azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing | https://github.com/glscopeclient/scopehal-apps | Logs: https://libera.irclog.whitequark.org/scopehal
laintree is now known as lain
Degi_ has joined #scopehal
Degi has quit [Ping timeout: 272 seconds]
Degi_ is now known as Degi
tiltmesenpai5 has joined #scopehal
benishor_ has joined #scopehal
miek__ has joined #scopehal
bgamari_ has joined #scopehal
josuah` has joined #scopehal
elms_ has joined #scopehal
cybernaut has joined #scopehal
balrog has joined #scopehal
balrog_ has quit [*.net *.split]
tiltmesenpai has quit [*.net *.split]
GenTooMan has quit [*.net *.split]
elms has quit [*.net *.split]
esden has quit [*.net *.split]
miek has quit [*.net *.split]
bgamari has quit [*.net *.split]
josuah has quit [*.net *.split]
benishor has quit [*.net *.split]
JSharp has quit [*.net *.split]
elms_ is now known as elms
tiltmesenpai5 is now known as tiltmesenpai
benishor_ is now known as benishor
JSharp has joined #scopehal
esden has joined #scopehal
cybernaut has quit [Quit: Leaving]
GenTooMan has joined #scopehal
<azonenberg> also oops
<azonenberg> i just realized the AV1's charge pump regulator, the LTC1983, is 5.5V max Vin
<azonenberg> well 5.5 recommended, 6 abs max
<azonenberg> i've been giving it 6 all this time :p
<azonenberg> oops
<monochroma> D:
<monochroma> toasty
<azonenberg> well there was some loss in the cable so it probably was actually seeing more like 5.75 or something
<azonenberg> but still
<monochroma> heh
<azonenberg> anyway, i'm now debating switching to the LM27761 i used on the AD4
<azonenberg> (on the latest ad4 rev that is, untested)
<azonenberg> which is an inverting charge pump plus LDO
<azonenberg> it will simplify the power stage BOM and also result in more parts commonality between probe designs
<azonenberg> And the new AV1 layout is 1mm skinnier
<azonenberg> 22.9 x 9.5 mm
<azonenberg> and there's some blank space that i think i could cut length off of if i really wanted to
<azonenberg> the 9.5mm width thoguh is about as small as i think i can go
<_whitenotifier-7> [starshipraider] azonenberg pushed 1 commit to master [+0/-0/±4] https://github.com/azonenberg/starshipraider/compare/d94e8b5ed70f...18366dbbfb24
<_whitenotifier-7> [starshipraider] azonenberg 18366db - AKL-AV1 v0.5 with new power supply, GSG tip, clearance fixes, and passive CTLE
<_whitenotifier-7> [starshipraider] azonenberg pushed 1 commit to master [+0/-0/±1] https://github.com/azonenberg/starshipraider/compare/18366dbbfb24...d03b1f2a5013
<_whitenotifier-7> [starshipraider] azonenberg d03b1f2 - Added file missed in last commit
<azonenberg> woop, just ordered the new AV1
massi has joined #scopehal
bvernoux has joined #scopehal
bvernoux has quit [Quit: Leaving]
<darthrake> irt AV1 on the old wireless router: did you do a comparsion to a near-field probe on a SA yet and see how much (if at all) the frequency moves?
josuah` has quit [Quit: WeeChat 3.4.1]
josuah has joined #scopehal
GyrosGei1r is now known as GyrosGeier
darthrake has quit [Ping timeout: 268 seconds]
darthrake has joined #scopehal
massi has quit [Remote host closed the connection]
nfbraun has joined #scopehal
<nfbraun> Hi!
<nfbraun> I looked into 2X-Thru deembedding in the last days, but I don't really understand the assumptions it makes.
<nfbraun> AFAICT, you have to assume that each half of the test fixture is symmetrical: https://gist.github.com/nfbraun/f08f602748dd11b9268218bb41149c82
<nfbraun> The IEEE 370 standard seems to actually state this assumption, but the implementation in scikit-rf gives unsymmetrical results, even with synthetic test data where the halves are symmetric.
<nfbraun> I also don't understand why this assumption should be valid; it obviously is if the fixture consists merely of a transmission line, but if something happens at the launch from the connector into the TL, I would think it becomes unsymmetric.
<azonenberg> nfbraun: So, the assumption is that you have three identical connector+TL networks
<azonenberg> the first two are back to back, forming a 2x thru
<azonenberg> the third goes to/from your DUT
<azonenberg> assuming they are equal, given a measurement of the 2x thru you can calculate the response of just one half
<azonenberg> then de-embed that from your DUT
<azonenberg> So it's OK if there is mismatch at the launch from connector to TL as long as the mismatch is consistent and reproducible across all three copies of the circuit
<azonenberg> Error will, of course, be introduced if the three copies are not identical
HexaCube has joined #scopehal
<HexaCube> ooooooh, well who knew this place existed!
<azonenberg> o/ HexaCube
<azonenberg> so anyway, re your question... there's a 100 ohm Vishay HML01 tip resistor which is on the order of $9
<azonenberg> a bunch of jellybean passives like 1% 0402 resistors and 0.47 uF 0402 caps, let's call that no more than a dollar combined
<azonenberg> the high freq trim cap is $9 @ qty 1 on digikey (assuming you're asking about low volume BOM not production qty)
<azonenberg> output filter is $6.73
<HexaCube> oh dang, $9 for a resistor alone, impressive :D
<azonenberg> the SMPM connector is $9.38
<HexaCube> what makes it so expensive if I may ask?
<azonenberg> HexaCube: more annoying is the 100 unit MOQ
<azonenberg> it's just an ultra obscure part thats made in low volume
<azonenberg> you are basically paying them to tool up the line and put molds on machines etc just for you
<azonenberg> so it's not actually $900ish per min order = $9 each, it's probably more like $2 each + $700 tooling
<azonenberg> lol
<HexaCube> hrmmm
<azonenberg> (But i haven't inquired about higher qty pricing yet)
<azonenberg> internally its an 0402 or MELF resistor with leads soldered/welded to it and then overmolded in plastic
<HexaCube> what's the reason for picking that specific one then? low capacitance/inductance, or the fact it's a special size etc?
<azonenberg> It's an axial lead resistor that's very tiny
<azonenberg> not much competition in that space (0.9mm square or so iirc)
<HexaCube> hmm, I see. Right I seem to recall you are using those to directly be soldered to what you're measuring? Or was that something else?
<azonenberg> Yes
<azonenberg> that is indeed the use case. this is how most big name solder-in diff probes work
<azonenberg> my ground lead is extra lead wire from the resistors (they come with like 30mm of wire on them which is gross overkill, i trim to 7.5mm overall length and use a 7.5mm length of the extra wire as a ground)
<azonenberg> anyway, moving down the bom, the 20k trimmer is $3.17
<azonenberg> LM27761 is $1.68
<azonenberg> the ADP7142 is $3.40
<azonenberg> the OPA210 is $3.24
<azonenberg> and the BUF802 is $5.06
<azonenberg> so all told you're looking at on the order of about $52 in components per probe head in low volume
<HexaCube> Reason I asked is because I keep having this idea of taking the basic idea of the PCBite probes and improving upon them heh
<azonenberg> dropping to probably high 30s or low 40s in higher qty
<azonenberg> Add to that the PCB (next to nothing, under a dollar each at oshpark) for probe head and mounting foot, a 55mm piece of 20 gauge copper wire
<azonenberg> (to connect the head to the foot)
<azonenberg> and the SMA-SMPM cable which is close to $100 in low volume but drops to $60ish in quantity
<HexaCube> in which case your design is drastically exceeding anything I'd want, but I figure I could start off your design and cheapen down from there considering even 1/10th of the bandwidth is still plenty :D
<HexaCube> huh, just the cable, $100-60?
<azonenberg> Yes
<HexaCube> interesting. Then again, at work I pay like 15 EUR for a plain M12 sensor cable hah
<azonenberg> These are not cheap cables. at this price they come with serialized VNA measurements
<azonenberg> (you can cut $5-10 off the price without that)
<azonenberg> for a solder in probe, a highly flexible cable is crucial to avoid putting forces on the solder joints
<azonenberg> but it also has to not be excessively lossy
<azonenberg> micro coax tends to have high loss
<azonenberg> KF047 is the best tradeoff i've found for this to date
<HexaCube> Right, that's where I kinda wanna see if the PCBites can be improved
<azonenberg> Anyway, so you could definitely cut a bit of cost by swapping the SMPM connector out for MMCX. MMCX is only good to 6 GHz which is fine for this probe
<HexaCube> As they're generally just "dropped" onto pads, but I wonder if one could add a nice, compact way to 'preload' aswell
<azonenberg> but i'm using SMPM on all of my designs for parts commonality
<azonenberg> since the AKL-AD4 and AKL-PT5 are good to 7 and 8.5 GHz roughly
<azonenberg> MMCX is iirc what the pcbite probes use
<nfbraun> azonenberg: "assuming they are equal, given a measurement of the 2x thru you can calculate the response of just one half": But for that, you have to assume that each half of the 2x thru is symmetrical?
<azonenberg> nfbraun: symmetrical meaning S21 == S12?
<HexaCube> azonenberg: I suppose I wouldn't need the 1.8GHz lowpass either
<azonenberg> The term I usually see used for that is reciprocal
<azonenberg> HexaCube: sooo
<azonenberg> that LPF is actually for EMC reasons
<nfbraun> S21 == S12 and S11 == S22
<nfbraun> The latter bothers me.
<azonenberg> nfbraun: yeah i dont think that assumption is valid
<azonenberg> most 2x thru algorithms i've seen use time gating on the total S11/S22 response to figure out the portion coming from one half
<azonenberg> HexaCube: basically, the tip is pretty good at picking up 2.4 GHz and the amplifier is high impedance so it doesnt take much
<HexaCube> azonenberg: ahhh, you're saying otherwise you're turning the probe/cable into a 2.4GHz radiator? :D
<azonenberg> No, a receiver
<azonenberg> the -3 dB BW of the probe is less than that anyway, but strong wifi can drown out or severely interfere with weak signals being measured
<azonenberg> by explicitly filtering out the 2.4 GHz ISM band, you get much lower noise floor when near a wifi AP
<HexaCube> right, but it shouldn't matter if the 'design goal BW' is way lower, no?
<HexaCube> or am I missing something
<azonenberg> well it depends on how high BW the scope you use it with is
<azonenberg> if you have a 1 GHz probe with a 4 GHz scope that still passes 2.4 GHz you'll see it
<HexaCube> right, that's what I meant
<azonenberg> if you only ever use it with a cheap scope you can get away with more but the filter is cheap, i dont see it being worth cutting corners there
<azonenberg> nfbraun: i think S11 / S22 *magnitude* have to match
<HexaCube> I mean you said $6, that's 10% of the BOM cost (of the probe itself :P) hehe
<azonenberg> but phase does not
<azonenberg> HexaCube: yeah you could probably get away with a simple like first order r-c LPF or something
<azonenberg> made out of a pair of 0402 passives
<HexaCube> right that was what I was gonna ask next
<azonenberg> i needed a sharp cutoff to get 1.8 GHz to pass and 2.4 to be notched out
<HexaCube> I wonder if I could convince some prof at university to let me use whatever gear required to characterize a selfmade probe
* HexaCube scratches head
<azonenberg> if you're targeting like 300 MHz BW it would be a lot easier
<azonenberg> anyway, so next up
<azonenberg> my probe design as it stands uses the HML tip resistor to provide damping, there's quite a bit more L-C ringing without it
<azonenberg> whether that is enough to matter at your freq of interest is hard to say
<azonenberg> The biggest challenge though is grounding
<azonenberg> you can put a pogo or whatever at the tip
<azonenberg> but then you have to ground it somehow
<azonenberg> that's where the pcbite stock probes fail
<azonenberg> And it's a problem i have not yet solved
<HexaCube> azonenberg: right, if I remember the slides of your probe-presentation correctly (and the theory behind it) - basically the closer that resistor is to the pad/point of measure, the better yeah?
<azonenberg> Correct. that's why it's in the lead rather than on the PCB
<azonenberg> But in the case of probes with capacitive inputs it also helps damp out ringing
<azonenberg> anyway so one idea i had was to have some sort of compass-like structure like fancy diff probes have
<azonenberg> where you have signal and ground points with adjustable spacing
<azonenberg> i experimented a bit but never reached something i was happy with during testing
<azonenberg> i eventually want to make an active voltage probe, a transmission line probe, and an active differential probe in pcbite form factor
<azonenberg> i have an early prototype transmission line probe that was basically just a few resistors, a SMA, and a pogo
<azonenberg> it fit a pcbite mounting arm, cost like $15 in parts to make, and was extremely flat to 3 GHz then had some peaking you could filter off
<azonenberg> but the ergonomics were awful, it just had a copper wire dangling off the back that you had to bend to touch the pcb just right
<azonenberg> if you didnt align it perfectly it wouldnt work
<nfbraun> Ok. I saw the time gating thing as well, but that is at best an approximation, right?
<azonenberg> nfbraun: I mean all de-embedding is approximations. the reference and actual trace will never be exactly equal in geometry etc
<azonenberg> that said, the math behind it is where i get lost
<azonenberg> it took me quite a bit of study of Dunsmore to even grok basic de-embedding given S-params of the other network
<nfbraun> Of course. What I mean is that it will not give perfect results even with synthetic data.
<azonenberg> HexaCube: anyway, so i am experimenting with a bunch of ways to get similar tip mounted damping without using the solder-in resistors
<azonenberg> for a handheld or positioner based browser probe
<azonenberg> i have three main tracks i'm pursuing at this time
<azonenberg> rod-shaped RF resistors soldered into a cut probe needle
<azonenberg> a flip chip 0402 soldered into a cut probe needle
<HexaCube> azonenberg: what would one actually need in terms of gear in order to characterize a probe? A network analyzer and I guess a fast-edge generator?
<azonenberg> and a full custom tip made of machined ceramic then metallized with nichrome and copper
<nfbraun> Or at least that's what I think I found from my testing of the implemention in scikit-rf.
<azonenberg> experimental ceramic tip, not metallized (yet)
<HexaCube> wow, now THAT is fancy hah
<HexaCube> did you machine that yourself?
<azonenberg> A friend did, she had better luck than me. my first 3 attempts all snapped :p
<azonenberg> you can see its not perfect, there's some roughness in the surface finish and then the point is not quite centered
<azonenberg> because the dremel we used as a "lathe" has nontrivial runout in the spindle at these scales
<HexaCube> so I seem to remember something about carbon tips
<azonenberg> it's very difficult to machine a material that is harder than tungsten carbide and super brittle :p
<HexaCube> was that also part of your slides or am I getting something mixed up
<azonenberg> you basically have to use diamond tooling
<azonenberg> There is. LeCroy has a patent on probe tips made of composite materials consisting of resistive fibers or particles in an insulating binder/matrix
<azonenberg> the actual probe they ship uses pultruded carbon fiber impregnated with epoxy to form a 100 ohm resistor that is also a structural element of the tip
<electronic_eel> is the ceramic tip sturdy enough not to snap easily when used for probing?
<HexaCube> yeah I was wondering if you could sinter something like that
<azonenberg> The patent expires in 2025 but one way i'm considering sidestepping it in the near term is this, using a fully insulating tip with thin film resistor deposited on the surface
<azonenberg> at that point the probe is not made "substantially of" resistive material
<azonenberg> nor is it fibers or particles in a matrix
<azonenberg> so i'm clear on that patent
<azonenberg> it would also eliminate solder joints and other potential weak points
<azonenberg> as the tip is one monolithic piece
<azonenberg> electronic_eel: Good question
<azonenberg> i dont know yet
<azonenberg> this is the only one that's survived machining and it's shaped wrong to fit in a socket
<azonenberg> we're exploring better tooling to get more precise geometry and will then do some actual tests on un-plated ceramic blanks
<HexaCube> I was also gonna ask, what's stopping you from in theory directly soldering a teeny tiny resistor to a pogo-pin
<HexaCube> (usability aside)
<azonenberg> You mean like this? https://www.antikernel.net/temp/Image010_ch00.jpg
<HexaCube> exactly :D
<azonenberg> That is the other option i'm pursuing
<azonenberg> i'm exploring two different routes here, one is a rectangular flip-chip RF resistor (Vishay CH0402)
<azonenberg> the other is cylindrical rod resistors
<azonenberg> the flip chip resistors do not have metallized end caps, only that top rectangle of metal is plated
<azonenberg> so they're super weak and need glue or something to support them
<azonenberg> the rod resistors in theory would be stronger
<azonenberg> but i'm not a fan
<HexaCube> (also followup question - couldn't you put the 100R directly on the board, at least with the boards you#re making yourself?
<azonenberg> one vendor hasnt even got back to me yet
<azonenberg> the other told me they could send me two 50 ohm samples
<azonenberg> if i wanted higher values i'd have to order 250 of them at like $28 each
<azonenberg> they do not have s-parameters, mechanical strength data, or any other design ifnormation
<HexaCube> heh, fun
<azonenberg> i would have to buy the whole production lot and hope it works
<azonenberg> you see why i'm starting to explore a DIY thin film route :p
<azonenberg> at least for R&D
<electronic_eel> that they don't have any proper data on their resistors, but then want $28 each with such a MOQ is quite strange
<azonenberg> electronic_eel: yeah
<HexaCube> yeah, understsandable azonenberg
<azonenberg> i asked specifcially for data on how strong the end caps were in shear
<azonenberg> and the peel strength of the plating to the ceramic body
<azonenberg> they just said "use data for alumina"
<azonenberg> idgaf about compressive strength of the resistor body
<azonenberg> thats not where it's going to fail
<azonenberg> i want to know about the plating to ceramic bond :p
<azonenberg> anyway this is active research with multiple paths being pursued in parallel
<azonenberg> i dont yet know which way i'm gonna go
<azonenberg> For lower BW you will likely be fine with a short pogo or rigid tip
<azonenberg> this kind of stuff matters more as you get in the 5+ GHz range
<azonenberg> Keep in mind i'm trying to push limits on probing tech here. my passive probes are already competitive with the best ones on the market in terms of bandwidth etc
<azonenberg> and i'm aiming to go even higher
<azonenberg> So you can get away with less crazy designs if you're not aiming to make the highest bandwidth passive probe in the world lol
<electronic_eel> azonenberg: be careful, you might end up building your own probe tip manufacturing plant in your backyard ;)
<HexaCube> thanks for going into so much detail azonenberg =) I appreciate it
<azonenberg> electronic_eel: not happening
<HexaCube> -yet :D
<azonenberg> it would be in my friend's garage
<HexaCube> LOL
<azonenberg> this was our janky ceramics machining setup lol
<azonenberg> we moved into the porch due to mosquitos later in the evening
<azonenberg> but basically dremel held in a vise with a little indoor fountain pump squirting cooling water over the workpiece
<azonenberg> then hand holding a diamond sharpening block against the spinning ceramic rod
<HexaCube> haha that is *quite* janky considering the very tiny sizes
<HexaCube> which makes it even more impressive
<azonenberg> as a proof of concept, it worked
<azonenberg> we got experience with alumina as a material and discovered some of the limitations of this setup
<azonenberg> namely we need a spindle with lower runout and a proper tool holder with like micrometer screws or something
<azonenberg> rather than freehanding the tool
<HexaCube> takes me back tot he time where I got really interested into how to build a precision spindle
<azonenberg> the cooling/dust removal setup worked very well
<azonenberg> no airborne dust at all, no overheating, everything was cold to the touch
<azonenberg> So that's a keeper basically as is
<azonenberg> although there was splatter from the chuck jaws flying in and out of the stream of cooling water
<azonenberg> (we had a very short extension from the chuck because the mateirla was 1/16" stock and we had no way to do a live center at these scales)
<HexaCube> azonenberg: you actually might be able to cobble together a spindle with way lower runout for not too much money
<azonenberg> we were looking at a cheap chinese PCB milling thingie
<azonenberg> she actually had one we had planned to use that night
<azonenberg> but it only had collets, no chuck, and we didnt have one of the right size for our stock
<electronic_eel> does the company that made your precision pcb rework mill also make lathes?
<HexaCube> you can buy these shanks that come with collet holders, should just need some bearings and a way to attach a motor via some belt :D
<azonenberg> electronic_eel: Yes. but i'd feel bad putting a sherline through this kind of abuse
<azonenberg> water full of highly abrasive ceramic dust getting everywhere
<azonenberg> i'd rather have a purpose built tool we dont mind abusing and can easily repair/replace parts on
<HexaCube> and with the right bearings / motor you could also get to the really high RPMs needed for "proper" machining of that size
<azonenberg> yes thats the other point against a sherline
<azonenberg> they max out at like 3000 RPM
<azonenberg> if we were gonna use a premade tool it would be like a watchmaker or jeweler's lathe
<HexaCube> could probably even use a nice (oversized :P) BLDC, just gotta find a belt that can take the centripetal forces hah
<HexaCube> for a janky-but-still-decent setup, I mean
<azonenberg> Yeah. i mean the forces involved are extremely low
<azonenberg> you'd be taking off microns per pass
<HexaCube> exactly
<azonenberg> (also ping monochroma in case you want to chime in)
<azonenberg> not sure if she's awake but we'll see :p
<HexaCube> or maybe you can find a used PCB-spindle on ebay/craigslist
<HexaCube> wonder if there's actually a market for "pcbite, but more bang for the buck" geared towards... sophisticated hobbyists
<HexaCube> via tindie or something :D
* HexaCube still needs to find a way to earn money whilst studying
<electronic_eel> i once played with the idea of buying a pcb drill/mill spindle. but they usually need specialized vfds and a ton of other custom stuff, like air pressure for the toolchanger, oil, cooling water and so on. so getting it up running would have been quite an involved setup. so i decided against it
<azonenberg> electronic_eel: this was something along the lines of the circuitmedic microdrill
<azonenberg> it looks like either the same thing or a clone of it
<azonenberg> this is a brushed dc motor and an extremely good spindle
<HexaCube> because it turns out designing PCBs for private individuals SUCKS
<azonenberg> i ran mine up to like 45K RPM under a microscope with a steel blank in the chuck
<azonenberg> it looked like it was standing still
<azonenberg> i couldnt see any runout at all
<azonenberg> no idea what kind of bearings they use but they're excellent
<azonenberg> So a tool like that would do great for this
<azonenberg> HexaCube: as far as a market, good question
<azonenberg> sensepeek is very much aiming at the lower end according to my conversations with them
<HexaCube> Because I generally think the PCBite idea is great
<azonenberg> My personal plans for these probes is to keep the designs open source and sell assembled units at digikey once i'm satisfied with them
<azonenberg> yes i love the positioner. i just think they need better probes in the same form factor
<HexaCube> I just don't think the average joe wants to drop 200 bucks on just 4 100MHz probes
<azonenberg> i actually had a nice conversation with iirc their CEO and one of their engineers
<azonenberg> around christmas 2019 iirc
<electronic_eel> what i was writing about was getting a used spindle off ebay. from some pcb fab that closed down here in germany. it was geared for industrial production, so in this environment all the custom stuff was ok
<azonenberg> or was it 2018
<azonenberg> They sent me some of the special solderable mounting nuts to play with
<azonenberg> i came up with a prototype and emailed them asking to have a meeting to chat about my work to date and how we could improve it
<azonenberg> then they ghosted me
<HexaCube> Ouch :/
<azonenberg> i keep meaning to reach out and see if maybe it just got caught in spam or something
<azonenberg> but i've had my hands full with so much other stuff i didnt want to add to my workload
<HexaCube> understandably so
<azonenberg> HexaCube: anyway, so my target here would be more of very high end hobbyists / independent consultants / small businesses
<azonenberg> people who might have scored a GHz scope on ebay
<azonenberg> but can't afford a bunch of active probes to use with it
<HexaCube> wait wait am I seeing it right that the only grounding option of those PCBites probes is that grabby-arm plus a dupont wire?
<azonenberg> Yes :D
<azonenberg> lol
<HexaCube> O_o
<HexaCube> that's a mighty weird decision
<azonenberg> there is a bare copper pad on the back of the PCB where you can solder your own ground on
<azonenberg> but they provide no stock accessories to go there
<azonenberg> these are targeted at very low end hobbyists
<azonenberg> low bandwidth i mean
<HexaCube> azonenberg: at which point I find 40-50 EUR still kinda expensive - then again I keep underestimating how much it costs to make stuff like that
<azonenberg> Yeah i think the price is very reasonable
<azonenberg> anyway so for my own probes my very rough model for pricing is, look at a commercial product in the same performance class
<azonenberg> chop a digit off the end
<HexaCube> reasonable - sure; But 'low end hobbyists' don't wanna pay reasonable prices hah
<azonenberg> and i want to come in at or below that
<HexaCube> guess I should just see what I can come up with (at what cost) after I'm finally done with this damn LED-Matrix project <.<
<HexaCube> after a freaking YEAR of having to redesign the damn board because chips go in- and out of stock
<azonenberg> so for example, the ZS1500 from LeCroy is 1.5 GHz BW (i measure it as more like 1.2), 1M ohm || 900 fF, 10:1, only works with lecroy scopes, $2500ish MSRP
<azonenberg> the AKL-AV1 is 1.8 GHz BW, 5M ohms || 400 fF, 10:1, works with any scope
<azonenberg> and i figure $250 is a reasonable MSRP for it :p
<azonenberg> given BOM of $50ish for the probe head, $60ish for the cable
<azonenberg> then PCB costs, import tariffs on some of the components, shipping, labor for assembly, cost of packaging, fees charged by the retailer, etc
<azonenberg> somewhere in the $250 range should be enough to make the project sustainable (i.e. not losing money) and bringing in just a little bit of profit per sale that i can sink into R&D for the next probe
<azonenberg> Which is expensive for a hobbyist, sure
<azonenberg> but dirt cheap compared to a new name-brand active probe
<HexaCube> aye, fully agreed
<HexaCube> and sorry for asking again I think the question kinda got swamped in chat - but if I were to make my own probe, how would I go to characterize it / gain any info about performance
<HexaCube> I assume at the very least a Network Analyzer and a Fast Edge generator?
<HexaCube> to measure flatness and er... what would you cause the other one, ringing?
<HexaCube> (and well a scope to actually do measure the edge? :D)
<azonenberg> So I generally do three different characterization procedures on my scopes
<azonenberg> VNA of the probe across a 50 ohm line (port 1 = probe tip or as close as i can get on the fixture, 50 ohm termination from that line to ground, port 2 = probe output or as close as i can get to it with adapters)
<azonenberg> fast edge step response
<azonenberg> and eye pattern of serial data
<azonenberg> on my probes*
<electronic_eel> HexaCube: i think you'd need a probe fixture to go with your vna
<electronic_eel> and the fast edge generator is only helpful if you also have a scope with enough bandwidth
<azonenberg> Yeah. Fixturing is a challenge
<azonenberg> I currently use a LeCroy PCF200 but there are losses in the fixture that i cannot de-embed
<azonenberg> i want to get a fixture that i can more fully characterize with a 2x thru or similar
<HexaCube> I'll just send them to you azonenberg harharhar
<HexaCube> :P
<azonenberg> lol i mean that is an option
<HexaCube> not for a poor student like me :P But I bet I could convince an EE-Lab Prof or whatever to give me a few hours if I ask really nicely
<HexaCube> they should have all the nice gear I'd hope
<HexaCube> but first I gotta get this mess made: https://i.imgur.com/qGWKcOo.png
<electronic_eel> the fast edge generator is quite cheap, you can get them from leo bodnar or even make them yourself
<azonenberg> I generally offer steeply discounted or pro bono lab services to noncommercial/open hardware projects
<azonenberg> if you're not making money on the project i'll do the lab work at no cost unless it's a major time commitment or needs expensive consumables
<azonenberg> all i ask is you reimburse postage if you want the prototype back
<azonenberg> otherwise i'll e-waste it here at no charge
<HexaCube> azonenberg: right but shipping alone to the US is around 50 bucks or so I'd wager. Though might be way cheaper if fit into a small envelope
<electronic_eel> the high bw scope to read the results with ... not so cheap or easily made
<HexaCube> a very generous offer azonenberg, thanks and I might consider it if I actually do end up building the thing :D
<azonenberg> HexaCube: you could also order parts to my place and ship me a bare board
<azonenberg> or something likethat
<azonenberg> if its not too complex i could assemble here
<azonenberg> anyway, basically whenever the lab is otherwise idle i'll gladly make my resources available to the community
<HexaCube> or even more interestingly, do both - measure it myself AND have you measure it, so I can see how close I can get to the proper measurement
<azonenberg> such stuff is done on a best-effort basis and slotted in around billable stuff for work and my more important stuff
<HexaCube> oh, I do have one last question, just out of curiosity - on your AKL-AV1, you're using a charge-pump for the negative voltage, and then two linear v-regs right?
<HexaCube> actually nevermind I just answered my followup question myself I think :D
<HexaCube> was gonna ask 'could one use a virtual ground too' but I suppose that won't work out due to the scope's GND reference eh
<HexaCube> anways, thanks again for all the info and help =)
<HexaCube> https://sensepeek.com/magnifier I gotta say this lens look pretty nice :D frameless and all
<azonenberg> HexaCube: so for the latest AV1 i actually am using a combo charge pump + LDO for the negative rail
<azonenberg> then a separate LDO for positive
<azonenberg> so 2 chips vs 3
<azonenberg> but same block diagram, just more integrated
<HexaCube> oh that's a thing? neat
<azonenberg> Yep. LM27761
<azonenberg> it's a 2 stage regulator with an inverting charge pump that just negates the input voltage, followed by a negative LDO
<azonenberg> since there's no regulation at the first step i can actually run with less headroom than the previous design
<azonenberg> so more efficient (less loss in the LDO)
<azonenberg> And of course it costs less and saves a bunch of space on the PCB
nfbraun has quit [Quit: leaving]