<bvernoux>
I have done some small improvements on it
<bvernoux>
The case is quite expensive as it requires a CNC and I have launched it on PCBWay and it cost >130USD with Aluminum 7075 Bead Blast
<tnt>
bvernoux: is that 130USD per case ?
<bvernoux>
yes 130USD for the full case top+bottom
<bvernoux>
For those who have a CNC it shall not cost a lot
<bvernoux>
I have not found anything cheaper online in fact with Aluminum 7075 Bead Blast
<bvernoux>
Also with threads
<bvernoux>
jankae have done an amazing job (again) on it
<bvernoux>
I will share my design modifications which are really minor
<bvernoux>
I plan to use special Via Covering with Copper paste Filled & Capped
<bvernoux>
for the PCB
<bvernoux>
To improve temperature spreading like in heat pipe
<bvernoux>
it shall also improve isolation and RF signal integrity in theory
<azonenberg>
there's something really weird about seeing a usb-c connector and a spartan-6 on the same board
<azonenberg>
it's like seeing a BGA and a uv eprom in the same place
<bvernoux>
spartan-6 ?
<azonenberg>
(the actual vna not the ecal)
<bvernoux>
ha yes
<bvernoux>
I do not plan to by the LibreVNA the performance are not good enough and frequency a bit too low (6GHz max)
<bvernoux>
by->buy
<bvernoux>
But the LibreCAL is clearly amazing and such things (2 Ports version) cost more than 1600USD
<bvernoux>
here we have 4 ports ;)
<azonenberg>
Yeah if it's any good :p
<bvernoux>
the BOM is not cheap anyway
<bvernoux>
PCB+Components+Assembly cost more than 150USD per unit
<bvernoux>
So 1/10 of the price of any eCal
<d1b2>
<david.rysk> 6GHz is pretty capable for lots of applications... but why an S6?
<bvernoux>
the problem is not only 6GHz the dynamic range and noise floor is not very good on LibreVNA
<bvernoux>
really worst than on my old boat anchor HP 8753D
<azonenberg>
yeah s6 is ancient tech
<azonenberg>
there's absolutely no reason to use it in a modern design
<azonenberg>
they were obsolete when i was in grad school
<bvernoux>
the FPGA was planned to be replaced by something more recent
<bvernoux>
but anyway that will do not change anything on the performance as everything shall be redesigned to have performance similar to PicoVNA ...
<bvernoux>
anyway so far I'm interested only in the eCal for my different VNA and to avoid exploding my costly Cal kit which cost more than 600USD
<bvernoux>
eCal use state of the art switch with PE426462 - UltraCMOS® SP6T RF Switch - pSemi
<bvernoux>
Those switch are really amazing for the price (about 10USD/unit)
<d1b2>
<david.rysk> picoVNA is too expensive, nanoVNA is not sufficiently capable
<bvernoux>
with IIP3 of 60dBm you can search you will never find a similar part available in a chipset (AD are very far behind)
<bvernoux>
Isolation is just amazing too min 30dB and up to 65dB
<bvernoux>
35dB isolation at 6GHz
<bvernoux>
with extended temperature range -55°C to +125°C
<bvernoux>
I'm still searching very good resistor for the LOAD 49.9Ohms
<tnt>
bvernoux: So are you selling complete units (assembled pcb + case + cal factors) ?
<bvernoux>
No
<bvernoux>
I just plan to build 5 PCB and test them and build 1 Case
<bvernoux>
As the production is the same price for 1 PCB or 5x
<bvernoux>
especially the cost for the P&P which are quite high at start
<bvernoux>
The aim is mainly to do that for my own use and potentially sell some spare PCB populated and fully tested
<bvernoux>
I can also produce the Cases to have something complete with full calibration of each eCal saved in flash
<bvernoux>
using my expensive Cal Kit + HP 8753D from 30KHz to 6GHz (I cannot go to 8GHzwhich in theory is supported)
<tnt>
Yeah, I'd be interested in one but thing is I have nothing to calibrate it and measureing it without the case would probably be useless since I'd expect the case to influence the calibration ...
<bvernoux>
Anyway there is no hurry as the part will take >10 days to arrive and the Fab will take probably 10days ...
<bvernoux>
The Aluminum case take 15days work
<bvernoux>
If anyone find cheaper/faster/better for the Aluminum case it will be great
<bvernoux>
It requires non magnetic aluminum of course
<bvernoux>
but there is some RF gasket planned to attenuate any RF signal too
<bvernoux>
Like described by jankae ManufacturingDrawing.pdf => RF gasket SG080160D-22
<bvernoux>
this can attenuate RF up to more than 100dB
<azonenberg>
woop i have 5 more boards inbound from oshpark now. the ethernet directional coupler probe, isolation test, new resistor characterization board, new SMPM test board, and the Beatty standard
<bvernoux>
ha great
<bvernoux>
Will be nice to find some good resistor to be used as LOAD for VNA calibration
<azonenberg>
Vishay FC or CH series
<azonenberg>
their 50 ohm are very close to ideal
<bvernoux>
in fact 50ohms is not mandatory
<bvernoux>
even 49.9 is good what is important is to have S11 very high
<bvernoux>
at >6GHz
<bvernoux>
with a stable impedance of course
<bvernoux>
The aim here is to reach up to 8GHz which is the specification limit of the switch
<bvernoux>
the SMA connectors are specified up to 26.5GHz and are the amazing Amphenol 901-10510-2
<bvernoux>
which are very good
<azonenberg>
bvernoux: these days I prefer Rosenberger 32K243-40ML5
<azonenberg>
but those amphenol are not bad at all
<azonenberg>
i just like the screw mount
<azonenberg>
more mechanically robust when torquing the connector
<bvernoux>
yes the price is not the same too ;)
<azonenberg>
also, vishay CH0402F 50 ohm has better than -20 dB return loss to 10 GHz
<bvernoux>
ha great
<azonenberg>
CH02016 is even better
<azonenberg>
better than -40
<bvernoux>
CH0402-50RGFT ?
<azonenberg>
yeah, or similar. CH02016 is best for low parasitics
<azonenberg>
you do of course have to pay attention to how you mount it
<bvernoux>
the eCal have a heater too
<bvernoux>
with a NTC thermistor to check the temperature to regulate it
<bvernoux>
all is managed with open source firmware with a RP2040
<bvernoux>
also the PC tool are amazing with a QT Application or python ones
<bvernoux>
I think the 0402 Vishay will be perfect
<bvernoux>
0201 is a bit small and quite sensible to solder ...
<bvernoux>
I see the impedance is the same for both
<bvernoux>
even the S11 is the same up to 10GHz
<bvernoux>
as anyway we cannot exceed 8GHz with the switch specification
<bvernoux>
In theory CH402F shall be better than CH402 N
<bvernoux>
but I do not see any difference in the datasheet with S11/S21 ...
<azonenberg>
Ignore the datasheet
<azonenberg>
look at the published s-params
<azonenberg>
the actual Modelithics data is much better than what's in the pdf
<bvernoux>
SParameters are not available for CH0402-50R