Performa 5200CD died. related to CUDA maybe but stuck.

[ImmortanJoe]

Fairly recently I bought a couple of PPC macs in my far-too-long quest to have a working Old World PPC Mac again. One of which was a 5200CD. I got it working mostly, except for sound. And the HDD was very tired. Then it stopped working. Powers on. No power light. No boot chime or other signs of life. CPU gets warm.
So far in no particular order I've had it apart many times. Cleaned everything on the logic board and some cards including with demineralised water. Did a temporary re-cap with through hole electrolytics (the originals seemed fine. No shorts or anything but I was running out of ideas), checked PSU +5 and +12 voltages (perfect), Went over the logic board with a USB microscope (found some tiny flake debris from the plating on the rear backplate in less than desirable places) and tried another CUDA battery which was admittedly a crude stand in using three button cells.
Almost nothing had an effect. Except the CUDA battery. Two things. It discharges way too quickly. When it's installed the computer won't turn on. Not a bump. Not even a twitch from the 5 or 12v on the HDD connector. However if I take the battery out, wait a minute or so with the power switch off then try again it'll do the dead powerup. If the time period is shorter the PSU will bump momentarily then shut back off.
Last night i tried something a little different. I used a buck converter set to 4.5v to supply power to the CUDA battery connector. No voltage sag so I know it was getting the right voltage to the board. Same results in the computer as with the battery. But having a steady input supply let me have a better look while it was on the desk because the inability to power up with a voltage from the battery is strange.

I don't suppose anybody has a schematic of how the board handles the battery vs PSU power for the Egret? Something's weird and I'm not convinced it's the Egret chip because I can still power it up via the keyboard. If the chip was nonfunctional that wouldn't work, would it? Anyway on pin 13 I'm seeing 3.6v which doesn't seem right to me. And while I press the CUDA button it goes back up to 4.5v. Really weird. I'm also not sure what surrounding components should be seeing voltage when on battery power but it feels like I'm seeing too many. I'd really like some suggestions here.

I'll be re-capping again with SMD caps later. Because of postage cost being so high I'm holding off until I work out what components I may need for this and the other dead Mac I got on the same day (7300/200 which is also a no-start).
All of this is just a part of my quest to replace my terminally battery bombed G3 Beige. It's done though. I got it working for a while but the damage was too extensive. This is as far as I've made it in about two years.

So any suggestions, schematics, anything. Please let me know. Please.

 

[GRudolf94]

Keyboard power means CUDA/Egret, whatever the hell the 68HC05 in the 5200 is, is alive. Switching that between batteries and the PSU's 5V standby (trickle) rail is done by the small SOIC-8 IC marked 0120 (U28, LS1176 in the schem). From the 9600 schematic that's floating around the website, here's the CUDA bits. They're the same across the board, Apple wired it the same way pretty much everywhere.

 

[ImmortanJoe]

Let's call it CUDA. I find the other names confusing and I'm not even sure I'm using the right one. I suspected that the 0120 did something like you say but I couldn't find a match for something so generic online.
Thanks for the schematic. That's super helpful. I'll be working through it.

Edit: I'm going to re-check the weird voltage change on pressing the CUDA reset that I reported. It makes even less sense now that I'm looking at the schematic.

 

[demik]

On a 5200 it's a Cuda, since it's LC630 based.

Your problem is weird. It would look like a PSU but all your test would suggest it's fine
Do you have a Mac around with similar boards to rule out the backplane connector and or PSU ? (LC630 / 6200 / 5300, etc ?)

 

[Arbee]

All Old World PPC machines are Cuda. Egret was only IIsi, the first 3 LCs, Classic II, and IIvx/IIvi.

 

[slomacuser]

If you cant make it work again I have a 5260 board for sale

 

[ImmortanJoe]

demik, If I had another compatible-ish board I'd have already tried that but sadly I don't. Also apologies I haven't quite got the hang of how this forum's formatting works so I'm keeping it plain. Weird things happen. I've done what I could with the PSU anyway. Having the logic board trapped inside when it's powered is extremely frustrating! But I know +5 and +12 are good from the HDD connector. +5v standby / trickle must be fine or it wouldn't turn on when I press the button on the keyboard. CUDA must be at least somewhat functional for the same reason.

The inability to power on when the PRAM battery is connected is however very strange. I spent a little time last nightt with an even simpler setup of an 18650 connected to the CUDA battery header so I could probe voltages now I have that schematic. My conclusion is that I have no conclusion. The voltage drop in the circuit seems to be coming from the 0120. It was late last night so my memory is a bit fuzzy but if I recall the 0120 was seeing 4.0v in (which was what the 18650 was providing), but the output on pin 1 (VRTC) was 3.2v. IIRC with my earlier test when I was driving it with 4.5v the result was 3.8v.

I just had a rubber ducky moment. Thanks for being my duck! While the 7300/200 I have here is also dead, GRudolf94 said that the CUDA circuitry is pretty similar across models. I'm pretty sure the 7300 has a different failure mode so I can use it to at least compare some unknowns (to me) like the voltage drop across the 0120. Maybe.

slomacuser, that's a complicated one. I would have bought a G3 Beige board to replace my dead one if I could. But the lack of availability + obscene prices people are asking have prevented me buying one. Prices are way lower overseas but postage to Australia is high. I mean usually far higher than the actual board itself. For complex reasons the majority of old equipment is either buried in landfill or rusting away on foreign shores. The 5200CD and the 7300/200 were the first time in a couple of years anything has come up that I could even get my hands on and even then at a premium price especially for untested crumbly things that had been sitting in a shed for twenty years. Besides the G3 beige being one of my favourite things to use and having software for it, it also acted as a localtalk bridge for my IIgs, and probably would for my recently repaired Mac Pluses. A tip on that. If you have a padded carry case be wary of storing one in it. It worked to absorb and concentrate moisture inside. I'm glad I saved it. A friend gave it to me a long time ago. The other... I didn't even know I had it until recently and it has an interesting video breakout that I want to investigate further.
I had a look at your ad. While I'm sure it's a good price, once I factor in the relative worth in AUD and my relative lack of worth I'll have to give it a pass.

 

[ImmortanJoe]

I pulled the 7300/200 board so I could poke at the CUDA circuitry briefly. I wanted to pull it to catalogue the electrolytics anyway. The 0120 on the 7300 doesn't exhibit the voltage drop like on the 5200. It was something negligible like .01v difference. So I guess I may have found the general vicinity of an issue. Not that I have a replacement chip for the 0120. Maybe I'll get lucky and it's just a passive at fault.

 

[slomacuser]

I have shipped Powerbook 165c to Tasmania

 

[GRudolf94]

@ImmortanJoe what sort of resistance do you see between 0120 pin 7 (aka battery +ve) and ground? Might be wise to compare to the 7300.

Edit: on a trayloader iMac G3, the only Cuda-thing I have handy, it's close to 1Mohm.

 

[ImmortanJoe]

I can't get a good reading off the 0120. I think the circuit is trying to run from the power from the multimeter. What I will try is a setup similar to what I used earlier and set up the buck converter to 4.5v then I'll stick the multimeter inline and measure the current.
The 7300 measurement will be a while. I have it assembled for safekeeping and half the CUDA / RTC related circuitry including the part that I need happens to be on the underside of the board.

 

[ImmortanJoe]

On the 5200 board the current drawn from the "battery" via the terminals is roughly 30mA. So if the schematic is correct for this model there may be a resistor in between. However when I measure the resistance between the positive on the battery header and pin 7 on the 0120 it comes out at 0.3ohm. I can speculate that it has something to do with this board using a 4.5v external battery instead of a 3.6v 1/2 AA battery but I don't know for sure. Also my experiment with the draw of this circuit was promising. When I press the CUDA button the current drops to effectively zero, stays extremely low for a few seconds then goes back up to 30mA so that effectively means the CUDA's reset circuitry is functional too.
What I could really use is a confirmed correct pinout of the power part of the edge connector. There were two pins that I'm unsure of so I'm not game to power it externally yet to see what's happening. Not that I have a known good power supply which provides the correct voltages I can use at this exact moment anyway. But I really want to see what the 0120 pins and surrounding circuitry do with trickle and normal power.

While I'm here, could somebody tell me what the MTP3055E MOSFET on the board does? I can't think of what it could be controlling.

 

[GRudolf94]

30mA is way too high. Those original brick batteries are 800mAh, so you'd see one last roughly a day. The 68HC05E1 datasheet tells us that, running from a 32.768kHz oscillator, such as is the case for Cuda, supply current at 5V is 100 microampere typical, 160 worst-case. That's off by two orders of magnitude. If I were you, I'd remove 0120 and measure around it, and then it itself out of circuit.

 

[ImmortanJoe]

30mA is way too high. Those original brick batteries are 800mAh, so you'd see one last roughly a day. The 68HC05E1 datasheet tells us that, running from a 32.768kHz oscillator, such as is the case for Cuda, supply current at 5V is 100 microampere typical, 160 worst-case. That's off by two orders of magnitude. If I were you, I'd remove 0120 and measure around it, and then it itself out of circuit.

That's what I observed when I cobbled together a battery for it using three button cells. I was literally watching the voltage drop via the multimeter. What's been bugging me is I haven't seen any indicator that the 0120 has actually failed. But you're absolutely right. The only way to work it out is to pull it. I've been avoiding this because most of my equipment is a long way away from me so I don't have access to it. I've only got a Pinecil and some basic gear here. I actually found myself looking online for an old scope yesterday out of frustration. That was a no because the reasonably priced ones are too far away. Same goes for the other stuff like soldering / desoldering / rework stations. I don't want to double up just for this.

 

[alectrona2988]

If you're having power issues it is likely due to a failing PSU. I'm personally not a huge fan of the 5200 series because it had a considerably rocky lifecycle. Be sure to take EXTREME care when handling it though! Ask me how I know...

 

[ImmortanJoe]

If you're having power issues it is likely due to a failing PSU. I'm personally not a huge fan of the 5200 series because it had a considerably rocky lifecycle. Be sure to take EXTREME care when handling it though! Ask me how I know...

I'm pretty sure it's not the PSU. 5v and 12v are rock solid. There's something wrong in the CUDA general region with the way it handles power. I haven't gotten around to doing more since the last post. I did have a chance to check a store for some basic supplies I needed but they didn't have anything.
Not sure I need to ask you. I named the computer "Crumbles".

 

[alectrona2988]

Ah, just wasn't sure if anyone had asked about that. Also, that's quite a fitting name for any PM5200 machine.

 

[68kPlus]

Hello fellow Aussie
Although I don't know enough about components (slowly learning), perhaps the eBay seller ComputerChange can help.
Shoot him a message and he might have something you can use.

Security measure

www.ebay.com.au

 

[ImmortanJoe]

Ah, just wasn't sure if anyone had asked about that. Also, that's quite a fitting name for any PM5200 machine.

Being careful isn't enough with this one. It's so brittle that everything just breaks and crumbles. I just finished printing a new hard drive sled for it. Yes I know it's for a non-working computer. But I figure if I can get some nice prints happening I might be able to sell them for a modest amount to people who want functional parts over originals. Mostly so I can afford parts to repair my Macs, and buy more filament.

Hello fellow Aussie
Although I don't know enough about components (slowly learning), perhaps the eBay seller ComputerChange can help.
Shoot him a message and he might have something you can use.

Security measure

www.ebay.com.au

Thanks for the heads up. I don't know why I've never seen this seller when I look on eBay. FB marketplace is an absolute bust. Everything is either too far away or too expensive. Being too far away also makes things too expensive quite rapidly. It really doesn't take far before I'm looking at $100+ extra in petrol.
While I'm here, does anybody know if the 0120 is a rebadged commodity IC? Where it's labelled in that schematic as an LS1176. That's no good. The only match I can find is for a differential bus transceiver.

 

[GRudolf94]

Nope, that's a custom part. 0120 is shorthand for 343S0120. You could theoretically replace it with some other battery voltage monitor IC, but that would need a bodge.