<d1b2>
<Hardkrash> I've looked over the schematic. - Q0403 has drain and source swapped.
<d1b2>
<Hardkrash> - For U401 you will want to power VCONN in case some cables have e-markers that use VCONN power. The chip should auto detect Ra and provide VCONN power. - Do you have the pinout for the mating connector of J401, to confirm the SS RX/TX and cables crossover? Is a 0 Ohm jumper to switch polarity reasonable? or 1x3 header with shunt jumpers? - Consider 6pin with FTDI compatible UART header. GND -> Pin 1, TX (cable to board) -> Pin 4, RX (Board to
<d1b2>
cable) -> Pin 5 - Verified common availability of 12V DC adapter in 2.5mmx5.5mm barrel plug.
<d1b2>
<Hardkrash> @david.lenfesty I reviewed the schematic at git SHA 4f517bd85a342a90e4deb5d205368f55694bac0f. I like the PNP with the base to ground to control the -7V. The only issue is the swapped source and drain on Q403.
<_whitenotifier>
[starshipraider] azonenberg pushed 1 commit to master [+0/-0/±1] https://git.io/JLFCQ
<_whitenotifier>
[starshipraider] azonenberg 85e2273 - AKL-PT2 v0.5c with extra vias
<azonenberg>
Got pricing back from Multech for the AKL-PT2
<azonenberg>
$1240 for 500, $975 for 250, $705 for 50
<azonenberg>
Bare boards, no assembly
<azonenberg>
This is using Panasonic F-775 polyimide (same as oshpark prototypes) but Shengyi SF305C coverlay instead of soldermask
<azonenberg>
I'm thinking 250 is the way to go
<azonenberg>
50 is only a little bit less, the cost per board is much higher that way
<azonenberg>
250 to 500 gives a lot more boards but I'm a long ways from being ready to sell 250 of them
<azonenberg>
If i do end up selling 250 i'll have easily made enough from doing so that i can afford another pcb fab run
<azonenberg>
lain, monochroma: thoughts?
<lain>
I agree, 250 sounds most sensible
<d1b2>
<Hardkrash> @david.lenfesty discovered that U404 will conduct from -7V to GND. Suggest using a bidirectional diode configuration. e.g. DPN 283-STS232120B301CT-ND or the STS232150B251 at a slightly higher voltage for revers breakdown.
<azonenberg>
Ok, ordered 250 AKL-PT2 v0p5c boards
<azonenberg>
Between now and when they arrive, i need to update the mold design for the stiffeners. I believe i can get away with only respinning the bottom half of the cavity
<azonenberg>
the top shouldn't change
<azonenberg>
Also ordering a minimum sized (100 unit) reel of the 200 ohm FC0402 resistors. I'm almost out, and paying $1.92 a pop in ten-packs seems a bit silly when I can get them for $1.47 per hundred
<azonenberg>
The price drops to $1.26 per 300 but i don't think i need that many quite yet
<azonenberg>
sorry not per hundred, per unit @ qty 100
juli966 has joined #scopehal
_whitelogger has joined #scopehal
_whitelogger has quit [Ping timeout: 260 seconds]
_whitelogger_ has joined #scopehal
<azonenberg>
woop, components for the diffprobe are out for delivery. Took their sweet time, but not a big deal because the PCB is a ways out
<azonenberg>
the amplifier board is currently expected to ship wednesday, and i believe the tip hasn't even gone to fab yet
<azonenberg>
it's on the "jan 1" flex panel but i guess not enough people ordered flex over the holidays
<azonenberg>
and it didn't fill up
<azonenberg>
Paste stencil is out for delivery too
<azonenberg>
It's 10 Gbps rated vs 20. not sure if this means it only has one pair of SSTX/SSRX lanes?
<azonenberg>
that probably won't matter for the amount of current we're drawing
<azonenberg>
but it's e-marked, and right angle which is probably good for the probes
<azonenberg>
I'm assuming the amplifier board will be fairly skinny, so putting the SMA and the USB connector side by side might be tricky if we want to allow enough space for you to tighten the SMA nut
<azonenberg>
which is why i'm contemplating having the signal flow be linear with probe at ont end and SMA at the other
<azonenberg>
but the usb connector on the side
<azonenberg>
Using a right angle cable so you can have the usb and sma cables go the same direction towards the scope
<azonenberg>
Also just put in the purchase order for upgrading my Sonnet seat to L3 Gold. Which is the first edition that can model S-parameter components, and has a large enough design size cap to be able to do full EM modeling of an entire probe board
<azonenberg>
And with a third layer, i should be able to model soldermask loss on a conductor over a non-solid ground plane
<azonenberg>
lain: what do you think about that mechanical concept for the diffprobe amp? long, skinny pcb with SMPM input at front, SMA output at back, and USB power/control on side?
<miek>
i think it would be better to make the board wider and put the USB on the back, it would probably be less bulky overall & you don't need a special usb cable for it
bvernoux has joined #scopehal
<azonenberg>
miek: it would have to be side by side with the SMA, and if you want room for a torque wrench around the SMA you need about +/- 7mm of clearance on either side. Plus the width of the USB connector overmold
<azonenberg>
ballpark, it would double the width of the PCB
<azonenberg>
It could be done, no doubt
<azonenberg>
(That's +/- 7mm from both the SMA and the USB, not overlapping)
<azonenberg>
you might be able to get away with slightly tighter spacing because you dont need to be able to actually fit the wrench body between the sma and usb
<azonenberg>
you can torque it from the end, since it's not 3+ connectors in a line
<d1b2>
<david.lenfesty> @Hardkrash thanks for the review! - fixed the MOSFET - I provided VCONN through a jumper to U401, shouldn't be an issue. - It crosses over because I'm using TX for +7V and RX for -7V, the pinout is designed to match TX1 w/ RX1 and TX2 w/ RX2 so they'll be a little crossed. - I'll switch to 6 pin. - ...I don't know why I didn't think U404 would conduct at a negative voltage.
<azonenberg>
Hmm. Seems like even with L3 Gold I can't *quite* simulate the entire AKL-PT2 in one go. I'll probably have to split it in half
<azonenberg>
interestingly, if i split it in half the ram usage for the solver doesn't go in half. it seems to go down more like a quarter. i wonder if there's an O(n^2) scaling somewhere
<d1b2>
<Hardkrash> @azonenberg I looked into that cable, it is 10Gbps rated coax with the display port support so it should have all lanes. I think that the only way we will know is that cable checks need to be accounted for in the system. When manufactures specify USB 3.2x2 or USB 3.1X2 then the cable is listed that way we should know works.
<d1b2>
<Hardkrash> What lenghts of these cables are you looking for?
<azonenberg>
I was thinking ~3 feet
<azonenberg>
typical scope probe length give or take a bit
<d1b2>
<Hardkrash> ok, i know at passive certified thunderbolt 3 cables would always work. but those lengths are at the 0.8m.
bvernoux has quit [Read error: Connection reset by peer]
<azonenberg>
hardkrash: specifically looking for right angle on one end
<azonenberg>
That seems to be the tricky part
<d1b2>
<Hardkrash> Ahh
<azonenberg>
I think it's not going to be viable and i will have to design the amplifier to have the SMA and USB connectors on the same edge of the pcb
<d1b2>
<Hardkrash> That seems reasonable, I know that you wanted spacing for tightening the SMA connector, but is there a reason the USB need to be attached when tightening?
<d1b2>
<Hardkrash> is this probe side in the repo?
<azonenberg>
I guess not. The probe side is in a PR from david.lenfesty that i haven't merged yet
<azonenberg>
I planned to merge the PR once he had signed off on it as ready to go to fab
<azonenberg>
sorry misread
<azonenberg>
the host side
<azonenberg>
TiltMeSenpai said they would work on the probe side. I haven't seen progress on that yet
<d1b2>
<TiltMeSenpai> I... need to make progress on that
<azonenberg>
Lol
<d1b2>
<TiltMeSenpai> I haven't really been feeling like laying out a pcb lately and it's gonna end up like < 1 square in so it'll be a bit before I manage to fab it... that being said I shouldn't just sit on this lol
<azonenberg>
Just heard from Shahriar, he's recording the AKL-PT1 video now