• Please review our updated Terms and Rules here

SWTPC MP09 CPU board troubleshooting

Witchy

Experienced Member
Joined
Oct 11, 2015
Messages
376
Location
Flatlands, UK
Hi folks,

I'm looking over an SWTPC S/09 6809 system for another museum, some of you may have seen me post about it on Twitter.

It was working up until around a year ago then something went bang, we don't know what. Since then the PSU caps have also given up - both 8V and 16V rails collapse at powerup. I don't think the two events are related. Currently I'm getting all 3 rails from bench PSUs and voltages are stable. There was a fault with the MC1488 transmitter on the MPS2 serial board which had taken out the +12/-12 regulators, this may have been the source of the bang. I haven't checked to see if the 6850 UART for channel A looks active.

Thanks to Michael Holley's documentation at deramp.com I've been able to do a lot of fault finding (clocks, RESET etc), but I'm still getting no signs of life on the 2812 terminal (which there is zero docs online for apart from a couple of posts here). Terminal is fine otherwise, if I put it in half duplex mode I can type and everything's echoed as it should be. With my breakout box on the RS232 port I can see it trying to transmit when I hit keys.

I put the MP09 CPU board on the bench running from an 8V DC supply and my scope showed active and correct-looking traces on all the address/data lines. However, if I put my HP1660a analyser on the CPU the signals just disappear like I'm sucking the life from it. Remove test clip and the board comes back to life. I've not seen that behaviour before.

Anyone know why my analyser may do this?

Cheers!
 
>>> Anyone know why my analyser may do this?

The 6809 has the crystal directly connected to pins 38 and 39 of the CPU. It is possible your logic analyser (if connected to these pins) is killing the oscillator by loading it up too much?

The other possibility is that your clip is on the CPU the wrong way round of course...

Dave
 
>>> Anyone know why my analyser may do this?

The 6809 has the crystal directly connected to pins 38 and 39 of the CPU. It is possible your logic analyser (if connected to these pins) is killing the oscillator by loading it up too much?

The other possibility is that your clip is on the CPU the wrong way round of course...

Dave
Hehe, I've been bitten by the 'clip the wrong way round' before so all my clips now have pin 1 clearly marked. The previous thing the analyser was used for was my Positron 9000 which is also 6809 based, but for the Posi I'm using the E clock as a base. For the MP09 I put the clock probe on the pin that gave me the strongest 'scope signal on the CPU, can't remember which now. That thought is ringing bells though, the last time I had an analyser (USB one in this case) drag a chip down was on my 40th anniversary 6502 badge which also has the crystal directly connected I think. Is there a way around this?

I might have a board you can try
That would be awesome. Would you risk posting it?

Cheers

Witchy
 
It's just occurred to me - bypass the crystal and feed it an external clock? Do I have a clock generator? No.
 
You can get a little 5V 4MHz TTL oscillator package quite cheaply.

Ground pin 39 (XTAL) and feed the TTL input into pin 38 (EXTAL).

Dave
 
You can get a little 5V 4MHz TTL oscillator package quite cheaply.

Ground pin 39 (XTAL) and feed the TTL input into pin 38 (EXTAL).
I should have one by Friday. Just in time for a weekend's attempted debugging.

Cheers,
 
Well I checked and the diagram I have looks the same as the Mike Holly one. Oddly all the documents I have refer to the MP-09B but the diagram says MP-09A. I have scanned it which needed 6 sheets on my scanner and stitched them together best I can. Not good I know but something to compare with..

 
Brilliant, cheers Dave. The more info I can amass the better! As my cabin workshop still isn't electrified the machine is on my dining table and it's taunting me every time I walk past it :)
 
I got it running today :D

Much bounce. In the end the single issues were the +12/-12 regs on the serial board had been cooked, the MC1488B on there had died as a result, and a 74LS640 on the MP-ID i/o board was upside down because that board was designed by a sadist who had every single other chip mounted correctly, and the 640 is mounted upside down. No wonder the one fitted got burning hot. Amazed it still works! Oh, the PSU caps are toast too so I've been running the whole thing on 3 separate bench PSUs. Those caps are expensive!

Cheers all!
 
Cheers Dave! I forgot to add I was also using the wrong port on the serial card. Naturally you go for port A don't you, but my logic tracing showed that the code was probing 0xE004 which is the status register for the B port. Change plugs over and I got life on the 2812 screen. Also one of the LS240 address buffers on the CPU card was dead. A fun fix...
 
Ah yes, the old "port switcheroony".

I have that at work - a PORT A and a PORT B. Naturally PORT A is TT01: and PORT B is TT02: (not...).

Dave
 
I got it running today :D

Much bounce. In the end the single issues were the +12/-12 regs on the serial board had been cooked, the MC1488B on there had died as a result, and a 74LS640 on the MP-ID i/o board was upside down because that board was designed by a sadist who had every single other chip mounted correctly, and the 640 is mounted upside down. No wonder the one fitted got burning hot. Amazed it still works! Oh, the PSU caps are toast too so I've been running the whole thing on 3 separate bench PSUs. Those caps are expensive!

Cheers all!
Those MC1588/1489 chips roll over and die and the slightest issue. When I worked on kit with those the technical guys always kept a good stock...
 
Its not so much that the 1488/1489 were sensitive. It's usually the result of "hot plugging" the serial cables.
(we all did that a lot)
If the data connects before the ground you can expose the chip to voltages in excess of the max rating.

joe
 
Back
Top