Don't see expected bandwidth in AD2

[miv2k]

Hi,

I've been playing with Analog Discovery 2 (totally new to it, so please forgive me if I ask stupid questions) and noticed that when a 5MHz square wave from WG1 was scoped by a 1x BNC probe on CH1, it looked like a triangular wave. That was somewhat unexpected to me, since the specs call for 30+MHz bandwidth @ -3dB.

I then dug out an AD2 Reference Manual, and on page 30 of it saw how they demoed AWG spectral characteristics by running Network Analyzer in WaveForms by feeding WG1 directly to CH1, and their plot was also suggesting -0.5dB @ 5MHz and -3dB @ 20MHz.

I did the same, and got way different results: -0.5dB @ 2MHz and -3dB @ 5.5MHz for 1x probe and even more bizarre plot for 10x.

Why are my results so much worse than the specs? What may I be doing wrong here? All screenshots are attached.

Thanks!

 

This is how how a 5MHz square wave looks when scoped by enclosed BNC 1x probe:

This is characteristic measured by 1x BNC probe:

This is characteristic measured by 10x BNC probe:

This is what AD2 reference manual calls for when BNC probes are used ("Figure 21 shows the typical spectral characteristic of the AWG. In the first experiment (solid line), a coax cable and a Digilent Discovery BNC adapter were used to connect the AWG signal to the Scope inputs")

Edited December 16, 2022 by miv2k

[attila]

Hi @miv2k

Various lines attenuate signals differently.
See BNC with 0R and 50R , 1X probe with 50R and 0R output

[miv2k]

Thanks for a quick response.

I'm trying to understand the legend in your screenshot. How was the orange plot obtained? I had a 1x probe connected to CH1 in DC coupling mode going straight to BNC of WG1 with 50R jumper.

[attila]

Hi @miv2k

The R1 is with BNC-BNC cable and 0R jumper on the BNC adapter board's AWG.

[miv2k]

Oh, you mean you skipped probes altogether and just used a BNC to BNC cable...

Here's my capture with 1x probe going to WG1 0R, and I think it matches your green plot.

Should I blame then the oscilloscope probes that came with AD2 for the reduced frequency range?

I then fail to understand marketing material... "When used with BNC probes the adapter increases the oscilloscope bandwidth from 9MHz to 30MHz". Even if we say that the bandwidth is given by -3dB cut-off (which for measuring equipment is kinda extreme), you and I are seeing only 8MHz with the BNC probes, far cry from 30MHz.

Edited December 16, 2022 by miv2k

[attila]

Hi @miv2k

The 30MHz+ BW is for the Scope inputs and not for the AWG outputs, these are 9...12dB, see:
https://digilent.com/reference/test-and-measurement/analog-discovery-2/hardware-design-guide#scope_spectral_characteristics
https://digilent.com/reference/test-and-measurement/analog-discovery-2/hardware-design-guide#awg_spectral_characteristics

[miv2k]

Thanks for taking your time to answer my questions. I don't have another waveform gen capable of 5-30MHz, but I'll see if I can come up with 5-10MHz square wave and check if it looks better on AD2 scope than the one in the first screenshot. If it does, then yeah, it will confirm that I'm hitting AD2's WG limits here, not the scope.

[miv2k]

Oh yeah! Tried 74HC14N to generate a square wave, and at 6MHz with 10x probe it looks way more square than the one from WG. Even some ringing. Good to know the limits.

[engi]

Why is the ringing only at the high level? Is there an impedance matching issue?

[miv2k]

IDK. What could be the reason?

Initially I thought that ringing could be because I just hacked it on a breadboard with long leads on the feedback resistor, but now I fed the generated signal to another trigger in the same chip and scoped right from the second trigger's output pin. I still see ringing. Higher frequency too now. Probably because I'm not loading feedback with the probe's capacitance anymore.

 

Edited December 17, 2022 by miv2k

[attila]

Hi @engi @miv2k

https://en.wikipedia.org/wiki/Ringing_(signal)
https://en.wikipedia.org/wiki/Overshoot_(signal)#Electronics
https://en.wikipedia.org/wiki/Step_response

[miv2k]

Well, the ringing as a concept is no secret, but what's causing it here? Inductance in breadboard? Too fast a transition?

For the heck of it I removed crocodile ground wire from the probe and added a thing like below, but it didn't change the situation much.