Saturn Power Supply Failing - Repair Ideas?

You gotta see the Japanese launch model power supplies with a hybrid circuit board hanging in the middle of the primary circuit. "Yamaha power supply".

 

That fix sounds very nice! I'll have to test it soon, my main machine has that problem and adding an extra fan doesn't do much to help anymore. Incidentally, the other type of power supply does have one more extra 2200uf capacitor on it, and from what I can see on my snapshots, it IS connected to the same pin of the photocoupler (PC123 on the "good" type power supply, NEC 2561 on the "bad" one).

Additionally, a while ago someone decided to change the TOP202 with a modern equivalent, and that also removed the buzzing. I think it was a TOP222 but I don't remember anymore. There should be a thread around here somewhere about it.
edit: it was a TOP224YN, and supposedly it also runs much cooler.

I suspect that there WERE some dieshrinks, or at the very least some re-spins / subversions. If someone had a guide to the Yamaha LOT numbers used on ICs, we could know for sure, but I doubt anyone has that.

There are 3 "types" of power supplies, with the last one having a pal variant that has an extra output pin (used solely for the SCART switching voltage).
They changed power supplies as the motherboard form factor changed.

 

Look forward to seeing your results!

Swapping the TOP202 out for modern ones seems like it's worth doing too.

 

Not a strange decision at all - at that point in time, there were still tradeoffs involved with SDRAM - you got bigger block sizes (mostly important because that allows you to fill an entire cache line in one memory transaction), but the more complicated access protocol meant that the overall address latency was inferior to contemporary page mode RAM. Using EDO meant that you got both the same first-word latency as page mode RAM and the ability to transfer large blocks at high speed of SDRAM (since all you had to do was toggle CAS to strobe the next word out). SDRAM did have some other advantages, such as being much easier to build large arrays out of, but this really isn't an issue in something like a console that has a fixed-size non expandable memory anyway.

This was for the early (57/60/66MHz) SDRAM - as the clock scaled up, the timing penalty imposed by the more complex protocol was reduced until by the time it hit 100MHz it was safe to say that SDRAM was "just better". Incidentally, Sony did use synchronous RAM (SGRAM) in the video subsystem - presumably because they concluded that the ability to have two pages open at once would win them more performance than they would lose due to the more complicated access protocol.

Sorry for the engineering rambling...

 

I've also heard mentions of the Playstation gpu having much better fillrate because it used SGRAM too, while the Saturn used SDRAM and was extremely fillrate limited.

 

Early PS units used a 160 pin "GPU" chip with a bank of VRAMs (you know, these memories invented by Texas Instruments on the 80s with dual port, one being serial for raster based video drawing) and a SONY made DAC chip. Probably because this arrangement was not as cost effective as the SGRAM arangement it got replaced...

You can look the early PU-7 PS1 units with S-Video port they all have this design

 

Yeah, I had actually forgotten about that - I think the early PU-8s were like that too. I wonder if they gained enough performance from the SGRAM to entirely offset the loss of bandwidth from no longer having a dedicated read port? My gut feeling is that most of the time the answer would be yes, but I bet there are quite a few (non degenerate) cases where it's no...

I suspect that a lot of the reported problems with the Saturn were not really inherent to the design, but just a result of it having an architecture that made it very hard for developers (even good ones) to get their head around it. I can't honestly think of any reason that using SDRAM would kill your fill rate - it's a little bit slower on the first word of a block, but that's about it.

 

I have one of the bad type power supplies is it safe to replace the top202 with top224

 

Do what I do, check the datasheets. If the specs meet or exceed it I'd roll with it, but make very sure the voltage input/output is the same.

 

You mean like this?





This isn't even from a launch model, it is from a 1996 VA7...

 

I said hybrid circuit. That's a normal PCB with standard parts. A hybrid module is usually potted in ceramic. Also the circuit board is made of ceramic with conductive traces etched in it. And the parts installed in it are usually SMD.

 

Oh, I see.
I do have a very early launch model (made in 1994 October, with a very low serial number), I'll have to check its power supply for that.