azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing | https://github.com/azonenberg/scopehal-cmake, https://github.com/azonenberg/scopehal-apps, https://github.com/azonenberg/scopehal | Logs: https://freenode.irclog.whitequark.org/scopehal
electronic_eel has quit [Remote host closed the connection]
electronic_eel has joined #scopehal
Degi has quit [Ping timeout: 260 seconds]
Degi has joined #scopehal
juli964 has joined #scopehal
<azonenberg> So i heard back from Sonnet support
<azonenberg> no definite answers yet but i'm trying to track down the sources of my sim-hardware mismatches so i can do better modeling of connector launches in the future
<azonenberg> we're probably going to have a call tomorrow to discuss in more depth
<azonenberg> in the meantime, i managed to import the actual kicad geometry (with some simplification of vias down to big boxes) into sonnet and i'm trying to model the whole thing
<azonenberg> the first attempt wanted to use 10 GB of RAM, my edition is only good for 128 MB. I managed to get it down to 111 MB with some finagling
<azonenberg> So we'll see if this looks like my original model or if it's closer to what i saw in the VNA
<azonenberg> Also continuing to make progress on the TCP code. I'm hopeful that within a day or so i will have "optimistic TCP" functional
<azonenberg> i.e. it won't retransmit dropped packets, but will work fine as long as you never drop anything
<azonenberg> not useful for production of course, but will be good enough for preliminary bringup testing
<azonenberg> latest sonnet model of the test fixture compared to the real VNA data
<azonenberg> it's the right shape but shifted
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
<azonenberg> Not sure if due to mesh size differences or what
bvernoux has joined #scopehal
<azonenberg> bvernoux: so i just got off the phone with an apps engineer from Sonnet, we spent almost an hour talking about various stuff
<azonenberg> Definite progress, and i have a model that is closer to reality now
<bvernoux> ha great
<azonenberg> but there's still a lot of experimenting and tweaking to do before it's really where i want
<bvernoux> does the model has changed ?
<azonenberg> VNA vs reality. Sim is wildly optimistic, by about 10 dB, vs the real world
<azonenberg> Sim is not modeling the connector itself which may be part of it
<bvernoux> ha ok interesting
<azonenberg> I'm thinking of picking up some ARF2500-ND
<azonenberg> see how those compare to the samtec part i've been using
<bvernoux> why you do not want to use the one I'm using ?
<bvernoux> they are cheap and nice
<azonenberg> which one is that?
<bvernoux> PowellInc 142-0771-831
<bvernoux> 4.1USD per unit
<bvernoux> with full spec
<azonenberg> i see them on digikey for $17.93
<bvernoux> BelFuse
<azonenberg> the amphenol one i linked is only $11.73
<bvernoux> just buy them at Powell ;)
<bvernoux> they are at 4.1USD per unit
<bvernoux> with VAT
<azonenberg> yes that is the same one digikey sells for $17
<bvernoux> I have bought 100 with that price
<bvernoux> but for 50 price is 4.87USD
<bvernoux> per unit
<bvernoux> so quite cheap for excellent connectors the price of 6GHz SMA on DigiKey ;)
<bvernoux> and the price was higher for me as it is imported from USA
<bvernoux> yes exactly the same sold for 17USD on DigiKey
<bvernoux> Genuine
<bvernoux> With Bell Connectivity package LOT CODE ...
<bvernoux> see picture
<bvernoux> each packet contains 25 units
<bvernoux> the only thing which is tricky is placement
<bvernoux> exactly the same but placement shall be simpler
<bvernoux> and I suspect the price is the same when bought from Powell Inc
<bvernoux> hmm 142-0771-821 version is 6USD min at Powell
<bvernoux> So I will probably keep the other ;)
<bvernoux> 142-0771-831
<bvernoux> I need to do more tests anyway
<bvernoux> I have received the glue and I shall assemble my TRL Board v0.1 using OSHPark FR408HR 4Layers
<azonenberg> I think i'm gonna try the amphenol one and see how i like it. yes, it looks like placement might be a bit tricky
<azonenberg> But it depends on how snugly it fits
<azonenberg> The samtec part will not move, at all, once you slide it on, it seems to be a slight interference fit
<azonenberg> so i put it on, jiggle it a bit to make sure it's centered, then just push on
<lain> wiggle wiggle
<bvernoux> my issue is also OShPark PCB are too thin ;)
<bvernoux> expected to be 1.6mm and they are 1.5mm ;)
<bvernoux> 1.56mm in fact ;)
<bvernoux> OShPark specify 4 Layers to be 63mil (1.6mm) nominal
<bvernoux> but they do not provide tolerance
<bvernoux> and I see clearly 0.04mm less than 1.6mm
<bvernoux> so 15mill less is not good for the thickness and to connector SMA as they are loose...
<Degi> Hmh yeah, I think its like 0.05 mm tolerance? At least at jlc
<Degi> Oh, its +- 10%
<Degi> JLC says "min. BGA 0.3 mm" but 0.8 mm kinda pushes their limits?
<Degi> Maybe for single layer
<Degi> Nice @TCP
<Degi> I think it should be possible to make the pos. and neg. rails with a two or three winding transformer instead of an inductor and a synchronous PWM buck converter
balrog has quit [Ping timeout: 265 seconds]
balrog has joined #scopehal
<bvernoux> 4 layers Tolerance are here https://docs.oshpark.com/services/four-layer/
<bvernoux> As i speak about OSHPark 4 Layers board
<bvernoux> if you cumulate each tolerance which are minus final result is not -15mill ;)
<bvernoux> final value is expected to be -6.84mil
<bvernoux> worst case
<bvernoux> in fact i obtain worst case 1.5mm so 0.1mm off it is in tolerance about -4mil
bvernoux has quit [Quit: Leaving]
futarisIRCcloud has joined #scopehal