<Bird|otherbox>
azonenberg: pong! am trying to figure out how to get gcc to dump its own commandline to stderr or something like that so I can get a handle on what this errant command line arg it's complaining about *is*
<azonenberg>
Bird|otherbox: so that is actually the lower of the two priorities re static analysis
<Bird|otherbox>
ah, what's the bigger fish in the kitchen?
<azonenberg>
(there are also cmake commands, i forget what, to add verbosity and show what you're actually executing)
<azonenberg>
i think it's make VERBOSE=1 or something
<azonenberg>
But the big fish is that when i run analysis, it doesn't run all the way through
<azonenberg>
i get failures on VICPSocketTransport.cpp with cppcheck giving an error exit code and no output
<azonenberg>
which aborts the build just like a compile error would
<azonenberg>
I'm using cppcheck 1.86 from the debian package, have not tried building latest from source
<azonenberg>
Can you reproduce with whatever version you're using? if so, please investigate
<azonenberg>
if not, i'll try latest and see if it goes away
<azonenberg>
(building latest cppcheck from github anyway to see if it finds more bugs)
<azonenberg>
now using 2.3 dev (latest from github as of today)
* azonenberg
goes off to do something waiting for the compile
Degi has quit [Ping timeout: 260 seconds]
Degi has joined #scopehal
<Bird|otherbox>
I'm using 2.1 from Debian testing
<Bird|otherbox>
and yes I have tried make VERBOSE=1
<azonenberg>
interesting. if this current analysis run works, we might have some issues with 1.86
<azonenberg>
given that analysis is a developer tool only i don't think it's unreasonably to require a recent-ish cppcheck
<azonenberg>
What do you think of having the cppcheck detection look for 2.x only and consider 1.x-only as "not found"?
<azonenberg>
maybe log a warning if you see cppcheck 1.x only, saying "cppcheck is too old, not used" or similar
Pretzel4Life has joined #scopehal
Pretzel4Ever has quit [Ping timeout: 260 seconds]
<Bird|otherbox>
hm, I'll have to look into it
<azonenberg>
confirmed 2.3 analyzes it ok
* azonenberg
files ticket
<_whitenotifier-f>
[scopehal-apps] azonenberg opened issue #236: Analysis fails with cppcheck 1.x on VICPSocketTransport.cpp - https://git.io/JTmCR
<_whitenotifier-f>
[scopehal-apps] azonenberg assigned issue #236: Analysis fails with cppcheck 1.x on VICPSocketTransport.cpp - https://git.io/JTmCR
<_whitenotifier-f>
[scopehal-apps] azonenberg labeled issue #236: Analysis fails with cppcheck 1.x on VICPSocketTransport.cpp - https://git.io/JTmCR
<_whitenotifier-f>
[scopehal-apps] azonenberg labeled issue #236: Analysis fails with cppcheck 1.x on VICPSocketTransport.cpp - https://git.io/JTmCR
<azonenberg>
and now that analysis is running to completion, and i'm using newer cppcheck, i have a whole slew of findings to address. yay
<azonenberg>
so far two potential memory leaks, plus it doesn't like that the SCPITransport classes doesn't have operator= or copy constructors. We don't use them so i'll just add =delete definitions
electronic_eel has joined #scopehal
electronic_eel has quit [Ping timeout: 260 seconds]
<Bird|otherbox>
yeah, I'd either =default or =delete them so that you're saying what you mean
<_whitenotifier-f>
[scopehal] azonenberg pushed 3 commits to master [+0/-0/±4] https://git.io/JTm8U
<_whitenotifier-f>
[scopehal] azonenberg 15797b5 - Updated to latest graphwidget
<_whitenotifier-f>
[scopehal] azonenberg 8024f4b - SCPILXITransport: removed unnecessary copies and replaced with const_cast since this is confirmed safe w/ older liblxi. More efficient, fixes static analysis warnings and a memory leak
<_whitenotifier-f>
[scopehal] azonenberg pushed 1 commit to master [+0/-0/±7] https://git.io/JTmXS
<_whitenotifier-f>
[scopehal] azonenberg 3739d78 - Various cleanup to fix static analysis warnings. Added field length to IBISParser sscanf calls. Overflow was impossible because of earlier bounds check, but good practice anyway.
juli965 has joined #scopehal
juli966 has joined #scopehal
juli965 has quit [Ping timeout: 272 seconds]
vup has quit [Quit: No Ping reply in 180 seconds.]
<bvernoux>
still waiting the 6GHz version with 100dB dynamic range ;)
<bvernoux>
to replace/compare it with my old HP VNA ;)
<bvernoux>
which cost 30x more ;)
<bvernoux>
it include
<bvernoux>
4" display
<bvernoux>
4x faster sweep speed (400 points/s)
<bvernoux>
it start to be very very interesting and it is portable
<bvernoux>
if anyone is interested I sell my xaVNA ;)
<azonenberg>
bvernoux: i got rid of my xavna too
<bvernoux>
yes clearly garbage
<azonenberg>
not that the picovna software isn't garbage too
<bvernoux>
it is fun just to play with VNA but it cannot be used for serious stuff
<bvernoux>
for that NanoVNA 2 is clearly a must have
<azonenberg>
in particular changing the sweep range or number of points takes like *20 minutes* to recompute... something. What i dont know
<azonenberg>
(with picovna)
<miek>
wtf
<bvernoux>
especially the latest NanoVNA v2plus4 very impressive to have 80dB dynamic range!!
<azonenberg>
miek: and sometimes it gets in a mode where at about 3.5 GHz all RX's output noise rather than valid data
<azonenberg>
you can go through a whole cal procedure and everything looks good
<azonenberg>
then you start a sweep and about 2/3 of the way through the range the trace just disappears and you get noise
<azonenberg>
the only fix i've found is to close and restart the software and re-cal
<bvernoux>
what a shame for such a VNA
<bvernoux>
Pico SW are very bad anyway
<bvernoux>
Picoscope 6 also is awfull
<azonenberg>
i hope the picovna 108 (8 GHz version) and the new version 3.0 software is better
<azonenberg>
But i don't plan to find out
<azonenberg>
my next vna will not be a pico
<bvernoux>
but I think their VNA SW is even worse ;)
<azonenberg>
at least picoscope runs on linux
<bvernoux>
yes picoscope have a good API
<azonenberg>
And that
<bvernoux>
I use it to stream stuff with my own sw ;)
<bvernoux>
but it cannot be used like a normal scope with Picoscope Software
<bvernoux>
it is also why I have bought the MSO5074 ;)
<bvernoux>
as it is a REAL scope
<bvernoux>
Piscoscope are NOT real oscilloscope and they cannot really be used like real oscilloscope
<bvernoux>
the only good things is their streaming feature over USB 3.0 which kill any other oscilloscope
<bvernoux>
it is a shame that even high end oscilloscope are ultra slow and does not support streaming over Ethernet Gigabit or USB 3.0
<bvernoux>
next step is to wait VNA2 6GHz version ;)
<bvernoux>
to replace my HP VNA ;)
<azonenberg>
yeah, hopefully i can fix a lot of that nonsense with my scopes
<azonenberg>
but probes are the higher priority for me
<bvernoux>
maybe I shall start to sell it ;)
<bvernoux>
do you imagine how the instrument industry is locked in paste and ultra expensive things
<bvernoux>
it is open source stuff which are moving that
<bvernoux>
with such NanoVNA and especially VNA2 jankae if the final version can reach 100dB dynamic range from 1MHz to 6GHz (or more)
<bvernoux>
that will give a good lesson to R&S, Keysight and so on on their ultra expensive VNA
<bvernoux>
Waiting the same for high end scope at correct price ;)
<bvernoux>
and also for good Spectrum Analyzer ;)
<bvernoux>
the same for the passive probes ;)
<bvernoux>
and all those locked SW features like SPI, I2C ;)
<bvernoux>
even Rigol is providing those for free now
<bvernoux>
R&S and others are still asking something 500USD to have SPI or I2C decoder what a joke
<azonenberg>
yeah its absurd lol
<azonenberg>
I like what signalhound is doing with the 10GbE specan though
<bvernoux>
it will change soon ;)
<azonenberg>
that seems to be a very good value for the money
<bvernoux>
yes signalhound is very nice as SW part are free
<bvernoux>
it is paid in HW
<bvernoux>
yes it is clearly a must have I was planning to buy it ;)
<bvernoux>
but glurps >16KUSD
<azonenberg>
Yes but look at what R&S etc charge for a 20 GHz specan
<azonenberg>
the price is actually quite good i think
<bvernoux>
need to buy a good PC and sell my old instruments before ;)
<bvernoux>
yes it is clearly the best price when compared to other
<bvernoux>
and it is top of the art HW
<bvernoux>
R&S, Keysight and other do not do anything compact anyway
<bvernoux>
it is the other very good point to something which weight >25KG and which is very big ;)
<bvernoux>
and noisy like hell ;)
<azonenberg>
right now my long term plan for upgrades is a signalhound specan, a copper mountain VNA, and probably sticking with lecroy for scopes for a while longer
<azonenberg>
exactly what i get for scopes depends on what's available secondhand
<azonenberg>
right now i'm eyeing three scopes
<azonenberg>
if any are available when i have budget i'll consider them
<bvernoux>
check before to buy those copper mountain VNA
<bvernoux>
I have heard very bad things on them
<azonenberg>
1) TRS-RenTelCo has a 6 GHz SDA, this is the least value for the money
<bvernoux>
like my calibration kit does not support it as the Dr Kirkby say Copper Mountain VNA is crap
<azonenberg>
it's $39K for only 6 GHz
<bvernoux>
yes crazy 39KUSD for 6GHz SDA in 2020
<bvernoux>
with the old diskette ;)
<azonenberg>
2) TestEquipmentBay has two LeCroy SDA8Zi scopes currently: 16 GHz 40 Gsps for $32.95K and 25 GHz 80 Gsps for $59.9K
<bvernoux>
clearly better than 6GHZ SDA for 39KUSD ;)
<azonenberg>
Both have the same software options: 8b10b, CBL-DBED (basic de-embedding, attenuation constants only), EyeDr (advanced de-embed with full S-parameter), and SDA2 (same single-lane serial data analyzer as my waverunner)
<bvernoux>
on the other side I like Tek MSO64B ;)
<bvernoux>
I want 3 year warranty ;)
<azonenberg>
Both run Win Vista SP2 and have an i7-2600 CPU
<azonenberg>
so they cannot run glscopeclient locally (not that this is a huge problem) and are generally old CPU
<bvernoux>
miek, yes the more recent seems better
<azonenberg>
The other nice thing about the SDA 8Zi is they have both ProBus and ProLink inputs so i can use all of my ProBus and BNC probes
<azonenberg>
but also switch to ProLink (BMA + active probe interface) for full bandwidth
<azonenberg>
anyway, those are my current top candidate for 2021-2022 scope upgrades if budget becomes available, but even the 16 GHz is pushing what i might be able to afford soonish
<bvernoux>
azonenberg, on the other end you have 2 chan 50GSPS with Tek6 MSO64B ;)
<azonenberg>
i'm not done
<bvernoux>
which can clearly go to 10GHz BW easily
<bvernoux>
and is SW option ;)
<azonenberg>
The third option is the LabMaster 10Zi-36-A from SiloTest, currently listed for $42K
<bvernoux>
remember that the 1GHz basic version can be upgraded by SW to 10GHZ BW ;)
<bvernoux>
I have not found anything which that spec which can
<azonenberg>
Which is 36 GHz, so significantly more bandwidth for significantly less money than the 25 GHz SDA
<bvernoux>
yes but 40GSPS is not serious to announce 25GHz
<bvernoux>
Niquist say /2 ;)
<azonenberg>
It has a Xeon X5660
<azonenberg>
Has Cbl-dbed, EyeDr, equalization, 10GbE compliance test, and the full multilane SDA3 package
<bvernoux>
will be interesting to know what there is inside TEk 6 series
<azonenberg>
It does 80 Gsps
<azonenberg>
as does the 25 GHz SDA
<bvernoux>
yes but 80GSPS 25GHz BW seems not credible
<bvernoux>
or it is for a square wave ;)
<azonenberg>
i mean its for serdes analysis
<bvernoux>
less than 4 points per period
<bvernoux>
you cannot rebuild any signal with that except digital signal
<azonenberg>
they assume you're mostly doing eye patterns or CDR recovery of digital serial data
<bvernoux>
it is why I say advertisment of 25GHz for 80GSPS is BS
<azonenberg>
For that, the sample rate is fine (and i think they do a few hundred Gsps in equivalent time mode)
<azonenberg>
If i got any of these i'd mostly use them for 10G SERDES
<bvernoux>
for me it shall be at least 5 or 10x GSPS vs BW
<bvernoux>
to reconstruct correctly a signal and analyze it
<Degi>
I think having BW > SPS is fine too, if you know what you're doing
<bvernoux>
else it is a sampling scope ;)
<azonenberg>
anyway, so the labmaster is more bandwidth for less money
<bvernoux>
but it is for very specific case with repetitive signal
<azonenberg>
there are two downsides vs the SDAs
<bvernoux>
azonenberg, I think Tek6 B is better for the price ;)
<azonenberg>
1) the labmaster is physically much larger, it will take a LOT of rack space between the acq module and the controller
<azonenberg>
2) There are no ProBus inputs. This is a pure high speed scope, the inputs are all... looks like probably 2.92mm
<azonenberg>
So it can't be used for anything else
<azonenberg>
while the SDAs have low speed 1M inputs too so it could be used as a "daily driver" scope
<bvernoux>
check calibration price for that also
<bvernoux>
they are crazy ;)
<azonenberg>
for labmaster? yes
<bvernoux>
yes
<azonenberg>
(otoh there's also the geek points of having a labmaster if i went that route, lol)
<bvernoux>
you shall take a yearly subscription for more than 4KUSD/year ...
<azonenberg>
I also see from "apiccorp" a 45 GHz labmaster 9, so one generation older, for $57K. That's also a brand new ebay seller while the others are from 100% rated sellers with a long history
<azonenberg>
so its either a scam or somebody liquidating
<azonenberg>
not seriously considering that
<bvernoux>
yes with high risk to havec something defective
<bvernoux>
I'm done to buy expensive things on Ebay with 1 week or 3 month (fake) warranty ;)
<Degi>
What kinda display do the labmasters have on the channels? LED matrix?
<azonenberg>
I think so
<azonenberg>
The SDAs on the other hand look like very viable options, especially the 16 GHz one. $33K for 16 GHz is a pretty decent price i think for a serious scope
<azonenberg>
It would be 4x the bandwidth and 2x the sample rate i currently have on my waverunner (I'd replace the HDO9204 with it)
<Degi>
So anyways, for the ADC there is "Last Time Buy Date (date the last purchase order must be received by): 10-JUN-2021"
<Degi>
They say that the EV10AQ190 can be an alternative, but it costs at least 645 $ for QTY 5. It has 5 GS/s and 10 bits, still better than the LMK, but way pricier than HMCAD
<bvernoux>
will be short to launch a crowd founding and design it in time
<Degi>
HMCAD: 50 $ / GS/s, this one is above 100 $ / GS/s but has 10 bits
<Degi>
For 2k you could probably still get two of the 10 bit ADCs and a scope around it, thats still double the cost of rigol but it has up to 3.2 GHz BW
<azonenberg>
Degi: ad9213: $600/GS/s :p
<Degi>
(compared to HMCAD 650 MHz and Rigol 350 MHz)
<bvernoux>
Degi, the challenge is to discuss with them as maybe with good negotiation they can sell the 10 bits for something like 200USD/unit if there is a big batch like 500units
<Degi>
Heh yes, I guess the upside to that is that you dont have to parallel as many converters
<Degi>
bvernoux: Do they do that? Might be worth it, but where to get 100 k$ from
<Degi>
Ah you mean in case we mass produce scopes
<bvernoux>
Degi, from a Crowd Founding ;)
<bvernoux>
to avoid crazy risks
<Degi>
The ADC has 8 bit ENOB at 620 MHZ and 7.7 bit at 1.2 GHz
<Degi>
In 4 ch mode. In 1 ch mode, it reduces by 0.1 bit
<bvernoux>
so clearly better than MSO5000 or MSO8000 ;)
<bvernoux>
or Siglent but Siglent are far in GSPS the best they provide is 5GSPS
<Degi>
Worst case is 7.2 ENOB. Idk, my MSO5000 looks pretty much less than 8 ENOB even at 70 MHz
<bvernoux>
Degi, in HD mode it shall have more than >8bits ENOB I think at 70MHz
<bvernoux>
HD mode is just decimation ;)
<bvernoux>
if you start from 8GSPS
<bvernoux>
you win 0.5LSB by /2
<bvernoux>
and it start at 5bits ENOB @8GSPS ;)
<Degi>
Kindaaaa
<bvernoux>
ENOB increase also with lower speed
<Degi>
Ah nobody has it in stock tho
<bvernoux>
yes it is so specific use case ...
<bvernoux>
do like Olimex and buy directly at source ;)
<bvernoux>
they bought 40k CPU last time ;)
<bvernoux>
they can even build obsolete stuff for that Qty ;)
<Degi>
Can I buy 1 QTY of that IC?
<bvernoux>
they will say MOQ 5000 ;)
<bvernoux>
I doubt for such ADC as it is niche market with low volume anyway
<bvernoux>
so maybe they ask 250 min ...
<Degi>
And how to ask that u get a response in the first place
maartenBE has quit [Ping timeout: 240 seconds]
maartenBE has joined #scopehal
<bvernoux>
hehe just bought the NanoVNA V2 Plus4
<bvernoux>
I want to see if it burn my MiniVNA Pro2 ;)
<bvernoux>
so like that I contribute to open source development
<Degi>
Hmm, it costs more € though now
<bvernoux>
I will check it also to compare my Calibration Kit which cost >600Euros vs the one provided ;)
<bvernoux>
and to compare it to my HP VNA ;)
<miek>
do the various nanovnas support cal kit definitions yet?
<bvernoux>
yes
<miek>
ooh cool
<bvernoux>
anyway the ones provided are crap ;)
<bvernoux>
without any parameter I could characterize it anyway for fun
<bvernoux>
to check
<Degi>
Hmh, the S11 noise floor is still the same
<bvernoux>
yes but dynamic range is much better
<Degi>
Does it help when dynamic range > noise floor?
<bvernoux>
IIRC S11 noise floor is better vs previous NanoVNA2
<bvernoux>
of course for 129USD you do not have something equivalent to an old HP 8753D
<bvernoux>
I will show it ;)
<bvernoux>
especially to check the limit
<Degi>
Though it can go down to 10 kHz
<miek>
is that noise floor spec supposed to dBm?
<Degi>
I think its dBc?
<Degi>
At least in 2 port analyzer mode, thatd make sense
<bvernoux>
using my genuine JFW attenuator ;)
<bvernoux>
it can go from 0dB to -110dB attenuation
<bvernoux>
I have caracterized it ;
<bvernoux>
also a good test is to check with a very good notch filter
<bvernoux>
all is in dBm
<bvernoux>
S11 noise floor (calibrated): -50dB (up to 1.5GHz), -40dB (up to 3GHz)
<Degi>
dBm or dB for the noise floor?
<bvernoux>
dB
<bvernoux>
as it is delta
<Degi>
Hm yes
<bvernoux>
I do not find the spec for my HP about S11 noise floor
<bvernoux>
I know dynamic range is about 100 / 105dBm
<bvernoux>
except on low freq < 50KHz it is 90dBm
<Degi>
Hmh, not sure if the V2 is worth it vs the V1... Mostly because the noise floor in both cases is well within the dynamic range. The other upsides of the V2 are higher frequency range, lower P2 return loss at 3 GHz and more pixels and speed, basicalyl
<bvernoux>
if you check SVA1075X
<bvernoux>
the Dynamic Range is like NanoVNA v2plus4 ;)
<bvernoux>
90dB in best case
<bvernoux>
100KHz to 10MHz 60dB
<Degi>
60 dB noise floor?
<bvernoux>
here it is dynamic range
<Degi>
It has a displayed average noise level specification hmh
<bvernoux>
depending on frequency
<bvernoux>
example of spec in latest Siglent VNA 7.5GHz
<bvernoux>
the siglent VNA is clearly not the best VNA ;)
<bvernoux>
my old HP is 20dB better
<bvernoux>
also if you check SVA1000X
<bvernoux>
they can do only maximum 751 sweep points
<bvernoux>
what a joke
<miek>
dynamic range should be in dB, absolutes like noise floor should be relative to something (like dBm)
<bvernoux>
I always use 1601 points to have something accurate
<bvernoux>
yes dynamic range is in dB
<bvernoux>
it is a mistake to say dBm ;)
<Degi>
Hm yes, noise floor in dBm would make sense, though I think dBc makes sense too, if the carrier causes an increase in noise
<miek>
a noise floor relative to the carrier is basically dynamic range :p
<Degi>
Lile -50 dBm noise floor and if the specan can do 30 dBm output would give 80 dB of noise free range, where 80-90 dB dynamic range would make sense if I understood that correctly
<Degi>
Yes, thats what im wondering about
<Degi>
When they say -50 dB noise floor and 90 dB dynamic range, what does that mean
<Degi>
That the ADC has enough bits to resolve 90 dB?
<Degi>
For example a 16 bit ADC would have 48 dB range
<miek>
yeah.. i don't really know. i think some of it is extra range from averaging
<miek>
but still it'd have to take >30dBm input power for that to work, which seems unlikely :p
<Degi>
Assuming it is dBm
<Degi>
On the other hand its S11...
<Degi>
Maybe P2 has 90 dB noise free range?
<Degi>
Like for example for testing filters
<miek>
hm yeah good point
<Degi>
And S11 / Port 1 reflectance meter only has 50 dB
<Degi>
Can it be used as a spectrum analyzer? bvernoux
<bvernoux>
important point is also max input level
<bvernoux>
azonenberg, yes it was so big that it was douable to measure it
<bvernoux>
Degi, I doubt you can push easily your rubidium to generate 1GHZ signal
<bvernoux>
Degi, it is why it is so good
* Degi
gets out hydrogen maser
<azonenberg>
bvernoux: I have a fairly nice Crystek SAW resonator to produce a 1 GHz tone
<azonenberg>
thats probably my best oscillator right now
<bvernoux>
Degi, the issue is not to have a very nice clean signal but to multiply it and reach 1GHz
<Degi>
Is 1,420,405,751.786 Hz close enough?
<azonenberg>
CCSO-914X-1000
<bvernoux>
azonenberg, will be interesting to compare it to the spec of SignalHound 1Ghz Phase Noise Clock Standard
<azonenberg>
bvernoux: this is -146 dBc/Hz at 10 kHz
<Degi>
The SignalHound is -147
<Degi>
Though its -149 at 1 MHz while the Crystek one is -171
<azonenberg>
my only complaint is the second harmonic is typically -20, max -15 dB from fundamental
<azonenberg>
so i added some filtering to get rid of that
<bvernoux>
also signalhound have +10dBm amplitude
<bvernoux>
I doubt Crystek have that
<Degi>
But at 10 Hz the Crystek one is worse. I think the low frequency parts can be handled by comparing it to a rubidium and adjusting it accordingly
<azonenberg>
The crystek does +8 dBm
<bvernoux>
ha +8dBm not bad
<Degi>
How much phase noise does amplification add?
<azonenberg>
The big issue is it's not temp compensated
<Degi>
Can u use the crystek one in a PLL loop
<azonenberg>
it can vary 150 PPM across -40 to +85C
<azonenberg>
Degi: It's not voltage controlled but they do have a version that is
<bvernoux>
yes
<bvernoux>
150ppm is huge ;)
<azonenberg>
but at a controlled temp it's super stable
<azonenberg>
I got it for stability, not absolute accuracy
<Degi>
You could lock that to a rubidium and set the PLL to 10 kHz BW and feed it with a rubidium, that should make it better than the SignalHound one
<bvernoux>
signal hound ones is temperature controlled