800xl not reading any cartridges

[_The Doctor__]

the chips I mentioned won't be the same 'u' number in the 400 but you should be able to find matching chips...

after the the other things I mentioned... pulling basic shouldn't hurt anything... try a cart plugged in with basic out...

next is to swap the OS chip... not interchangeable really as XL has pbi and 400 doesn't they just aren't the same... so swapping those don't work for the long haul.

[_The Doctor__]

check the flip side of the socket where you put sophia with a smart phone or magnifying glass for solder blobs and bridges across the pins and pads... I can't see then well enough in the picture but it sure does look like they've been worked on and such things are common happening for a number of happy hackers

[GabeShack]

Thanks for the suggestions.  I found what looked liked a dried piece of grass in the cartridge slot, but still not reading anything.  According to Simius, the MMU looks to be bad if its not reading the input from the card slot.  I reviewed the soldering for the bottom of the sophia board riser, and each pin is separated without any bridging as tested with continuity.  I just need an MMU to swap with to test Simius's theory.  Ive also cleaned the slots extensively, even using a very fine emery board to clean each contact head with no effect.  I have tested each contact using the pins outs referenced earlier by touching the actual cartridge port pins to the legs of the respected chips (or ground), and each seems to connect properly which again leads me to the chips themselves.  I just need to get my hands on a replacement MMU to test with then move on to other chips.

 

BTW with the basic chip pulled, the system boots right to the self test.

 

[+DrVenkman]

58 minutes ago, GabeShack said:

BTW with the basic chip pulled, the system boots right to the self test.

You might also have a bad PIA. The built in self-test really isn’t a particularly useful tool, but does it show anything amiss? For that matter, with the BASIC ROM chip pulled, do carts work then?

[GabeShack]

@DrVenkman

I had tried swapping the PIA with the one from my 400 and nothing changed.  The self test shows all green on the ram and roms.  And with the basic chip pulled and a cartridge in, it just boots to the self test.

Im hopeful that Simius is correct and its just needing a replacement MMU, but still have to find one to test with.  I just reached out to best-electronics in hopes of ordering one.  Looks like its only about $14 so not too bad.

[_The Doctor__]

star raiders with basic out should have yielded something other than the self test I would have thought.

[GabeShack]

I just received the A/V cable set and was able to test with the original GTIA chip and its the same issue.  So at least its not the Sophia 2 chip.

Let me try star raiders with the basic out.  I was testing with 2 known good carts that worked on my 400, empire strikes back and caverns of mars.

 

[GabeShack]

Tested with star raiders with basic chip out still only boots to self test.  So could it possibly be the OS chip U5?  If thats also a possibility then I'll order a spare of that as well with the MMU.

[_The Doctor__]

Okay basic chip back in, does it boot to basic?

holding down option while turning on, does it boot to basic?

Edited February 8, 2021 by _The Doctor__

[GabeShack]

Basic chip back in it boots to “ready” so yes basic boots to basic. 
But holding option down boots to self test

[GabeShack]

I spoke with the best electronics shop who suggested removing the MMU and cleaning it again and putting it back 3 or 4 times to help clean out the sockets. Going to try that next. Last resort is buying MMU and OS chips.  

[+warerat]

I suspect the MMU, but you’re going to need some other tools (hardware and software) besides a continuity test to validate this.  There’s nothing particularly special about the MMU, it’s simply an address decoder with lots of inputs and all logic equations are known.  People like to throw shade on the built-in self test as the RAM tester isn’t particularly rigorous, but at the very least it will checksum the OS.  So if your self test shows green on both ROM blocks, you can reasonably assume your OS chip is OK.  If it fails, it can be the MMU or the OS chip.  I'm going to assume it's not the OS because you can jump into BASIC.  If you have a device programmer you can make your own MMU with a GAL or a PAL16L8.

 

If you examine the schematic of the cartridge slot and the logic associated with it, the relevant pins to focus on the MMU are 7, 8, 12, and 19.  Pins 7 and 8 are the “ROM Detect” lines that disable the onboard RAM when there is an 8K block of ROM present.  These are pins pulled high on the cartridge PCB to change the memory topology depending on whether it needs an 8K or 16K hole in the address space.  RD4 (pin 7) and RD5 (pin disable the system RAM at $8000-$9FFF and $A000-$BFFF, respectively, and also enable the active-low ROM chip selects S4- (pin 19) and S5- (pin 12) for cartridge.  They both have pull-down resistors to ground in their normal (no cartridge) state.

 

Step one is to see if the RD lines are working from the cart slot to the MMU pins.  I’m assuming you know how to use a voltage meter or better yet, have a logic probe.

No cartridge: pin 7 low, pin 8 low

Star Raiders: pin 7 low, pin 8 high

Pole Position: pin 7 high, pin 8 high

 

If you don’t see those logic levels on the MMU pins, you have either have an internal short in the MMU or a bad connection from the cart PCB <-> cart slot <-> MMU.  To isolate the MMU, simply remove it from the socket and test those same pins without it present.  If don’t you see those levels without the MMU in the socket (the machine obviously won’t work), then you have a physical connection issue.  Otherwise, if you see the right levels with the MMU removed, then the MMU is at fault.

 

Step two is to see if the ROM selects are actually triggering.  With Star Raiders, you should see activity on pin 12.  You have a use a logic probe or oscilloscope for that, as the transitions will be too fast to register a voltage meter.  If you don’t see pulsing on pin 12 after you push reset, then the MMU is most likely bad or there is a connection issue (or the cart is toast, but you've tried too many).  Alternatively, if the machine is otherwise working and you have a way to boot something, I’ve used a tool called DEBUG+ for years that I modified to load into low memory (by default it loads into $A100, which obviously won’t work with a cartridge inserted so the custom version loads into $2100).  You can actually use whatever tool you want really, as long as it doesn’t load into the cartridge area.  The premise here is you want to examine the memory to trigger accesses to the cartridge address space and see if there are pulses on pin 19 for $8000-$9FFF, or pin 12 for $A000-$BFFF.  The pulses will only be present in those ranges if the corresponding RD line is high.  The benefit of using a memory debugging tool like this is you can also see if there is valid data when the ROM is enabled, so you could also conceivably diagnose a data bus or address bus fault if you know what is supposed to be there.  If the selects aren't enabling the ROM, the bus will float and you’ll see data values corresponding to the upper address byte that will change when you cross a page boundary.  What this looks like can be simulated by pulling the ROM chips on a cartridge.

 

The moral of the story is: troubleshoot before shotgunning chips.  I’ve got a stack of arcade boards I’ve repaired where previous owners replaced dozens of chips but completely missed the blatantly obvious problem that would have made it a ridiculously simple repair (along time and money saved) had they spent 5-10 minutes with a logic probe in a critical area. 

 

[GabeShack]

Thanks for the through response! Is the cost of a logic probe greater or less than the MMU chip?  If less, I’ll grab one and start probing lol. Any recommendations for an inexpensive model?

[manterola]

You can check if a pin is pulsing by reading a significant voltage both in DC and AC mode with the multitester. 

So you can get a probe but meanwhile still can use a multitester. Less than ideal, but better than nothing. 

[+DrVenkman]

13 hours ago, GabeShack said:

Thanks for the through response! Is the cost of a logic probe greater or less than the MMU chip?  If less, I’ll grab one and start probing lol. Any recommendations for an inexpensive model?

A logic probe is about $20 on Amazon. 

[GabeShack]

@DrVenkmanAny recommendations for a good one?

[+DrVenkman]

9 minutes ago, GabeShack said:

@DrVenkmanAny recommendations for a good one?

This is the one I have. Bear in mind, a logic probe isn’t necessarily very intuitive to use. Once you kind of get it and understand what you’re hearing and seeing with it, though, it can be surprisingly useful to check for stuck bits on a data or address line, a reset signal that doesn’t trigger, etc. I do suggest that you get to Atarimania and download a copy of the Sam’s Computerfacts for the 800XL. It has logic charts for all the IC’s on the board. The book as a few typos but you can figure those out from context when examining the board, the schematics and the charts.