Author Topic: Tektronix 2445 - bad microprocessor S68A08?  (Read 17401 times)

0 Members and 1 Guest are viewing this topic.

Offline stephunk

  • Contributor
  • Posts: 24
  • Country: sk
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #50 on: June 16, 2015, 06:02:03 am »
MarkL
I am watching your progress and now, i´m loosing words :o. You´ve got my all respect!
Calibration procedure cannot be done here, because i don´t have the proper time mark generator. I do have only Philips function generator <1MHz.
I am working on a simple circuit, controlled by buttons, which gives necessary clock and data impulses into EAROM and giving the proper logic levels to C1, C2, C3 mode inputs. I need to modify only one address, so i can do it manually only with buttons..
I already found, that word 0300(HEX) on this specific address location makes the checksum same as stored on 0x00.(using your algorithm).
Everything in the scope is OK, but channel A shows 10x amplitude. This has to be something wrong with sense ring circuitry, i assume nothing serious.
Again, thank you for your help, I appreciate it much! :)
 

Online MarkL

  • Supporter
  • ****
  • Posts: 1592
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #51 on: June 16, 2015, 02:06:42 pm »
Thanks, glad to know you're still engaged!

Here's another thought.  I noticed that TEST 04, before it goes through all the locations to check parity and the checksum, does a write test to EAROM location 99 (0x63).  On my 2465 it erases 99, writes 0x1555, and then reads it back.

It should be fairly easy to find this piece of code and modify the EPROM so it will write whatever you want to any location in the EAROM.

Do you have access to equipment to erase and reprogram a 27128 or 2764 (depending on which you have in your scope)?  You could create a modified EPROM, boot the scope so it writes the value you want to EAROM 0x4C, and then put back the old EPROM.

If you want to do this instead of building an external circuit to control the EAROM let me know.  I would need to know what version of EPROMs you have in your unit, and possibly get a copy of the image from you if I don't have it or can't find it anywhere.


On the x10, I think you're right and I'd start looking at ring sense circuitry.


EDIT: This is a little bit wrong.

Actually what the write test does is read the current value at 0x63, invert the bits, erase, write the new value, then read back.  This makes more sense since just writing and reading the same value each time won't reveal a bunch of possible problems.

So, the value at 0x63 alternates between 0x1555 and 0x2AAA.  I didn't catch it before because by chance I was capturing every other run.

It's probably not of major importance to anyone, but I don't want to leave bad information out there if I can help it.

« Last Edit: June 16, 2015, 04:07:28 pm by MarkL »
 
The following users thanked this post: santosp

Offline stephunk

  • Contributor
  • Posts: 24
  • Country: sk
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #52 on: June 16, 2015, 04:06:56 pm »
MarkL
I don´t want to bother you. But thanks anyway, you are very kind. I try to make my programmer. It´s easier for now.
 

Online MarkL

  • Supporter
  • ****
  • Posts: 1592
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #53 on: June 16, 2015, 04:08:14 pm »
Ok, good luck!  Let us know how it goes.
 

Offline mian2zi3

  • Contributor
  • Posts: 30
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #54 on: June 17, 2015, 04:11:41 am »
EDIT: Added pic of 2465 being probed.  Fun!

Fun!  Would you mind describing your setup?  I'm currently in the market for a LA for exactly this type of thing.
 

Online MarkL

  • Supporter
  • ****
  • Posts: 1592
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #55 on: June 17, 2015, 02:11:02 pm »
The two pieces of equipment I used was an Agilent MSOX3104A mixed-signal scope and an HP 1631D logic analyzer.

In the picture, on the right, the MSO digital inputs are connected to the EPROM address lines and the analog probes are connected to the EAROM.  The analog inputs were used because the EAROM inputs are at 33V.

On the left, the 1631D is connected to the processor address and data bus lines, plus clock, VMA, and RD/WR.

At first I was only looking at the EAROM with the MSO to see what was being read and written.  Triggering was set up on specific serial EAROM addresses and data bytes.  Then I wanted to know what code was doing the access, so then I connected the MSO up to the EPROM address lines.  Unfortunately, the MSO only has 16 digital inputs so I couldn't see the data bus at the same time.

So, then the 1631D was then connected up to all 16 address and 8 data lines so I could do some stateful tracing of the code and trap values as they were being stored to SRAM.  At times the serial decode trigger output from the MSO was used as the arming trigger for the 1631D.


If you're starting out wanting to do this kind of digging, I honestly wouldn't recommend the 1631D.  It has a miniscule storage of 1024 events and the UI is slow.  I used it because I had it laying around from Z80 development I did years ago.  What I do like about it is that it has a 43 bit wide input and a plain-language multi-level trigger set up.

The scope has good serial triggering abilities, but it's really bad at stateful analysis.

If I wanted to do more of this type of work, I would probably look seriously at getting an HP/Agilent 16x00 series modular analyzer, most likely a 16702B, and a state/timing module with lots of capture space like the 16741A with 4Mpts.  There's scope modules available for it too, so you can get correlated analog traces.  If buying used, make sure you get any cables and probing pods that go with it since those can be difficult and/or expensive to replace.

If you're looking lower cost, I always thought the advanced triggering capability of the OLS logic sniffer was pretty good, but there's no GUI for it that I'm aware of:

  http://www.mygizmos.org/ols/fpga.html

(And I'm talking about an ADVANCED trigger GUI, not the standard SUMP GUIs that are out there.)  But given the way used LA prices have been dropping, you might be able to pick up something would support a 16550 module for not much more then the OLS anyway.

Whatever you might choose, the key is being able to capture all the address and data lines plus some control lines.  And the more limited it is on memory, the more sophisticated the triggering should be, at least in my opinion, so you can preserve those events that are important to the task at hand.

EDIT: Fix mangled URL, scope model num.
« Last Edit: June 17, 2015, 06:03:28 pm by MarkL »
 

Offline DC1MC

  • Frequent Contributor
  • **
  • Posts: 981
  • Country: de
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #56 on: December 11, 2018, 09:36:01 pm »
I want to quickly thank to all the people that contributed to this post, I've got a 2465 with the exact same crappy CPU model, mine is made by AMI ?!?!, same picture, same corroded crap and dead as a brick.
I've ordered some 68B02 and hopefully I'll have a nice Christmas surprise :).

I'll definitely  plan to use a solid box knife and pry open the dead CPU, it seem that the upper and lower part of the package were joined in a miserable way and that caused the problem.

 Cheers,
 DC1MC
 

Offline DC1MC

  • Frequent Contributor
  • **
  • Posts: 981
  • Country: de
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #57 on: December 27, 2018, 10:31:08 pm »
Here is my modest contribution for other people if they need them, my Tektronix 2456 A5 board EPROM 4x8KB dumps with the following ID:

160-1625-06
160-1626-06
160-1627-06
160-1628-06

S/N of the scope: B020XXX

In this thread there is another set, from an older S/N, with xxxx-04 at the end, a binary comparison showed that they are really strongly different.

Maybe it helps somebody.

 Cheers,
 DC1MC
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf