White screen on new opentendo build, no signal on M2

Discuss hardware-related topics, such as development cartridges, CopyNES, PowerPak, EPROMs, or whatever.

Moderators: B00daW, Moderators

vinnieb
Posts: 8
Joined: Mon Jan 18, 2021 10:58 am

White screen on new opentendo build, no signal on M2

Post by vinnieb » Mon Jan 18, 2021 11:16 am

Hi All,

I've been having fun trying to put together an opentendo build, but starting from scratch hitting a few issues. First off I fitted a 21MHz crystal despite ordering PAL chips (UA6527P and UA6528P), and had a broken picture stumping me for a while. New crystals turned up, and now I just have a solid picture. Colour varies slightly from yellowy white to pinkish white.

Following another thread around a similar sounding problem, I looked at testing M2 and A0, and I have nothing on either pin. I'm having to make do with a 24MHz logic analyser, and if I try this on the CPU-CLK pin I at least get a measurement, but its 2.6MHz - suspiciously out by 10x but I suspect because its above the analyser capture capability.

The VCC pin is coming in at 4.3V, and RESET stays high. I've had to deviate from original parts in a few places, so the CIC chip is replaced with the workaround here https://github.com/Redherring32/NullCIC ... ullCIC.jpg, and I don't have the RF board so I've instead connected 5V directly, and put the video out through a little amplifier. As best as I can figure out, I wouldn't then need the front panel power+reset switch(?). Just to add too, I've tried another CPU in the socket with same result (I ordered a few of each just incase!)

Attached are a couple of pictures of screen output and closer up of the board.

Hopefully I've not made any daft assumptions in the setup, but am doing this as a bit of fun learning.

Many thanks for any pointers!
Attachments
IMG_20210118_180801744.jpeg
IMG_20210118_180817537.jpeg

lidnariq
Posts: 10435
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: White screen on new opentendo build, no signal on M2

Post by lidnariq » Mon Jan 18, 2021 12:38 pm

vinnieb wrote:
Mon Jan 18, 2021 11:16 am
Following another thread around a similar sounding problem, I looked at testing M2 and A0, and I have nothing on either pin. I'm having to make do with a 24MHz logic analyser, and if I try this on the CPU-CLK pin I at least get a measurement, but its 2.6MHz - suspiciously out by 10x but I suspect because its above the analyser capture capability.
For a repetitive input like the main system clock, you are likely seeing aliasing, i.e. 26.6MHz - 24MHz = 2.6MHz. (For the NTSC crystal you would instead see 24MHz - 21.477MHz = 2.52MHz)
RESET stays high. [...] the CIC chip is [the standard "two wires" CIC disable] [...] and I don't have the RF board so I've instead connected 5V directly, and put the video out through a little amplifier. As best as I can figure out, I wouldn't then need the front panel power+reset switch(?). Just to add too, I've tried another CPU in the socket with same result (I ordered a few of each just incase!)
The CPU won't work unless /RESET stays low for some time after power is supplied. Double check that, possibly adding a switch if you have to.

vinnieb
Posts: 8
Joined: Mon Jan 18, 2021 10:58 am

Re: White screen on new opentendo build, no signal on M2

Post by vinnieb » Mon Jan 18, 2021 3:30 pm

Many thanks for the quick reply. Just to check, taking it to ground would be sufficient, for a couple of seconds? Or something more advanced needed to replicate the original reset switch?

Looking at the schematics and trying to reverse engineer from photos, it looks like the original reset switch when pressed would connect orange and yellow cables that run to it, which looks like the +5v line and 4th pad up on the board, which goes down to where the CIC chip would be? Just a bit wary of applying 5v based on what I can see in various online pics.

lidnariq
Posts: 10435
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: White screen on new opentendo build, no signal on M2

Post by lidnariq » Mon Jan 18, 2021 3:42 pm

vinnieb wrote:
Mon Jan 18, 2021 3:30 pm
Many thanks for the quick reply. Just to check, taking it to ground would be sufficient, for a couple of seconds? Or something more advanced needed to replicate the original reset switch?
About 1/10th of a second. Maybe less? Don't need more (but more is also safe).
Looking at the schematics and trying to reverse engineer from photos, it looks like the original reset switch when pressed would connect orange and yellow cables that run to it, which looks like the +5v line and 4th pad up on the board, which goes down to where the CIC chip would be? Just a bit wary of applying 5v based on what I can see in various online pics.
The 2A03 has a weak pullup on its /RESET pin. The Famicom connects that pin via a capacitor and momentary button in parallel to ground. This is all you need; you can do something fancier if you want, but don't need to.

In the NES-001, the CIC has weak pulldowns on its pins, so the button there pulls CIC pin 7 high, and the CIC inverts and stretches it before it goes to the CPU and PPU (and the power LED)

pakosup
Posts: 47
Joined: Mon Apr 24, 2017 10:23 pm

Re: White screen on new opentendo build, no signal on M2

Post by pakosup » Mon Jan 18, 2021 9:30 pm

All UMC CPUs are quite pickie to clock signal. So you can play around with capacitor between generator and cpu.

vinnieb
Posts: 8
Joined: Mon Jan 18, 2021 10:58 am

Re: White screen on new opentendo build, no signal on M2

Post by vinnieb » Tue Jan 19, 2021 2:11 am

Thanks both. Trying both approaches, a quick connection of ground to /RESET does black out the screen, but then on release almost instantly back to the white / slightly off white screen. Will need to have a think what the long term solution is for switches, hopefully I'll get to that stage :)

I've tried changing the trim capacitor a bit as suggested, but it doesn't seem to have made much difference, there's a very slight picture change when the screwdriver head makes contact, but after that changing either direction has no effect on picture or measurement at M2. Just incase, I reset a few times between turns too, incase it needed to reset against the slightly changed frequency.

From what I understand, this early on in the circuit there's not that many variables that could be impacting it? Frequency presented on CLK not being right, CPU chip itself having an issue, or /RESET line issue?

I've got a bunch of spare components, I guess I could try to recreate a striped down part of that circuit in a breadboard, might help me understand what's going on if nothing else.

lidnariq
Posts: 10435
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: White screen on new opentendo build, no signal on M2

Post by lidnariq » Tue Jan 19, 2021 12:44 pm

vinnieb wrote:
Tue Jan 19, 2021 2:11 am
I've tried changing the trim capacitor a bit as suggested,
Not the trimcap! Pakosup meant C44, between the crystal and the amplifier Q3, and the CPU.
vinnieb wrote:
Tue Jan 19, 2021 2:11 am
From what I understand, this early on in the circuit there's not that many variables that could be impacting it? Frequency presented on CLK not being right, CPU chip itself having an issue, or /RESET line issue?
Yeah, basically that's it.

Because the PPU is drawing a solid-colored picture, that means that the PPU thinks the 26.6MHz clock input is fine. I'm not really clear why the CPU disagrees, but the source of M2 (pin 31) doesn't care about anything other than /RESET and the main system clock as input.

Can you check whether M2 is floating or is driven? If it's floating, the /RESET input is somehow invalid (don't know how, but...).

If you do put the CPU in a breadboard by itself, the clock divider should be static logic so you could be able to manually use a button and a resistor to drive pin 29 high and low and watch M2 change every 7.5-ish button presses.

vinnieb
Posts: 8
Joined: Mon Jan 18, 2021 10:58 am

Re: White screen on new opentendo build, no signal on M2

Post by vinnieb » Tue Jan 19, 2021 3:18 pm

Ah right, got the wrong end of the stick there, looks like C44 is the one just off the emitter of Q3, at 220pf? I've got spares of the other caps from the board, would trying a 100pf or 330pf be in the right ball park?

When you say, floating or driven /RESET, do you mean changing state of its own accord? When I had the logic analyzer on it, it stayed high, but are you suggesting something to test more than that?

Sorry for basic questions, more of a software guy really but learning alot here! Thanks again.

lidnariq
Posts: 10435
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: White screen on new opentendo build, no signal on M2

Post by lidnariq » Tue Jan 19, 2021 3:26 pm

vinnieb wrote:
Tue Jan 19, 2021 3:18 pm
Ah right, got the wrong end of the stick there, looks like C44 is the one just off the emitter of Q3, at 220pf? I've got spares of the other caps from the board, would trying a 100pf or 330pf be in the right ball park?
I don't know, unfortunately. The bright side is you can touch another capacitor in parallel to the one that's there and the two capacitances will sum (i.e. 220pF in parallel with 100pF will become 320pF), so it's easy to test increases without desoldering anything.
When you say, floating or driven [M2], do you mean changing state of its own accord? When I had the logic analyzer on it, it stayed high, but are you suggesting something to test more than that?
"Floating" means nothing is pulling the pin high or low: it's just left to its own devices and that usually sits around doing nothing.

The logic analyzer has its own pullups, so you can't tell the difference between "floating" and "CPU is driving the pin high".

If you have a voltmeter, and it has a logic tester, it should distinguish between "pin is driven low" and "pin is floating and happens to be low".

If you don't have a logic tester but you do have some LEDs that can work too.

vinnieb
Posts: 8
Joined: Mon Jan 18, 2021 10:58 am

Re: White screen on new opentendo build, no signal on M2

Post by vinnieb » Wed Jan 20, 2021 12:24 pm

Ah that's very cool, yeah have a few LEDs and assembled that. All the bottom row pins of the CPU apart from /RESET light up the LEDs (2x green) on the +5v side LED. The /RESET, pin 3, lights neither.

On the upper pins, more mixed but of interest, M2 pin 31 lights up the LEDs (2x red) on the ground side of that diagram. Pin29 CPU-CLK is lighting on the green +5v side.

More interesting developments on the capacitor C44. I took a couple of wires from C44 on the board to a breadboard, to try around different combinations. Through an accidental shorting when two capacitor legs were touching, I started seeing a signal on M2, and trying it properly with no capacitors and direct connection across pads, it loads up a game. Photo attached which whilst a massive step forwards, makes me wonder if there's something still not right, text is broken and some characters/numbers are wrong.

I've also tried quite a few variations in C44 now from 10pf to getting up to around 1000pf, but no luck if there is any capacitance in there so far.

Pretty satisfying to have something on the screen at least.
Attachments
IMG_20210120_191011902.jpeg

lidnariq
Posts: 10435
Joined: Sun Apr 13, 2008 11:12 am
Location: Seattle

Re: White screen on new opentendo build, no signal on M2

Post by lidnariq » Wed Jan 20, 2021 12:40 pm

That glitch implies something's going on with PPU A0, but not the multiplexed AD0. Check the soldering around the latch to the PPU RAM. on the PPU RAM specifically.

pakosup
Posts: 47
Joined: Mon Apr 24, 2017 10:23 pm

Re: White screen on new opentendo build, no signal on M2

Post by pakosup » Thu Jan 21, 2021 8:25 am

vinnieb wrote:
Wed Jan 20, 2021 12:24 pm
Through an accidental shorting when two capacitor legs were touching, I started seeing a signal on M2, and trying it properly with no capacitors and direct connection across pads, it loads up a game.
I know the case when CPU starts only with C44 2n2. And if you short pads then CPU can start heating. So maybe you should use at least resistor.

About PPU, looks like shorten address legs somewhere.

vinnieb
Posts: 8
Joined: Mon Jan 18, 2021 10:58 am

Re: White screen on new opentendo build, no signal on M2

Post by vinnieb » Thu Jan 21, 2021 9:05 am

Thanks both, I've got a USB microscope turning up tomorrow so I can take a proper look at the soldering around PPU and PPU RAM.

I've tried 2.2nF and then worked around variations from there adding others I've got in parallel, 1nF, 4.7nF, 10nF, 22nF, without any luck so far.

Will try some more, and report back once I've had a good look with some magnification around the PPU.

vinnieb
Posts: 8
Joined: Mon Jan 18, 2021 10:58 am

Re: White screen on new opentendo build, no signal on M2

Post by vinnieb » Fri Jan 22, 2021 10:39 am

Well, have to admit not quite sure what the problem was. I did a continuity check of every pin from RAM and PPU, and all seemed to go to intended place. Also did a continuity check to see if any adjacent pins were shorted. That was fine too. Turned it back on again, and have a nice clear picture. I can only guess, that maybe when playing around with capacitors for C44, it needed the capacitor out and a reset to right itself.

I've still got to sort out audio and make some controllers so long way to go, but thats a pretty solid basis now so thanks alot of the help.

You were right about CPU starting to get warm, any suggestions for resistance value? I had another round of trying increasing from some lower increments incase it was a bit below 1nf, and no luck.

vinnieb
Posts: 8
Joined: Mon Jan 18, 2021 10:58 am

Re: White screen on new opentendo build, no signal on M2

Post by vinnieb » Sat Jan 23, 2021 7:18 am

Just to update, anything >= 470ohm was too much and was on the usual white screen, 220ohm its working fine.

Post Reply