Memory Problem with Apollo 1230 MkIII

[Rocs9005]

Hi

I'm having problems with my accelerator card in my A1200, and sad to say it's not the first time.

As best I can tell, the card is an Apollo 1230 M III.
It has a 030 -25 on it, but I believe it's actually clocked at 33
It has 2 simm sockets.

My problem is this. Every once in a while my A1200 crashes (This recent one was trying to load the game "History Lines" using WHDload) and when it comes back up, it reports no fast ram. Doesnt seem to matter what I do from that point, it works fine, but cant see any of the memory on the card. Up to a few mins ago, I was happily using 1x 16MB stick. The 2nd ram slot is a bit tight to the floppy housing, so I dont use it. I think 16MB is enough for what I do with my A1200.

Things I've tried:

Switching the Amiga off for a while.
Taking the Accelerator board out
Booting with the Accelerator in with no memory, then trying again with memory.
Changing the "wait" jumper on the board (I believe this is for 60 or 70ns ram)
Changing the mem jumper (so it should report no ram) then changing it back.

Nothing seems to make any difference. It's as if History Lines told it to "forget" it had the memory, and nothing I do can get it to look again.

Would really welcome any advice or suggestions.

Thanks,

Rocs.

[hit]

cant say for sure, but i recall people talking in such case about power supply issues (maybe to weak), or contact problems of the parts to the sockets (try cleaning the contacts). are there additionaly devices, like flickerfixer, ide things, cf cards/pcmicia (unplug them).
just in case, here is a ramtest by Stingray: http://www.a1k.org/forum/showthread.php?t=11986&page=5 (ville9 made some additon, in the middle of the page - memtest.pdf )
remove the extension (.pdf) its used coz the board upload needs an extension.

welcome to the board.

[Rocs9005]

Hi.

Thanks for the welcome, reply and suggestions.
Yes, there are a few other things plugged into my A1200
It has an external Scandoubler/Flickerfixer
and an Eyetech EZ-CD SE Buffered IDE interface (with a 2.5 IDE HDD on it)

I am using the PSU from my A500+ (my first Amiga ) , as I understood they have a higher output than the A1200s original one. I don't know how much higher though, so perhaps it is still not enough to run my setup

I did try plugging the accellerator into my spare A1200, and it behaves exactly the same, suggesting my A1200 isn't broken. However it could still be both the powersupply (I was using the same one) and the contacts.
In all my swapping boards, SIMMs etc, things eventually did start working again, so for now I am back up and running with 1 16MB stick of FPM ram.
I will run the memtest you linked to and check it out.

I was considering replacing the 2.5 HDD with an 44pin IDE-CF adapter and a CF card. If I understand right, they use less power than a HDD, so perhaps it will help.

Thanks,

Rocs.

[hit]

cant say much more. just gave some hints. maybe some of the tech experts can give some more detailed advice.
regarding the memtest, use the NOWA option: "memtest startadress endadress NOWA" - this seems to perform faster. good luck

[Rocs9005]

I've managed to find out a bit more about my system which may shed more light on the problems.

My system board is a 1D3 revision.
My PSU is from my original A500+, which has 4.3A rating on the 5V line, and 1.0A on the 12V line, according to the label on the underneath.

I am fairly sure now my Accelerator is an Apollo 1230 Mk-III Turbo, also known as a Winner 1230. It looks exactly the same as the 4th pic down on this page... http://amiga.resource.cx/photos/apollo1230mk3 which lists it as a Winner 1230. The only difference is mine has the optional 2nd Simm slot.
It is a 25MHz full 030 (just readable on the chip, although someone has tried to scratch the number off!!) running at 40MHz.
I got it from eyetech a long time ago.
Sysinfo shows
CPU 68030 /40
FPU 68882 /40
MMU 68030

Searching the forums I found this comment by ant512...
"The card probably uses an overclocked CPU (28MHz chip clocked to 40MHz), which renders the MMU unstable. Apollo, masters of cheap and unreliable hardware, did this a lot."

Yep, sounds like what I've got, and perhaps the cause of the crashes/memory issues. I can confirm that if you enable the MMU on WHDload options, nothing works, almost instant lockup/crash.
So, for WHDload, solution is simple - leave MMU disabled.

My question is, is there anything else that anyone would suggest I could do? Can I / Should I disable the MMU somehow for all software (WHDload already does it, but I do other stuff too.)

I know a bit about overclocking PCs, and know that it comes down to luck, quality components, good power supplies (including increasing voltages) and good cooling.

To this end I have removed as many things as I can from my Amiga, replaced the HDD with a 4GB CF + IDE adapter (amazing result, small fast, silent, and low power) disconnected the IDEfix97 board and connection to the externally powered CDROM (only ever used it to install OS3.9 anyway...). I am still using the external flickerfixer/scandoubler (as I use a VGA screen). So far seems ok with the same 16MB (60ns, not EDO) SIMM I was using before. Installed ClassicWB 3.9 edition yesterday, and it's very nice.

I have stuck a copper ramsink I had spare from a PC gfx card on the CPU on the accelerator, incase that helps it keep its cool. Overclocking a cpu from 28 to 40 with out even a heatsink - I'm surprised it didn't die long ago!! Would it be possible to actually slow the thing down by replacing the thing that says "40.00000MHz" or something?

Any other suggestions? (someone will probably say buy a decent accelerator card and throw mine away... )

Thanks in advance,

Rocs.

[meega]

That's why I don't like Apollo cards... not that there's necessarily anything wrong with overclocking, but they seemed to go out of their way to try to disguise the fact that they've done it - even to the extent of scrubbing off identification marks. They also did sometimes fit heatsinks though - but like as not that meant that you could no longer fit the trapdoor back in place. It's just not nice.

I should think changing the crystal to a slower one might well work, and you might also be able to remove the heatsink that you ought never to have needed.

[Zetr0]

@Roc9005

any chance you could take some pics of your card m8y, i may be able to help with a faster rated 030 pending on the one you have installed.

[DoogUK]

My very first accelerator was an apollo 030@25 overclocked to 42mhz and it was a complete nightmare....lockups, crashes and losing the memory as you described.
I changed the xtal and clocked it down to 33mhz and it was rock solid...food for thought.....

[Rocs9005]

Hi, thanks tons for the replies people.

@meega Yep, not only was the speed number at the end of the chip scrubbed off, it wasn't even done properly, so you can clearly see that it starts with a 2, so I am guessing at the 25. Sadly I can't go have a look at it right now, as the heatsink covers almost the entire chip, and I don't really want to peel it off, and have to clean the gunk off the chip, the sink, and then re-attach it. I'm lucky, the ramsink I used is 20mm square and only 6mm
high, so just misses the bottom of the keyboard when its in place.

@Zetr0
It really does look near identical to this...

With 3 differences;
1. I have a 2nd (not populated) white simm socket in the gap above the one in this pic.
2. There is a brown chip socket in the gap on the bottom right, with what I think is the 68882 chip in it. (Is it worth putting a heatsink on that?)
Looks from the number on it like that is a 33Mhz rated chip.
3. There is a heatsink on my CPU at the bottom left corner.
If you want pictures of anything specific on mine, let me know.

Thanks for the offer of the faster chip, but to be honest, that CPU doesn't look replacable to me (well, certainly not with my lack of soldering skills!!)

@DoogUK Thanks for that, very encouraging. Forgive my ignorance, I'm still learning! xtal = crystal = the silver component with C015050 40.000000MHz RALTRON written on it? (just below the middle of the simm socket in that pic) How hard is a 33MHz one to get hold of and install?
If that could be changed for a 33MHz one or something similar, that would be a great option - much rather have a stable 33 than a flakey 40!!

Thanks again people.

Rocs

[DoogUK]

Yeah xtal meaning crystal...this determines the speed the cpu runs at.
33mhz ones are easy to find and very cheap to buy.

[DoogUK]

In fact i have one here unused...pm me your details...i'll pop it in the post for you.

[meega]

If the fpu is in fact a 33MHz part and you run it at that speed then it most certainly does not require a heatsink. It is designed to run without one - at rated speed. It would probably benefit from one if you o/c it by any serious amount, and I'm not at all sure that PLCC chips will take overclocking gracefully.

[Rocs9005]

@DoogUK Thanks tons, PM sent.

@meega I think the FPU is a 33, because the full number on the chip is
MC68882FN33A. I have no desire to overclock it, I'd rather get it and the CPU to run at 33.

Having just read some of the soldering advice on the forums, I'm now completely scared I'm going to kill my accelerator completely!!
I think it's only 4 pins, but it's been a long time since I soldered anything delicate! Opinions - is it worth getting a socket and solder that on, so I can put the crystal chip in the socket (less risk of cooking the crystal I guess)
I notice Amigakit sell them.

Thanks again,

Rocs

[prowler]

Originally Posted by Rocs9005

Having just read some of the soldering advice on the forums, I'm now completely scared I'm going to kill my accelerator completely!!
I think it's only 4 pins, but it's been a long time since I soldered anything delicate! Opinions - is it worth getting a socket and solder that on, so I can put the crystal chip in the socket (less risk of cooking the crystal I guess)
I notice Amigakit sell them.

Rocs

Hi Rocs,

Yes, I'd go for it, mate! That's not really a very delicate job, and you've got the advantage of having read those soldering tips.

Once you've done that you'll get the confidence to try something more delicate next time.

Good luck!

prowler

[meega]

Originally Posted by Rocs9005

MC68882FN33A

That is indeed a 33MHz PLCC fpu.

[Chain]

Originally Posted by meega

I'm not at all sure that PLCC chips will take overclocking gracefully.

ive got in my hands about five of these, all of them runs great slightly overclocked like 25>33, 33>40, 40>50

edit:
25Mhz Crystal Oscillator (8-pin DIL) at amigakit http://amigakit.leamancomputing.com/...roducts_id=409
socket http://amigakit.leamancomputing.com/...roducts_id=416

[meega]

The one on the shown card looks like 14-DIL to me.

[Zetr0]

Apollo's by ACT use a 14 Pin DIL socket for thier A1200 Oscilators where as Blizards by Phase5/DCE use 8 Pin DIL format.

[Rocs9005]

Hi, sorry about the post necromancy, but it is on topic..,

I finally got round to asking Dimlow to desolder the 40MHz crystal on my Apollo 1230 Mk III Turbo and put a socket in instead. Got the card back from Dimlow (with the 40MHz crystal in the socket) and it worked as before, eg a bit flakey, not admitting any memory exists at all, no matter what stick(s) I try.

I got a 33.3MHz crystal from ebay, and tried that today. Sadly, it doesnt work any better. Still refuses to admit there is any memory on the card. Tried several different sticks, including ones that had worked previously in this card (until it crashed). I tried to find some kind of util that would confirm the speed the CPU is running at now, but with no memory, my Classic WB wont even start up, as it hasn't got enough memory to load the things at startup.

I booted off a plain WB3.1 floppy, and tried to run sysspeed from the hdd, but it crashed. Now it wont boot at all with the card in (dim power LED and dark grey screen). This is with no memsticks in. Tried taking the card out and reconnecting it, cleaning the edge connector, cleaning the card contacts as best I can etc, putting the 40MHz crytal back in.

The Amiga still boots fine if you take the card out.

Tried the card in my other A1200, and got same results - dark grey screen, dim power light, nothing else.

I'm wondering if the card has finally died for good.
Beginning to hate this thing...

Do I give up, bin the card and get a mem card instead?
I have a mem card with 4MB in the 2nd A1200 that I bought from someone here, and it works perfectly.

I can't afford another accellerator, but I dont want to spend loads more money trying to get this one working.
The CPU is soldered onto the board, so it would be a fair amount of work for someone to remove it, fit a socket into the empty gap and fit a new CPU... If that is even what is wrong!!

Any suggestions very welcome.

Rocs9005

[8bitbubsy]

Unstability and spontaneous error messages can be a sign of a weak PSU... This happened to me when I used the A500 power brick. Using an ATX instead solved all problems. Stable as a champ!
Have you tried another stronger PSU, like an AT/ATX? You can get an AT/ATX to Amiga-cable on ebay, or make one yourself. It's really worth trying, believe me.

EDIT: After reading a bit more in this thread I found out that you also use the A500 power brick. It's stronger than the A1200/A600 one, but in many cases it's a bit too weak...