Wonky Apollo 1260

[KillaByte]

Hi guys,

since this board seems to have the most Apollo-savvy hardware nerds maybe one of you guys can help.

I got a Apollo 1260 I bought from a friend about a year ago. When I tested it a few weeks ago all I got was a black screen. At first I thought it was dead because this was reproducible on three different A1200s.

On Friday we decided to give it a last shot and tried it in the A1200 of the friend who sold it to me and lo and behold it worked without a hitch.

I then tried it with my A1200 that I had brought with me and it worked there, too. We then noticed that he was using a 4,5A PSU. So we switched that with a 3A A1200 PSU and got the same "dead as a nail" behaviour as before. So we came to the conclusion that the 1260 is too power hungry for the wimpy A1200 PSU.

The next day I fetched my PSU collection from the cellar for some additional testing. I had three lightweight 3A A1200 PSUs, a heavy 3A A600 PSU (the molded one), a lightweight 3A A500 one (with the big case) and a lightweight 4,5A A500 one (big case, too).

With the A1200 ones it didn't work as anticipated. With the 4,5A one it worked - no surprise there, also. BUT it also worked with the other two which were rated at 3A also.

So I decided to measure the output voltages of the PSUs with a voltmeter. Some of the PSUs didn't show stable results - especially on the -12V line. I measured a pulsating voltage between 0V and -2V. Since I measured without load I assume that these were switching mode PSUs. The other ones showed stable voltages without load.

Funny enough the PSUs with the wobbly -12V output were exactly the ones the Apollo wouldn't work on. So my thesis is that my 1260 doesn't like switching mode PSUs. This seems kind of weird as I haven't heard of a thing like that before. I guess I'll build myself a measurement adapter for the video port since it carries all three voltages. This should enable measuring the PSUs under load.

There's another strange thing about the card. It's fitted with a 68060@50 MHz but doesn't work with the CPU jumper set to 060 (regardless which PSU is used). When I got it the CPU Jumper was set to 040 instead of 060 which seemed like the wrong setting according to the manual. In some forum threads it was suggested that the CPU jumper affects only the clocking of the RAM slot (full crystal speed at 060 setting vs. half speed at 040). According to this it should work at both settings with no RAM inserted (and the RAM jumper removed) to eliminate the possibility of a low quality DIMM. But it didn't work with the 060 setting even without RAM. It showed the same "black screen all over" behaviour as with the wrong PSU.

Now this all looks quite wonky to me. Does any of the Apollo gurus have an idea what's going on with my card?

[DoogUK]

Is the crystal a 50mhz one? i mean is the cpu actualy running at 50mhz.

[KillaByte]

Yes - the crystal is a 50 MHz one.

EDIT:
Oh - beneath the CPU the CLKEN resistor is soldered to the CLK setting. The GND position doesn't have a resistor fitted. That might be of interest, too.

[DDNI]

Originally Posted by KillaByte

Yes - the crystal is a 50 MHz one.

EDIT:
Oh - beneath the CPU the CLKEN resistor is soldered to the CLK setting. The GND position doesn't have a resistor fitted. That might be of interest, too.

Photos are always helpful.

[DoogUK]

Mines running at 80mhz.
CLKGEN setting is the same as yours..there's also no resistor on the ground and CLK jumper is set to 040.

I believe you set it to 040 for 60/66mhz and 060 for 50mhz.
It halfs the ram speed as most edo's wont run at 60/66mhz.

[KillaByte]

Originally Posted by DDNI

Photos are always helpful.

*sigh* what's it with you guys and the hardware pr0n?

Okay - here goes (click to enlarge):

[DDNI]

Your CPU Jumper is set to 040?

[KillaByte]

Yes - as I wrote the card will only work with the jumper set to 040.

[Zetr0]

Originally Posted by KillaByte

Yes - as I wrote the card will only work with the jumper set to 040.

thats becauase you have CLKEN enabled as such =)

byt putting it into (060) its trying to run the Ram at 50mhz.... this is NOT gonna happen!!

if you move this resitor back then you can select the 060 setting.

This would be completely pointless and counter productive as it will reduce the higher *clock* frquency the card can run at (between 66 - 88mhz)

you Apollo is perfectly fine, just some crappy PSU's get rid of them, or better yet retro-fit them with Baby ATX (MATX) 120watt + psu's for £5 (or there abouts...)

oh.... another thing..... were not Apollo nerds.... were Geeks... theres a HUGE distinction =D

[Paul_s]

C= power supplies = carp, they tend to smoke a bit nowadays due to age. I would never use one now

ATX PSU's are cheap enough and it takes minutes to wire one up to the original power lead

[KillaByte]

@Zetr0:
But doesn't the Apollo manual say that the CPU jumper should be set to 060 for 50 MHz and to 040 for 66 MHz? I'm confused...

With the jumper set to 040 my RAM will run at 25 MHz (with a 50 MHz crystal), right?

And how do you explain the card not running with the jumper set to 060 even *without* RAM?

I wonder how they could sell these cards back in the days when they wouldn't work with the stock 1200 PSUs - I have tried three of them, mind you.

@all:
Could someone try to reproduce some of these scenarios to confirm that this is Apollo specific behaviour?

[musashi5150]

Originally Posted by KillaByte

I wonder how they could sell these cards back in the days when they wouldn't work with the stock 1200 PSUs - I have tried three of them, mind you.

I used to run a Blizzard060, external drive and a HD on a C= PSU back in the day

[StevenJGore]

But doesn't the Apollo manual say that the CPU jumper should be set to 060 for 50 MHz and to 040 for 66 MHz? I'm confused...

Don't be confused, that's correct!

My Apollo 1260 has a 60MHz 68060 and the jumper is set to 040. It won't run at all when set to 060, just like yours.

Also, FWIW, I'm using a 120W picoPSU to power my A1200. This fits inside the A1200 casing and provides plenty of nice clean power!

Steve.

[KillaByte]

@Steve:
Thanks for the confirmation. I just wondered because I got a 50 MHz crystal on my card and the manual says that the jumper should be set to 060 with a 50 MHz crystal.

[KillaByte]

About the CLKEN Jumper - take a look at these two pictures from the BBoAH:
50 MHz Version
75 MHz Version

On the 50 MHz Version the CLKEN Jumper is set to GND, on the 75 MHz Version it's set to CLK.

On my card it's set to CLK with a 50 MHz crystal fitted. So which setting is the right one then?
I'd still like to understand how CLKEN and the CPU jumper are related.

EDIT:
I built a measurement adapter for the video port to compare the voltages of the "good" and the "bad" PSUs. The results were a little surprising because there's nothing alarming to be observed:

Good PSU: 4.975V, 12.47V, -11.4V
Bad PSU: 4.92V, 12.9V, -12.45V

[Zetr0]

@KillaByte

This is all about memory speeds not CPU speed.



I believe its not *flakey* Apollo card, more slow memory you have, whats its ns speed?

[KillaByte]

It's 60ns BUT when I set the CPU jumper to 060 the card won't work even when there's no memory inserted (and RAM jumper open).

And I still don't get the relation between the CLKEN resistor and the CPU jumper because according to the manual the CPU jumper should do what CLKEN does according to your description.

[Photon]

Get a $5 80 MHz crystal and keep it as it is. Memory will run at 40MHz and CPU will run at 80...vroom!

Fixing unstable -12V supplies, btw - as easy as adding a cap across 0 and -12? This would be nice to know for all, I think it's a common problem with C= and non-C= switching PSUs. But pulsating between 0 and -2V, that sounds weird. Cap leak near audio, or short? Have you tested those PSUs on a known good mobo?

I've had black screen with my Apollo 1240 upgraded to 1260, but it was either the 5V to 3.3V adapter or the trapdoor insertion that was a little slanted in their sockets (the card had budged).

[Damion]

And I still don't get the relation between the CLKEN resistor and the CPU jumper because according to the manual the CPU jumper should do what CLKEN does according to your description.

Basically this is because the 68040 and 68060 have slightly different pinouts and bus operation. 68040 is half speed bus (not changeable AFAIK), unlike the 68060 which is normally 1/1, but can be divided.

Looking at the pinouts, CLKEN doesn't exist at all on 68040. (Also, 68040 BCLK is missing on 68060.) CLKEN seems to be the signal on 68060 that enables/modifies the divider - full, half, or quarter speed operation.

My guess (could be totally wrong here!) is that if CLKEN goes to GND with a 68060 installed, the mechanical jumper does absolutely nothing - bus is 1/1 regardless. With CLKEN fed to CLK, the jumper enables half-speed bus operation when set to '040.

Take a look at the 68060 user manual, it goes into great detail... way over my head.

FWIW, my Apollo *will* boot with the jumper set to '060, but only without RAM installed. My understanding is that some cards won't boot either way, some *might* work at 50MHz bus with good (50ns?) RAM. I tried every stick I own, all 60ns, and couldn't get it working. However, there were tales of a guy with a madly overclocked Apollo 1240 @ 100MHz (50MHz bus), so it must be possible.

In addition to a good PSU, I also have to supply power to the floppy connector, otherwise the computer is very flaky, especially when cold. Using a 120W PicoPSU here which works great. :-)

[KillaByte]

@Photon:
That pulsating on the -12V rail was measured without load attached to the PSU. When I measured on the video port of my A1200 it was rock solid. That's what really puzzles me - I get "good" voltages on all the PSUs when measured at the video port - the working ones and the non working ones.

@Damion:
Thanks for the clarification.