Re: Is my Tek 468 beyond repair?


Ray
 

So there is a very easy albeit brutal sounding way of dealing with removing DIL type chips:

Take same slim side cutters and snip through every single leg as close as you can to the body.

Then you heat each leg with the solder iron and pull from the other side. Suck out the hole with a solder sucker. Works all the time.
Off course the chip is done for.

I use a crescent brand cutter from home depot, found in the electrical tools section right nex to the zip ties, for which it was probably designed.

You could call it the divide and concur method ;)
-Ray

On 09/20/2020 5:42 AM John <subs@qcontinuum.plus.com> wrote:


He was trying to tell you that until you have a working CPU,
all of your other tests are wasted.

Fair point.

Does anyone have the bin files for the service ROMs, or knows where there is a link to them? I have the bins for the normal ROMs but am looking for the service ROMs to try out on my 468.
I would be interested in the same and a "test ROM" was mentioned in a discussion by Reed Dickson posed by Daveolla posted in message 171596.

https://groups.io/g/TekScopes/message/171596

I was going to go back with a request but only once I got the computer working to a suitable point.

Just my opinion but 100mv of ripple on the digital circuits seems excessive. When I checked my 468 yesterday, I don't think I had more than 20mv of ripple on main PS +5V line.
Agree and must admit to being suprised that 80mVpp was the expected signal. Since I had everything unplugged except the transformer connection from the fuseboard, these indications were without load attached other than the logic chips onboard the A19 board.

Have you tried pulling out pin 1 and connecting it to a 5Mhz clock from a function generator?
I monitored the clock while I was taking the LA trace to make sure that a signal wwas present. Nevertheless, tried an experiment this morning with the CPU board reconnected and a lead clipped to the output of the crystal at pin 1 of U128, and one to pin the output at pin 10 of U120 (couldn't get the clip on pin 5 of U128) to allow me to connect a scope. First thing I observed was that there was no signal at all from the crystal. I set the sig gen for a 10MHz sine, 5V amplitude, 2.0V offset so that the floor was at -0.5V as per the trace at point 151 and injected into the first lead at the crystal output. This gave me almost exactly 2MHz square signal at the output with an amplitude of approximately 4V. I checked pin 1 for the 8085 for the clock signal and it was there. Still no ALE though. I pushed the signal to 20MHz (the max of this particular sig gen), got a 4MHz output, but still no ALE on the CPU.

One thing that puzzles me is that if 50MHz in should give 25MHz out (20ns square = 1/2 cycle = 25MHz) then presumably 10MHz in should give 5MHz out? So although the output using the sig gen seems stable, it is not proportional to the injected frequency.

I am guessing that the 50MHz oscillator is the type in a DIL metal package with just four external pins. The output is driving four Schottky high speed TTL gates which is quite a load (each gate about four times as much load as the usual 74LS devices), but no doubt within the spec of the oscillator package when new. It is possible that one or other 74S112 is moving out of spec and presenting an even larger load which is causing the oscillator to heatup and fail.
Its quite possible as you suggest that one of the 74LS112's is dragging the crystal down. It also seems likely that the CPU is faulty. The logic ICs are not in sockets so I'm not particularly looking forward to trying to remove them from a board with plated through holes. If I go ahead and manage it, then I will probably install sockets.

I assume you will be pulling the crystal and trying it out on the bench before replacing the crystal. The output should be able to drive a 470 ohm load if the crystal is ok.
Yes, I think that will be the next task on the list. Remove the board and extract the crystal + 74LS112s. Might as well rig up a test for the crystal while I'm at it. Not sure how to set it up though. Will need to research.

Is there any possibility of testing the 8085 on a breadboard with minimal hardware? I can hook up a sign gen and power rails. Not sure about memory though.

Looks like the shopping list now consists of an 8085, a crystal, two 74LS112's and IC sockets + parts required for the EEPROM adapters.


Join TekScopes@groups.io to automatically receive all group messages.