Amiga 3000UX CSPPC stability problem

[Lianna]

Hi,

In a bout of some retro computing I've recently "rescued" my Amiga 3000UX from storage, and I have one particular annoying issue remaining.

My "base" config: A3000UX (68030+68882/25 on board), 2MB chip memory, OS 3.9, 4.3GB SCSI HDD, Ariadne Ethernet, CyberVision 64/3D, FDD, original 240V 135W PSU with a new Noctua fan upgrade. I've removed the NVRAM battery (fortunately no leak damage) and replaced it with a Lithium coin cell. I've cleaned everything with PCB cleaner, and repaired the broken LED front panel. All the caps are original and look in okay shape (I might still replace them though, for a sense of security).

I've got a Cyberstorm MkII (68060/50 rev. 1, 128MB 60ns memory) which has always been stable. I also have a Cyberstorm PPC with 604e 200MHz, which I've always had difficulty with (in the past I never could get it to work well, I may have sent it for an RMA to Phase 5 back in the day to check for any problems ... can't exactly remember). As this 3000 was my daily driver back in the day I never spent a massive amount of time on troubleshooting the problem, and I simply stuck with the CS MkII. Until one day late in the 90's I jumped ship to a Pentium ...

Anyway, yesterday I started testing the old Miggy with the CSPPC to see if I could figure out what the problem exactly is ... swapped the '060 and memory over, switched the J100 and J102 jumpers, put the CSPPC in instead of the MkII which was still working fine, let the system boot to a CyberGraphX screen mode ... and got quite reliable lockups, such as when doing something as simple as opening a folder with several files in.

Trying the same "folder opening" while it was still booting and stuck in an ECS mode with CacheCDFS nagging me about the external CDROM not being connected, I discovered I was not getting these same crashes!!

I've taken the CV64/3D out today and haven't had a lock up yet (so far only really doing backups through Amiga Explorer though).

Testing and configuring things has been slow and difficult as I can't find my keyboard (already gone through most of storage) and thus I've had to do almost everything by prepping floppies on my A500+, I've had to make use of an on-screen keyboard application which often doesn't work at all, and sometimes have to use Amiga Explorer to e.g. locally edit tooltips in Amiga Forever and then send the icon back across. All quite tedious! I've now got a keyboard adapter cable on the way, which at least should allow me to run the CSPPC test disk (it needs keyboard input, unless I'd modify it on Amiga Forever I guess).

Google hasn't been all that much help, except for one thing which I've never considered: the original PSU may have been the culprit with the lack of CSPPC stability all along. My current suspicion is that the PSU can't cope well enough with all this hardware, and I may need to replace it to fully test the stability with the CV64/3D added. Btw, I've not yet tried a watt meter to check how much it's pulling at the wall.

But maybe I've missed something obvious? Any suggestions on what to test?

[patrik]

135W should be enough and I would say that your level of expansion is fairly normal. The card is sold as A3000/A4000 compatible, so it is designed to fit the power budget of an A3000, as you were not generally expected to replace the PSU in big box amigas to be able to run an accelerator back in the days.

Have you checked the capacitors in the PSU? In the PSU in my A4000 they leaked through the bottom of the capacitor, so the top of the capacitor looked ok, but you could see that it was dirty on the pcb around the capacitors.

The difference in load that the missing CV64/3D makes should not make much difference regardless. Could something perhaps work a bit differently with the Zorro3 bus when using the CSPPC? What buster chip version do you have? The CV64/3D manual recommends buster11, but if not I agree it would be a long shot to upgrade given that it works fine with the other accelerator.

[Lianna]

Thanks for your reply patrik.

I thought I had checked the capacitors "carefully", guess I hadn't





I did a couple of years of electronics way back in school, I should've known better

My buster is an -11, so no further upgrades to be had there.

Given that these large caps have spilled their guts (I'm amazed they're still working, though the goop has solidified - probably long ago) I'm not expecting the PSU to be able to deliver sufficient, stable power.

I've ordered new capacitors and a high wattage resistor so I can discharge them safely, it will be at least a week before I get them though.

Not all that hopeful it'll fix the issue, but the power supply hadn't been opened before ... these leaks could have happened before I got the CSPPC.

[ajk]

That yellow stuff is probably just some glue they have used to fix things in place. I have at least never seen a cap leak like that - usually the electrolyte is a clear, somewhat oily substance.

Doesn't mean they might not still need replacing, though.

[Lianna]

Originally Posted by ajk

That yellow stuff is probably just some glue they have used to fix things in place. I have at least never seen a cap leak like that - usually the electrolyte is a clear, somewhat oily substance.

Doesn't mean they might not still need replacing, though.

Hmm hadn't considered that. There's quite a lot of it though, and it looks like it went there by leaking rather than being applied to it ... but it's all quite hard to tell, and even more difficult to get it on camera. I'll get more pics once I properly disassemble everything and replace the caps.

[patrik]

In the first picture it is definately glue and nothing else. I don't know why, but components are often glued to the pcb in power supplies. Anyhow, dust tends to stick in the leaking electrolyte so it will look dirty around the capacitor.

[mech]

Originally Posted by Lianna

Hmm hadn't considered that. There's quite a lot of it though, and it looks like it went there by leaking rather than being applied to it ... but it's all quite hard to tell, and even more difficult to get it on camera. I'll get more pics once I properly disassemble everything and replace the caps.

That is glue you are seeing, they do this for several reason, its easier to hold the caps to the board for wave soldering it and or they do this to keep vibration down on the caps which could lead to cracked solder joints etc.

cap leakage is usually liquid and many times will smell like dead fish. a tell tale sign is alot of oily residue and dust collecting heavily in the area of leakage.

having said that thought, all amiga caps are way past design life and MTBF so its a good idea to recap the amiga and psu.

As for your 3000 problem, what buster chip version do you have and has the INT2 hack been performed on the A3000?

I would check the power supply under load,you can do this by probing the psu connector at the motherboard from the back side with a multimeter. +5 is the big concern,since anything under 4.92 volts will cause erratic operation i find. -5 can show up as video problems, -12 audio problems and rs232 if memory serves. +12 usually for harddrive motors and floppy motor etc.

[Lianna]

Originally Posted by mech

That is glue you are seeing, they do this for several reason, its easier to hold the caps to the board for wave soldering it and or they do this to keep vibration down on the caps which could lead to cracked solder joints etc.

Cheers, makes me feel a little better about still using it.

Originally Posted by mech

cap leakage is usually liquid and many times will smell like dead fish. a tell tale sign is alot of oily residue and dust collecting heavily in the area of leakage.

Yeah, I can't exactly remember what it smells like, but it's a distinctive smell when an electrolytic cap explodes.

Originally Posted by mech

having said that thought, all amiga caps are way past design life and MTBF so its a good idea to recap the amiga and psu.

As for your 3000 problem, what buster chip version do you have and has the INT2 hack been performed on the A3000?

Buster 11, and yeah, the INT2 mod is also in place. Though I don't yet use the SCSI interface on the card (which I'll need to use for AmigaOS 4.1 FE Classic, as I understand it ... scsi.device is apparently not supported). Thought that would've been a good way to really test the card ...

Originally Posted by mech

I would check the power supply under load,you can do this by probing the psu connector at the motherboard from the back side with a multimeter. +5 is the big concern,since anything under 4.92 volts will cause erratic operation i find. -5 can show up as video problems, -12 audio problems and rs232 if memory serves. +12 usually for harddrive motors and floppy motor etc.

That's a good idea, though getting under the motherboard will be tricky as it'll need to have the daughter board in place to even boot. I'll start with probing a molex connector I guess.

[mech]

Originally Posted by Lianna

Cheers, makes me feel a little better about still using it.



Yeah, I can't exactly remember what it smells like, but it's a distinctive smell when an electrolytic cap explodes.


Buster 11, and yeah, the INT2 mod is also in place. Though I don't yet use the SCSI interface on the card (which I'll need to use for AmigaOS 4.1 FE Classic, as I understand it ... scsi.device is apparently not supported). Thought that would've been a good way to really test the card ...


That's a good idea, though getting under the motherboard will be tricky as it'll need to have the daughter board in place to even boot. I'll start with probing a molex connector I guess.

You dont need to get under the board, probe thru the top psu power connector,you can reach the pins inside.

UWscsi is the best thing about the csppc.. 35MB/s(27MB/s with acard sata adapters) or so with good drives vs stock a3000 5-6MB/s Os 4x will need csppc scsi.

[Lianna]

I've finally got around to do some probing, voltages look fine:

+12V is +12.13V on the PSU connector, +12.06V on a molex connector.
-12V is -12.22V on the PSU connector.
+5V_USER is +4.94V on the PSU connector.
-5V is -4.92V on the PSU connector.
+5V is 4.98V on the PSU connector, +5.05V on a molex connector.

(Molex voltage measured with ground on the molex as reference, while PSU connector voltages were measured with the chassis as reference).


Different question: I've ordered an ACARD AEC-7720UW adapter, and it would be really helpful if someone could tell me exactly what cables and terminators I should order to allow me to connect this adapter plus a 50 pin SCSI CDROM to the CSPPC. My understanding is I'd need 2 active terminators, but besides that I'm quite lost on what I need (it's not surprising SCSI never was a big hit with consumers).

[Lianna]

Pfew, that was a ton of work. I've just finished recapping the PSU and the 3000, and did a quick boot test. I've yet to test extensively, and I've yet to try the CSPPC, but I haven't got the right cabling yet to try the 7720UW.

The goop on the capacitors was indeed glue, and what a joy to get those capacitors off with that glue. I've tested the capacitors as I desoldered them, and most of them tested quite okay (only the ESR was a bit high, generally in the 1 ohm range). None of the caps on the motherboard or PSU were way out of spec, and I couldn't find much in the way of signs of leaking except a few small spots. The bigger caps on the daughterboard though had all lost a significant amount of capacity, and one of those had probably leaked a bit.

[A4000Bear]

The CSPPC has a surface mount 68060 socket, and I have found that there are almost always dry solder joints associated with it.

To test for this, try pressing firmly on the top of your 060 chip while powering up the Amiga. If the Amiga works OK while the 060 is pressed down, then you have dry solder joints.

I repaired mine by doing the following. Get a soldering iron with a small tip. Make sure the tip is absolutely clean with no excess liquid solder on it at all. Press the tip into the 'cup' of each individual socket. Hold it there for about 3 seconds. To check the solder has melted, give the tip a very slight gentle wiggle. If it moves, then you can remove the tip and move to the next socket. It is time consuming and you have to be very careful, but it does fix the problem.