H89 Troubleshooting

[Bradley Marshall]

Hey everyone, I have recently obtained a Heathkit H89 in pretty good shape, and I am in the process of trying to get it working.

The terminal portion of the computer seems to work perfectly, the screen is nice and sharp and every key works in offline mode. when I got it I couldn't get any output from the CPU board, and I tracked it down to a couple bad tantalum caps pulling too much current through the -12v regulator.

After replacing those, it shows the "H:_" prompt on power up, and I can hit 'B' to try and boot. I don't yet have any boot disks, so I was just running the memory test to see if everything was working.

If I leave the computer on long enough, the screen freezes and the computer stops responding to the terminal. It doesn't take the same amount of time to lock up each time, and if I switch it off and back on, I don't get the "H:_" prompt unless I leave it off for a while and try again later. That makes me think that the fault has something to do with a part heating up, but I haven't been able to tell what part is the problem. Is this a failure that anyone else has experienced?

I have some more capacitors coming in the mail, I figure if I've already had two go out, the others aren't far behind.

Thanks in advance for any help you guys can give me.

-Bradley Marshall

[Mark Garlanger]

Hey Bradley,

    Welcome to the group, not sure why, but gmail decided to file your email in my spam folder. 

  That definitely sounds like a heat issue. Is it staying up long enough to make it through multiple memory tests?

You got lucky with those tantalum capacitors, I've had multiple ones short and pop, including one time where the traces on the back of the board acted like a fuse and peeled off. 

Mark

--
You received this message because you are subscribed to the Google Groups "SEBHC" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sebhc+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/acbade89-f677-4e95-8732-2a53bfc38477n%40googlegroups.com.

[Glenn Roberts]

Perhaps a silly question but is the fan plugged in and working?

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/CAAjkm793LkWa6dW6FfMCYWuf8bRjAxFCSq5X-NHCzARu2u4G%2BA%40mail.gmail.com.

[Joseph Travis]

Assuming all power supply voltages are correct, the next time the problem occurs, press the OFFLINE key and see if the TLB (Terminal Logic Board) responds correctly to key presses.  If so, the CPU board or one of the accessory boards has a problem.

You can remove the Serial I/O board and the Floppy Disk Controller (with the power off of course) to further isolate the problem. Run the memory test until your convinced it is good / bad.

If the memory test passes, install the accessory boards, one at a time and rerun the memory test.

If there is a memory failure, the address,  data written and data read is displayed. From this info you should be able to identify what memory chip/location is bad.

I hope this information helps.

Regards,

Joe Travis n6ypc

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/02ec01d82983%24e73c1a40%24b5b44ec0%24%40gmail.com.

[Bob Groh]

I'm late to this thread but that exact problem happened to the early H-89s - I owned one, had that same problem (the terminal board stopped communicating with the CPU board after a couple of minutes - I was working at Heathkit at the time (design engineer in the Ham Radio Group) so I approached the computer guys and said I would trouble shoot the problem if they would cover my 'a** if I blew up my new computer.  Being very busy, they snarled at me and said '.... sure .. just don't bother us ..".

Well I found the problem and fixed it - (excuse my memory of what went down but this is what my 40 year ago memory remembers) - the problem was the clock on the UART on the CPU board - there was two clock signals on the board - one clock (crystal) on the UART IC itself and a much stronger 4 MHz clock used for the CPU et al.  As the UART chip warmed up, the gain of the on-chip oscillator decreased and the stronger CPU clock took over (literally - you could see on a scope) - this meant the baud rate changed and, bingo, the two boards no longer talked to each other!  The fix was to cut the long and meandering leads to the crystal on the UART and wire a short lead in place of it.  Problem solved (for me and a bunch of other folks). The pC board layout was promptly changed - this all took place a couple of weeks after initial production began.

Anyways that might be of help. There may be a file in this group on that change.  Contact me if you need any other information (e.g. it all happened a loooooong time ago).

Bob Groh

Principal Engineer, Heathkit, 1977 to 1981

[Bradley Marshall]

Hey everyone, sorry for the slow response, life has been busy.

 I've got new tantalum capacitors installed on the main CPU board, and that seemed to make it run more stable for a while, but now it's back to inconsistently hanging.

 To answer some questions, yes the fan is working and I don't suspect the power supply area to be overheating. I can also verify that after the computer hangs, I can put the terminal in offline mode and I can still type on the screen and move the cursor around, which leads me to believe the problem is with the CPU board.

I also have been checking the voltages at the CPU board power connector when it hangs, and I'm getting the expected ~+20v, GND, and ~+5v. I have also been checking the linear regulators at the top of the board, and I am getting a consistent -12v, +12v, and +12v as I would expect. I've been checking those specifically because of the capacitor problem I had at the beginning pulling up the -12v line.

While I had it running better the other night I was able to run the dynamic ram test for a while and it errored out with the message "Error @ 040036 = 000" and a different time with "Error @ 040106 = 000". I'll add that along with a bunch of other expansion cards (which I will dive into once I get it running consistently) I have the 16k ram expansion board. It has a ribbon cable that plugs into the CPU board next to all the other ram chips. Does anyone know if I can just pull that out and go without it for troubleshooting purposes, or will I need to change some jumpers to get it working without that? I'll also note that the memory test always says LWA = 040111 at the top, which seems to be 16kb written in octal, what is LWA an acronym for?

And does anyone have a list of all the functions you can do in the bootrom without loading an OS? I have been going through all the documentation I can find and haven't been able to find a comprehensive list. From what I can tell, the original Heath manual doesn't even mention the memory test. I can't figure out what I did to cause it, but the other night I got it to run the memory test in hex instead of octal (that was strange because it had "LWA = 2049" which is 8kb? and it errored out at "2047 = 00"

I appreciate the warm welcome and hope no one is bothered by my infodump here, I've just been having trouble getting my head wrapped around what troubleshooting step I should take next.

Thanks everyone,

-Bradley Marshall

[Mark Garlanger]

If you take out the 16K board, you will need to place one of the chips from the board into the cpu board, where the ribbon cable was.

And if you change the size, you will need change the jumpers, their is a document here for setting it - https://sebhc.github.io/sebhc/documentation/hardware/HZ89/H-88-89-90_Cnf.zip

Mark

--

You received this message because you are subscribed to the Google Groups "SEBHC" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sebhc+un...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/20978f56-5094-4978-bc4f-2797a14fc1fcn%40googlegroups.com.

[Joseph Travis]

Bradley,

The memory errors that you gave indicate a bad D0 bit in the lower 16K bank of RAM which would be U542.

LWA = Last Working Address (377377=64K, 277377=48K, 177377=32K, etc.)

The ROM Monitor commands depend upon which ROM that you have.  If you were able to execute the memory test by typing 'T' at the H: prompt, you have the MTR-90 ROM which was the latest version (444-84 or 444-142). The previous ROM was the MTR-89 (444-62).

As my memory serves, the command list follows (* MTR-90 only):

 B(oot) [[SD]0-3]  -  Boot from specified device, default is: 0.  SD is the Secondary Device (dependent upon H/W configuration)

*C(onvert)  - Converts a decimal number to Octal or Hex depending on R(adix)

 G(o)  [addr]  -  Execute program at address.  If no address given, executes from PC

 I(n) port  -  Where port is an I/O address in octal or hex depending on R(adix)

 O(ut) port,value  -  Where port / value is in octal or hex depending on R(adix)

 P(rogram Counter) address -  Set the PC register to address in octal or hex depending on R(adix)

*R(adix)  -  H(exadecimal) or O(ctal)

 S(ubstitute) addr  -  Begin data entry at addr in octal or hex depending on R(adix)

*T(est Memory)  -  Displayed data is in octal or hex depending on R(adix)

*V(iew) from,to  -  Dump memory from, to in octal or hex depending on R(adix)

There is a (H17) disk drive speed test available by typing:  G7372

The memory test can be executed by typing:  G7375

I hope you find this information useful.

Regards,

Joe Travis n6ypc

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/CAAjkm78mKpEGHoZ%3DeZEmP1z0ns0E7jyBHAQounXUXoVfF-CG_w%40mail.gmail.com.