Do self-tests ever lie?

[MaximRecoil]

The self-test screens in Super Punch-Out look like the ones attached below.

When I entered the service mode on my SPO machine, it looked exactly like those screen shots from MAME, with everything marked as "OK" on both the RAM and ROM screens.

So is this an absolute guarantee that both my RAM and ROMs are fine? Why does it only show 5 locations for ROM when there are 33 EPROM chips on that board set? I'd imagine that there are more than just 2 RAM chips on the board set as well.

If this means that my RAM and ROMs are definitely OK then I can eliminate them as a possible cause of my audio issues; which I suppose could only leave the N2A03 sound chip as the culprit?

[ChadTower]

Because there aren't 33 ROM chips?  The others can be CPU, RAM, DAC, VRs, PIAs, any number of things... but I guarantee you there are not 33 ROM chips on that machine.

[USSEnterprise]

Could those letters indicate banks of chips?

[ChadTower]

If you built your own adapter, absolutely suspect that first.

[MaximRecoil]

If you built your own adapter, absolutely suspect that first.

LOL. There is no possible way, in theory or in reality, for an adapter to cause a punch sound effect to be replaced with a "ding" sound, and leave all the other sounds intact. Audio is simply a hot and a ground for each amp coming off pins 51 (upper amp), 52 (lower amp), and 54 (ground to both amps). They are not even going through the adapter I made, they are going direct to the amps. Now, a circuit like that either works or it doesn't, it can't replace specific sounds with other sounds. It is the same ultra-complex (lol) circuit that is made when you plug one end of your RCA cable into the audio jack on an NES and the other end into the audio input jack on the TV. Now, give me your theory on just how an issue with that circuit could cause 2 specific sounds to be replaced with a dinging noise on your favorite NES game?

Your suggestion here is akin to suggesting that someone who just changed the tires on his car must have done it wrong and that is why the carburetor is acting up.

[MaximRecoil]

Because there aren't 33 ROM chips?

[RayB]

BOOYAH!

[Flinkly]

woah...

no offense to either of you two guys, mostly not chad, since he's a cool guy anyways, but that was a major smack right there.  i'm glad i traveled over to the misc category for a bit.

[AmericanDemon]

Wow.  Balls of Steel. 

[ChadTower]

Hey, I can be wrong, it has happened before... don't see much reason for an intelligently designed board to have 33 ROMs on it.

That's probably where I fell down, assuming an intelligent design.

[RayB]

Take a look at your own Atari boards Chad.

Max: I think those tests usually report on a bank of chips. Also, consider that maybe the DATA on a chip is wrong? The test wouldn't know if data was wrong, only if writing/reading to the chip failed. I am wondering if maybe someone did the SPO conversion and mixed up two chips.

[ChadTower]

Yeah, but those boards are a lot older than the Punch Out boards.  Bigger chips were available later, and still, the Atari boards have nothing like 30+ ROMs from what I remember.

[RayB]

The chips got bigger but then the amount of data also got bigger. (better graphics, better sound...)

[ChadTower]

Right, but an intelligent design is at odds with 33 potential failure points.

[RayB]

Word.

[Ed_McCarron]

Right, but an intelligent design is at odds with 33 potential failure points.

Maybe it evolved?

'k.  I'll shut up now.

[2600]

Right, but an intelligent design is at odds with 33 potential failure points.

You say that as if you knew they had a choice.  Do you know that to be true?

[ChadTower]

There is always a choice.  When you're designing a custom PCB, you can do things a LOT of different ways.

[2600]

There is always a choice.  When you're designing a custom PCB, you can do things a LOT of different ways.

That is true.

Even though they are doing there own PCB, there may have been some restrictions.  Like the CPU and peripherals being defined.  Now if the CPU they used can only address up to a certain size, then that could restrict them on ROM size choices.  So instead of using 32k ROMs they have to use 16k, well that just doubled the # of chips for the same amount of data.

You also have to think about it with the restrictions of the time as well.  If you look at games from that year (I only did a brief look), Super Punch out doesn't have the most roms.  Also, I only saw a few games with 32k roms.  Maybe they had just came out, hard to get, expensive, etc. 

Also, since Super Punchout has 2 monitors.  One could assume it has a few more graphics to store then a single monitor.

[ChadTower]

Those are all valid possibilities.  It is also just as possible to change the CPU so that it can address a larger ROM, therefore using less of them. 

My best guess is that they reused the architecture of an existing board, and this is the way it turned out.

[MaximRecoil]

Take a look at your own Atari boards Chad.

Max: I think those tests usually report on a bank of chips. Also, consider that maybe the DATA on a chip is wrong? The test wouldn't know if data was wrong, only if writing/reading to the chip failed. I am wondering if maybe someone did the SPO conversion and mixed up two chips.

I have already pulled and reseated all of the ROMs but that changed nothing. I didn't confirm their locations when I did that though. So, I took it out again a few minutes ago and verified that all the locations were correct, both according to the locations marked on the board and confirmed in the manual. Assuming that the labels are correct (the little label stickers on the ICs themselves), then they are all in the correct location and all accounted for.

I took apart a NES afterwards to get a look at its 2A03 chip but it was through hole soldered directly to the board, while the same 2A03 chip on my SPO board set is socketed. I would have liked to have tried a simple swap with another 2A03 sound chip but this was not to be. Is there any source for purchasing socket mount 2A03 chips?

There are 2 sound chips on the PO/SPO board, the 2A03 like in a NES for general sound effects and music that goes to the left speaker's mono amp and the faster custom VLM5030 that generates the announcer's speech and a few other sound effects that goes to the right speaker's mono amp. These do not mix. The audio issues I am having are solely coming from the 2A03 side of things. I have no idea if it is the chip itself that is the problem, but it is definitely somewhere along that 2A03 path. I would like to know which ROM(s) hold the instructions to the 2A03 side of things and exactly what components are in the path between the instructions and its audio pinout.

If I had a 100% working regular Punch-Out board to swap the SPO ROMs onto, it would be very enlightening I'm sure.

[ChadTower]

Are these EEPROMS?  If so, how can you be certain of their contents?  Maybe some of them have the wrong contents.

[MaximRecoil]

Are these EEPROMS?

[2600]

I would like to know which ROM(s) hold the instructions to the 2A03 side of things and exactly what components are in the path between the instructions and its audio pinout.

If you look at one of your other threads, I've answered part of that.

chp1-c.4k

[MaximRecoil]

I would like to know which ROM(s) hold the instructions to the 2A03 side of things and exactly what components are in the path between the instructions and its audio pinout.

If you look at one of your other threads, I've answered part of that.

chp1-c.4k

Yup, I just saw that. Thanks for the reply.

[2600]

When you say the chip is socketed, is it still the same size package?

Couldn't you just unsolder the one from the NES and put it on your board?

[MaximRecoil]

When you say the chip is socketed, is it still the same size package?

Couldn't you just unsolder the one from the NES and put it on your board?

That's possible, they appeared to be the same size, though the through hole version in the NES came down to tiny needle points at the base of its legs in order to fit into the holes it was soldered into. I don't think the socketed version on my SPO board does that, though I could pull it to be sure.

Unsoldering that NES chip would be very difficult, and at the very least I would need some solder wick (which I don't have right now) to clean up the legs after getting it out of the board. I used to work in a PCB factory, building and repairing boards, both through hole and surface mount; and even removing a single terminal block with 4 relatively large legs in relatively large holes was enough of a pain, because even after solder sucking the fillet out of there, the legs were still sticking to the sides of the hole...only legs that happened to be perfectly centered in the hole came out easily, any that were touching the sides of the hole you had to wiggle around with the iron and try to get it away from the side of the hole, and stay away from the side of the hole long enough for the solder to harden so it wouldn't stick again when it went back into its natural rest against the side of the hole. Now, when you are dealing with something that just has one leg, that's easy because you can keep the heat to it and pull it out without worrying about other legs that are still barely soldered to the sides of the holes.

It can be done, and I can do it (tweezers help) but damn. I wish there was someone who sold socket-type 2A03's. If I knew for a fact that the 2A03 was the problem, and I knew that the desoldered NES chip would slide right into that socket on the SPO board, I would be more anxious to be sacrificing the NES and taking the time to desolder that chip with forty-eleven tiny legs.

[Dav]

The rom test is run by the main cpu, so obviously it can only test the roms it can get to.  In this case there are only 5 roms that can be checked.  Same story on ram, often most of it can't be checked.  The sound cpu rom and ram can not be checked by the main cpu and may not be checked. 

[tommy]

Hey, I can be wrong, it has happened before... don't see much reason for an intelligently designed board to have 33 ROMs on it.

That's probably where I fell down, assuming an intelligent design.

It must be a horribly designed pcb if chad made a mistake and looked the fool. 

[MaximRecoil]

It must be a horribly designed pcb if chad made a mistake and looked the fool.

[tommy]

It must be a horribly designed pcb if chad made a mistake and looked the fool. 

That's the only time I ever make a mistake; you know, when the item that I am expounding and elucidating upon turns out to be inproperly designed in the first place.

In such a case of egregious error on the part of the real engineers, I generally skip the formality of presenting my own "correct" design for said item, and simply [circularly] claim that if the design had been correct in the first place, then what I have been saying about it all along would have also been correct, by default.

Chad, stop loggin in as maximrecoil.

[ChadTower]

Take shots at me at EE if you want, that's cool, but don't threadcrap in this guy's project thread.