or is that where the problem originates, that the higher harmonics are relatively close to the original signal, and are not cut-out by the simple filter in analog front-end of the Rigol DS1054Z?

[pa3bca]

If the source in my electronic design has a max. frequency of 100 MHz, then where can I expect to get the higher frequencies from? Interference, noise, higher harmonics?

Everybody is discussing here that the low-pass filter in the analog front-end of the Rigol scope is no good, as it is not a higher order filter and does not have a steep slope behind the cut-off frequency.

But who cares if the signals in the electronic design are below or at max. 100 MHz?
I have a hard time to understand where the higher frequencies can come from, and if there are any, that their impact is noticeable.

Would like to have a better understanding on unexpected higher frequencies that can show up, their origin (interference, noise, harmonics) and their impact?

Aren't the higher harmonics far enough away, even if the analog filter isn't perfect?
Or is that where the problem originates, that the higher harmonics are relatively close to the original signal, and are not cut-out by the simple filter in analog front-end of the Rigol DS1054Z?

But how would you know / be sure that there are no higher (i.e. >= Fnyquist) in your signal? You certainly cannot check that with your scope.
Look at the output of my DG1032Z. Output at 24 Mhz and look at the spurious frequencies up to 100 MHz. Granted, in this case they are low enough not to be a problem, but I only know that because I have checked it with a spectrum analyzer. En more often than not You would have NO idea.
200 MHz into the 100 MHz DS100Z produced a 50 MHz signal of almost half the amplitude of the real 200 MHz due to aliasing. So, without being absolutely sure that no > Fnyqust signals are present you must allays distrust your scope.
Know what you are doing.
And a sharp roll-of filter (which the Rigol demonstrably does not have) is some safeguard that you are not seeing false-positives on your screen.

[marmad]

I like the paper but I am surprised someone wrote one about this and I do not understand how it applies in this case.

As I've stated a number of times over a few posts, leakage occurs because of a mismatch between the fundamental frequency and the sampling frequency. It's not really an issue when dealing with the time domain because no one attempts to sample and reconstruct that close to Nyquist (e.g. the original 44.1 kHz CD sample rate is ~fs/2.2) - but it's a regular problem in the frequency domain due to the nature of the FFT process. I only posted the link to the paper to illustrate the work going on to find solutions to the problem of leakage in that respect.

[pascal_sweden]

I want to come back to my question: if none of the signals in the electronic design, either the source, the intermediate signals in the signal path or the output signal has a frequency above the limit, then the higher frequencies can only come from interference, noise or higher harmonics.

Interference: Can this really go up to 100 MHz with a noticeable impact? Don't believe so.

Noise: Can this really go up to 100 MHz with a noticeable impact? Don't believe so.

Higher harmonics: Let's say we have a 100 MHz square wave. With a square wave at 100 MHz, the fundamental frequency is at 100 MHz, the 3rd harmonic is at 300 MHz, the 5th harmonic is at 500 Mhz. Doesn't the low-pass filter cut-out frequencies above 300 MHz? I believe it does!
If not, it is a design failure I would say, if a 100 MHz scope does not even handle a 100 MHz square wave (DS1104Z with 100 MHz BW, 250 MS/s per channel).

[marmad]

Everybody is discussing here that the low-pass filter in the analog front-end of the Rigol scope is no good...

No one said that. The low-pass antialias filter is absolutely fine for the maximum 1GSa/s rate - which is what manufacturers generally design their filters for (they can't design them for every sample rate since you will ALWAYS get lower sample rates sometimes when you're using slower time bases).

We've only been attempting to point out that, while most 4-channel DSOs have at least two ADC blocks (one devoted to each pair of 2 channels), the Rigol only has the single ADC block for all 4 channels, which will cause the sample rate to drop to 1/4 maximum when 3 or 4 channels are used - so you have to be aware of the limits of the filter when using 3 or 4 channels (even at the maximum sample rate) - just as you should be aware of the same thing when using any amount of channels at slower time bases (i.e. slower sample rates).

[Fungus]

Higher harmonics: Let's say we have a 100 MHz square wave. With a square wave at 100 MHz, the fundamental frequency is at 100 MHz, the 3rd harmonic is at 300 MHz, the 5th harmonic is at 500 Mhz. Doesn't the low-pass filter cut-out frequencies above 300 MHz? I believe it does!

 

Yes, but it doesn't solve aliasing, which is what we're discussing.

As you approach the Nyquist frequency there's this thing called "aliasing" which you willfully seem to be ignoring.

[DanielS]

As you approach the Nyquist frequency there's this thing called "aliasing" which you willfully seem to be ignoring.

I think the point he was trying to make is that the first harmonic of a 100MHz square wave (300MHz) would be a fair bit beyond the DS1054Z's front-end roll-off and not cause much aliasing.

But I agree that in the grand scheme of things, such a specific case is of rather limited use.

[Fungus]

As you approach the Nyquist frequency there's this thing called "aliasing" which you willfully seem to be ignoring.

I think the point he was trying to make is that the first harmonic of a 100MHz square wave (300MHz) would be a fair bit beyond the DS1054Z's front-end roll-off and not cause much aliasing.

But I agree that in the grand scheme of things, such a specific case is of rather limited use.

I'm not sure what he's getting at any more. He seems to be trying to find some sort of mathematical imperfection in a 'scope when looking at 100MHz signals with 250MSa/s.

But ... of course it's imperfect. The Nyquist limit is a theoretical limit, not a practical one. It only applies under ideal conditions.

(And we're discussing a $40 'scope)

[Mark_O]

I want to come back to my question: if none of the signals in the electronic design, either the source, the intermediate signals in the signal path or the output signal has a frequency above the limit...

And again, as pa3bca already pointed out, you can't know that, in general.

Interference: Can this really go up to 100 MHz with a noticeable impact? Don't believe so.
Noise: Can this really go up to 100 MHz with a noticeable impact? Don't believe so.

We're discussing physics, not religion.  Thus what you believe or do not believe will have no impact whatsoever on what will occur.

If not, it is a design failure I would say, if a 100 MHz scope does not even handle a 100 MHz square wave (DS1104Z with 100 MHz BW, 250 MS/s per channel).

And you'd be wrong.  Again. 

[Fungus]

If not, it is a design failure I would say, if a 100 MHz scope does not even handle a 100 MHz square wave (DS1104Z with 100 MHz BW, 250 MS/s per channel).

And you'd be wrong.  Again. 

Yep. A big part of "design" is ... (drum roll) ... "compromise"!

It's not a design failure at all to design a 'scope with the capabilities of the DS1000Z and sell it for $400 (in fact I'd call it a monstrous design success!)

If you want less compromise, ie. a 100Mhz 'scope with a very sharp cutoff filter for higher frequencies, then you have the option to pay a more money.

Or ... buy a 200MHz scope to look at your 100MHz signals - it's probably cheaper than paying for that filter.

Welcome to "reality". Please wipe your feet before entering.

[pascal_sweden]

According to all your "uncertainties" about not knowing for sure if there is a higher frequency or not, that suggestion about buying a 200 MHz scope should not hold either to be consistent with your "uncertainty". Because your "uncertainty" can not guaranty either that there is not a higher frequency than 200 MHz in the signal path, although you are measuring 100 MHz signals.

So I believe it is not me who should enter "reality", but you guys who should leave "uncertainty"
There are no things such as ghosts caused by noise, interference or higher harmonics that cause unexpected higher frequencies in the signal path that are higher and close enough to the frequency of interest which you are trying to measure.

And when Nyquist says that the sampling frequency should be two times, I believe that 2,5 is already an extra margin. But the "uncertain" guys here want to have a whopping margin of 10x.
No wonder that you need to reach out in your pocket for a scope which is way over dimensioned for the kind of electronic toys which you are designing

[Fungus]

According to all your "uncertainties" about not knowing for sure if there is a higher frequency or not, that suggestion about buying a 200 MHz scope should not hold either to be consistent with your "uncertainty". Because your "uncertainty" can not guaranty either that there is not a higher frequency than 200 MHz in the signal path either, although you are measuring 100 MHz signals.

That's true, but:
a) I wasn't promising "certainty" with the 200MHz scope.
b) You now have 4 samples per wave and the cutoff filter is twice as sharp as before (relatively speaking). You're four times better off than you were with the 100Mhz scope.

If you want even less uncertainty than that (nb. I said "less uncertainty", not "certainty"), buy a 500MHz scope ... or whatever your budget allows.

This is what "reality" is.

Edit:

But the "uncertain" guys here want to have a whopping margin of 10x.

The general consensus is that 10x is an acceptable level of uncertainty, given the budgets and constraints people tend to work in.

The people working at the Large Hadron Collider (for example) probably don't think that's enough. They'll use even more.

[marmad]

And when Nyquist says that the sampling frequency should be two times, I believe that 2,5 is already an extra margin. But the "uncertain" guys here want to have a whopping margin of 10x.
No wonder that you need to reach out in your pocket for a scope which is way over dimensioned for the kind of electronic toys which you are designing

You seem to be unwilling (or unable) to grasp the very basics of sampling - even though they have been laid out here time and time again. The Nyquist theorem is just that: a mathematical theorem. It is not real world usage - such as trying to get an accurate image of a waveform you're trying to look at. It doesn't take much imagination to see what kind of image a waveform sampled at fs/2 is going to deliver with linear interpolation:



Yet even though the concept of 10x oversampling (which, BTW, was developed by DSO manufacturers) has been explained here more than once, you can't seem to fathom it.

[pascal_sweden]

Is 10x margin enough for sure, or only expected to be enough? Does that 10x margin still depend on having an acceptable higher order low-pass filter in the analog front-end of the scope?

Let's say I want to go for the 10x margin, and select the Rigol MSO2000 series instead.

Rigol DS2000 series:

If I go for the Rigol MSO2072A scope and patch it to a Rigol MSO2102A.
This scope has only 2 channels. So then I will have 1 GS/s per channel.
If I only expect to have 100 MHz signals in my application I would meet the 10x margin, and should not expect any problems. Is that correct?

So although the Rigol DS2000 series goes up to Rigol DS2302A, in reality the best configuration which can be used and meets 10x margin when using 2 channels at the same time, is the Rigol DS2101A. Is that correct?

If I use only 1 channel on the Rigol DS2000 series, the Rigol DS2202A makes sense, as I can then measure up to 200 MHz signal, with sample rate of 2MS/s (10x margin).

If I use 2 channels on the Rigol DS2000 series, only the Rigol DS2102A makes sense, as I can then measure up to 100 MHz signal (on both channels), with sample rate of 1MS/s (10x margin).

The Rigol DS2302A scope does not make sense, as even if I only use 1 channel, I can at max measure 200 MHZ signal, to meet the 10x margin.

Rigol DS1000Z series:

The Rigol DS1054Z can be patched to a Rigol DS1104Z scope.

The Rigol DS1104Z scope can act as a reliable 100 MHz scope when you only use 1 channel (to meet 10x margin - 1 GS/s).

The Rigol DS1104Z scope only acts as a reliable 50 MHz scope when you use 2 channels at the same time (to meet 10x margin - 500 MS/s).

The Rigol DS1104Z scope only acts as a reliable 25MHz scope when you use all 4 channels at the same time (to meet 25x margin - 250 MS/s).

[a1976888]

Hi,
Probably I'm OT but I wonder if it would be possible to group buy this oscilliscope to have a bigger discount for large quantities.
I think a lot of people here would like to buy one of these...
Any ideas?
Thanks!