Added AnalogQuantity.square_wave(). #77
Conversation
|
Seems like a good idea!
|
|
...just realised you covered that in your message. Oops sorry. Ignore me! |
|
No worries, |
|
So, I think this is probably pretty easy to get merged and would love to do so. @zakv, did you come down on a choice for how to paramterize? My 2 cents (feel free to ignore), is to use amplitude and offset. That is most familiar to me as it closely approximates the standard behavior of every function generator I've ever used. As you say, a docstring should be more than enough to stamp out any potential confusion. Another option (that hopefully doesn't require too much work) is to just have two versions: one for each parameterization. If manageable, I suppose it could be one and you just do the conversion between them internally. I'd still personally just do the former and be done with it, but if you can't decide, that's probably a good sign that both are needed. Besides, many function generators let you parameterize both ways anyhow. |
|
Sounds like that's two votes for amplitude/offset rather than level_0/level_1! I think the two-method option seems reasonable given that function generators have that behavior and a similar thing was done with I don't have a function generator handy so I think it'd be good to double check the expected behavior. Suppose the amplitude is 0.5 and the offset is 1. In that case, what would the high/low levels be? And should it start low or start high? And if the duty cycle is 0.1 it should be in the low state 90% of the time and in the high state 10% of the time (assuming amplitude is set to a positive value) right? And how should negative amplitudes be handled? |
|
Here is an example screen from a Tektronix AFG that I clipped from the manual.
It defines the amplitude as the peak-to-peak value, the offset as the DC component. So 0.5 and 1 would give a high/low of 1.25/0.75. As for start, I'd say start high since you typically code the start for when something actually happens. That can be tuned by adjusting the phase parameter (default being zero, starting high at time zero). So if you wanted it to start low, you could just set phase to pi. I think it is convention to define duty cycle as percentage on, so 10% duty cycle is 10% on, 90% off. Personally, I think it should enforce positive amplitude and leave setting negative values to appropriately setting the offset. Finally, this may be a bit too much, but you could probably add a debug kwarg that prints the equivalent high/low values. That should help users ensure what they have programmed is what they actually want without having to fire up runviewer. |
|
Actually, looking at that screen clip and thinking a moment, maybe a delay setting instead of a phase would be easier to parse for a square wave? |
…then added square_wave_levels() method which uses the old parameterization. This PR also inverts the previous meaning of duty cycle in the old parameterization. Now duty cycle is the fraction of the time spent outputing `level_0` rather than `level_1` in square_wave_levels().
|
Thanks for the suggestions! I've updated the code to have to parameterizations: For now no error is raised if amplitude is set to a negative value since that is probably an easy/convenient hack for starting LOW rather than HIGH (although you can do the same thing by adjusting the phase, you have to think for ~30 seconds about whether you need to set Currently the methods are setup with a |
commit f6409e0 Author: David Meyer <[email protected]> Date: Tue Dec 7 10:11:56 2021 -0500 Update setup.cfg to show python 3.9 support. commit 4d26344 Merge: 62455b5 deca5d8 Author: David Meyer <[email protected]> Date: Wed Nov 10 18:42:45 2021 -0500 Merge pull request #77 from zakv/square-wave Added AnalogQuantity.square_wave(). commit deca5d8 Author: Zak V <[email protected]> Date: Wed Nov 10 18:00:24 2021 -0500 Changed the parameterization of AnalogQuantity.square_wave() inputs, then added square_wave_levels() method which uses the old parameterization. This PR also inverts the previous meaning of duty cycle in the old parameterization. Now duty cycle is the fraction of the time spent outputing `level_0` rather than `level_1` in square_wave_levels(). commit 62455b5 Merge: 13239cc bda942a Author: David Meyer <[email protected]> Date: Wed Nov 10 08:53:04 2021 -0500 Merge pull request #84 from chrisjbillington/full-unitconversion-import-path Save the full import path of unit conversion classes commit bda942a Author: chrisjbillington <[email protected]> Date: Tue Nov 9 16:50:14 2021 +1100 Save the full import path of unit conversion classes This allows unit conversion classes located outside of `labscript_utils` to be used. Closes #71 commit 13239cc Merge: 83e10e5 4716ceb Author: zakv <[email protected]> Date: Fri Aug 20 16:48:15 2021 -0400 Merge pull request #72 from zakv/save-git-info Save git info commit 83e10e5 Merge: 9fbd24c ea17d77 Author: zakv <[email protected]> Date: Fri Aug 20 16:47:49 2021 -0400 Merge pull request #73 from philipstarkey/feature/performance-improvements Performance improvements commit 9fbd24c Merge: 04db899 736cb32 Author: zakv <[email protected]> Date: Fri Aug 20 08:33:57 2021 -0400 Merge pull request #79 from zakv/fix-78 Fix #78: labscript.py accidentally overwrites dedent() commit 04db899 Merge: a805b69 9915e50 Author: David Meyer <[email protected]> Date: Fri Jul 16 14:54:58 2021 -0400 Merge pull request #74 from dihm/labscript-docs Initial pass at API docs for labscript commit 9915e50 Author: David Meyer <[email protected]> Date: Thu Jul 15 10:17:58 2021 -0400 Even more docstring coverage in the API. commit a4057e5 Author: David Meyer <[email protected]> Date: Wed Jul 14 10:38:13 2021 -0400 Adding more docstrings to the API. commit 846377d Author: David Meyer <[email protected]> Date: Wed Jul 14 10:34:02 2021 -0400 Remove colorama stuff. Doesn't work on RTD anyway. Now everything will be more consistent. commit 073ceb4 Author: David Meyer <[email protected]> Date: Wed Jul 14 10:08:19 2021 -0400 Add missing dependency that sphinx should bring in automatically but doesn't on RTD. commit 08071b1 Author: David Meyer <[email protected]> Date: Wed Jul 14 09:51:36 2021 -0400 Update sphinx pin to newest stable version. commit 4b6f805 Author: David Meyer <[email protected]> Date: Wed Jul 14 09:41:26 2021 -0400 Quick hack to get a coverage percentage of the API documentation. We can't easily use the coverage extension because it counts things documented via the `undoc-members` flag as documented! Actual items that need documenting can also be printed if desired by setting the `undoc_print_objects` flag in conf.py. This is automatically done for RTD builds. commit 09091e3 Author: David Meyer <[email protected]> Date: Tue Jul 13 18:02:21 2021 -0400 Add docstrings for all of the defined functions. commit c1960c0 Author: David Meyer <[email protected]> Date: Fri Jul 9 11:45:36 2021 -0400 Clean up warning messages and dead code. commit b9be2c3 Author: David Meyer <[email protected]> Date: Fri Jul 9 11:23:00 2021 -0400 Version bump RTD theme. commit 7e086b6 Author: David Meyer <[email protected]> Date: Fri Apr 2 13:23:33 2021 -0400 Version bump sphinx build requirement to 3.2.1 commit 689bc40 Author: David Meyer <[email protected]> Date: Fri Apr 2 12:19:55 2021 -0400 Adding the connection diagram figure from Phil's thesis. commit f5b8eed Author: David Meyer <[email protected]> Date: Tue Nov 17 07:29:34 2020 -0500 Version bump sphinx to 3.1.2 This fixes issue where autodoc does not correctly handle decorated function signatures. commit 6512ac6 Author: David Meyer <[email protected]> Date: Mon Nov 16 20:17:19 2020 -0500 Version bump sphinx Fixes issue with using autosummary templates commit 0e91b09 Author: David Meyer <[email protected]> Date: Mon Nov 16 18:39:30 2020 -0500 Initial pass at API docs for labscript. This relies on recursive features and custom templates for autosummary, as described in https://stackoverflow.com/questions/2701998/sphinx-autodoc-is-not-automatic-enough/62613202#62613202 commit 736cb32 Author: Zak V <[email protected]> Date: Thu Feb 25 19:23:29 2021 -0500 Fixed issue 78 in which labscript.py accidentally overwrote labscript_utils.dedent() with matplotlib.cbook.dedent(). commit 054e281 Author: Zak V <[email protected]> Date: Thu Feb 25 18:40:41 2021 -0500 Added AnalogQuantity.square_wave(). commit 4716ceb Author: Zak V <[email protected]> Date: Thu Oct 15 05:47:31 2020 -0400 Reduced _vcs_cache_rlock contention in save_labscripts(). commit ea17d77 Author: Phil Starkey <[email protected]> Date: Thu Oct 15 18:50:45 2020 +1100 Performance improvements These small changes seem to improve compilation time significantly. Tests (done with saving of hg info disabled) indicate compilation time for a complex experiment is decreased somewhere between 25-33%. commit 903cc43 Author: Zak V <[email protected]> Date: Wed Oct 14 03:59:16 2020 -0400 Worked on some thread safety issues with vcs caching. commit 0f3f187 Author: Zak V <[email protected]> Date: Tue Oct 13 02:01:49 2020 -0400 Results from git/hg are now cached for faster shot compilation. commit 6022d97 Author: Zak V <[email protected]> Date: Sun Oct 4 12:52:10 2020 -0400 Added '--tags' to call to git describe. commit 2025755 Author: Zak V <[email protected]> Date: Sun Oct 4 12:38:10 2020 -0400 Updated git/hg error message. commit dc8d917 Author: Zak V <[email protected]> Date: Sun Oct 4 12:35:39 2020 -0400 Added "git describe --always HEAD" to git commands. commit 9a20221 Author: Zak V <[email protected]> Date: Sun Oct 4 12:30:44 2020 -0400 Removed "--verify" from git rev-parse call. commit 58ed6db Author: Zak V <[email protected]> Date: Sun Oct 4 12:28:35 2020 -0400 save_labscripts()'s calls to git/hg for a given file now run in parallel. commit df4e17f Author: Zak V <[email protected]> Date: Sat Oct 3 14:35:59 2020 -0400 Added support for save_hg_info and save_git_info options in the labconfig [labscript] section. commit e0073ea Author: Zak V <[email protected]> Date: Sat Oct 3 13:48:15 2020 -0400 Added support for saving git repo info.
This PR adds support for a
square_wave()waveform for analog outputs. It is of course already possible to achieve a square wave output usingAnalogQuantity.customramp(), but getting it right takes a bit of thought, particularly accounting for rounding errors in the timings. Therefore it's useful to have a built-in method for them.The arguments that set the output values are called
level_0andlevel_1which isn't the prettiest. I went back and forth over a few different ideas on how users should set the output levels but decided that was the most clear and simple way. I thinkamplitudeandoffsetwas somewhat ambiguous (is offset the midpoint or low level? Is amplitude peak-peak or is it half that like a sine wave?). I also thinkhigh_levelandlow_levelisn't ideal because then you'd probably want another option likestart_high=False, which can be misleading if the initial phase isn't zero, and what should be done if the user setshigh_levelto something smaller thanlow_level? Also,initial_levelandfinal_levelaren't great since the actual final output value may be equal toinitial_levelif there is a non-integer number of cycles. I'm all ears though if someone has a better suggestion for those arguments.