And also less resistance on the way by having it closer since people might use shitty wire.
An option to solder it on the borad would be great. Will make installs much cleaner
ok, thanks for the explaining, so the eddy currents, have no effect at all? nor the timings?
It will increase resistance and inductance. Keep in mind the mosfet is shorting this voltage rail for a moment which means there will be many amps flowing with the capacitors still in place. And the inductance will cause ringing which has the potential to destroy your SoC if it goes out of hand.
If the PCB would be sitting between RAM and SoC it's probably fine but the extra wire length when it's not can be problematic.
Well, not so simple. The FETs responsible for performing the glitch on the APU could reach currents up to 10A. The via's on the CPU flex are thick enough to support this.
If the FET would be moved to the chip PCB, thicker wires would be needed for the install, nontheless there's a high probability for signal delay by using longer wires, so not sure if the glitch would work properly.
I'm currently working on a SwitchME port to the adafruit tinyusb library, which opens up a massive array of microcontrollers including the RP2040 to supporting RCM injection, does anyone have any experience with low level calls to the TinyUSB library as it's shockingly poorly documented?
The wires needed to handle the current wouldn't be able to fit through the RF shielding without cutting, the flex cable doesn't move them that far away in the grand scheme of things, as you've said putting the module inside the shield will probably be alright but the mcfly wont fit inside the shield and i dont want to be held responsible for people using magnet wire for the main glitch signal, it turning into a fuse and them causing damage. It's easier to just use the hwfly flex or one of the other flexes people are working on
Since you're making a new chip, might as well not use the flex cable that they can and will monopolize. Its the logical approach tbh. Not to mention, the flex cable is not that accessible to everyone with its only source, aliexpress.
If you're opposed to magnet wire, you could use ribbon cable laser for ps2 which can be cut in half and bridging the 2 halfs for both capacitors.
That's just 1 of the possibilities. Flex cable can be designed and purchased without the mosfets in built from pcbway too. You wouldn't want people to solder mosfets to flex cable, high risk procedure.
There are alternative open source flexes ive seen in development. I'm not opposed to magnet wire as a concept and indeed it will be very useful for soldering the EMMC lines however for the currents that it will conduct during the glitching it is not of sufficient gauge. my board also supports direct soldered FETs so it is not dependent on aliexpress hwfly cables
The Mcfly will be open source, so once it's complete as is you could add the mosfet and test it yourself. I'm deliberately embarking on a conservative design to guarantee success
Yeah, that flexex have Thick wires, and really are not so "appart" from caps, are at the very edge of the CPU, the flexes also have a design for avoid eddy currents because have the GND wires along the path.
Quick update: still waiting for components unfortunately, LCSC are being frustrating. I'm working on other fun things in the background i'm hoping to announce soon
At this time "far" is a maleable concept. you need to have it as close as you can to the capacitors, is like the capacitors they must be close as can to the CPU.No it's not, there's a reason it wasn't included on the chip.
The ideal point is exactly by the cap, but that is for the skilled ones. the flex cable was designed to be easy of install and still be closest to caps.
Put the mosfet outside of the CPU enclosure will not work at all. search about eddy currents and clock timings.
Yea this is incorrect, again. It will work outside the cpu enclosure. People tend to exaggerate the distance of the mosfets thinking its so far that it will get loss signals.
Mcfly/RP2040 is right next to the SX flex cable original position.
Out of interest have you actually tested this to demonstrate that it properly works?
Post automatically merged:
I sacrificed another rp2040 project for this but here it is, at functionally assembled (no FFC connector) a total thickness of 1.5mm on the dot at the thickest part! I'm very happy with that.
Next step is testing, see if it takes firmware and works n all that. I also need to get my extra components and build up the rest of these to send out to testers
Attachments
Last edited by Saliciae,
Board is now fully assembled, added the ffc connector. Had to change the RGB led on this one so it's not as thin as it could be but that will get sorted in the next revision, as a bonus the sk6805-e-j led I plan to use will allow me to make it even smaller!. Gonna build some more and start shipping them out.
Thanks to all who messaged me offering to test, I'll start messaging a few of you about where to send them.
This project has progressed super quickly and I'm excited to see how it performs!
Thanks to all who messaged me offering to test, I'll start messaging a few of you about where to send them.
This project has progressed super quickly and I'm excited to see how it performs!
Attachments
Similar threads
-
-
@
Sicklyboy:
@Xdqwerty, to answer your question, they're a fusion Brit-pop/J-pop/electronic band with a woman vocalist. Flamingo is hands down their best known song but they've got a ton of other really good songs+1 -
@
Sicklyboy:
For example, one of my other favorite songs from them, with some massive house music influence -+1
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
