azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing |,, | Logs:
juli965 has quit [Quit: Nettalk6 -]
Degi has quit [Ping timeout: 260 seconds]
<azonenberg> New dense-packed attenuator test board is at fab now, ETA end of moth ish
<azonenberg> month*
Degi has joined #scopehal
<azonenberg> High-res sonnet sim of the v1.0 and 1.1 probes are still running
<azonenberg> about 54 hours so far on this run lol. On 32 cores
deltab has joined #scopehal
Degi has quit [Ping timeout: 265 seconds]
Stary has quit [Ping timeout: 265 seconds]
Stary has joined #scopehal
Degi has joined #scopehal
electronic_eel has quit [Ping timeout: 260 seconds]
electronic_eel has joined #scopehal
Nero_ has joined #scopehal
Nero_ is now known as NeroTHz
_whitelogger has joined #scopehal
<azonenberg> Woop, first big sim finished. Second one should be done by Thursday afternoon
<azonenberg> already have another two (quite a bit smaller) queued up for after that
<azonenberg> There's still a weird dip in the sim response at around 800 MHz I'm not seeing in the real board
juli965 has joined #scopehal
juli965 has quit [Ping timeout: 256 seconds]
bvernoux has joined #scopehal
<azonenberg> electronic_eel: ping, just wanted to let you know your probes should be shipping out this weekend or so
<azonenberg> They're sitting on my shipping table now, just need to print out a cal cert and box it up
<electronic_eel> ah, nice, thanks
<azonenberg> But given the ongoing plague i try to minimize leaving the house
<azonenberg> So it might not actually get mailed until i go grocery shopping
<electronic_eel> what carrier do you plan to use?
<azonenberg> USPS small flat rate box
<azonenberg> That's what i based the KS prices off. I could do something faster like fedex/ups if you wanted to pay the difference
<electronic_eel> ah, ok, so given what you read in the press about usps right now, it may take some time
<azonenberg> but flat rate USPS is by far the most economical option
<electronic_eel> yes, I'll stick to that. no hurry
<azonenberg> (I made that decision back in April before this all happened)
<electronic_eel> fedex and so on also charge additional fees for collecting import duties. when you send it with usps, it ends up at Deutsche Post where you just have to pay the duties themselves, not additional fees for collecting them
<electronic_eel> azonenberg: you are currently running full sims for the older probe layouts, trying to match the sims to your vna results?
<azonenberg> electronic_eel: Yes, and more
<azonenberg> First thing i'm doing is running full board EM sims at fairly small mesh sizes to try and match the VNA results. They're close but not quite perfect
<azonenberg> The second thing i'm doing is simulating proposed modifications
<electronic_eel> yeah, I read that with the smaller r distance
<azonenberg> Yeah
<azonenberg> Early results suggest i should have a ~3.5 GHz probe with that change
<azonenberg> I should also be able to sell upgrades to current AKL-PT1 owners fairly cheaply. PMK overcharges for the accessories by a *lot*
<azonenberg> So if you already have tips and grounds, a new probe body actually only will cost me $50ish
<electronic_eel> you shouldn't have epoxied them, then it would be even cheaper
<electronic_eel> nylon screws?
<electronic_eel> doesn't look that nice, I know
<azonenberg> that would mean holes in the pcb or a much larger enclosure
<azonenberg> and actually earlier versions just friction fit
<azonenberg> But the tolerances on the 3d printed enclosures were too loose
<azonenberg> some of them didn't stick
<electronic_eel> if a probe fell apart during measurement it wouldn't be a professional experience for the user too
<azonenberg> Yeah
<azonenberg> i actually switched to epoxy from UV polyurethane at the last minute
<azonenberg> (after having assembled a bunch)
<azonenberg> because when i started cal'ing them the uv glue cracked on a bunch of the probes just from the pressure of inserting tips
<azonenberg> i figured if the shear strength was that low there was no way it would survive real use
<electronic_eel> ui, I'd have thought uv pu would be superior to epoxy
<azonenberg> It depends on what you are sticking to
<azonenberg> mechanical strength of the bulk glue matters, but so does shear strength on particular surfaces
<electronic_eel> yeah, proper glueing is a science in itself
<monochroma> 2part full enclosure case next time ;)
<azonenberg> So what i'm using now is AA-Bond 2104 which is a thixotropic 2-part epoxy that is filled with silica microspheres
<azonenberg> monochroma: the very early enclosure was a 2-part clamshell
<azonenberg> i had better results with the unibody
<monochroma> just need to re-engineer it
<azonenberg> I'll worry about such things once i get electrical performance better
<azonenberg> The other change i want to make for DFM reasons, although it will increase the PCB cost
<azonenberg> is changing the stackup
<electronic_eel> if you could swap out the pcb from the case, it would also allow the user to try different resistor combinations or attenuations
<azonenberg> Current stackup is a thin ro4350b core and thin fr4 core laminated together with a fr4 prepreg (and maybe another core in there too)
<azonenberg> but more specificlaly rogers on layer 1-2 only
<azonenberg> i plan to switch this to have rogers on 3-4 as well, using fr4 for 2-3
<azonenberg> 3-4 is an electrically inactive layer, but i had a bunch of flexure during reflow from CTE mismatches
<azonenberg> So a symmetric stackup should help with that
<azonenberg> i dont like the idea of using rogers for that and might look at other materials to try and find similar CTEs
<azonenberg> But we'll see
<azonenberg> it wasnt HUGE bowing but it was enough to bother me
<azonenberg> and actually that might kinda be a possibility with the next board rev
<electronic_eel> could bad flexing and internal mechanical stress also be the reason that some of the probes performed worse?
<azonenberg> not with board swapping
<azonenberg> but the new, closer spaced resistors mean almost all of the resistors will be reachable from the exposed tip end of the probe
<azonenberg> so you could plausibly replace them without removing the pcb from the enclosure
<azonenberg> It could be, yes
<azonenberg> Some of the probe to probe variability could also be due to flexing in the VNA leads and the probe cable
<azonenberg> Some recent data i did showed a surprising amount of variability when i bent the cables and jiggled stuff
<azonenberg> considering switching to higher end cables to reduce the effects of this
<azonenberg> Possibly even semi-rigid
<miek> i'm not sure there's many people who would buy the finished & cased version, then want to crack it open to mess with it?
<azonenberg> That too
<azonenberg> certainly not the pro as half what you pay for is the data
<azonenberg> NeroTHz: what do you think of using hand formable semi-rigid cable for these VNA measurements?
<azonenberg> in terms of reducing variability
<azonenberg> it might not be labeled as phase stable, but if it holds its shape then that shouldn't matter much?
<monochroma> azonenberg: that's fairly common, using rigid fixturing for VNA measurement
<azonenberg> I have some 12 inch Crystek Microwave semi-rigid cable that i bought a while ago and never really ended up using for anything
<azonenberg> i vna'd a few recently and they had the lowest loss of any of my cables
<azonenberg> although they were shorter than the minicircuits ones and i think the loss/foot was about the same
<miek> there's advice on tests you can run to check out the cables in p54 onwards
<electronic_eel> do you have good connectors that are designed for this kind of cable?
<azonenberg> electronic_eel: these are premade cable assemblies with SMA male on both ends
<azonenberg> Crystek Microwave part CCSMA-MM-086-12
<electronic_eel> ah, even better. then you don't have to worry about correctly attaching the smas
<azonenberg> specced at <0.82 dB loss to 12 GHz
<azonenberg> interestingly the center conductor is *steel*
<azonenberg> it's copper clad and silver plated so i guess all of your signal ends up in that outer layer
<azonenberg> and the steel core is just for rigidity
<azonenberg> The shell is a copper braid soaked in tin
<azonenberg> also i'm eyeing a Mini-Circuits ULC-2FT-SMNM+
<azonenberg> Very affordable, $87.95 per cable
<monochroma> ooo nice
<azonenberg> Specified to 0.69 degrees of phase stability and 0.01 dB of insertion loss variation from a 2-inch radius bend
<azonenberg> By comparison the $279 Pico (rebranded Jyebao) cables i have
<azonenberg> are specified to 2 degree phase stability and 0.1 dB
<azonenberg> And total loss of 0.7 dB at 6 GHz
<azonenberg> which is on par with the minicircuits one
<azonenberg> oh, and the pico cable is specced to 6 GHz max and the minicircuits cable is rated out to 18 GHz
<azonenberg> i think i'm gonna pick up a pair of the minicircuits cables and see how i like them
<azonenberg> i've been very happy with the quality of minicircuits products in the past
NeroTHz has quit [Read error: Connection reset by peer]
<_whitenotifier-3> [scopehal-apps] azonenberg opened issue #160: When using dual cursors on a FFT view, display in-band power (integrated from cursor 1 to 2) in the popup text -
<_whitenotifier-3> [scopehal-apps] azonenberg labeled issue #160: When using dual cursors on a FFT view, display in-band power (integrated from cursor 1 to 2) in the popup text -
<bvernoux> azonenberg, yes minicircuits are clearly a must have for the price and quality
<azonenberg> bvernoux: yeah i might still pick up one of those nice ones nero suggested at some point
<bvernoux> whih one ?
<bvernoux> which
<azonenberg> but for less than half the price, getting 10x the amplitude stability and 3x the phase stability?
<azonenberg> (vs the pico one)
<azonenberg> quite the no brainer
* miek just ordered a bunch of minicircuits cables too
<bvernoux> FLC-2FT-SMSM+ is very nice
<miek> the cheap sma gauges have hidden costs :p
<azonenberg> miek: lol
<azonenberg> My main workhorse cable these days is FL086-24SM+
<bvernoux> yes for 20USD ;)
<bvernoux> I plan to buy some too
<bvernoux> as they are cheap and good for the price
<bvernoux> FLC-2FT-SMSM+ is much more better but 4 time more expensive ;)
<azonenberg> bvernoux: how is it better? it's actually slightly lossier
<azonenberg> and a lot fatter
<azonenberg> If you don't need phase stability i don't see it being a good choice
<bvernoux> yes it is very stable
<bvernoux> and flexible
<bvernoux> mainly for phase stability and freq > 18GHz too
<bvernoux> I'm using it with my KC908 ;)
<azonenberg> ah ok yeah it's rated to 26 not 18
<azonenberg> my VNA goes to 6 and my scopes top out at 2 and 4
<azonenberg> so 18 vs 26 doesnt matter to me
<bvernoux> you know the fun with a cheap UWB-2 antenna specified from 2.8GHz to 11GHz
<bvernoux> I can listen FM radio ;)
<azonenberg> lol
<bvernoux> at -80dBm ;)
<azonenberg> Looool
<bvernoux> as the noise floor is -125dbm
<bvernoux> totally crazy radio especially with latest fimrware 1.2.x I can use the 40dB amplifier ;)
<azonenberg> also i just heard back from lecroy service about my other scope. They confirmed there was a problem (duh), removed the main acquisition board from the scope, and it's in the factory area for an unspecified component level repair
<azonenberg> Presumably swapping the CH4 front end attenuator and possibly other stuff too
<bvernoux> ha and it is in repair ?
<azonenberg> expected to ship late this / early next week
<bvernoux> ha ok
<azonenberg> This is the 2 GHz scope with the flaky channel 4, you remember that?
<bvernoux> you have sent them the whole scope ?
<bvernoux> yes I remember I have seen a tweet about that
<bvernoux> few weeks ago IIRC
<azonenberg> So back in january i bought a retired HDO9204 from TRS-RenTelCo. The LeCroy factory warranty was expired
<azonenberg> but TRS provided a 1-year warranty for it with the sale
<azonenberg> A few weeks ago i noticed channel 4 wasnt giving correct results at one particular gain setting
<azonenberg> Sent it back to TRS because the warranty is through them
<azonenberg> They determined it was beyond their ability to repair in house and sent it to lecroy
<azonenberg> I could have sent it to lecroy myself but then i'd have had to pay for the fix since it's out of warranty, instead of making TRS foot the bill
<bvernoux> yes
<bvernoux> it is why there is such delay
<azonenberg> no idea if it's gonna have to bounce through TRS for the return shipment too or not
<azonenberg> On the plus side, i finally got the software options for my 4 GHz scope that should have been installed from the factory :p
<bvernoux> I'm waiting final descision from Tek USA about the Tek 6 Series refurbished for a "very very" good price ...
<bvernoux> But I'm pretty sure they will decline
<bvernoux> as they have provided a counter offer in worst case with a Tek 5-Serie ...
<bvernoux> But it is clearly not the same beast ...
<bvernoux> 6.25GSPS Serie-5 vs 25GSPS Serie-6 ;)
<bvernoux> also noise is 4x time bigger on Serie-5
<bvernoux> as it use the older ADC ASIC ...
<bvernoux> azonenberg, do you think they limit the BW also on 50 Ohm input mode ?
<bvernoux> as it is clear they switch by SW for 1MOhms/10Mohms but I have not checked if that affect also 50Ohms mode ....
<bvernoux> depending on the path ...
<azonenberg> you mean in lower end models?
<azonenberg> yes absolutely
<bvernoux> in all models ;)
<azonenberg> normally the 1M mode bandwidth is only a few hundred MHz
<bvernoux> those who have 50OHms mode
<bvernoux> to know if they limit both normal probe 1MOhm and 50Ohms ?
<bvernoux> as 50 Ohms is more direct to the chip ADC ...
<bvernoux> for tek anyway all is integrated in their asic ;)
<bvernoux> there is not just an AD or TI chipset in the chain which is here to limit the BW over I2C/SPI ;)
<bvernoux> like Rigol scope ;)
bvernoux has quit [Quit: Leaving]