Q700 ADB or board kaput? Mouse clicks but no movement

[pb3623]

So, my Q700 that was only in my hands for about a month is suddenly having hardware/ADB issues as well as the 950 I've written about here.





Background is above. Pardon the crosspost but I wouldn't normally hang out in the Peripherals section so maybe others wouldn't either... and this graduated from a problem isolated to the video card to impacting the entire machine!

Here it is in its glory/gore. Thank you in advance for following along!

Above PC 24x jacked up my Q700. Machine at the time had a upgraded/soldered 33 MHz crystal, PPC 601 PDS card, PowerPump NuBus card and companion SIMM that was supposed to increase the clock from 33 to around 40 MHz but I never could get it to work.

The Quadra came to me without an 040 CPU so I bought a MC68040RC40A (note the 40 MHz so it should handle the overclock just fine) and installed it, even though the 601 runs most of the time.

I have a 3-port ADB/HD15 KVM switch that works (seemingly) perfectly.

Plugged in the Radius card to revisit the issue I noted in the above post and it basically the mouse went out shortly thereafter (after a couple of reboots, there wasn't a particular moment of failure). Clicks work but no x,y movement. The pointer is grounded in the upper-left corner, regardless of boot disk (SCSI2SD or original HD), OS version (7.6.1, 8.1, 8.5) or H/W configuration.

1. Booted from Legacy Recovery CD (7.5 IIRC?) and FloppyEmu (various versions).
   
2. I've cleared PRAM countless times
   
3. I have 64 MB matching 30-pin RAM (+4 MB onboard), that I've taken out
   
4. I have 2 MB VRAM that I've taken out
   
5. I plugged in a known working PrecisionColor 8xj NuBus from another machine
   
6. I left in/took out the PPC 601 card
   
7. I pulled the 68040 and forced it to let it run on PPC 601 only 
   
8. I took out all NuBus cards and ROM SIMM
   
9. I *think* I tried to run a stock Quadra 900/700 ROM onto a ROM SIMM I burned but I'm not sure if I did it right (just got the programmer last week) 
   
10. I bypassed the ADB KVM by connecting wedge mouse and AEK II to both Quadra ADB ports
    
11. I disconnected AEK II and connected multiple mice directly to both ADB ports
    
12. I verified AEK II and mice work in other machines
    
13. I removed and replaced the (tested good) PRAM battery
    
14. I've used ADB Renewal and Mouse Jolt utilities that are supposed to reset the ADB/bus and don't require a mouse to do so
    
15. I've had success with TechTool PRAM zapping as it claims to be more thorough (beyond the initial ~192 bytes they claim is wiped with a regular Cmd-Opt-P-R?) but can't navigate TechTool without a f-ing mouse
16. I get an 0000000F/00000021 occasionally when I try to boot from SCSI2SD, usually if I don't pull the PRAM battery AND power for ~30 sec.
    
17. I have a Snooper NuBus card that I'm not even sure is working. It lights +5V, +12V and -12v light up but no ADB or Activity - the orange clock LED blinks rapidly ... but it does the same thing in my working Iici too.
    

The main symptom is STILL: the mouse clicks but doesn't move. (which is great if the Apple menu were a "FIX ME AND GET FILTHY RICH IN THE PROCESS" button). 

I'll buy a Q700 board if someone has a working one for sale but that's a last resort. But I also want to get this solved and put my front bedroom/lab back together soon. 

Is this a lost cause?

And thank you again for hanging in there with me!

 

[trag]

The "memory" (or delusion) is very old, but I think I had the same symptom (or completely unrelated mouse symptoms) on a similar age machine.

At some point, Apple moved the ADB control into a dedicated CUDA chip.  The part number on the thing is either 341S0788 (newer) or 341S0060 (older).   It is a 28 pin SOIC (14 pins per side) chip made for Apple by Motorola.  I'm just not certain that the CUDA appeared as early as the Q700.    I know it was on the AV Quadras.   And it wasn't on the 68030 Macs.    So it first appeared with one of the three generations of Quadras (Q700/900;  Q610/650/800; or AV Quadras). 

I would check the Q700 board for one of those, clean thoroughly around it, and if necessary, touch up the solder on its pins.   Worst case, you might need to replace it with one scavenged from another board.  Unfortunately, it's a custom Apple chip.   It is said to be based off of a Motorola 68HC705, so in theory, we should be able to figure out how to reproduce one, but as far as I know, no one has.

If you look at this image:

https://www.prismnet.com/~trag/S900_MB_1.jpg

and look to the left of the second PCI slot from the bottom, you will see a CUDA chip.    Yes, the CUDA chip had a long life from the Quadras to the PCI Power Macs.  I'm just not certain that it appeared as early as the Q700.  

 

[pb3623]

Trag - thanks for the thorough reply on this.

I wasn't able to find a CUDA chip on the Q700 board (nor a reset for the CUDA like I'm used to seeing on my Power Macs & G3.

For what it's worth, I was trying to figure out how to get the Newer Technology PowerPump working on this machine (came pre-installed) and I assumed it wasn't working because it lacked an 040 when I bought it. So I bought a 40 MHz 040 but it still didn't work, but maybe it had to do with the 66 MHz crystal the previous owner had used to replace the stock 50 MHz one.

Anyway, one of the things you're supposed to do is to attach the black PowerPump hook to pin 1 of the crystal and the red claw to pin 1 of a chip label U53 (Motorola XAA219/ part #F803)

I don't know what this particular IC does and maybe it's a red herring but when both connectors are attached, no chime (never did). 

Right now, the board is bare, leaving the stock/soldered VRAM, 4 MB RAM, 040 CPU and ADB peripherals.

 

[MOS8_030]

Regarding the PowerPump I'm sure you've seen this video?

https://www.youtube.com/watch?v=lkPnI5lRVqM

 

[pb3623]

Most definitely. The original owner never used it - I got the full box w/ manuals, etc. Everything you see in the video. Shame because, even though the 601 card is running at 66 MHz by way of the clock chip, could manage to get 80 out of it with the PP. Anyway, maybe it never worked because of the 66 MHz crystal (even though I set the ROM SIMM jumper for the 33 MHz Q950), but maybe, just maybe there is a bad part that went from flaky to failure, resulting in this "partial" ADB outage.

It's just really weird to see the keyboard hits and mouse clicks register but not 

Also, I noticed it often boots to a Happy Mac twice each time it's powered on.

Wonder if this https://mac68k.info/wiki/display/mac68k/Diagnostic+Mode can be used on a Quadra?

 

[powermax]

I wasn't able to find a CUDA chip on the Q700 board.

Please post a bigger picture of your Q700 board here and we'll try to find the CUDA chip on it. If your board does have CUDA, it should have the PRAM battery as well. It's a small, 1/2 AA size lithium 3,6V battery. CUDA MCU is usually located near that battery...

Unfortunately, it's a custom Apple chip.   It is said to be based off of a Motorola 68HC705, so in theory, we should be able to figure out how to reproduce one, but as far as I know, no one has.

Everything correct. CUDA is a custom model of the 68HC705 MCU Motorola made for Apple. Its Motorola part number is 68HC05EG.

I have some confidential information that Motorola has released a nearly identical chip to the public, known as 68HC05E1. Its datasheet is located here: 68HC05E1GRS.pdf

I have no clue if it's still buyable. But if it is, it would be possible to reproduce CUDA simply by flashing the already downloaded CUDA firmware to the fresh purchased MCU. It would only require the right programming device

It sounds like a possible project for BMOW )

 

[powermax]

I wasn't able to find a CUDA chip on the Q700 board

Because Q700 doesn't have CUDA, it uses the "Bit-bang" ASIC instead, Apple part number either 342S0440 or 342-0440-A or 342S0440-B, see Mac Technical Notes.

 

[trag]

Thank you for the info, Powermax.   Very cool.    Am I reading you correctly that someone has already downloaded the CUDA firmware out of the CUDA?

pb3623, apologies for the mini-hijacking of your thread.  

Powermax, Should we start a CUDA thread in hacking?

So I got smart and checked the Hardware Developer Note for the Q700.   It does not use a CUDA chip.   It has a VIA1 and VIA2.  I'm not sure what those look like on the logic board, but the goal should probably be to identify VIA1 and then clean/examine pins and traces between that and the ADB port.   Edit:  Looks like Powermax identified the part number.  ACcording to the Developer Note, ADB is controller by a chain of VIA1 and then GI ADB transceiver.  I'm not sure which of those is the chip referenced above.

Again, a good scan/picture of  your logic board that allows us to read the markings on most of the chips would help with that.  We might be able to identify the VIA1. 

I have at least one Q700 at home, but it's not very accessible right now.  I know I have some boards in a box in the sun room; I was removing caps and cleaning them, but doesn't the Q700 have tantalum caps from teh factory?   So it wouldn't be in that batch   It's been so long since I started that cleaning project, I can't remember if there's  a Q700 in the mix.     It's mainly a bunch of very gooey IIcx's and a couple of Q840AVs and a couple more boards that might be Q700 or Q800, I can't remember.

 

[powermax]

Am I reading you correctly that someone has already downloaded the CUDA firmware out of the CUDA?

Yes. IIRC, three different CUDA firmware dumps are currently accessible: 341S0060, 341S0417 (CUDA v2.35 from Color Classic) and 341S0788 (CUDA v2.37 used in Quadras AV and PowerMacs). EgretManager has a call called read 6805 address allowing the programmer to read values from an arbitrary MCU memory address. There is even a tool for making Egret/CUDA dumps, see here: http://www.gryphel.com/c/minivmac/extras/egretrom/index.html

Powermax, Should we start a CUDA thread in hacking?

Yes.

 

[pb3623]

Here’s a picture of the above-mentioned ASIC 342S0440-B. I poked around the chip - each pin’s solder joint appears to be intact. I did notice the fuse above it but assume that blowing would be obvious AND be all or none?

 

[powermax]

I did notice the fuse above it but assume that blowing would be obvious AND be all or none?

Those are resettable SMD125 fuses similar to this one. The theory on their operation is well explained in this Wikipedia article.

The fuse ABOVE the 342S0440-B ASIC is probably not the right one. It should be a fuse sitting between the ADB connector and the ASIC. Moreover, you should see a 4-wire EMI board filter made by Steward (similar to this one although your's could be entirely black without any markings) sitting right behind the ADB connector. I'd check this one for damage...

 

[powermax]

Could you shoot a picture showing the whole ADB circuitry (the connector and the 342S0440-B ASIC)?

 

[powermax]

Some more to check:

• is the failing mice attached directly to computer's ADB port?
  
• did you try to attach the Keyboard instead? Does it work then?
  
• tried with only one device at a time?
  

I know these are very basic steps I assume you already tried out. Please double-check them anyway before fiddling around with the circuitry...

 

[pb3623]

Sure. The ADB ports are J3 and J4. The ASIC is U12 for reference. 

Good thing I can’t sleep tonight! (US-Florida)

 

[pb3623]

Some more to check:

• is the failing mice attached directly to computer's ADB port?
  
• did you try to attach the Keyboard instead? Does it work then?
  
• tried with only one device at a time?
  

I know these are very basic steps I assume you already tried out. Please double-check them anyway before fiddling around with the circuitry...

Yes - I usually use a 3-way ADB KVM but did away with it for this troubleshooting effort. 

I’ve tried:

* keyboard + mouse in separate ADB ports 

* keyboard + mouse daisy-chained as usual to single ADB port

* just keyboard

* just mouse

Both devices work fine on other machines. Only mouse click is registered in each of the above scenarios. 

 

[powermax]

Hmm, if the Keyboard works as expected (different keys recognized) then it's safe to assume that the ADB circuitry is intact. I'd look for software-related issues. Does the mouse pointer respond to mouse movements during boot or is it frozen?

 

[pb3623]

Frozen and I’m booting from SCSI2SD, the original Barracuda and even a 7.1 floppy. And done so with and without the 601 PDS card when possible. What I haven’t done is to boot the Legacy Recovery CD. 

 

[powermax]

Last question: what happens with a keyboard attached? Does it respond when you press the TAB key/ any other key?

 

[pb3623]

CD is booting now. Same exact behavior. 

Not that I think this is relevant but this is all onboard video up to this point. Recall this stated with an apparently bad/haywire Radius 24xj.