65XE boot issues

[guus.assmann]

Hello Ziggy,

Don't worry, to learn one needs to ask questions.
And you're right, a lot of this stuff is easy for me. I've been repairing the 600XL and 800XL since they were introduced in Netherlands. (As an Atari employee)
Moreover, the MMU file I sent is in German. That may complicate matters.
So from the top?
The OS is the only 28 pin IC in the board. And Basic is 24 pins, same with as OS.
Pin 18 of Basic is the CS line. And what I would do, is make sure that pin is insulated from the board.
There's a couple of ways to do that. You could desolder pin 18 and (gently) pull it out of the PCB. Reconnect later with a piece of wire.
A safer way would be to cut the pin, as close to the PCB as you can. Later, use a piece of wire to reconnect the pin.
And the design file below, tells that you could also insulate pin 13 of the MMU.

Compare, and yes a multimeter will do, the signal on pin 6 of MMU to that of a working board in the self-test mode.
The working board will show you, if it's 5V or 0V when Self-test is active. (Honestly, I'm not sure about that right now. And yes I could also find out by reading the GAL-design file... )
Pin 6 of the MMU is activated by the PIA (6520)
I would strongly advise to remove the MMU and place a socket in the board for that. (Hope your soldering skills are up to that, it's not very easy)
Once the MMU is in a socket, you can bend out a pin so it's not in the socket anymore. (The MMU will forgive this a couple of times until a pin may brake off after some 10 of more times)

@rubber_jonnie Correction Basic and Selftest do not ever occupy the same space. Upper 1/2 Cartridge and Basic do indeed exchance places, it's one or the other in that area of memory.

And now for the file with the MMU design. (MMU is 24 pins and narrow size)
A11 /* 1 Adress 11 */
A12 /* 2 Adress 12 */
A13 /* 3 Adress 13 */
A14 /* 4 Adress 14 */
A15 /* 5 Adress 15 */
MAP /* 6 MAP, Selftest Rom activate */
RD4 /* 7 RD4 from Modulport */
RD5 /* 8 RD5 from Modulport */
REN /* 9 REN, ROM Enable from PIA PB0 */
GND /* 10 Ground */
REF /* 11 Refresh from ANTIC */
S5 /* 12 CS-Signal for $A000-$BFFF Cartridge */
BASIC /* 13 CS-Signal for BASIC ROM at $A000-$BFFF */
MPD /* 14 Mathrom (in OS-ROM) use $D800-$DFFF */
OS /* 15 CS-Signal for OS-ROM at $C000-$FFFF */
CI /* 16 deaktivates CAS Line e.g. when Signal PBI-EXTENDB active */
IO /* 17 DXXX to 74138 address decoder */ IO means Input Output
BE /* 18 BASIC ENABLE, from PIA PB1 */
S4 /* 19 CS-Signal for $8000-$9FFF Cartridge port */
VCC /* 20 +5V Supply voltage */
Following the /* is the pin number and I've translated the text to Englisch.

By interrupt I do mean the "normal non computer" englisch. More commonly nown as "cut the line or trace"
And fooling the MMU would be to reomve or disconnect a pin and put 5V or GND directly to it. If you connect only this particular pin to GND, the PIA won't mind very much and it will survive thus a brutal deed. (Just 1 pin at the time! more could thermally stress the PIA)
I know, some of the terms I've used are "Hardware nerd lingo"

I hope I've clearified some thing.
If not, don't give up and ask again.
In my current profession I'm a trainer and I've learned that in general there's no dumb questions, just dumb answers.

BR/
Guus

[rubber_jonnie]

Hello Ziggy,

@rubber_jonnie Correction Basic and Selftest do not ever occupy the same space. Upper 1/2 Cartridge and Basic do indeed exchance places, it's one or the other in that area of memory.

Guus

@guus.assmann I did realise about the OS and Basic are in different memory locations, but happily stand corrected on the cartridge upper half only taking the place of basic.

I'm asking for my own understanding now, so when a cartridge is plugged in, the upper half exchanges place with Basic, I assume the rest of the cart mapped is mapped into RAM? Am I right in saying 8KB cartridges normally get mapped between $A000 & $BFFF (40-48K) and 16K cartridges are mapped between $8000 & $BFFF (32-48K). I presume there some sort of bank switching going on for larger cartridges that exceed the amount of physical RAM available? Thanks in advance

@ziggy Listen to Guus, he clearly knows his stuff, he'll get you there

[ziggy]

@guus.assmann

Hey Guus, thank you very much for this in-depth explanation. It's amazing to see some real Pros active on this forum.
The 65XE board has all ICs socketed, so disconnecting pins is easy and painless. Just the two 130XE boards are non-socketed, that's why I can't easily use their MMU & ROM for testing.

Here's what I got so far:

Disconnected Pin 18 of the 24-Pin BASIC ROM (CO24947A-01): No change, black screen on boot.
Additionally disconnected Pin 13 on the MMU (CO61618-20) (You said it's 24 pins, but that's a typo, right? Mine is 20 Pins, same with your Pinout file): No change, black screen on boot.

Pin 6 on the working 130XE: 5V when booted to READY prompt, 0,13V when in Diagnostic mode.
Pin 6 on the 65XE: 0,8V with BASIC pin 18 & MMU pin 13 disconnected (Seems weird to me - That not really 5V nor 0V...)

Now, I resocketed the Pins on the ROM and the MMU. I disconnected pin 6: Nothing changed, black screen. Pin reads exactly 0V on my meter, yet no self-test is triggering.

This board is an enigma....

[guus.assmann]

Yes, you're right its only 20 pins for the MMU.

OK, some more TTL logic.
All voltages between 0 and 0,8V are logical zero.
And depending on the exact logic family, anything abouve 3,5V is logic one. (Mostly, >4,8V is more common)
In between is "not determined". (Can also be an active signal or a floating input)

Having 0V on any pin exept GND pins is very suspicious. Check with an ohm meter between GND and any suspect pin. (Machine not powered ! )
In general there should be above about 1KOhm between GND and a non-GND pin.
In this stage, I would strongly suspect either Pia or MMU. Either those chips or a short.
Any short can be very sneaky, calles "whiskers" of solder. That's a nasty habbit of solder.

@rubber_jonnie
The MMU playes the mayor role here.
It has a couple of inputs that determine the memory map. Obviously of course. (For those who intimatly understand computer logic)
Now for the 8-bit Atari, all memory between 0 and $7FFF is Ram. (or 0 to $3FFF for the standard 600XL and 400)
There's also an MMU pin that totally disables all memory (pin 16, used for external access)
The rest is determined by the MMU input pins:
RD4 and RD5 switch the cartridge 8K parts. RD4 for $8000-$9FFF and RD5 for $A000-$BFFF.
When active low, the cartridge CS signals will be activated. These are called S4 and S5.
Basic lives at $A000-$BFFF but only if there's no cartridge at S5 and no self test. (Determined by MMU and PIA, where PIA needs to be programmed by OS)
Larger cartridges will be bank switched by accessing an area that's also present in the cartridge port. $D100-$D1FF, this is I/O, R/W is also on the cartridge connector.
The Self test is active in the area $5000-$57FF but is alo part of the OS-Rom. (If MMU doesn't disable the Ram in this area, the selftest will not run)
The pin MPD swithes $D800-$DFFF for the Mathemetics pack.
$D000-$D7FF is I/O area, decoded into 8 chuncks by a 74LS138.
The MMU also allows for using the RAM below the OS-Rom, using the pin REN (Pin 9 MMU from PIA PB0 )
This is usefull if you want to test your own OS. (Also used in an upgrade by Bob Wooley to get a 65816 @7Mhz at full speed)
For memory exceeding the 64K, bankswitching is used. The area $4000-$7FFF is banked. Switching done by PIA (and extra circuits.)
There's 4 pieces 16K bank with the 130XE (or extend 65XE) and more if there's 256K or even 512K Ram instead of 128K.

This felixibility makes it a bit harder to trouble shoot a machine.
BTW: The 65XE has the same MMU as the 130XE and 800/600XL

One more tip, write an E-Mail to: juergen(at)van-radecke(dot)de.
He developed a part called SYSCHECK. This is specially made to troubleshoot 8-bit Atari's and I think he has a 130XE version as well.

BR/
Guus

[rubber_jonnie]

@guus.assmann Thanks for the detailed explanation, that was a really interesting read

[ziggy]

@guus.assmann

When I mentioned Pin 6 showing 0V, it was not connected to the socket. Not sure if that was clear from my explanation.
I measured all pins on the MMU (with connected pins, machine turned off)

1 - 7,9k
2 - 7,9k
3 - 7,9k
4 - 7,9k
5 - 7,9k
6 - 3,3k
7 - 1k
8 - 0,7k
9 - 3,3k
10 - 0
11 - 1,3k
12 - 7,7k
13 - 7,7k
14 - 3,3k
15 - 7,7k
16 - 7,5k
17 - 7,7k
18 - 3,3k
19 - 7,7k
20 - 0,3k

I'll leave the interpretation to the experts

[guus.assmann]

Hello again,

Sorry it took a while. I'm not in the habit of checking a Forum very often.

The value on pin 8 of MMU seems a bit low. This is a line from the cartridge connector to the MMU and has a 1K resistor to GND.
And nothing more.
It's the line RD5 and a cartridge may use that to disable Ram (and Basic) in the area $A000-$BFFF.
It's likely there's something wrong with this trace, the cartridge connector or again, the MMU.
Compare with another working board as well, to get some more certainty.
These pins (8 for RD5 and 9 for RD4) are also (kind of) static, meaning you can use a Multi-meter to check the voltages when the computer is active.
Again, compare with a working unit.
Without cartridge and with either Basic-prompt or Self-test, these lines should have (almost) 0V. Below 0.8V in any case.
If pin 8 has more than 3,5V, than that's stopping the computer from booting into Basic.
I'm not sure if / how it would effect the Self-test. But I wouldn't be surprised if it does.
RD5 is also used by the cartridge that you got when the computer is working.
If RD5 is stuck to logic 1, it would explain a lot of the symptoms.

BR/
Guus

[ziggy]

Hey @guus.assmann , no worries. You're in no way obliged to answer my calls

Here's my new measurements:

65XE, the non-working machine. Powered on, no cartridge:
Pin 8: 0,65V
Pin 9: 5V

130XE, working machine, powered on, no Cartridge:
Pin 8: 0,8V
Pin 9: 5V
Resistance while OFF:
Pin 8: 0,65k, compare to 0,3k in 65XE
Pin 9: 3,1k, compare to 3,3k in 65XE

Every time you mention a possible problem I regain some hope, only to have it crushed by my new measurements

[guus.assmann]

Sure I know I don't have to answer fast.
But I do know how frustrating it can be if a computer problem eludes you.
The problem on this one seems to do so.

My next point of action would be to desolder the MMU.
But for me that's not such a big deal, I've done it many times.
And I hate to destroy possible good parts.
The Pia is not very likely anymore, that has no direct "interaction" to the cartridge

The safest way would be to get another MMU. That could just as well be a Gal. (May be easier to get)
Then just cut all pins next to the housing of the MMU, heat and pull every pin and clean the holes.
Next put in a quality socket and test again with a known good MMU or Gal.
The MMU in the 65XE and 130XE is identical by the way. (Double jepardy...)

Why this way?
The PCB is not a very strong one and if you're not too experienced with desoldering an IC, you might pull of some traces of the PCB.
That's also easy to repair, if you notice it.
Again, bin there, done that. But I´ve learned from that.

So I also once removed 18 pieces 20 pin IC´s from a $30.000 processor PCB. These were micro/code Roms. And the second board I took the Roms from was even more expensive. After having the Rom´s copied, the value of the board was $10.000 more. (And this was not 100% legal of course.)
Data General won´t sue me for that anymore I guess.

BR-
Guus

[ziggy]

HI @guus.assmann, your positivity sure helps me keep going, thank you

I'll take the board with me to a friends place with an desoldering station and try the MMU of the 130XE. If that works, great! Just need to program a GAL.
If not... I don't even want to think about it


Greetings