MicroMac MultiSpeed Accelerator won't boot

[AndyDiags]

I could use some help getting my Accelerator fixed.

I recently got a Mac Plus with this MicroMac MultiSpeed Accelerator installed. I'm, however, having issues booting from it.
The initial boot screen is fine, but when the logo is supposed to animate off-screen, I get a snowy screen instead, and the system locks up.

Below are the images of the board from when I first got the system and pulled it out.

Initially, when I received the system, it wouldn't boot at all. Upon close inspection, I noticed the CPU pins were bent on the socket. My guess is that the previous owner tried to do some troubleshooting by pulling the board off and back on, which caused the pin to be bend.

I carefully managed to bend the pins back into place.
I noticed some minor cold joints at the cable connector, so I resoldered those.
I pulled the ROM chip out to verify it was reading well; when I did so, I noticed one of the legs was barely hanging on due to rust, so I burned and put in a new eprom.
This finally takes me to where I'm stuck now.

The system boots with the logo as shown in the video (no bong), then locks up with snow instead of animation. I expect a bong and the MicroMac to animate horizontally off the screen, then the system should bong and continue booting. Here is a clip of the expected behavior.

Here's some of the troubleshooting I did so far.

• I've tested with a different Mac Plus logic board and get the same issue
• I've reseated the CPU
• I've tested the four 4MB ram sticks in a ram tester and all the stick pass
• I've put in a different set of 4MB RAM sticks and get the same issue
• I've recapped all the electrolytic capacitors, still same issue (there were no signs of any leaks under any capacitors)
• I've cleaned the board and carefully checked for any damaged traces but don't see any
• I've tested with the original ROM eprom chip and get the same issue

I'm stuck here.

One interesting thing is that if I replace the 4MB sticks with 1MB sticks, I get a RAM error on boot.

The error reads

MEMORY ERROR
TEST ADDRESS  $00000039
TEST DATA     $00000000
RESULT DATA   $0000007B

The error varies on every boot; "test address" is between 1 and 31, while the "result data" is between 1 and 7B. I initially thought this was probably expected as the jumpers are most likely configured to only accept 4MB sticks, and I can't find any documentation on the jumpers. I recently found this post where the seller states he has 1MB sticks in there, and the jumper settings are the same. So perhaps this clue is meaningful, and there's some error with the RAM (e.g. bad trace?)


Final note, this Mac Plus also came with a working internal hard drive, I see that this system was last working 15 years ago. Seems this issue is what caused the previous owner to sell the system.

 

[AndyDiags]

Small update.

I figured out why I was getting the MEMORY ERROR when trying the 1MB sticks.
It turns out all the different RAM sticks I tried were too slow, the RAM on the acceleration card needs to be really fast.

I tried with these 1MB IBM sticks, and the system was able to show the MicroMac logo instead of the MEMORY ERROR message.

Unfortunately, the system still hangs at the logo.

My guess is that one of the other components on the board has gone bad. Here's a quick lick of all of them:

2x - N74F543D - Scientific Solutions Octal registered transceiver, non-inverting (3-State)
1x - F244 - Motorola MC74F244DW Octal Buffer Line Driver
1x - LS09 - Motorola 74LS09 Quad 2-Input AND Gates with Open
1x - 74F74D - Scientific Solutions Dual D-Type Positive Edge-Triggered Flip-Flop
3x - F574 - Octal Edge-Triggered D-Type Flip-Flop with 3-State Outputs
1x - 68030 @ 50 mhz
1x - CALMOS CD130005A

These boards use the same CALMOS chip and software used on the NewLife accelerators. I did manage to find instructions for those. It seems the current jumper settings is set to support 4MB or 16MB total. If I want to use 0MB or 1MB, I'd need to change the jumper settings.

 

[KennyPowers]

Do you have an SE you can try it in without the adapter boards for the Plus, just to verify the problem is on the accelerator itself?

My Multispeed (that's my video ) had some cap leakage, but surprisingly the two data line continuity breaks I found were bad vias that weren't really near obvious cap leakage. I guess my point is maybe do some continuity sanity checks even if you don't see anything questionable. And check your soldering on your new caps. You can probably forego checking continuity on connections to the SIMM slots or I would think that accelerator's RAM test would display an error before showing you the splash screen.

Also, that 030 being labeled as a 50Mhz part makes me wonder if that's the original CPU. I'm pretty sure your accelerator actually runs at 25Mhz even though there's a 50Mhz oscillator on it. I think the Multispeed was only available in 16Mhz, 25Mhz, and 32Mhz speeds. A lot of these accelerators overclocked their CPUs, not underclocked. If MicroMac had enough 50Mhz 030s laying around that they could put them on 25Mhz accelerators, then I wanna know why they put 33Mhz 030s on two of my 50Mhz accelerators If that's not the original CPU, it could be a remarked one. Even if it is, I wouldn't think it would have trouble running at 25Mhz, but trying a different CPU is an easy thing to do just to rule that out. My Multispeed has the same 50Mhz oscillator, but the 030 is marked as a 20Mhz part.

Also, I'm pretty sure these are NewLife accelerators that MicroMac bought/licensed. Here's the top of ROM 3.1:

I ran across someone at VCFMW last week who had one of these in an SE with ROM 2.7 and his splash screen said "NewLife" instead of "MicroMac", even though the PCB said "MicroMac" like yours and mine.

 

[KennyPowers]

Is R102 connected to the oscillator's ground pin? I can't see if there's a trace in that gap there.

Mine doesn't even have R102...looks like a pulldown resistor on the oscillator's output? Yours looks newer than mine...

 

[JDW]

While I understand what you may be going through now, I nevertheless feel that text display to be very slick in that they didn't just product the standard System Error. It seems to be a custom error message generated by the accelerator board. That's the first accelerator I've seen which does that. Pretty neat.

 

[AndyDiags]

Do you have an SE you can try it in without the adapter boards for the Plus, just to verify the problem is on the accelerator itself?

I've already checked with two different working Mac Plus logic boards, this should narrow down the issue to the accelerator itself. Testing with an SE would however provide the benefit of knowing whether there is an issue with the Mac Plus adapter board. Unfortunately, I don't have a Mac SE on hand.

My Multispeed (that's my video ) had some cap leakage, but surprisingly the two data line continuity breaks I found were bad vias that weren't really near obvious cap leakage. I guess my point is maybe do some continuity sanity checks even if you don't see anything questionable. And check your soldering on your new caps. You can probably forego checking continuity on connections to the SIMM slots or I would think that accelerator's RAM test would display an error before showing you the splash screen.

I already checked for continuity between the RAM and the FPU (this was before I could get the 1MB sticks working). As you state, the continuity there is likely fine as I don't see any RAM test errors.

Also, that 030 being labeled as a 50Mhz part makes me wonder if that's the original CPU. I'm pretty sure your accelerator actually runs at 25Mhz even though there's a 50Mhz oscillator on it. I think the Multispeed was only available in 16Mhz, 25Mhz, and 32Mhz speeds. A lot of these accelerators overclocked their CPUs, not underclocked. If MicroMac had enough 50Mhz 030s laying around that they could put them on 25Mhz accelerators, then I wanna know why they put 33Mhz 030s on two of my 50Mhz accelerators If that's not the original CPU, it could be a remarked one. Even if it is, I wouldn't think it would have trouble running at 25Mhz, but trying a different CPU is an easy thing to do just to rule that out. My Multispeed has the same 50Mhz oscillator, but the 030 is marked as a 20Mhz part.

Great callout! The one thing that catches my eye is the date code on that oscillator, 9631, which is 31st week of 1996. The other chips are labeled between 1992~1993. This leads me to believe that someone tried to upgrade the CPU on this unit.
The following seems to be the oscillator-to-CPU speed mapping:
32mhz oscillator -> 16mhz CPU
40mhz oscillator -> 25mhz CPU
50mhz oscillator -> 32mhz CPU

I guess the first thing I could try is to order a 32mhz oscillator as well as a replacement CPU, as these components may have been meddled with.
Honestly, I was going to leave the CPU as one of the last items to test, but this just bumped its priority here.

Is R102 connected to the oscillator's ground pin? I can't see if there's a trace in that gap there.
Mine doesn't even have R102...looks like a pulldown resistor on the oscillator's output? Yours looks newer than mine...

Nice find, I've put the board away for now. I'll double-check next time I get the board out for further testing.

 

[KennyPowers]

32mhz oscillator -> 16mhz CPU
40mhz oscillator -> 25mhz CPU
50mhz oscillator -> 32mhz CPU

I'm curious how you arrived at this mapping? I'd always just assumed the CPU was being run at half of the oscillator's frequency, but I've never verified that with a scope. I did find the below photo (not mine) of a MultiSpeed with a 40Mhz oscillator though, and 40/25 and 50/32 are similar ratios

Awhile back on 68kmla, some people were trying to figure out what jumpers JP0-JP4 do. There was initial speculation that they might control the multiplier, but that was eventually abandoned, especially since those jumpers appear to be in the same configuration on all 3 variants. Here are the best photos I could find of MultiSpeeds with all 3 oscillator frequencies (the last two are yours and mine):



It looks like that F74D dual flip flop to the right of the oscillator is being used to divide the frequency because its second output pin appears to be connected directly to the 030's clock pin (at least from squinting at some front/back photos). Maybe I should get my SE out and verify the CPU frequency with a 50Mhz oscillator...is it 25Mhz or 32Mhz?

 

[KennyPowers]

Mine with the 50Mhz oscillator is running the CPU and FPU at 25Mhz:

Both the CPU and FPU clock pins are connected directly to the output of the flipflop IC next to the oscillator, so the flipflop is halving the frequency. That implies that, unless there were significantly different board revisions, the CPU/FPU frequency on these is always half of the oscillator. I guess that picture of a MultiSpeed with a 40Mhz oscillator is an unadvertised 20Mhz version?

Also, while I had it apart, I took out the 4MB of RAM I had on the thing and put in the only four 4MB SIMMs I had to try maxing the RAM, but it crashed with an illegal instruction error at the OS boot screen with and without extensions enabled. It did pass the accelerator's RAM tests though. Guess it doesn't like those SIMMs

 

[KennyPowers]

Also, while I had it apart, I took out the 4MB of RAM I had on the thing and put in the only four 4MB SIMMs I had to try maxing the RAM, but it crashed with an illegal instruction error at the OS boot screen with and without extensions enabled. It did pass the accelerator's RAM tests though. Guess it doesn't like those SIMMs

If you get it working and want to max the RAM, I can confirm that this 16MB kit works in the MicroMac MultiSpeed: https://www.tindie.com/products/siliconinsider/purpleram-16mb-4mb-x-4-30-pin-dram-simm-kit/