Repair help: OSI 600

[dave]

Your new board is a Rev C or D, which adds color capability and can select between nominal 24x24 and 12x48 displays. I believe that U6 and U7 are only needed for the expansion. The on-board circuitry uses the unbuffered data lines. You will probably want to install those chips to expand the machine later.

I second Mark's assessment about the video RAM. I bet you get the proper display when you swap out the 2114 at U40.

About the keyboard on the Rev B board: The problem with using a multimeter for digital circuits is that the actual signals are pulsing between 0 and 5V (or close) and the DMM gives you an average value, which is only an indirect clue of what's happening. The logic probe helps by telling you that something is pulsing, but the scope gives you the best picture.

I think this is a great time to break out your Analog Discovery. It is a good enough scope for troubleshooting these old computers. This is a great opportunity to get some practice with it, even if you don't know how to use triggering yet.

Using either the logic probe or the Analog Discovery (which I'll call a scope from here out), verify that there are pulses on pins 9,10,15, and 16 of U2 and U3. If not, then we know the problem is either with those chips, or farther back in the selection circuitry.

If you see the pulses, then check the keys. Check for pulses on pin 2 of U4 with the shiftlock key pressed (you should get pulses) and released (no pulses). If that is correct, then check pin 3 (the output of U5) for pulses with shiftlock depressed, no pulses when released.

If that's all good them move on. Check for pulses on pin 5 of U4 with M pressed (pulses) and released (no pulses). Then check pin 6 of U5 (the output) with M pressed and released, as for the shiftlock.

Then check the D anc C keys the same way, for each one, first checking the U5 input (pin 5) and output (pin 6).

Last, check the W key the same way, at U5 pin 9 (input) and pin 8 (output).

You can do all that with just the logic probe, but since you have a scope, I encourage you to try it.

I suspect both your machines are very close to working.

Dave

[ianoid]

OK so although U2 and U3 put out a waveform for pins 9,10,15, and 16, the logic probe just lights up on (1). Those should be lighting up PULSE, correct? Since I have not used a logic analyzer before, I’m not sure if it’s me doing something wrong. The logic probe is being used in pulse mode, rather than mem modem and is set to TTL rather than CMOS.

What is weird is that I was testing it and I got U4 pin 2 to change from 0 to 1 but not pin 3, and then the computer started behaving a little differently. Usually the display of the 4 letters was flickering subtly when I touched most keys. It stopped doing that and stopped registering caps lock at U4 pin 2. It also started to refuse to reset upon break some of the time, instead giving me more screen gibberish (that looks like stacked race cars to me). I also noticed the right 8 RAM chips are running noticeably hotter than the left 8. At this point the computer had been on for maybe half an hour. I’ll let it cool off until tomorrow and see if it’s happier. Either way, I probably need to figure out why those RAM chips are running hot as well.

I was beginning to suspect that U4 74LS125 was the issue, but since I can’t replicate it right now, I’m reluctant to move forward with that. Nor can I confirm U2 and U3 are outputting correctly based on my interpretation of the logic probe.

Dave, thank you for the troubleshooting path. I look forward to running all the way through it once I iron out the above.

[dave]

I didn't realize the analog discovery 2 has a logic analyzer. That's great! It looks like it has 16 inputs, 5V tolerant. So, try hooking up inputs 0-3 to pins 16, 15, 10, and 9 of U2; inputs 4-7 to pins 16, 15, 10, 9 of U3. Trigger on input 0 going low. What you should see is a short low pulse appearing on each input in sequence.

The logic probe might be lighting the 'LOW" LED, but only so briefly that you can't detect it. However, the pulse LED should be blinking. In a contest between a scope or logic analyzer, the logic analyzer wins. (and in a contest between a logic analyzer and a scope, the scope wins).

It is sounding as if U2 and U3 are going to be fine. You whould also check WKB (pin 4+13 of U2 and U3) along with the other pins. You should see a short high pulse for each low pulse seen on one of the outputs.

Next check the inputs and outputs of U4 and U5, along with ~RKB. You should see ~RKB briefly pulse low after each high WKB pulse.

Definitely take advantage of the logic analyzer!

[ianoid]

Just to add to the confusion, I went back to testing the board that has chips in the expansion space, referred to as my second board previously. I did this because I had ordered some 2114 RAM and I received that in the mail and I wanted to try your suggested repair.

I added the 8T28 chips that were missing (someone had said those were needed), and I swapped out the RAM at U40 for a new chip and voila!, I get the D/C/W/M prompt. However, similarly, this unit does not respond to any of those keys. It actually boots straight to the prompt and does not require me to push break to get to it. The only key it responds to is “=“, which clears the screen and shows “00 0000”. I believe once I tried it and it showed “00 2000” although I can’t replicate that. Any suggestions for this unit? The caps lock is down and the screen flickers when I hit most keys, which makes me think the keyboard is mechanically function, at least partially. Not sure if that is a valid thought. Break does not get me back to the prompt or do anything I am aware of.

I will try out the other unit that is requiring more complicated troubleshooting for the keyboard when I have a moment.

[ianoid]

Going back to the first unit, the one with keyboard issues from the start, the one with the unpopulated expansion area.

This unit is acting a little differently than before. After about 5 minutes of use, the RAM chips U31-U34 and U45-U48 heat up (hot to touch) and it stops resetting. That is, it will go to a gibberish screen on BREAK but never be able to exit it.

Before it reaches this state, Checking U2 and U3 pins 9,10,15,16- U3 pin 9 is “0” (can I say “off”?) and all other pins on both chips are “1”. None of them come up as “pulse”.

And before it seems to overheat, U4 pin 2 changes with caps lock from “0” to “1” but pin 3 of U4 and pins 2 and 3 of U5 do not change. After a little while of messing with the computer, U4 pin 2 stops changing with caps lock and just stays at “1”.

Any suggestions moving forward with these findings, both the unresponsive keyboard and the possible heating issue?

I realize that the Logic Probe is probably closer to my level than the scope. If this repair can be done with the Logic Probe, I’d like to try. Just to be sure, should the Logic Probe be set to TTL or CMOS. Even just getting waveforms using the scope function is tough for me to do confidently. I’m constantly messing with settings to try to make the waveform visible- it sort of makes we wish I had traditional scope dial controls to mess with rather than stuff buried in software.

If I am to start messing with triggering with any confidence, I’ll need to complete a basic project or some tutorials. That is going to be an impediment for me getting through this project as time and energy are limited. I’d like to get these working before I move onto another electronics project or else these could collect dust for years before getting the attention they deserve, if ever. Plus, I fear some of these communities and their helpful knowledge may not exist in 10 years.

[dave]

Regarding the first board:

Let's assume, as it appears, that the video circuitry is good.

If the RAM is getting hot and you can no longer reset the machine, then the first concern would be a bad RAM chip. These chips should not get that hot. A bad RAM chip could keep the firmware from running properly, if page 0 or page 1 is bad. If the ROM routines are not running, you won't see pulses on the keyboard latches.

Regarding the second board:

It seems a lot is working. The keyboard pathway could definitely be the problem. If any keys work, then the addressing is probably OK, and you would want to follow the process outlines above to "divide and conquer" the problem.

You can figure out a lot with just a logic probe, but it takes either good intuition about the circuit, or experience, or both to troubleshoot complicated circuits with a logic probe. The scope makes the problems much easier to understand, especially for a beginner. My suggestion is to use this as the project to learn an oscilloscope. Start out by just putting the probes on the circuit, and not trying to trigger, but either run a single capture (triggered by your keypress), or have the scope acquire waveforms continuously and just see what they look like.

Activating a trigger will be a matter of looking at the waveform, seeing what range of voltages are displayed, and telling the scope which voltage to trigger at, and whether the voltage should be rising or falling. For 5v logic, you can trigger at 2.5V (not a valid logic level, but a valid trigger option). Just see what happens.

The nice thing is that you can trigger from one pulse, and view other pulses with other probes. For example, trigger on the chip select signal of a chip, and put probes on each output, and see the pattern.

I think if you just try hooking up the probes and turning on the scope, you'll get an idea of what to do next. Then, if you have questions, ask them here or google them. You won't break anything. And once you have a feel for what the scope output looks like, then you would get much more from a short tutorial video, etc.

Dave

[ianoid]

OK so I tackled the RAM swapping first for the first system, the one without the expansion area populated, despite the appreciated encouragement to troubleshoot the keyboard path.

I only have 9 new 2114 RAM chips that I just received, presumably new old stock (since 1 went to the other unit to get that started).

I started from the right, just removing the old chips and seeing basically no difference until I reached U35 and U49. At U35, there was a change during one power up, one character was displayed, so I felt I was on the right track. With U49 replacement, I was able to get the memory test working and C worked to get me to the BASIC prompt! I was able to run a simple program.

When following the memory prompt, it basically reports whatever I type as the amount of Bytes Free, even up to 40K. That doesn’t make sense. Does it normally report 255 bytes free when you don’t type anything at the memory prompt?

Are these 8K units? Is there any RAM other than 2114 that will work in these without further modification?

I see there is a more detailed RAM troubleshooting section in the manual, so hopefully that can help me narrow down RAM chip is a problem so I can use most of the original RAM included with the unit. I’ll work on that tomorrow.

Very exciting development, especially to finally make some progress with the unit after a good deal of help from you guys. Thanks again.

4FBDCD00-AE84-4172-8161-F400B9E25FF8.jpeg (3.28 MiB) Viewed 6449 times

8952B3B5-019E-4FA7-94AE-1DA662161EB8.jpeg (3.17 MiB) Viewed 6449 times

D69D3C3B-D5F9-44C1-99DC-9ABAAB61A933.jpeg (3.27 MiB) Viewed 6449 times

[Mark]

Hey, good news for your first board!

When you press <RETURN> at the Memory Size? prompt, the OSI boot code does a memory test. The value returned can be used to determine where bad memory is located. A fully populated 600D board normally has 8K of RAM, the memory test should return with 7423 bytes free. You are free to override this with your own value, OSI will not verify the number you have entered is correct. Primarily this is used to reduce the space BASIC is using in order to reserve memory for some other purpose.

If you look at the SAMS service manual for the OSI 600D, you will find a table on page 2 of the Trouble Shooting Guide that correlates memory size to working RAM chips. According to the table, 255 bytes free means you only have 1K of working RAM and U31/U45 are OK. There is an error in the 2nd 1K bank so that means U32 or U46 is not operating correctly.

You can further isolate the memory chip now that you have a booting system by entering the ROM Monitor at startup and testing memory locations manually. For instance to test the start of the 2nd 1K of memory: at the D/C/W/M? prompt, press 'M' then enter .0400/FF this sets the current address to $0400 and changes the value there to FF. Then enter .0400 again and look at the value displayed. If it is something like 0F then you can tell that the upper 4 bits of that memory location are not correct and the error is in U46 (according to the schematic). Try other values like 00 A5 5A.

Another way to test memory is using a memory test program like this one that fits in 1K RAM. It's better used to find intermittent bit errors or marginal RAM that works most of the time.
Simply swapping memory chips is probably the easiest and fastest way to fix your system at this point.

Good Luck,
-Mark

[ianoid]

Argh! The drama continues.

I was happily swapping RAM and I had isolated a couple of bad chips, got the ram to display 5375, when the computer just want dark, as in it didn’t start up. There was no mistake or incident in particular that resulted in this, just some repeated power cycling and RAM checking. I tried a new 6502, mainly because I have those sitting around, and it didn’t make a difference. Specifically I was in the process of swapping chips in the U36-U38 and U44-46 locations. I checked the power supply, which was still putting out around 5v.

Any suggestions to resurrect this unit at this point?

[BillO]

Don't take this the wrong way, but we all make mistakes.

Did you possibly remove or replace a chip while the power was on?