Troubleshooting A1200 w/ overclocked b1260

[hall]

Hi,

I could use some help troubleshooting my 1200. It has a Blizzard 1260 fitted, treated by stachu (though not bought directly from him). This card was sold to me with a 77.xx MHz crystal, but I can't get my system stable above 66 Mhz.

At 77 MHz I get all kinds of seemingly random crashes, reboots, gurus and freezes. However, memory always tests OK. I have two SIMMs which both should be up to the task. One is the module the b1260 was sold with, the other is the very same type of 64 MB module that Novacoder successfully runs at 80 MHz.


Some examples of strange behaviour I recall:

* Sound distortion, low frequencies more than high. Gets progressively worse from power on. However, it only seems to happen with newer demos and not with oldschool WHDLoad demos or 4-chn modules.

* Not executable-error. Last night I successfully ran the Human Traffic-demo (at 77 MHz). When it ended I thought "Great! Let's try it again!". So I did, but then I got the file is not executable-error. So i tried protect +e. Still the same error. Rebooted, and then it was fine again.


Hardware info:

Commodore A1200 1d1 motherboard (bought from Amigakit about 3 years ago)
3.1 roms
Phase5 Blizzard 1260 (with latest mask MC68060)
Indivision AGA mkII
Lyra2
Micromys v3
FastATA mkIV
8 GB disk-on-module (with PFS3_aio)
ATX power supply (also from Amigakit)


I've tried cleaning mobo expansion contacts and SIMM contacts, tried removing everything but the Indivision, tried a CF-card + CF-adapter which I used when this was a 030 system. Obviously without success. The fact that the memory modules passes the memtests combined with the sound distortion makes me suspicious towards the motherboard. I've been given the impression that the 1d1 mobo needs no timing fixes etc. though. Bad caps? On visual inspection they do look fine, and all solder joints are nice and shiny.

Any ideas? All help would be deeply appreciated.

[Turran]

Tried with the trapdoor open? Elevate the amiga and try with a fan blowing up onto the card from underneath.

Can also try with the top off, blowing air from the top.

That should at least tell you if its overheating..

[hall]

Thanks for the suggestion Turran.

I forgot to mention I've modded the trapdoor with a large hole with a mesh covering it. But yeah, I've tried with the trapdoor completely removed too. Currently the CPU is fitted with a low profile heatsink and a fan blowing air through that. At the moment it also has a fan fitted (by stachu i guess) on top, which I plan to maybe remove later as i prefer the fan on the CPU side.

So, fans on both sides . The CPU gets warm of course, but i can put my finger on it and keep it there without a problem.

[demolition]

Like with any other troubleshooting, I would start by disassembling the system down to the bare minimum, i.e. the A1200 MB with the 1260 card in place and nothing else (ok, maybe a CF card or it'll be hard to test) just to make sure the fault lies in one of those two.
When that has been verified, the easiest would be to test the card in another A1200.

That being said, if the supply voltage the card gets from the A1200 is not clean enough, it could compromise the stability of the 1260, particularly when overclocked. Bad caps lose capacity (even before they start leaking), thus increasing the ripple on the supply voltage.

[hall]

Originally Posted by demolition

Like with any other troubleshooting, I would start by disassembling the system down to the bare minimum, i.e. the A1200 MB with the 1260 card in place and nothing else (ok, maybe a CF card or it'll be hard to test)

That's kinda what I did, except i left the Indivision in, as no image also makes it hard to test . I suppose though I could bring down my old 1084 from the loft and pop out the Indivision too, just to be really sure.

[demolition]

Originally Posted by hall

That's kinda what I did, except i left the Indivision in, as no image also makes it hard to test . I suppose though I could bring down my old 1084 from the loft and pop out the Indivision too, just to be really sure.

You don't have some TV standing around with Composite input?
But yeah, I don't think the Indivision has anything to do with it, it's just a general rule of troubleshooting as you can't be 100% sure.

[hall]

Allright, done some more testing. I removed *everything* - the only two things connected to the a1200 mobo was the b1260 and the disk-on-module thingy. It made no difference whatsoever I'm afraid. If anything, it made the system even a little more unstable, but that may be my mind playing tricks on me.

So... anything else to try out if you, like me, don't happen to have a spare a1200 laying around?

[demolition]

Could try a different PSU.

At this point though, I'd probably change the caps if it hasn't been done before, at least the ones for the 5V rail.

[Turran]

If you live near Stockholm you can pop over with the card or borrow an A1200 for a few days..

[hall]

@Turran Thanks for the offer, that's very kind, but I'm nowhere near Stockholm.

I'm quite sure the caps are the original ones, and they are of course not getting any younger. The sensible thing is probably to have them changed and I might as well do so before they have made a mess inside.

[hall]

Time for an update.

The good news is that the cap-change went fine and as it turned out, a few of them had actually started to leak. The bad news however is that it didn't solve the problem. I'm having more or less the same behaviour as before.

Thinking about my next move...

[hall]

Good evening fellow Amigans. I thought I'd revive this thread of mine and share with you my story of success. It has been a long and frustrating journey, during which I almost gave up a few times, but I persevered, learned a lot and got there in the end!

So, in my last post I had just recapped my 1d1 board. Oh, and I went for ceramics, as per this thread. That didn't solve anything, but at least the board is now in very nice condition, I won't have to think about those caps for quite a while AND I bumped up my soldering skills/confidence a few notches!

Out of frustration I then tried my A600 PSU (the 3A one) just so that I would have tried all possibilities I currently had available. Now surely that underpowered brick would render the system even more unstable right? Wrong. Things did actually improve by a hair, just enough for me to take notice. Up until this point I had been confident power was not the issue at hand as I had the ATX conversion from Amigakit. Next thing, I switched the Amiga power cable in the ATX PSU back to the one it was originally supplied with (I did put a longer one there a few years back to accommodate my desk setup at the time and it had served me very well, pre-060). With the shorter cable the ATX was doing a little bit better than the A600 brick, but still far from a stable system.

Now convinced that power was the problem after all, I went out and got a meter. I also added power to the floppy power connector which bumped up the voltage readings a tad, but still no cigar. This is when I started going through power supplies... Over the last two months I've been through quite a few, well known brands as well as cheap sh*t. The best I could manage was I think 4.87V measured on the B1260. From what I've gathered, above 4.8V should be OK but clearly not in this case, albeit a lot better than from where I started out.

What's left to swap out then? The mobo of course! By chance I stumbled upon a 1200 locally a few days ago, so I'm happy to announce another Amiga rescued back into the community and at a fair price too. Case and keycaps cleaned up nicely, slightly yellowed, but the overall impression is “white”. The motherboard turned out to be a 1d4 in good condition after a bit of IPA+toothbrush. Did this board solve my problems then? No, it did not. In fact it made everything a little bit more unstable again. BUT! This Amiga came with something else, namely an A500 PSU – the 4.5A one. Now this PSU improved things significantly! Voltage readings were around 4.92A, again measured on the B1260. Now, slightly above 4.9V really should cut it, no? Well not my particular stubborn Blizzard. The system was almost stable, but I could still provoke crashes by running demanding applications. At this point I'm starting to believe that power is not the problem any longer, there has got to be something more to it...

So tonight I decided to try the very last thing I could think of regarding power. Yep, I swapped out the cable from the A500 PSU for a shorter one... Lo and behold! I now get rock-steady voltage readings of 5.10V everywhere I measure and the system appears to be completely stable. At the time of writing this it has been successfully running memtests for quite a while prior to running Quake for almost an hour now!

All of this brings a couple of questions to the table. At some point I might have to replace this PSU. Does there actually exist an ATX (or other) PSU that can deliver the goods on the +5V rail? None of the ones I have tried at least, including two Seasonic ones. Is my particular Blizzard unusually picky?

This turned out to be a long post. Thank you for reading – I hope you found it at least somewhat worthwhile.

[demolition]

Since it is definitely a power issue, there are a few things which can be done. I'd strengthen the connection between the power connector (or rather, just after the power filter coil) and the expansion port using a wire on the back side of the MB. This will make sure it can get a bit cleaner power.
Also, I'd add some extra decoupling caps somewhere close to the expansion port to reduce the supply ripple which may be the source of your problems. It should reduce the sensitivity to the power supply.
Also, check the decoupling caps on the Blizzard.

I just got a 120W PicoPSU which I will be integrating into a dead Amiga PSU case and I think it should be more than capable to run a setup like that, however, since you did try some Seasonic ATX supplies already (which are usually good), it is hard to say if it changes anything.

[Jope]

Get an arcade PSU with a tweakable +5V rail. :-) Unfortunately they don't supply -12V, though.

[Sandro]

Originally Posted by hall

Good evening fellow Amigans. I thought I'd revive this thread of mine and share with you my story of success. It has been a long and frustrating journey, during which I almost gave up a few times, but I persevered, learned a lot and got there in the end!

So, in my last post I had just recapped my 1d1 board. Oh, and I went for ceramics, as per this thread. That didn't solve anything, but at least the board is now in very nice condition, I won't have to think about those caps for quite a while AND I bumped up my soldering skills/confidence a few notches!

Out of frustration I then tried my A600 PSU (the 3A one) just so that I would have tried all possibilities I currently had available. Now surely that underpowered brick would render the system even more unstable right? Wrong. Things did actually improve by a hair, just enough for me to take notice. Up until this point I had been confident power was not the issue at hand as I had the ATX conversion from Amigakit. Next thing, I switched the Amiga power cable in the ATX PSU back to the one it was originally supplied with (I did put a longer one there a few years back to accommodate my desk setup at the time and it had served me very well, pre-060). With the shorter cable the ATX was doing a little bit better than the A600 brick, but still far from a stable system.

Now convinced that power was the problem after all, I went out and got a meter. I also added power to the floppy power connector which bumped up the voltage readings a tad, but still no cigar. This is when I started going through power supplies... Over the last two months I've been through quite a few, well known brands as well as cheap sh*t. The best I could manage was I think 4.87V measured on the B1260. From what I've gathered, above 4.8V should be OK but clearly not in this case, albeit a lot better than from where I started out.

What's left to swap out then? The mobo of course! By chance I stumbled upon a 1200 locally a few days ago, so I'm happy to announce another Amiga rescued back into the community and at a fair price too. Case and keycaps cleaned up nicely, slightly yellowed, but the overall impression is “white”. The motherboard turned out to be a 1d4 in good condition after a bit of IPA+toothbrush. Did this board solve my problems then? No, it did not. In fact it made everything a little bit more unstable again. BUT! This Amiga came with something else, namely an A500 PSU – the 4.5A one. Now this PSU improved things significantly! Voltage readings were around 4.92A, again measured on the B1260. Now, slightly above 4.9V really should cut it, no? Well not my particular stubborn Blizzard. The system was almost stable, but I could still provoke crashes by running demanding applications. At this point I'm starting to believe that power is not the problem any longer, there has got to be something more to it...

So tonight I decided to try the very last thing I could think of regarding power. Yep, I swapped out the cable from the A500 PSU for a shorter one... Lo and behold! I now get rock-steady voltage readings of 5.10V everywhere I measure and the system appears to be completely stable. At the time of writing this it has been successfully running memtests for quite a while prior to running Quake for almost an hour now!

All of this brings a couple of questions to the table. At some point I might have to replace this PSU. Does there actually exist an ATX (or other) PSU that can deliver the goods on the +5V rail? None of the ones I have tried at least, including two Seasonic ones. Is my particular Blizzard unusually picky?

This turned out to be a long post. Thank you for reading – I hope you found it at least somewhat worthwhile.

good info thanks for the explanation
so basically you say that a short cable solved voltage problems?
how long is the cable of your psu now? from where you get it ?
which psu are you using now ? A600, A500 or the ATX PSU?

[hall]

Originally Posted by demolition

Since it is definitely a power issue, there are a few things which can be done. I'd strengthen the connection between the power connector (or rather, just after the power filter coil) and the expansion port using a wire on the back side of the MB. This will make sure it can get a bit cleaner power.

Good advice. Any particular points close to the expansion connector you have in mind? Like I (perhaps a bit unclearly) mentioned, I already have wires from the connector going to the floppy power connector, but you are talking more or less straight to the appropriate pins on the expansion connector right?

Originally Posted by demolition

Also, I'd add some extra decoupling caps somewhere close to the expansion port to reduce the supply ripple which may be the source of your problems. It should reduce the sensitivity to the power supply.
Also, check the decoupling caps on the Blizzard.

This is a bit beyond my knowledge. Where would be appropriate to add decoupling caps, and what sort of values would be suitable? And exactly which caps on the Blizzard are decoupling caps? Sorry for my noobishness - you'll have to answer like I'm a complete idiot on this one.

Originally Posted by Sandro

good info thanks for the explanation
so basically you say that a short cable solved voltage problems?
how long is the cable of your psu now? from where you get it ?
which psu are you using now ? A600, A500 or the ATX PSU?

Yes, it's quite logical if you think about it. A wire has a resistance of its own, so the longer the wire the greater the voltage drop. The current length of the cable is approximately 40 cm. It was taken from a dead PSU if i remember correctly, but you could easily make your own if you can source the square plug. The PSU I'm using now is the A500 one (4.5 amps version). Sorry if this was unclear.