Rogol 1054Z actual bandwidth

[Tepe]

The Nyquist for the DS1054Z displaying 3-4 channels  is 125 MHz.

And for the DS1104Z?

[tooki]

The Nyquist for the DS1054Z displaying 3-4 channels  is 125 MHz.

And for the DS1104Z?

Same. There’s no difference in the hardware. The 1104Z just comes with the 100MHz analog filter enabled instead of the 50 or 70MHz one.

[radiolistener]

Same. There’s no difference in the hardware. The 1104Z just comes with the 100MHz analog filter enabled instead of the 50 or 70MHz one.

100 MHz filter leads to worse aliases than 50 or 70 MHz one. So 100 MHz will have even worse aliases.

[tooki]

Same. There’s no difference in the hardware. The 1104Z just comes with the 100MHz analog filter enabled instead of the 50 or 70MHz one.

100 MHz filter leads to worse aliases than 50 or 70 MHz one. So 100 MHz will have even worse aliases.

While that is true, my answer is 100% correct. The question was what the Nyquist frequency was. And it’s absolutely unchanged at 125MHz (1/2 of the 250MS/sec sample rate), regardless of what filter is ahead of it. And there is no hardware difference between the three models — they differ only in which filter the firmware enables.

[Fungus]

The Nyquist for the DS1054Z displaying 3-4 channels  is 125 MHz.

And for the DS1104Z?

The same - it's the exact same hardware.

Remember that with 2 channels on, Nyquist is 250Mhz and with 1 channel it's 500Mhz. Hence my comment about turning off one of the other channels if something is really critical. You can always get a good display on the DS1054Z, it's just something you need to be aware of when you're using the scope with all channels enabled (ie. that any Gibbs phenomenon will be more pronounced, etc.)

Back in reality: You'll find your probing technique is far more critical for reducing artifacts than this problem is.

[rhb]

Has anyone actually determined that there are 3 analog anti-alias filters?  It seems quite unlikely.  I'd expect the change to be in the FPGA.  It certainly is in the Instek GDS-2000E scopes.

I've not sat down to work out the implications of aliasing if you stay in the time domain and don't do trace interpolation.

I did verify that my LeCroy DDA-125/LC648DLX display of a <40 ps step function does not change whether I am sampling at 2 GSa/s, 4 GSa/s or using the random Interleaved Sampling mode.

Reg

[rstofer]

Dave did a teardown and pointed out a Varicap to select between 100 MHz and 50 MHz - around 21:51

[Fungus]

Has anyone actually determined that there are 3 analog anti-alias filters?

Yes. It connects/disconnect different capacitors to the the analog input circuitry depending on the bandwidth option in the firmware.

It seems quite unlikely.  I'd expect the change to be in the FPGA.

Nope.

Dave did a teardown and pointed out a Varicap to select between 100 MHz and 50 MHz - around 21:51

He also did a reverse engineer of the input circuitry:

[tooki]

Has anyone actually determined that there are 3 analog anti-alias filters?  It seems quite unlikely.  I'd expect the change to be in the FPGA.  It certainly is in the Instek GDS-2000E scopes.

Yep. All of us that have answered you have done so based on actual knowledge of this hardware, not speculation. The DS1xx4Z series has been dissected and discussed to death here on eevblog, such that practically none of its operating principles and design is unknown.

Also, they’re not antialiasing filters per se. They're just bandwidth limits. Since all of them are below Nyquist, they’re not just there for antialiasing. Also, it’s 4 of them: 100, 70, 50, and 20MHz. The first three are used to capture consumer surplus (i.e. enable multiple price points), while the 20MHz is the user-selectable bandwidth limit.

[rstofer]

He also did a reverse engineer of the input circuitry:

To cut to the chase, look around 32:00 where Dave describes exactly how the bandwidth selection works.
This firmware selection of 50, 70, 100, 20 MHz is a settled issue.  It's shown in the schematic, it's proven by testing, it's settled!