AFG3000: Add support for arbitrary waveform generation

[mgunyho]

Hello again.

This PR adds support for uploading arbitrary waveform data to the AFG3000 series devices. It also contains an example notebook demonstrating this functionality. Tested on a real AFG3252.

I have one comment and one question:

• The waveform data is expected to be in the range 0..1, which is then scaled to the range voltage_low1...voltage_high1. Now, for example if voltage_low1 is -0.5V and voltage_high1 is +0.5V, 0.5 gets mapped to 0V which might be a bit confusing. But I think this is still the most sensible way to do it if the voltage range is asymmetric.
• Should we check for overflow in the waveform data? Now if the data goes outside of 0..1, it wraps around, which is probably undesired. Should we issue a warning or an exception?

Note that the commit history also contains a failed attempt at reading back the waveform data from the AFG memory, mostly as a curiosity.

[astafan8]

please disable the execution of the notebook, for details on how to do it see here https://qcodes.github.io/Qcodes/examples/writing_drivers/Creating-Instrument-Drivers.html#Documentation

[astafan8]

@mgunyho

Now, for example if voltage_low1 is -0.5V and voltage_high1 is +0.5V, 0.5 gets mapped to 0V which might be a bit confusing. But I think this is still the most sensible way to do it if the voltage range is asymmetric.

could you explain why? in particular, why 0.5 gets mapped to 0 and not to 1? i thought that +0.5 is on the right side of the range, hence it should be mapped to 1, no? or perhaps there's a typo in your example, perhaps you meant "voltage_low1 is +0.5V and voltage_high1 is -0.5V" (low votlage is higher than high voltage)?

[mgunyho]

So the values we upload to the waveform are in the range 0..1, so 0.5 (note: not a voltage!) is in the middle of the range. The center of the range -0.5V..+0.5V is 0V, so that's why 0.5 is 0V. (Maybe using +-0.5V as an example range was confusing.)

[mgunyho]

please disable the execution of the notebook

Ach, I always forget that! Done.

[astafan8]

So the values we upload to the waveform are in the range 0..1, so 0.5 (note: not a voltage!) is in the middle of the range. The center of the range -0.5V..+0.5V is 0V, so that's why 0.5 is 0V. (Maybe using +-0.5V as an example range was confusing.)

aah! understood, thank you! then it's all correct :)

but do you think it would be beneficial to also add function that would take an array of actual voltages, and then normalize it using voltage high/low settings and send the normalized 0..1-ranged data to the instrument? perhaps this will be less confusing for users (even though they still have to be aware of voltage high/low settings that they are using)

[codecov-commenter]

Codecov Report

Merging #132 (650af77) into master (fe59328) will decrease coverage by 0.02%.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##           master     #132      +/-   ##
==========================================
- Coverage   19.24%   19.22%   -0.03%     
==========================================
  Files         123      123              
  Lines       14629    14650      +21     
==========================================
  Hits         2816     2816              
- Misses      11813    11834      +21     

Impacted Files	Coverage Δ
...codes_contrib_drivers/drivers/Tektronix/AFG3000.py	0.00% <0.00%> (ø)

📣 Codecov can now indicate which changes are the most critical in Pull Requests. Learn more

[mgunyho]

add a function that would take an array of actual voltages

Hm, this sounds like a good idea, I'll do that later today or tomorrow. Let's not merge before that.

[mgunyho]

Should we then skip the 0..1 range completely? So we would have upload_waveform_raw() that uploads raw code values 0..16382 and upload_waveform that directly takes voltages and does the rescaling. And then I think it would make sense to throw an error if voltage values are outside of the voltage_low/high.

[astafan8]

Should we then skip the 0..1 range completely? So we would have upload_waveform_raw() that uploads raw code values 0..16382 and upload_waveform that directly takes voltages and does the rescaling. And then I think it would make sense to throw an error if voltage values are outside of the voltage_low/high.

agree :)

[mgunyho]

I started implementing this, but I realized that maybe passing the values as voltages doesn't actually make sense. When you upload a waveform, it's not associated with either of the channels, so we can't know what the voltage limits are. It's possible to output the same waveform on channel 1 and channel 2 with different limits by setting voltage_low1 to be different from voltage_low2! So I think this 0..1 solution is the least confusing way to do it.

[mgunyho]

I added the bounds checks, IMO this can be merged now.

[astafan8]

So I think this 0..1 solution is the least confusing way to do it.

indeed, that's fair!

[astafan8]

could you fix the mypy errors https://github.com/QCoDeS/Qcodes_contrib_drivers/runs/6573274474?check_suite_focus=true ? they are largely caused by typos or applying operations to a "List[float]" that it does not support (a solution could be to np.array(that_list_of_floats) before doing the operations)