Somewhat of a mystery what happened and what all broke, but when powering on with the 240p Test Suite on a modified cartridge, I have blown up my hi-def NES (USA front-loader). A sad day indeed.
I found that my NES RF box is providing voltage as intended, yet no power makes it to the NES 5V rails. Inspecting the Hi-Def NES board, I find that L8, right next to the main power input, is what blew. How could I go about finding the value of this part? It seems that it is there for the purpose of U10, which would provide clues to the L8 value. However, I have no idea what U10 is either. U10 seems to be a DC-DC converter in a QFN-16 package. It has this marking:
30012
77926
.505F
I know that L8 is what went bang though. If anyone has some secret schematics somewhere, I would love to know the specs of this inductor instead of making a wild guess.
I think I am going to stick in a 1 uH or less inductor tonight or tomorrow and see what happens. I like how the old one acted like a fuse, I worry I am going to put something too beefy in there now and something else harder to replace will blow next time.
I know what I did wrong in the 240p cartridge now. I only used /ROMSEL to disable the PRG EEPROM, I did not disable it when CPR R/W = 0, so every time it was trying to do a bankswitch, it created a CPU data bus conflict. So it seems that the hi-def NES blows up when there is a bus conflict. I am not too impressed by that at all if that is true.
No standard commercial discrete logic cart - neither licensed nor unlicensed - disables the ROM on bankswitching writes. And I'm confident tepples included a bus conflict prevention table anyway.
You might try PMing game-tech.us, or getting in contact with them in some other way.
As I understand it, the HDNES board is supposed to replace the original 5V regulator altogether, and it should take the unregulated ~+13V from the wall wart and - I thought - provides 5V back to the NES also.
lidnariq wrote:
No standard commercial discrete logic cart - neither licensed nor unlicensed - disables the ROM on bankswitching writes. And I'm confident tepples included a bus conflict prevention table anyway.
You might try PMing game-tech.us, or getting in contact with them in some other way.
As I understand it, the HDNES board is supposed to replace the original 5V regulator altogether, and it should take the unregulated ~+13V from the wall wart and - I thought - provides 5V back to the NES also.
Oh wow, I always knew that some carts had the potential of bus conflicts, but I never knew that they were happening all the time... So I guess I still don't know why this blew up.
Sorry, I am a little frustrated atm, and still really searching for an explanation. I don't mean anything against the Hi-Def NES, it is a very awesome thing, and I do use it as one of my main machines. I do have my original 7805 disconnected and feeding in directly into the Hi-Def NES. Will try powering it up carefully with a current limited bench supply tonight when I put in the new inductor.
Edit:
So as long as during a bus conflict, you write the same value that is stored at that location in ROM, you should be okay? Is that the idea behind this?
lidnariq wrote:
No standard commercial discrete logic cart - neither licensed nor unlicensed - disables the ROM on bankswitching writes.
Exceptions: ANROM has a circuit to disable ROM during writes, and AOROM uses mask ROMs with a +CE, which is connected to R/W.
I avoid bus conflicts in 240p Test Suite by overwriting the immediate value in the preceding load. Now I'm worried that I might have made a mistake. I'll probably have to figure out how to set up Mesen to break on bus conflict.
tepples wrote:
lidnariq wrote:
No standard commercial discrete logic cart - neither licensed nor unlicensed - disables the ROM on bankswitching writes.
Exceptions: ANROM has a circuit to disable ROM during writes, and AOROM uses mask ROMs with a +CE, which is connected to R/W.
I avoid bus conflicts in 240p Test Suite by overwriting the immediate value in the preceding load. Now I'm worried that I might have made a mistake. I'll probably have to figure out how to set up Mesen to break on bus conflict.
My EEPROM had an extra /CE, so added a hex inverter and fed it /(CPU R/W) and it still has the sprite problem on my front loader with this modification. So at least we rule out bus conflict as potential source of the sprite thing.
I found a similar looking 1.5uH inductor and replaced the blown one on the Hi-Def NES board. I turned on my bench supply with voltage set to 0 and current limit set low. I then hooked it up to the bare Hi-Def NES power input and increased the voltage, which stayed at 0 due to current limit. I then increased the current limit slowly. It drew tons of current and the voltage stayed right at zero. So, something else is shorted on this board, and the inductor blew because of that short. Will see about investigating carefully with a thermal camera tomorrow, I bet it is that mystery QFN-16 chip... But I will try to determine exactly how far into the board the current flows this way.
I have always had the bottom of my RF box covered in capton tape for as long as I have had the Hi-Def NES mod in there, so I know that it did not short out to that. This is really a mystery what happened here.
I figured out that U10 is a TS30012 5.0V Buck Converter, package 16-QFN 3x3. Cheap and easy to replace!
https://www.digikey.com/product-detail/ ... ND/5456418Later tonight I will try to figure out if that is what broke and possibly further speculations as to why.
It is very likely that my TS30012 is broken. I applied 1 Amp input for this capture. (My new inductor is rated 1 Amp.) Nothing warmed up after the TS30012. I will remove it and try applying 5 Volts to where its output is supposed to go.
Attachment:
IR_0076_2.jpg [ 115.58 KiB | Viewed 9905 times ]
Edit:
The LED does turn on and blink like normal when I manually apply 5 Volts. Seriously considering putting in a 7805 at this point.
Still could be something blown up in the NES but I did not find a short on the 5V rail when I checked. Will be taking taking this 1 small step at a time.
Edit 2:
Reading through the TS30012 datasheet, it has "Cycle-by-cycle currrent limit with frequency foldback" and also "Full protection for over-temperature [...]". It also apparently has internal ESD protection. So this symptom could not likely be caused by any sort of short circuit or ESD, which also eliminates anything I might have screwed up in the 240p test suite cartridge, or the software itself, bus conflicts, etc, all eliminated. However, in the absolute maximum ratings of the TS30012, input to GND voltage range is -0.3 to 26.4V. This makes me wonder if my bridge has failed in the RF box and I am getting AC instead of DC. Or if somehow a negative charge built up on +13V rail. Will check for those things.
I checked my RF box and the bridge rectifier is working fine. I hooked up DC +13V and DC -13V and both charged up the main cap positive. There were no bad connections or solder flash or conductive bits floating around. Kind of a dead end there.
Now I have the original 7805 hooked up, and feeding its 5 Volts directly to the Hi-Def NES board, but still not connecting the 5V to the RF box. I slowly turned it on, nothing crazy happened. With a cartridge inserted and HDMI cable not connected, the NES works normally out the composite AV jacks, with the usual 5-second delay that the Hi-Def NES has before it reverts to Composite mode. Sound and video are like usual in this mode. However, if I hook up an HDMI cable to my TV, the NES never starts and audio and video does NOT come out of the AV jacks... Normally there is modified "shadow" video coming out of composite in this mode. So the Hi-Def NES card is partially working, but something is still broken.
Since the LED can blink, and the 5-second timer works, we know that the FPGA and its ROM are at least partially working. It can sense whether or not I have an HDMI cable plugged in, but I am not sure what is involved for that to happen. I think that the next step is to probe the various power pins of the FPGA and the HDMI Transmitter chip and verify those voltages.
Edit:
the 1.8V and 1.2V linear regulators are both working. :\
Thanks! I did not know about the troubleshooting page. I read through that and realized that U11 and U8 are 2 more voltage regulators. These 2 are already working correctly though. All 5 of my voltage rails are good now:
5V -> U10 -> Broken, I removed it and am feeding 5V directly from the original 7805 into L6. 5V still disconnected from the RF box. Added 22uF Tantalum from R35 to GND for good measure.
3.3V -> U11 -> OK
2.5V -> U8 -> OK
1.8V -> U9 -> OK
1.2V -> U7 -> OK
I did not notice what the LED was doing when my HDMI cable is plugged in. I know that it was not blinking the same as it does without the HDMI cable. I will need to check that out. I also did not check my interposers. I had removed them in the process of debugging this, so it is possible that I damaged something while doing that. Will remove them again and check them out too.
I carefully removed the CPU and PPU interposers and found no damage or solder bridges. I removed both ribbons and inspected all 4 ends, everything looked fine. I put everything back together after this and the unit still functions the same, runs in composite mode but will not run if I attach an HDMI cable. It does not send an HDMI signal to the TV. If I attach an HDMI cable while it is on and running in composite mode, it crashes.
The internal Hi-Def NES LED:
When I turn on the NES without the HDMI cable, the unit runs in composite mode, the LED starts its slow fade in/out for a moment, then goes to full brightness with short blinks off each second or so. I believe that this is all normal. When I plug in the HDMI cable and then power it on, I get a very dim LED for a moment, followed by a somewhat brighter (but still dim) solid LED, no blinking, no fading. I could hook a scope to the LED if exact measurements would be helpful.
Do these symptoms narrow down what parts could be affected? I presently suspect these parts:
- Any of the 5 level translator chips (I have some spares already)
- The HDMI chip ($10)
- The FPGA (expensive, would be breaking new ground for me to solder one of those)
I actually happen to have exactly 5 spares of these particular level translator chips. They are on breakout boards but I could make use of them if there is any reason to think that could be the problem. Any guidance at this point would be appreciated!
Could you maybe update the fpga firmware? Just a shot in the dark, but it'd help check the fpga condition.
calima wrote:
Could you maybe update the fpga firmware? Just a shot in the dark, but it'd help check the fpga condition.
I am pretty confident that this is not a firmware problem, because the composite mode delay works and the LED is still able to blink correctly. These things are controlled by the firmware. I know that I do not have the latest firmware in there actually, but trying to update it right now will just add more variables and suspicions I think.
I was experimenting one day with replacing regular 7805 with a step-down one based on LM2596. Unfortunately, I forgot to solder one leg of the output capacitor. As a result (this is just my suspicion) - output voltage has risen very quickly over 5V before the feedback loop of this chip could react and it burned my PPU.
Since that time I am sceptic to non-linear regulators, especially that when I soldered the cap correctly and replaced PPU, i found out a little horizontal bars in the video which were not present when I replaced it with 7805 - probably the 100kHz switching freuqency adds significant noise.
It turns out the level shifters I have are similar but actually different, so I am not able to replace them... Not sure where to go next with this. It would be great if I could eliminate the CPU interposer card somehow. Right now I do not know if there is something wrong with the main board or with the interposer. It seems unlikely that this is the PPU interposer but still not impossible I guess.
FINALLY I got back to this. I replaced all 5 of the level shifter ICs. It had no effect at all. When running without an HDMI cable attached, it still works fine through the composite. When attaching the HDMI cable, it still hangs with dim LED. Not sure what to try next!
I know that all of my power supply voltages are good from the FAQ, and also I plugged the same HDMI cable into my Xbox 360 to verify the HDMI cable and the TV. That was all good. The only possible things remaining that I can tell:
- Additional logic on each interposer card (best case scenario)
- FPGA (hopefully not because it's expensive)
- AD9889B HDMI chip (Hopefully not because it is obsolete now, hard to find)
- The flash chip that runs the FPGA (seems doubtful)
My theory is that 1 of 2 things happened:
- The 5V regulator shorted itself out somehow (which is hard to explain.) This blew up the little inductor, which generated a high-voltage spike, which broke something else that I have not found yet.
or
- Something drew lots of current, which blew the little inductor, which generated a high-voltage spike, and the spike is what wiped out the 5V regulator and broke something else that I have not found yet. Whatever drew lots of current no longer does that once getting the 5V repaired (which is also hard to explain).
From what I can tell, the Hi-Def NES has no protection for bus conflicts, which theoretically should be okay, and theoretically is the same as a normal NES... These level shifters are modern though, and they may be more sensitive or drive more current or something. It seems a little far-fetched to me that this caused enough current to blow up that inductor, especially because the same cartridge that was in there doesn't break a normal NES (other than sprite artifacts) and replacing the level shifters had no effect.
Given the fact that there is a web page posted with instructions what to do if this inductor explodes, it must be a common problem. Do we have any idea the root cause of this inductor problem when it happened to everyone else?