Reply 100 of 115, by Kahenraz
- Rank
- l33t
Good luck!
Good luck!
Discovered and corrected a few minor alignment issues while working on the processor tweaker project. nothing really significant. been a while since I found a issue with this one of significance therefore I think I'm gonna order this one to prototype.
jlcpcb is currently manufacturing a prototype per their website. now to wait. it's of a older but basically same version as this minor update.
Mainly played with the silk screen on this one.
It's nice that, with the chip shortage, there doesn't seem to be a PCB shortage to go along with it.
After beginning production on one without issues, I ordered another one and they wanna argue about edge plating on the new one. Ugh… just make the thing…. Sheesh
Next thing you know they'll start offering to plate it with gold!
Kahenraz wrote on 2022-04-30, 01:07:Next thing you know they'll start offering to plate it with gold!
If you select gold fingers with a specific production option it will come gold plated
Do you hear that whooshing sound? 😜
Prototype should arrive soon! Can't wait!
Good news, they are indeed removable (this is the previous prototype), but bad news, not all the pads looked like they bonded. May want to consider lots of flux.
Given the age of these boards, it's probably a good idea to thoroughly clean and then pre-tin all of the necessary pads with new solder, since they will most likely be heavily oxidized.
Flux is incredible. My favorite is tacky flux. Let the joint swim in it and you'll get great results every time.
Just tested the latest prototype. No shorts. The adjustments seem to have totally solved the bridging problem I wasn’t even that careful while soldering either.
I used a whole bunch of flux (which I haven’t cleaned up)
This is on a dead s7 board.
The edge plating turned out okay. It comes with a bunch of shavings there and jlc sometimes mails you and tells you they can’t make it but it turned out fine so don’t listen to them.
Next I will install various processors and make sure none of those lead to shorts between vcc2 vcc3 and vss
Next I will try this on a working board. Most likley fic 503+ or gigabyte ga-5aa both are ss7 and don’t need it, but are among my lesser valued boards atm.
My final test. (Which I may wait for final revision) will be on my treasured tyan s1564d
It’s looking very good that I may graduate the latest version out of beta and call it ready. More testing to come though.
Kahenraz wrote on 2022-05-22, 23:32:Given the age of these boards, it's probably a good idea to thoroughly clean and then pre-tin all of the necessary pads with new solder, since they will most likely be heavily oxidized.
Flux is incredible. My favorite is tacky flux. Let the joint swim in it and you'll get great results every time.
Good point!
Just ran a bunch of processors through the motherboard and observed the ohms between vcc2 vcc3 and vss
All looked as expected. All dual plane processors showed no shorts vcc3, to vcc2, to vss
And only single plane cpus showed short between vcc2 and vcc3 (expected/normal)
Next step is to install on a good motherboard and try it out for real.
Okay, testing went well.
One problem found: lack of trace links between pads for pull up block (easy fix). not sure when that happened. Those used to be there 🤣.
I found that this motherboard was using 500 ohm pull ups so on my first round of testing when I tried to use 500 ohm pull downs. it didn't work. a little bit of investigating revealed the problem to be specific to this motherboard and probably some others but not all. (was expecting intel spec 2.2k pull up) My fault for not measuring what the pull ups the mobo uses are before assembling. 🤣. Anyway no harm can come from this error really, it just won't read a pull high or a pull low when you want it to. The fix for this is changing the pull downs on the tweaker to 100 ohm (instead of usual recommended 500 ohm) to work with the motherboard pull ups. (I've seen as low as 25 ohm/500 ohm split before. Anyway you want your pulldowns to be lower than the pull ups so it can over ride the pull up.
Anywho! that resistor thing was just a reminder that there is no one resistor value that users will choose for this. You'll have to inspect your specific mobo and decide. luckily once installed you have good measuring points for vcc3 and the bf functions so you'll be able to measure what your mobo is using easily.
Okay. So this concludes a good test
I was able to verify most functions and concerns and resolve the one defect that I found.
Test results:
BF0 pull down: pass
BF1 pull down: pass
BF2 pull down: pass
intel BF2 pull down: pass
BF0 pull up: pass but see note about now resolved defect
BF1 pull up: pass but see note about now resolved defect
BF2 pull up: pass but see note about now resolved defect
intel BF2 pull up: pass but see note about now resolved defect
I was able to change all BF0/1/2/i2 settings on live processors from the tweaker.
Continuity checked all function paths
Verified that the bridging problem is now gone
I don't have the ability to check the p5a mod on this gigabyte ga-5aa but it looks good on continuity test
Tillamook mods continuity check looks good. again, I can't fully test all of those on this mobo. There probably aren't any mobos out there that can use ALL of the functions built into the Tweaker. However, I wasn't able to get past post before enabling the two outer tillamook mods and l2 cache is working now.
So good test!
I present to you,
Version π.0
We are out of beta!
*claps*
Looks great!