• Please review our updated Terms and Rules here

Profile Hard Drive PSU giving low voltages, any usually suspect caps?

mikerofone

Experienced Member
Joined
Aug 9, 2020
Messages
94
Location
Switzerland
Hi all,

(TL;DR: My Profile got more and more unreliable in the last two years: It worked fine two years ago, then sat for the better part of a year after which it required multiple power-off/power-on cycles to initialize properly. After an additional 20h of runtime Lisa started crashing more and more while loading / putting away programs, and now the drive no longer initializes or performs its selftest at all. Now the drive spins up and may move the heads for a second, the busy LED might be on or off, and maybe after 20-60 seconds it will move the heads a bit and the LED resets as if the init routine restarted and failed again.

I measured the voltages recently and the 5V rail was at about 4.39V - way too low and that might explain why the drive controller is malfunctioning. I assume the voltages kept sinking over time which is why the drive gradually became more unreliable. I took some pictures of the PSU internals and there is some suspicious black gunk or charring on the PCB close to / under whatever that big heatsinked flat silver component is. My first thought was bad or leaky caps, but before I go recapping the entire thing I wanted to check whether there's any usual suspect caps in Profile PSUs that typically fail. (Besides the RIFAs obviously, which I discovered to my great horror were not replaced by the seller even though he mentioned "necessary caps replaced" when seeling the kit.)
I see that most if not all are Nippon Chemicon SXA series caps, which I've seen a lot in similar vintage gear, and I don't remember seeing these spill their guts. Can't say the same about those Nichicon PL(M) series caps I've met in various places...

There also seems to be some discoloration of the PCB under and around that gunk. As someone who knows way too little about PSUs, could that indicate a problem somewhere else in the circuit which maybe damaged the caps, instead of the caps failing on their own?

Thanks for any insights you might have!

Cheers
mikerofone
 
The board in the charred area looks like typical "wet" from leaking electrolyte. All of the caps need to be replaced, including the RIFAs.

Bad capacitors can be the source of the charring. The electrolyte is often conductive, so it can bridge different parts of the circuit that shouldn't be bridged and start cooking. Bad capacitors can also become electrically leaky and effectively turn into resistors, which can overload components both upstream and downstream of them and cause hideous amounts of damage.

The large TO-3 package IC screwed to the large heatsink is either a transistor or a mosfet, not sure. Someone more familiar with the SMPS from a Profile drive may know what it is. A common part used was a 2N3055.
 
I second that. One or more of the 47uf 250v primary DC filter capacitors are leaking and the electrolyte is conducting in contact with nearby components.

(Besides the RIFAs obviously, which I discovered to my great horror were not replaced by the seller even though he mentioned "necessary caps replaced" when selling the kit.)
Which caps??? The RIFAs are not done and if I'm reading the Nippon Chemicons correctly the date codes are nearly 40 years old. This power supply looks completely untouched.
 
Thanks for taking a look and confirming my suspicion! So I'll replace them all and won't take any chances.
Which caps??? The RIFAs are not done and if I'm reading the Nippon Chemicons correctly the date codes are nearly 40 years old. This power supply looks completely untouched.
To clarify, the seller said he replaced caps "where necessary". Apparently, Rifas don't automatically make the cut for him. I checked my message history with him and he thinks that if they last 2h when testing, they'll last forever so he doesn't replace them. /o\ I did check the Lisa PSU afterwards and those Rifas were replaced by different models already, but I didn't think to check the Profile. Now I'm happy that it failed in this unspectacular and very fixable fashion, instead of letting out the stink in my home or during an exhibition. :)

Cheers
mikerofone
 
That PSU will let out the stink when you start removing those capacitors. Touch any of that leaked electrolyte with a soldering iron and your whole house will smell like rotting fish for hours.
 
To clarify, the seller said he replaced caps "where necessary". Apparently, Rifas don't automatically make the cut for him.
No, like unless he carefully peeled the warning label off the old capacitors in the primary side and stuck them to the new caps, all the capacitors have the same date codes. Every capacitor looks to be original with a 1984 date code. He flat-out lied to you and didn't change them.
 
No, like unless he carefully peeled the warning label off the old capacitors in the primary side and stuck them to the new caps, all the capacitors have the same date codes. Every capacitor looks to be original with a 1984 date code. He flat-out lied to you and didn't change them.
Well, he did qualify it with "necessary" caps replaced. He could argue none were necessary.
 
That PSU will let out the stink when you start removing those capacitors. Touch any of that leaked electrolyte with a soldering iron and your whole house will smell like rotting fish for hours.
Yep, I'm familiar with that lovely odor and "looking forward" to that! -_- need to order the replacements first though.

No, like unless he carefully peeled the warning label off the old capacitors in the primary side and stuck them to the new caps, all the capacitors have the same date codes. Every capacitor looks to be original with a 1984 date code. He flat-out lied to you and didn't change them.
"Whoops, forgot to remove that bit in the listing, that was left in from the previous Lisa I sold." Possibly still a lie, but at least the Rifas in the Lisa PSU were replaced.
oldpcguy is kinda right, the seller replaced whatever was necessary to get about 60h of runtime out of the set before it broke down. Not quite "A++ would recommend!" and the price in my eyes was less justified after seeing this, but not a complete ripoff either. Oh well, live and learn! :)
 
I finally got around ordering the caps and recapping the board. I checked what was installed on the existing board (my list, which matches the listing on bitsavers - page 33) and replaced only the electrolytics with Panasonic FR (low ESR) series caps, except for C6-C9 for which I used Panasonic ED (high ripple) caps. The new caps are all much smaller than the old ones, but googling suggests that this is to be expected simply due to newer technology being more compact.

I now regret not testing the PSU out of circuit before recapping, as I didn't want to do the first test on my actual ProFile hardware. Without any load connected, the PSU "hiccups", which is probably expected. I connected a 3.5" PC HDD to the +12, +5 and GND pins as a dummy load which is likely much smaller than the real hard disk, but the hiccups remain. I have not run the PSU for more than 2-3 seconds as I don't know if doing so could harm anything, but the connected HDD doesn't start spinning up. Connecting an automotive dashboard light bulb (12V, 12W) to the 12V rail makes no difference.

The voltages I read at the HDD during that time are +15V on the +12V rail, and the 5V rail fluctuates between 1-6.5V - yikes! So I assume the hiccup is the overvoltage protection kicking in. The HDD survived, by the way. ;)
Edit: I cannot find any variable resistor or something to dial in the voltages, is there none?

Any suggestions what could cause this behavior? I've double checked the capacitor values and polarities of all new caps and they are fine. I've added pictures of the PSU with the caps removed and new ones installed to the photo album (scroll down a bit), in case you want to visually inspect my work. :)

Cheers
mikerofone
 
That power supply is likely group regulated off of the +5v rail, so the +5v rail would need to be sufficiently loaded for a stable output.

I would recommend checking the other components, like resistors, diodes and the small transistors to see if any of those are out of spec.
 
Thanks for the suggestion! I may need to rig up some bigger load then. I'll try connecting the light bulb to the 5V rail, or two in series. Maybe that'll draw enough current. I'll check the ratings of the 5V rail and experiment a little on the weekend. :)

Cheers
mikerofone
 
Success! \o/ This was a good reason to finally set up this cheap variable load I got a while ago, but so far haven't yet used. I hooked it up to the 5V rail, and given the 1.5A marking next to it on the PCB, set the load to 1.0A. The PSU no longer hiccuped, but the load only reported a voltage of ~4.50 V. :( It went up and down only by about 0.1V when tweaking the load to be between 0.8A-1.5A, which was disappointing - that's were I started! Oddly enough, the PSU started hiccuping again when I went below 0.6A, at which point the load only reported about 5.2V.

Just to be sure, I set it to 1.0A and got my multimeter out. And that measured ~5.1V across the stable range. So, the lesson is: Always test new measurement devices before you deploy them to production. :p (Especially if they were cheap. Edit: On closer thought, the very thin leads that probably made a pretty poor connection might be to blame. I measured the voltage at the PSU, not at the variable load. Either way, still a good lesson was learned! /Edit)
I've jammed an automotive 12V, 12W lightbulb (the type is apparently called W5W) into the socket for the 12V, 2A line and it runs a little hot at 12.35V, but I think that's just ever so slightly acceptable. Possibly the 1A this lamp draws is on the low side, so I assume that the HDD motor draws more and will be fine.

Excellent, so the repair seems to have been succesful! Next up: Testing the PSU connected to the original hardware.
 

Attachments

  • PXL_20240603_161018440.jpg
    PXL_20240603_161018440.jpg
    2.7 MB · Views: 6
Last edited:
Hm, so with everything reconnected, the voltages are fine (4,95V on 5V rail and 11.6-11.95V on 12V depending on motor startup or idle) but the controller doesn't initialize. The drive spins up, does a head seek, but then after ~15-20s, when the surface scan should start, the LED just blinks 1-3 times, sometimes accompanied by a head stepping noise. I've connected the Profile to the Lisa but it fails to detect the drive after hanging for a long time. Resetting the Lisa seems to trigger a reinitialization of the controller, sometimes with a *clack* noise which might be heads maybe hitting an end stop. I've added some (long and boring) videos to the album. The last and longest video has me turn on and off the ProFile and resetting the Lisa in various ways, with all resulting in what seems like a self-test that immediately aborts.

I get the impression that the initialization test fails, hence the controller stops. I found https://apple3.org/Documents/Manuals/Profile_Level_2_Manual.pdf and will dig through the debugging steps, but if in the meantime someone has a good idea of what might be wrong, I'm open to all ideas!

The drive was (sporadically) working before the PSU got too bad. Page 17 in that service manual suggests that the drive did not park but remain at track 0, as that's where the head stays when I shut the ProFile down. Since the drive died at an exhibition and was transported home by car afterwards, maybe there's a chance that the surface got damaged in transport. That'd be a real bummer. :(
 
I've let the drive run for a while and it is performing the self-test, but most sectors seem to fail and the controller spends a minute or so trying to read the track. I've waited for about 30-45 minutes and the drive is progressing through the sectors. Some tracks seemed to pass as the drive stepped to the next track in the usual speed (0.3-0.5 seconds) but on most it lingers. Towards the end, it does a lot of returns to the 0th and parking tracks. Eventually, the test finished and the ready light came on, but turning on the Lisa reset the controller, and the self-test restarted.

I do not have a solution yet, as low-level formatting seems to require a special Z8 chip and ROM, and possibly also an Apple /// computer. I'll take the drive assembly out of the case and will wiggle and reseat all the cables, as it was pretty much working fine before the PSU kicked the bucket. Hopefully it's something simple as that, and not a broken head or something.
 
Does the hard drive use a stepper motor to move the heads? You may try putting some oil in the bearing. If you're getting errors on most of the drive, you can only hope it's something as simple as that, otherwise it's a more serious hardware fault.

Also, get that RIFA X2 capacitor out of there. Those things tend to go on fire from the housing failing and letting air inside to degrade the capacitor. It sort of looks like it's already getting hot with the brown dot on the side.
 
The heads are moving freely, and the disk is seeking normally. In its test routine it does a couple of back and forth movements and they look and sound perfectly fine. So I'm afraid it's not going to be sth trivial like that, but I'll give it another look. I also parked the disk manually by moving the rotor with my finger, and it was easy to manipulate.

Thanks for the warning about the Rifa! I'm well aware of the risk and replaced all others. The one still in there was a replacement by the previous owner and should be good for another two decades. :)
 
I'd like to try that next, but so far I could only find instructions that require special ROMs and a "piggyback-z8" in the Profile.

I came across some hints that the Lisa toolkit for the Profile (forgot the name) might then be able to emulate the commands an Apple /// would send to kick off the LLF, so at least I wouldn't need to hunt down an apple /// too. :)

I now wonder though whether the erratic behavior the Profile has shown before it stopped working completely was the same problem slowly manifesting itself: it did require a few repeated power cycles to initialize a few hours before that no longer worked either. Is it possible that the out of spec (too low) voltages might have damaged some logic on the controller or the HDA over time?

I feel like I need to break out the scope and measure the signal from the heads. There are instructions in the service manual I linked above, but the example waveforms are all lost due to the poor b&w scan quality. :/
 
instructions that require special ROMs and a "piggyback-z8" in the Profile.
I've been circulating piggyback Z8s amongst the Lisa community for just this purpose. Let me know and I can communicate one for you to borrow. You will need to supply and burn the ROM unless I can find one kicking around here.

at least I wouldn't need to hunt down an apple /// too. :)
The BLU program running on a Lisa will initiate a format. If I recall correctly, it's not a very complicated process: it just sends one or two commands to the ProFile that effectively mean "format yourself".
 
Back
Top