<_whitenotifier>
[scopehal-apps] azonenberg 6056adc - Standalone multimeter serialization now works. Not yet tested with multimeters that are part of a scope.
<azonenberg>
ok i think multimeter serialization is done, other than autorange mode
<azonenberg>
but in general range control isn't fully implemented in the API so that's something i'll worry about once i've implemented those APIs
<azonenberg>
I guess that means loads and BERTs are next and i should have serialization for all the instrument types
<azonenberg>
Also i'm starting to think about how i'd implement scopevna as a filter graph
<azonenberg>
the basic idea is, feed a stairstep waveform into a function or rf signal generator
<azonenberg>
at each point, measure amplitude of the incoming signal on two scope channels, one direct and one through the DUT
<azonenberg>
digitally downconvert and filter to extract only the frequency of interest, then measure amplitude of the signal
<azonenberg>
(this is what i do already in the scopevna application)
<azonenberg>
ratiometrically calculate gain/loss of the DUT
<azonenberg>
as a vector, so you can measure group delay and phase response too
<azonenberg>
But then the output of this is going to be a single scalar measurement
<azonenberg>
so my thought is, you can trend it over time to get separate plots of freq/time and response/time
<azonenberg>
then have some kind of sampling filter that snapshots those two separate trend waveforms and makes an x/y plot or something like that. i'm not sure exactly how that will work
<azonenberg>
in particular how it will enforce things like strict monotonicity in the x axis which most scopehal waveform types (other than density plots like eye patterns) require
<Darius>
azonenberg: isn't this a TDR? ""_)
<Darius>
:)
<Darius>
I am interesting in measuring group delay at work, still ruminating about it though
<azonenberg>
Not a TDR
<azonenberg>
TDT
<azonenberg>
i'm looking at S21 not S11
<azonenberg>
The advantage of using a scope as your receiver is that you have switchable 50/1M ohm inputs which means you can measure the vector response of things that are not 50 ohm impedance on the output
<azonenberg>
e.g. active probes, or 10x R-C probes
<azonenberg>
its the only way i've found to remotely usefully measure S21 of a 10x R-C probe
<Darius>
azonenberg: OK sure, but conceptually very similar
<Darius>
luckily for me I am only measuring 50Ω stuff
<azonenberg>
yeah exactly. the other key bit is using ratiometric measurements
<azonenberg>
which means as long as your splitter is well characterized and has good isolation (or your DUT has reasonably good return loss)
<azonenberg>
you do not need a calibrated level for your signal source
<azonenberg>
or even calibrated *frequency* as you can measure that scope side
<azonenberg>
you also dont care about the scope frontend response as long as channels A and B are fairly similar (and you can even cal THAT out if you put some more work in)
<azonenberg>
basically you're making the two paths as identical as you can except for the DUT
<Darius>
it's calibration all the way down
<azonenberg>
and you only care about the shift in time and amplitude
<azonenberg>
almost nothing else in the chain needs to be calibrated is my point
<azonenberg>
you can use the cheapest function generator you have as long as it can cover the frequency range
<azonenberg>
it can be 30% off in frequency and have levels all over the place
<azonenberg>
cable loss doesnt matter as long as the two cables are equal loss or you put the effort into calibrating out the delta
<Darius>
azonenberg: don't siglent have that feature for their scopes/siggens?
<Darius>
(of course they expect you to buy their gear to do it..)
<azonenberg>
siglent has automatic level correction for the SSG5000X series if you add an external power meter
<azonenberg>
they do not actually sell power meters
<azonenberg>
they support an R&S model i'm eyeing and a few others, i forget by which vendors
<azonenberg>
I have an open feature request for them to support importing a touchstone file for a cable if you've VNA'd it and want to flatten out its loss curve
<azonenberg>
But again, with this ratiometric setup you can completely ignore that cable loss
<azonenberg>
there is some amplitude X hitting the splitter. you dont care what it is
<azonenberg>
after the splitter, it has Y dB of attenuation in the cable from splitter to reference input on the scope
<azonenberg>
then you have another cable in parallel with Y dB of attenuation going to your DUT
<azonenberg>
then the DUT, then the scope
<azonenberg>
Y doesnt matter, all you care about is (Y+DUT) - Y
<azonenberg>
and X is so far upstream it also calibrates out without even trying
<azonenberg>
the only part where it matters is setting the scope input v/div to give you good dynamic range without clipping
<azonenberg>
and you can just AGC by looking at the received signal amplitude
bvernoux has joined #scopehal
<azonenberg>
In other news, i think it's been a while since we've had a dev call. Will probably try and schedule something in november
_whitelogger has joined #scopehal
bvernoux has quit [Read error: Connection reset by peer]
<azonenberg>
@louis, lain: ryan is trying to get ngscopeclient to install on macos (as in actually using "make install") and filed a bunch of tickets related to that
<azonenberg>
i dont have a mac system to test on so please assist as possible