search

jankae / LibreCAL

4 port eCal module
GNU General Public License v3.0
60 stars 19 forks source link

Update to latest version LibreCAL A+ 2023-02-01 #8

Closed bvernoux closed 1 year ago

bvernoux commented 1 year ago

It is a minor update just to reflect my latest changes on my 5 produced units (in test waiting aluminum cases to fully characterize the boards with my HP 8753D VNA on 4ports from 30kHz to 6GHz)

Tell me if you are interested in those modifications / improvements else just tell me you do not want them and I will keep that in my own branch for reference with my small 5 boards production batch

jankae commented 1 year ago

Hi, thank you for the changes.

Optimized RF Traces for JLCPCB 4Layer JLC04161H-7628 Stackup with CPWG Z=50 Ohms Track width 0.34mm, Conductor Gap 0.2mm

Do you have any data to back up that this is actually an optimization? I have gotten pretty good results with 0.3mm width and 0.18mm gap so far.

Changed 50 Ohms Load Part from CH0402-50RGFTA to CH0402-50RGFPT

I am contemplating switching back to a generic 50 ohm resistor here. I have measured cheap (0402) resistors as 30dB return loss all the way to 6 GHz and it won't get any better than that due to the return loss limits of the switches (and probably the transition from connector to PCB trace). I just don't think these expensive resistors are worth it here.

Add NetInspector_PortSwitch.ods (LibreOffice Calc) to have an estimation of RF Loss on each RF traces

That looks rather useful. Is it generated by KiCad or have you created that yourself?

bvernoux commented 1 year ago

Hi, thank you for the changes.

Optimized RF Traces for JLCPCB 4Layer JLC04161H-7628 Stackup with CPWG Z=50 Ohms Track width 0.34mm, Conductor Gap 0.2mm

Do you have any data to back up that this is actually an optimization? I have gotten pretty good results with 0.3mm width and 0.18mm gap so far.

I have computed them with Saturn PCB Design PCB Toolkit V8.22 Also the fact that the Track width is a bit bigger is always better for the RF loss even if that shall not change a lot between 0.3 vs 0.34mm image In comparison with your values 0.3mm width and 0.18mm gap Saturn PCB Design PCB Toolkit V8.22 give an Impedance of 53.72 Ohms

Changed 50 Ohms Load Part from CH0402-50RGFTA to CH0402-50RGFPT

I am contemplating switching back to a generic 50 ohm resistor here. I have measured cheap (0402) resistors as 30dB return loss all the way to 6 GHz and it won't get any better than that due to the return loss limits of the switches (and probably the transition from connector to PCB trace). I just don't think these expensive resistors are worth it here.

The main advantage of RF 50 Ohms Load is the stability of the Impedance which is always the same even after 20GHz maybe it is not mandatory for the eCal. Anyway anyone can change the BOM and use a cheap 50 Ohms Resistor but the PCB+Assembly+Components+Aluminum Case are already very costly so adding about 40USD to the BOM to have the best possible RF Resistors is not a big addition.

Add NetInspector_PortSwitch.ods (LibreOffice Calc) to have an estimation of RF Loss on each RF traces

That looks rather useful. Is it generated by KiCad or have you created that yourself?

I have extracted the length of Tracks from KiCad then exported them in LibreOffice Calc which compute the estimated RF Loss for FR4 using 0.3 FR4 Loss in dB/cm @6GHz (which is quite good and based on https://www.edn.com/loss-in-a-channel-rule-of-thumb-9/)

jankae commented 1 year ago

Thank you very much for the additional feature :) But would you mind putting it in a new PR? I don't think I am going to merge the hardware changes, I'd like to keep my 0.3mm trace width and 0.18mm gap. I am aware that these values are not ideal if you plug them into a calculator but I got good experimental results with that and am very hesitant to change that.

(I can also try to cherry-pick if it is too much work to move this to a new PR)

bvernoux commented 1 year ago

Thank you very much for the additional feature :) But would you mind putting it in a new PR? I don't think I am going to merge the hardware changes, I'd like to keep my 0.3mm trace width and 0.18mm gap. I am aware that these values are not ideal if you plug them into a calculator but I got good experimental results with that and am very hesitant to change that.

(I can also try to cherry-pick if it is too much work to move this to a new PR)

I understand your point as you prefer to keep your hardware version (that you have manufactured & validated on your side). I need to check on my side for the future (as I plan some LibreCAL Firmware and LibreCAL-GUI update) how to keep my hardware version changes but to be synchronized with your LibreCAL repository for the Software parts. I have create a fork with hydrabus account for all Software Parts see https://github.com/jankae/LibreCAL/pull/13 This Pull Request is closed