Author Topic: SNES black screen at startup (Read 29670 times)

tyguy2 · « **Reply #25 on:** June 15, 2015, 10:17:57 pm »

What game are you using? Does it use Mode 7 or the Super FX chip?

SaabFAN · « **Reply #26 on:** June 15, 2015, 10:51:29 pm »

Maybe this will help: Full Schematics http://forums.nesdev.com/viewtopic.php?f=12&t=10470

I think, the next thing to do would be to get a logic analyzer and check what is really happening on the Address- and Data-Lines. At 3,58Mhz, the SNES runs slow enough for a Arduino-Zero based logic-Analyzer. Maybe even slow enough for a normal Arduino. There are ready to use Arduino-Sketches and Readout-Software on the Net available.

Also check all the lines on the CPU for the LOW and HIGH-Levels. If there is something wrong with the Cart-Connector, or the decoupling-caps of the Carts themselves, this is where it should show. The Data-Lines are probably the most likely suspect.

poot36 · « **Reply #27 on:** June 16, 2015, 05:15:32 am »

I have made a short (under 1 inch) jumper from pin 4 to pin 9 (ground) and can confirm that the SNES stays running. It however has not solved the problem. Sometimes on power up using the power switch it still does not start trying to read the cart. I do have the schematic for the second version of the SNES (1993 and onwards) but not for my first version SNES (1990 to 1992). Mine is from 1990. I am using Super Mario World as a test cart as this is the only one that I have had any luck getting anything to display on screen or output sound. It does not have a super FX chip or other accelerator chip in it. Not even a memory maper chip. I have tried other games but I get even less of a result with them. I have tried turning my Arduino UNO into a logic analyzer but I am not sure that I am doing it right. I have one program that uses digital reads and stores them and then converts them into hex and outputs them from the serial port and another that reads directly from the port B and D registers and outputs that as a 0 to 255 number from the serial port. Tomorrow I will try to attach the two programs and then you people can tell me which would run faster. I may end up getting one of those cheap (~$10) USB 8 bit 24Msps logic analyzers off eBay but I am not sure what software to use with it. Would the SUMP software work or is there a better free and legal one out there that preferably can convert the 8 bit data to hex?

SaabFAN · « **Reply #28 on:** June 16, 2015, 09:05:46 am »

With the cheap Thingies from ebay, you mostly use this software: https://www.saleae.com/logic/
If you can get one for 10$, go for it. I have one of those things here and its really helpful, especially as with the PC you basically have virtually unlimited memory compared to a DS1054Z.
The Scaleae-Software also has a lot of decoding-options. BUT The massive drawback: These USB-Thingies are only 8-Bit, whereas the SNES uses 32bit (16bit Address-Bus and 16bit Data-Bus). To really check what's going on in there, you need something more expensive and powerful.
What I would do instead: Check if the Voltage-Levels for LOW and HIGH are within spec when a Cart is plugged in. That way you can quickly rule out any broken connection between the processor and the Memory inside the Cart.

poot36 · « **Reply #29 on:** June 17, 2015, 03:51:17 am »

Ok, I have downloaded the latest SNES schematics (V1.5 instead of my V1.1 version). I have attached my Arduino logic analyzer files to this post they are rar but relabeled as zip. I will double check the voltage levels as well. Also in reading the SNES schematics there is a 8 bit data bus a 16 bit A address bus as well as a 8 bit B address bus on the cartridge slot. I am going to be focusing on the data bus as well as the 16 bit A address bus for my meter measurements. I have also used continuity mode on my meter to double check all of the cart connections to the eprom chip inside and they were all good. I even made my own cart connector just to eliminate any connection problems and it did not change any of the results.

SaabFAN · « **Reply #30 on:** June 17, 2015, 08:08:34 pm »

What? Only 8 Bit Data-Bus? So the SNES isn't a true 16bit-System?
No wonder (SEGA) Genesis did what Nintendon't

I think, after you've ruled out dodgy connections between CPU and Cart, the next step would be to work your way back from the Encoder-Chip to the PPUs and eventually the CPU itself.

McBryce · « **Reply #31 on:** June 17, 2015, 08:30:22 pm »

The SNES has a CPU based on the 65C816, so it's a 16bit core, crippled by an 8 bit databus.

Bryce.

tyguy2 · « **Reply #32 on:** June 17, 2015, 08:37:05 pm »

I think the 8 bit data bus was included to allow backwards compatibility with the NES, but it was never implemented.

poot36 · « **Reply #33 on:** June 18, 2015, 04:33:52 am »

Here is a scope picture of the D5 pin on the cartridge slot with out a cart in it.

poot36 · « **Reply #34 on:** June 18, 2015, 04:35:12 am »

Here is a scope picture of the A5 pin on the cartridge slot with out a cart in it.

poot36 · « **Reply #35 on:** June 18, 2015, 04:36:50 am »

And here is the test setup.

poot36 · « **Reply #36 on:** June 18, 2015, 05:19:40 am »

All of the scope shots were taken with the 20 Mhz bandwidth switch turned on, 2 Volts per division, DC coupling and DC trigger turned on, positive trigger, and I think the same timebase but I am not sure about that ( I know it was in the µs for sure) on a Tektronix 465. I don't know about you but the data bus seems very messed up! Unless my scope does not like properly triggering on certain digital waveforms I don't think there should be a flat portion at around 1 to 2 volts! Or is this caused by the pins trying to read data but picking up noise? I do know after turning on the console with the power switch it will usually not even start running through the address and data lines but will hold them high or low or in that 1 to 2 volt region and showing small square or sine wave action on the scope. If I monitor the pins with the scope and press and release the reset button I can sometimes get what you see in the pictures. If I am really lucky I will get random video output and then I do also see action on the data and address lines on the video ram chips. If there is no video output the data and address lines on the video ram chips usually stay high but sometimes will stay low or a combination of the two. I have also noticed that there is almost 0.5V of ripple on the reset line that goes to the CPU and a few other chips. Is this normal? This reading was taken on my non RMS multimeter set on AC mode.

Anks · « **Reply #37 on:** June 18, 2015, 07:10:01 am »

Looks like you have some form of bus contention on the D5 line or whatever is driving it is failing.

McBryce · « **Reply #38 on:** June 18, 2015, 08:08:30 am »

Yup, I'm with Anks. There's two sources writing data to the bus at (almost) the same time.

Bryce.

poot36 · « **Reply #39 on:** June 19, 2015, 04:15:13 am »

I have found that I can not get consistent scope traces at power up with the power switch but if I press reset I usually will get a consistent trace but not all the time! I am now really thinking there is a reset problem but I can not track it down. I have tried finding the 0.4V of ripple on the reset line but I have tried lifting all of the chip pins that it goes to and it is still there. Can a bad resistor cause ripple? I know that R74 on the schematic shows 1K but mine is 2.2K. This resistor is between the U3 (S-PPU2) chip and all of the other chips that need reset. I have also figured out that the CPU is sending the reset signal not U3 (S-PPU2). I am not sure if this is normal or not. I have attached pictures of the same signals (D5, A5 and screen) under the same conditions (I have removed the Arduino on the data pins but this shoud not make much difference) but a different power on and reset procedure as well as what I managed to get on screen. There is no cart in the SNES.

tyguy2 · « **Reply #40 on:** June 19, 2015, 12:21:33 pm »

Quote from: poot36 on June 08, 2015, 03:05:53 am

I have checked the 3 individual reset lines and they appear to be working but /resout1 does have some noise on it (0.1V to 0.2V).

Why don't you try bypassing the PCB all together and just put each reset to what you want them to be. Try soldering each reset pin high and low. Make sure there is a cart in the SNES, it only boots if a cart is inserted.

tyguy2 · « **Reply #41 on:** June 20, 2015, 09:54:47 pm »

Have you tried just using the A/V input? The S-Video may be messing up.

poot36 · « **Reply #42 on:** June 21, 2015, 05:57:32 am »

Well I think I am an idiot, the noise that I was seeing on the reset line was from the 24 Mhz clock line opposite to the reset lin in the cart connector. I think my extended cart lines are causing some interference. I have tried with the S-Video and Composite outputs and I have verifyed with a scope that I only get an image when there is a analog signal on the RGB pins. I have compared the noise on the reset line with a working SNES and there is actually more high frequency noise on the working SNES then the non working one so I think that rules that out. I still do not know why the SNES will usually not start with the power switch but requires using the reset switch to start activity on the address lines most of the time. I wonder if the method of disableing the lockout chip can cause startup problems. I am in the process of making a simple test cart that will just show a sold color on the screen and that is it. I think I have burned the eeprom with the *.smc file correctly ( I changed the *.smc extension to *.bin to get the programmer software to see it but I am not sure if this is the right way to do it) and am deciding whether to get a cheap cart and remove the original rom chip or wire up 2 connectors to plug into my floppy drive cable ports and attach them to a 32 pin socket for the eeprom. The first option would be faster but would also have more chips on the board to potentially cause more problems. The second option would take more time but should also work assuming that the schematics are correct (for some reason the address lines on the pinout for the cart rom chips do not match the cart connector pinouts at least on Super Mario World). Which one of these methods would you recommend? Quick question should the output of the crystal oscillators look like a sine wave on my scope? After one goes into a buffer chip it does look like a proper square wave on my scope. Also have just figured out that at least one of my scope probes is having a funny problem that causes the waveform to have curved edges on a square wave. On the output of the compinsation trimmer block that has the BNC connector on it if I wiggle the cable there the waveform will go funny. I have noticed that this section can be unscrewed and I have checked that it is tight. Is there something that I can do to fix the problem or do I need to buy a new scope probe cable or compinsation trimmer block? This is a Tektronix probe with the 10x id pin on the BNC connector.

justanothercanuck · « **Reply #43 on:** July 05, 2015, 02:43:29 am »

Just throwing it out there... You might want to check out nesdev for this kind of stuff.

That said, if you've checked the main input capacitor and you're still getting a black screen on startup, you might want to look at the (main PCB) SRAM chips, mainly the one for the CPU. If swapping the PPU and CPU SRAMs around doesn't change anything, then it might be the extended traces after all. Throwing some buffers into the mix might help a bit.

edit: yes sorry, I was confusing the NES and the SNES (the NES RAM chips are identical, not the SNES). I would still try the input cap though.

poot36 · « **Reply #44 on:** July 05, 2015, 02:59:12 am »

I will have a look at that forum as well. The SRAM chips for the CPU and the PPU are not interchangeable. The SNES was having problems before I extended the cartridge lines. I don't think they could be causing problems though because I have seen in portable SNES mods cables around this length as well.

poot36 · « **Reply #45 on:** July 13, 2015, 05:52:21 am »

Well I have done more testing and can confirm that the power on reset chip appears to be working properly. The chip will cut out at 4.05V or so. However in the datasheet it is speced to cut out at 4.2V and the min level is 4.1V with a maximum of 100mV play in the voltage presumably to prevent it oscillating. I think that is in spec still (but may be on the low side). Can someone confirm that I am interpreting the data sheet correctly and if possible test what there SNES browns out at? ~4V seems rather low for 5V logic. The good news is that I should be getting my hp 1661C logic analyzer within 7 days or so. Then I can provide you guys with more useful data (I hope). Or I am going to be asking either how to properly use it or fix it.

tyguy2 · « **Reply #46 on:** July 14, 2015, 05:54:39 pm »

I just rembered this. Are you trying to use a PAL connection on an NTSC screen (or visa versa?). Some carts are not cross compatible, among other compatibility issues. Also, the SNES has a 50hz mode and a 60hz mode, so try looking into that. Have you checked the audio coming from the A/V port? If there is audio, then there may be hope. Also try connecting to a screen via the RF connector. Good luck

poot36 · « **Reply #47 on:** July 15, 2015, 12:22:37 am »

I have checked that I have a standard North American SNES and cartridges. The Sony broadcast monitor is capable of accepting all video formats and it does indicate that it is receiving a NTSC signal. I have only occasionally been able to get sound out but it is usually one effect and not related to the normal startup of the game. I am starting to suspect the PPU1 chip or the RAM chip for the CPU. Is it possible to run a SNES without the PPU1 and PPU2 chips active? I currently have the reset lines so that the two video chips should stay disabled but I am not sure if this is the right solution.

poot36 · « **Reply #48 on:** July 29, 2015, 05:11:14 am »

Update: I got a HP 1661C logic analyzer for less then $100 and have connected it to the SNES. I have found that even with the cartridge and sound module removed I can not get it to consistently start up in the same way. I am not sure if this is because the SNES is very broken or I am using the logic analyzer incorrectly. The SNES is currently configured with the lockout chip enabled but the reset signal from the PPU chips is disconnected from the rest of the system. The reset line for the rest of the system is pulled up by a 100K resistor and monitored by a brownout detection chip. Got any ideas as to where to go from here? If you want I can post screen caps from my logic analyzer and an entire ascii dump of the 8K of data that it has captured. I do find it weard that the address lines show activity even before the sysclock output of the cpu has become active.

poot36 · « **Reply #49 on:** August 10, 2015, 06:23:27 am »

This topic has been continued on the nesdev forum under the user name Poot36. Link here: http://forums.nesdev.com/viewtopic.php?t=13088


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: SNES black screen at startup (Read 29670 times)

Share me