Author Topic: Tektronix 2445 - bad microprocessor S68A08?  (Read 22684 times)

0 Members and 1 Guest are viewing this topic.

Offline stephunkTopic starter

  • Contributor
  • Posts: 24
  • Country: sk
Tektronix 2445 - bad microprocessor S68A08?
« on: April 11, 2015, 08:17:02 am »
Hello

I have bought a 2445 for a reasonable price, checked visually, uncovered whole beast, cleaned it from dust and powered up.
The scope surprisingly started, WOW. I made some measurement and turned it off. Few days after i tried to power up again, but the scope didn´t started properly. Sometimes just diagonal traces and unreadable readout characters appeared and few leds were lighting. Sometimes only some LEDs were present and no display. Another time just the point appeared(timebase not running). So i digged and discovered some bad caps in PSU and mainboard too. The PSU now gives correct voltages. And then begun a series of troubleshooting. Finally i focused into A5 digital board, because all timing, control signals should going from there.  But they don´t. I found(according to numbers on the image below) that:

1. ONE signal from address bus of a MCU is strange. On pin25 of U2092,(TP506) i measured the signal which can´t drop to logic L due to some reasons. I tried to test the U2480(74LS138), but this was good. The data line A15 is going only into this decoder, and to unused J500 connector. This is the first reason for me to think, that the MCU has a faulty pin 25 :'( The signal from the pin 37 of MCU is OK.
2. I also measured the signal on MCU IRQ(pin4 on U2092). Straight line, no pulses. They should come from 4020 counter and inverted by 74LS04. The clock signal on U2668(4020) was present, but no pulses on output. Reset input was HIGH. I tested the 4020 and it was also faulty. Replaced with new one, but no change. The reset signal is coming from U2034 which should be clocked by decoder U2480. This decoder doesn´t give any clock signal, probably because of bad signal from the MCU(TP506)
This is the second reason why i suggest the MCU is faulty
3. The main clock for the MCU is OK.
4. I found that one data line(D0, pin 33 on MCU) has strange oveshoot, different from another 7lines of data bus. This is the third reason why i suggest the MCU is faulty

Guys, do you think my suggestion of faulty MCU(U2092) is right? Is there any chance to get somewhere the main MCU?
I hope somebody can help.
 

Offline mij59

  • Frequent Contributor
  • **
  • Posts: 693
  • Country: nl
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #1 on: April 11, 2015, 09:04:02 am »
The microprocessor not running properly could also be caused by a faulty eprom.
 
 

Online PA0PBZ

  • Super Contributor
  • ***
  • Posts: 5121
  • Country: nl
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #2 on: April 11, 2015, 01:52:30 pm »
I tried to test the U2480(74LS138), but this was good.

How did you test it? I'd start by replacing it.
Keyboard error: Press F1 to continue.
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #3 on: April 11, 2015, 01:59:52 pm »
1) I agree address line  A15 (pin 25) looks suspicious.  What test did you perform on U2480 (74LS138) to determine it was ok?

2) 4020 RESET forces all outputs high, therefore no pulses on output.

3) Ok.  The fact that the E line (pin 37) is a square wave would confirm that.

4) What "strange overshoot"?  Screen shot of that?

Did you try the "Kernel NOP Diagnostic Procedure"?  It's in the troubleshooting flow chart section and there's a description under "How to Verify the Control Data and Control Clock Lines".

Also, your voltage levels on your screen shots would indicate you're using a x1 probe.  The curves on what normally should be good square waves say that you're probably loading (and interfering with) the signal too much.  Use a x10 probe.
« Last Edit: April 11, 2015, 02:21:26 pm by MarkL »
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 13148
  • Country: gb
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #4 on: April 11, 2015, 02:03:07 pm »
The good news is that microprocessor failure is rare unless they are abused in some way. I agree with others, do not blame the microprocessor yet.

Aurora
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline stephunkTopic starter

  • Contributor
  • Posts: 24
  • Country: sk
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #5 on: April 11, 2015, 03:33:03 pm »
Thanks to all, dear members.

1. 74LS138 was tested separately on the breadboard-i just checked if the performance matches the truth table. I also measured input currents of control inputs(were as small as the datasheet says). I always test the possible faulty logic IC separately on the breadboard.
2. I did not perform the kernel NOP diag procedure yet, but i try this.
3. The picture of "overshoot" i can upload this evening. Overshoot is not the exact word for this phenomenon. I´ll do this right after kernel diagnostic.
4. sure I hope(and PRAY) the MCU is Ok. Rarely, but also on modern MCU´s there is possibility, that entire chip is working fine, but some pins are blown due to various reasons, reason no1 is ESD. Maybe I underestimated this, when brushed the board from dust with toothbrush. However, this never happened to me(maybe this is the first time)  :( But the failure appeared before brushing.
 

Online PA0PBZ

  • Super Contributor
  • ***
  • Posts: 5121
  • Country: nl
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #6 on: April 11, 2015, 03:47:46 pm »
Even if the A15 line is blown it still shows some activity. I guess it could be fixed by being a bit creative with buffering and pulling.
Keyboard error: Press F1 to continue.
 

Offline stephunkTopic starter

  • Contributor
  • Posts: 24
  • Country: sk
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #7 on: April 11, 2015, 04:37:09 pm »
So the diagnostic is done. Every address line gives nice rectangular signal. Only A15 has incorrect logic zero level. It drops only to a half of the supply voltage. So this is similar when a mother asked her little daughter if father is sleeping, she answered: Yes, whole daddy is sleeping, excepting his nose  :D

Now it´s time for pulling it down  ;)
Thank you again :-+
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #8 on: April 11, 2015, 06:54:46 pm »
What I would do at this point, if the processor is socketed, is unplug it and then put it back in with A15 hanging out.  Then you can test A15 in NOP mode with the assurance that nothing else is interfering with it.  You'll have to be the judge if the package can tolerate bending the pin sufficiently to do this.

If the processor is soldered, then I would carefully snip the A15 pin to isolate it, and then do the NOP test.  Make sure you leave enough pin so you can solder it back together again.  This method is a bit crude, but if you're looking at a processor replacement you may end up doing this to all the pins anyway.
 

Offline edavid

  • Super Contributor
  • ***
  • Posts: 3381
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #9 on: April 11, 2015, 07:25:31 pm »
FYI, if you need to replace the 68A08, you can also use any of  68A02, 68B02, or 68B08... 68B02 is probably the easiest to find.
« Last Edit: April 11, 2015, 09:36:03 pm by edavid »
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #10 on: April 11, 2015, 07:51:06 pm »
The manual lists the processor as a Tek #156-1342-01 which is a Motorola SC67127P.  From what I can find that's a Mil spec 6802.

A 156-1342-00 is a MC-6802P, so that should work too, to add to your list.

I'm not sure where where the OP got S68A08.  Maybe read it off the chip.
 

Offline edavid

  • Super Contributor
  • ***
  • Posts: 3381
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #11 on: April 11, 2015, 08:46:36 pm »
The manual lists the processor as a Tek #156-1342-01 which is a Motorola SC67127P.  From what I can find that's a Mil spec 6802.

A 156-1342-00 is a MC-6802P, so that should work too, to add to your list.

I'm not sure where where the OP got S68A08.  Maybe read it off the chip.

If there's a 1.5MHz CPU in there now (68A08), I wouldn't really want to downgrade to a 1MHz part (6802), when the faster parts are readily available.
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #12 on: April 11, 2015, 09:23:37 pm »
If there's a 1.5MHz CPU in there now (68A08), I wouldn't really want to downgrade to a 1MHz part (6802), when the faster parts are readily available.
From the scope trace on the E line, it looks like it's running at 1.25MHz (5MHz clock input).  And Tek has the internal RAM disabled so '02 doesn't matter.

You're right.  My mistake for not digging into the clock speed more closely.  Was just reading off the parts list.
 

Offline stephunkTopic starter

  • Contributor
  • Posts: 24
  • Country: sk
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #13 on: April 11, 2015, 09:48:52 pm »
Yes, the MCU is clocked on 5MHz.

MarkL: You´ve got th point. If pin25 is hanging free, this only one gives a logH 3,8V and log L is really near 0V(so the pin25 is probably good!). Other pins are giving 4V(LogH) and near 0V(LogL). If we follow the schematic, this pin goes only to 74LS138 pin3, which is one of three select inputs.
Next interesting thing, while pin25 unconnected, some voltage on pin3 of 74LS138 was present.. I unsoldered it and put in onto breadboard. I can´t believe my eyes. With just Vcc and GND connected, i measured various voltages on all its inputs which is around 4V(Vdd is 5,01V and perfectly stable). I connected the resistor 10k between GND and any input. The voltage dropped to 3,7V. With 680R resistor I measured 175mV. With 2k7 1,5V. I found this on 2 new(!) pieces of DM74LS138. There is some voltage on every input, and this voltage  depends of the value of resistor connected between input and GND. It yields, that input currents are 200-500µA...This is a big surprise for me. Never see it before..i should to keep the original TI SN74LS138 and don´t throw it..
The new chips have an old logo of National Semiconductor. It´s time for rest and beer here :) Again, thanks to you all  :-+
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #14 on: April 11, 2015, 10:26:39 pm »
The LS138 has a logic-low drive current max of 0.4mA @ Vinput of 0.4V (TI LS138 datasheet).  Keep in mind these are intrinsically bipolar devices and not CMOS, so they're going to draw noticeable current on the inputs.

It hasn't been ruled out that the processor A15 output is not driving hard enough, or at all, to ground.  The output should be able to sink 1.6mA and keep the voltage below 0.4V (according to the 6802 datasheet).

When you're done rest and beer, I would next try connecting a 3.3k resistor from A15 to +5V and see if you get <0.4V for logic low while running the NOP test.  It should actually do much better than 0.4V, but 0.4V is the max spec.

If it can't sink the voltage to below 0.4V, I'm afraid it's time to find a new processor or try PA0PBZ's workaround buffer idea.  If the '138 is the only thing on the A15 pin, I suppose you could also try a 74HCT138, which *is* CMOS and would draw close to zero input current.


EDIT: 3.3k resistor to +5V, NOT ground.  Maybe I need rest and beer too.
« Last Edit: April 11, 2015, 11:47:31 pm by MarkL »
 

Offline stephunkTopic starter

  • Contributor
  • Posts: 24
  • Country: sk
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #15 on: April 12, 2015, 07:02:22 am »
Yes, you are absolutely right. The pin25 can´t sink enough current so with 2k7 resistor it drops only to 3,6V. I built a small schmitt trigger circuit with 2gates of 40106 and 2 resistors. I can set the treshold voltage easy. Next thing is to put this into the circuit. There is some delay what is caused by relatively slow cmos circuit, maybe 74HCT14 would do the job better..

EDIT:the most effectively way would be to use 74HCT138, i run to shop and pick few. I don´t want to put some nest into such beautiful instrument. ;)
« Last Edit: April 12, 2015, 07:44:57 am by stephunk »
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #16 on: April 12, 2015, 03:37:26 pm »
You must have a good shop that has 74HCT138 in stock.

If you go with the 74HCT138, you might need a pulldown resistor on A15 to help it get low enough, depending on how damaged the output MOSFET is inside the processor.  You could try something from 35k to 100k, but you might need to experiment to find the right value.

Really the right fix for this is to replace the processor, if at all possible.  A search on possible replacements (from edavid) show that they can be had, and are not outrageously priced.
 

Offline stephunkTopic starter

  • Contributor
  • Posts: 24
  • Country: sk
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #17 on: April 13, 2015, 04:37:52 pm »
Shop is quite good, but not the best :) I work there as seller, therefore i can get almost everything also at midnight :)

I found only 74HC138 but for our purpose it works. No pull-up is needed. Scope is already starting, but there is an error message(test04 fail, error 11), which indicates some trouble with EAROM and associated IC´S and transistor. This is what i was worrying the most, right after MCU failure. So, the challenge goes on! EAROM contains data which represents the last working configuration of the scope right before turning the scope off. So if the scope is powered up again, the front panel settings are loaded. Wondering what else will jump on me..
Hmm, in normal operation i found noisy data line D0(but there is some activity too). In diag mode, this only one is in High state. Im going to test U2194 74LS244.
« Last Edit: April 13, 2015, 05:51:58 pm by stephunk »
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #18 on: April 13, 2015, 08:01:28 pm »
Hmm, in normal operation i found noisy data line D0(but there is some activity too). In diag mode, this only one is in High state. Im going to test U2194 74LS244.

Diag mode executes NOP instructions.  The 6802 NOP instruction is 0x01, so in diag mode D7-D1 are 0, and D0 is 1.

I would take a look at the suggestions in the manual for Test 04 Error 11: U2118, U2208, U2108, Q2025.

If the scope is running enough, you could also try the EAROM exerciser (exerciser 02) to see if there's any data in the EAROM.

Not only does the EAROM contain your last settings, it also has your calibration data.  You want it to be working.
 

Offline stephunkTopic starter

  • Contributor
  • Posts: 24
  • Country: sk
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #19 on: April 17, 2015, 12:01:53 am »
Hmm, i have to say that i am not so familiar with the digital stuff like MCU and memories, but i found something what is for me totally weird. There is a HM6116P-3 which is 2kx8bit fully static RAM. In the breadboard i connected the control inputs:
20. OE(neg) to Gnd
21. WE(neg) to +Vcc
18. CS(neg) to Gnd    So the device should be selected, and in read mode. Right?
Address inputs are not connected(so i estimate the input address is choosen 000h). The device doesn´t require any timing signals, is in the read mode and outputs should show the DC voltage 0V or Log1(am i right?).
Is it possible to measure a squarewave signal of period 20ms ONLY on I/O0(pin 9)?? It´s like inducing the 50Hz into the chip structure, but other I/O pins remains at Log0. The period is slightly changing and the duty cycle of the signal rapidly changes, when i make a pressure on the package with my finger. I took a pictures of this..it´s time for bed i think. Do the address lines need a pull-up resistor? I think these are schmitt buffered inputs..
BTW, tomorrow the EEprom´s are going to be tested also.

 

Offline edavid

  • Super Contributor
  • ***
  • Posts: 3381
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #20 on: April 17, 2015, 12:33:38 am »
Floating the address lines is very bad practice and can lead to strange behavior.
 

Offline stephunkTopic starter

  • Contributor
  • Posts: 24
  • Country: sk
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #21 on: April 21, 2015, 12:12:15 am »
You are right Edavid. I´m trying to improve myself so i build a TTL logic tester with ZIF socket. Each pin can now be connected directly to Vcc, Gnd, or to Gnd via 4k7 resistor. With this tester i found that HM6116 has its outputs OK(i programmed 3 addresses with 3 different words and read it back). However i can´t prove, that whole memory matrix is OK. My good friend has a programmer for 2764 UV-EPROMS and we´ve made a dump of each one. No errors during reading of these. Has anybody information if dups of these 2764´s are available somewhere, just to compare if mine are OK. It would be fantastic ;)
 

Offline MarkL

  • Supporter
  • ****
  • Posts: 2120
  • Country: us
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #22 on: April 21, 2015, 12:50:45 am »
There's a set of ROM images for the 2445 here:

  http://www.ko4bb.com/manuals/download.php?file=01_ROM_Images_and_Drivers/Tektronix_2445_Eproms.zip

In the zip file there's a notes.doc.  You should read that to make sure you're comparing the same versions.

By part number, the 2445 appears to have the same EPROMs as the 2465.  So, you could look at some of these:

  http://www.ko4bb.com/manuals/index.php?dir=Tektronix/Tektronix_-_2465_Oscilloscope

If none of those match, try searching for "rom image" and the number on the sticker on your EPROMs.

Why do you think the EPROMs might be bad?  The scope was complaining about the EAROM, unless something new has cropped up.
 

Offline stephunkTopic starter

  • Contributor
  • Posts: 24
  • Country: sk
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #23 on: April 21, 2015, 11:20:12 pm »
MarkL: thank you again for the link, although my EPROMS have different numbers. This forum is fantastic, because of its kind and willing users. I found the faulty EAROM driver MC1413-output of driver B is blown(pin15 always conducts to Gnd) and one driver doesn´t work(pin16). This could be the problem, because this particular pin drives inverts the pwr down signal from PSU for the MCU. Tomorrow i´ll bring some ULN2003 and see if this will work.
 

Offline mskobier

  • Regular Contributor
  • *
  • Posts: 97
  • Country: us
  • Test Equipmentaholic!
Re: Tektronix 2445 - bad microprocessor S68A08?
« Reply #24 on: April 28, 2015, 02:53:47 am »
stephunk,
    Its been a few days since your last post. Have you made any headway in resolving your issue with the 2445? I ask because I am having a very similar problem with the second 2465 I will start troubleshooting on again soon.

Thanks
Mitch
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf