Blew up my first 5A-75E... what could have gone wrong?

[dixi83]

Hi All, My first attempt replacing the buffers with the 74LVC245 failed... after replacing them the CPU gets really hot and does not respond any more to anything.... so I guess I fried it.

The idea was to make the J1-J8 input and leave the J9-J16 output I don't know what went wrong, I tested every IC after replacing for short circuit and between every pin. There are 3 options I see:

• my multimeter fried something while testing
• the new 74LVC245 came from china and they are b/c/d (or maybe z) -stock
• The PCB is a 4 or more layer PCB and I cut to much

I ordered a new one but I don't want to fry my new one.

Thanks!

[dixi83]

The version I have is a 8.2. As far as I could determine it has the same pin assignment as documented for the 7.1. I have only tested the RGB pins not A-E...

[kittennbfive]

Hi,

at least on my board which is V7.0 some FPGA-pins are connected to multiple output buffers. This could lead to some short-circuits if not taken care of. Just a thought. Also maybe the buffers are crap, i would recommand buying them from a reputable source like Mouser, Digikey, Newark/Element14/Farnell, ... depending on your location in the world.

I have a file !!!FOR VERSION 7.0!!! which shows what is connected where. Il will add it to my github, might be useful for other people.

Also i would suggest removing all the buffers with hot air and see if the FPGA (not CPU) works again.

EDIT: Other thought: Did you remove the FLASH-memory that contains the original/default bitstream? This bitstream might configure some FPGA-pins that are now inputs on your board as outputs, this is a problem. You can remove the FLASH-memory so the FPGA will power up unconfigured and you have to flash it first via JTAG (NOT persistent). I did this and it works fine (V7.0). Details on my github.

[dixi83]

@kittennbfive thanks for your reply! I ordered a new card from Ali and also some new IC's from Farnell. I will do a retry next week is I have more time. I will use kapton tape to isolate Pin 1 and 20 it looks really neat. I did not removed anything else then the buffers with a wire attached on the photo... what I do know now I should not have removed the buffers I wanted to keep as output. I did used a hot air gun. I only used a iron for cleaning the pads and some solder removal lint (not sure if it is the right English word) Sorry for shouting "CPU" to a FPGA... I was not aware 🤣

@everyoneelse this is the repo kittennbfive is talking about: https://github.com/kittennbfive/5A-75B-tools I received my new card last week. I also ordered a mesa card to get started with my machine and getting used to the way of configuring LinuxCNC.

The thing I did not mentioned in my previous post is that I did not even test the card before changing anything. so another option would be that my first card was death on arrival...

[kittennbfive]

The word you are looking for is "solder wick". ;-)

As i said, you really (imho) should remove (or erase at least) the FLASH containing the default bitstream so there won't be any FPGA-pins fighting against the new installed buffers. Also you should check for FPGA-pins connected to multiple buffers for the same reason. I am afraid i can't give more details because i don't have a V8.

[dixi83]

Thanks for your comment and extra warning. I think this is what went wrong.... I finally found what you meant: https://github.com/kittennbfive/5A-75B-tools/blob/8f4c14c88fb531b183f87d95d1defc01eb2f0712/I_want_inputs/README.md The next question is a noob question 😉 ... Can I "remove" the flash by compiling "empty" firmware and load this using openFPGAloader or is there a way to do this in another easier way?

EDIT:

openFPGAloader -c ft2232 --bulk-erase will probably do the trick. Am I correct?

[kittennbfive]

I would guess that simply erasing the flash should be ok, but honestly i don't know for sure. :(