Tektronix 2465B oscilloscope teardown

[MarkL]

...
Page 461 (front panel troubleshooting flow chart) seems relevant. I entered the Exerciser 01 mode and all the switches and pots are getting picked up. However, I can't find anything in the manual on the switch codes referred to on that page. Also the four 2-digit hex values seem to change in a funny way. Hard to describe without doing a movie but pressing a given switch once registers on the switch code display (in the first two groups), pressing it again registers too (this time in the second two groups) and subsequent presses don't change anything. This may well be right but it's not what I would expect from a keypad scanned in the usual matrix way. Why the two button presses?
...

Take a look at page 6-15.  It describes the format of the line and the codes that show up for Exerciser 01.

As james_s recommends, definitely start with power supply checks, and I would pay particular attention to the A5 board.

If the flowcharts don't get you anywhere, you could look at the LED DATA and LED CLK signals.

Do changes to the vertical and horizontal knobs show up on the readout?


EDIT:  Also, let us know if your serial number is <=B049999 or >=B050000 so we can play along with the right set of schematics.

[Michael Lloyd]

Much appreciated. I checked the service manual and was able to figure out how to retrieve the cal data. I video'd it and put it on YouTube. I don't know if it's useful for anyone else. If it is I'll post a link.

Wish I ran the test and recorded the data before replacing my Dallas. Something went wrong and I lost all the calibration data. @#$*$*($. Why don't you go ahead and post the link

The video turned out to be crap. The  numbers are out of focus or over exposed, I'm not sure which. I'll leave the link anyway. I'll shoot another video when I get home

https://youtu.be/iHlCVdnBvBQ

[Bryan]

Wonder if those hybrids could be replicated by modern SMD components on a small PCB?

Apparently one of the hybrids was. I believe the U800 I think...for awhile could be returned to a company in Japan that was able to fix them.

[peter-h]

S/N is not on the back! See pic: http://peter-ftp.co.uk/screenshots/2017-03-16_071429.jpg

It says B057050 on the front.

There isn't a fan on the back but there is a very noisy one inside.

I will open it up in the next day or two.

I did find that switch code table - thank you. I reckon there is something wrong with the switches. As I posted, pressing a button once shows up as a change in the four 2-digit values (only the left two change) and pressing it again shows up as another change (only the right two change). Subsequent presses don't change anything. I would be amazed if that was right. Can anyone confirm this? I can do a little video later.

It would be fairly common to use the same lines as are used to scan the switches for driving (multiplexed) LEDs. In fact I am using an ST LED display driver chip right now which does exactly that.

[tggzzz]

I did find that switch code table - thank you. I reckon there is something wrong with the switches. As I posted, pressing a button once shows up as a change in the four 2-digit values (only the left two change) and pressing it again shows up as another change (only the right two change). Subsequent presses don't change anything. I would be amazed if that was right. Can anyone confirm this? I can do a little video later.

It would help if you would relate your observations to the definition in the service manual p6-15 and p6-15, viz:

POTS AND SWITCHES (Exerciser 01). This routine displays the values that the Microprocessor detects asvarious digitized pots and switches are activated. Thehalf of the top line of the display appears after turningpot. The right half of the top line of the CRT display appears after pressing a switch. The top line of the CRT display has the following format:
AA BB CC DEEE FF GG HI JJ KL

The format is defined as follows:

"AA" is the code of the most-recently-activated potentiometer (see Table 6-9 for definition of pot codes).

"BB" is the current value (in hexadecimal) of pot AA. See Table 6-9 for the approximate range of codes for the CCW  (counter clockwise) and CW (clockwise) potentiometer rotations.

"CC" is the previous value (in hexadecimal) of pot AA.

"0" is the DAC Multiplexer code used to select pot AA (see Table 6-9).

"EEE" is the 12-bit DAC value (in hexadecimal) associated with pot AA. See Table 6-9 for the
approximate range of codes for the CCW (counter clockwise) and CW (clockwise) potentiometer rotations.

"FF" is the code of  the previously-activated potentiometer (see Table 6-9).

"GG" is the row code of the most-recently-activated switch (see Table 6-10 for definition of row codes).

"H" is the switch-position code: 0 for open; C for closed.

"I" is the column code of the most-recently-activated switch (see Table 6-10).

"JJ" is the row for for the previously-activated switch.

"K" is the switch-position code: 0 for open; C for closed.

"L" is the column code for the previously-activated switch.

[MarkL]

...
It would be fairly common to use the same lines as are used to scan the switches for driving (multiplexed) LEDs. In fact I am using an ST LED display driver chip right now which does exactly that.

It's not the case on this scope.  Take a look at schematic <3>.  There are 6 cascaded 74F164 shift registers that drive the LEDs directly and are loaded by LED CLK and LED DATA.

I'm not ignoring the other symptoms you're stating with the subsequent presses.  If the flowcharts don't help, I'm suggesting picking one obvious symptom that's easy to track backwards or forwards in relation to the processor.  There aren't that many other things involved in controlling the front panel LEDs, so that would seem to be the simplest choice.

You said everything was working and now it's not.  So it's likely we're looking at one failure somewhere.  Find the problem with the LEDs and you're probably going to find the explanation for all the issues.

When you get the lid off, examine A5 closely for signs of cap leakage.  This era of 24xxA/B scope is notorious for leaking caps which corrode and eventually sever PCB traces.

Some people will have you replace ALL the caps immediately even if everything looks fine.  (Wait for it....)  I'd recommend the approach of finding the actual problem and then deciding from there whether wholesale cap replacement is warranted.

[peter-h]

As a quick interim reply, tggzzz's explanation is clearly what I am seeing, so the switches seem OK.

I need to fix the LEDs. If they are driven from a shift register that should be easy, though I don't see how you could get them all lit up at some low brightness unless you are either clocking garbage data fast, or you have lit up the lot and there is a power issue (a low supply rail).

Thank you for the time taken to reply. This is a brilliant forum! I go back many years to sci.electronics.design and comp.arch.embedded but this is much better!

[MarkL]

It's hard to judge in the photo, but if they are that dim then it could be a power issue somewhere.  Low supply rail, or could also be a bad ground.  Check Vcc and Gnd potentials directly on the chip pins in case a trace got eaten somewhere.

I'd highly encourage you to read the theory section and maintenance section of the service manual before tackling anything.  It's among the best ever written (at least in my opinion).  There are descriptions on how the LEDs and switch/pot scanning are implemented.

Good luck and let us know what you find...

[BravoV]

Wonder if those hybrids could be replicated by modern SMD components on a small PCB?

Apparently one of the hybrids was. I believe the U800 I think...for awhile could be returned to a company in Japan that was able to fix them.

Thomas Lafay did that here -> http://www.davmar.org/TE/Tek2465/lafay.html

[tautech]

Just dropping this link to another thread where a member has seemed to overcome a U800 issue.
https://www.eevblog.com/forum/repair/2467b-with-a-strange-horizontal-issue-video-link/

[Satbeginner]

Hi All,

I recently got my 2465B through ePay from Italy, and it arrived safely.
I asked the seller to carefully pack it, surrounded by soft foam, and he did. Kudo's to him for that!!

Last week I started working on it.
It has the "all SMD" A5 board with the Dallas chip, and it started with no beam and stopped it selftest with the "ADD" light lit.

As it turned out, I could see the 05 FAIL 44 after I tweaked the GRID Bias.

I went to look for the 4x SMD caps, and yes, they all had been leaking and all of their values were down in the 10-30 nF range... ;-)
After replacing these (not really nice, because there were PCB-tracks eaten away as well) it still gave the same message.
It turned out the R2013 (10kOhm, part of the front panel DAC reference divider) was up to 35kOhm, this explained the "high voltage" error.
Somehow this type of 0,1% resistors are more susceptible to the electrolytes than the normal SMD ones?

I replaced it with a new 10kOhm et voila, problem solved!!

BTW, some people report this being a 0,1% resistor for accuracy, but I replaced it with a normal 1% metal film resistor. (true, I selected one of 10,002 kOhm with my Keithley 2000)
In my humble opinion this reference is "only" used to select the front panel settings, and not for measurements, so I think it is not really super critical.
It is true it must be stable, to not unwillingly change scope settings based on temperature changes...

Note: feedback I got from the Tekscopes group at Yahoo indicates my assumptions about the use of this DAC divider are not completely true. It appears this DAC -and so the divider!- are involved in measurement, so also in accuracy. At least a check of the calibration is necessary, so maybe a calibration as well.

Today I decided -after reading ALL of this thread!! and nothing else to do on a sunny Sunday- to go all the way, and recap all the electrolytic's and also replace the Dallas chip.
so I ordered the caps, heat sink, socket and Dallas chip at Mouser Europe, and now I have to wait.... (last time I ordered at Mouser Europe, they still shipped from the US, but no tax/customs issues though..)

Yes, although my scope has a Tek U800 thats is not getting hot, I will put an heatsink on it.
I ordered a few of these to see if they will fit:  http://www.mouser.es/search/ProductDetail.aspx?r=532-507222B00

Mine was build in Holland, Heerenveen, the serial-number starts with H706xxx, so in a way, it found its way back home.
(Not really, although I am Dutch, I do live in Spain.... ;-)

Will keep you all posted,

Un saludo,

Satbeginner (Leo)

[BravoV]

Will keep you all posted,

Hi Leo, welcome to the club  and thanks for the info.

About the damaged PCB traces caused by leaking electrolytic caps, did you take any close up any photos on that part ?  That will be a great addition to this thread. 

Looking forward to see your work on this gem, post lots of photos documentation please if you don't mind

[BravoV]

With this thread -> https://www.eevblog.com/forum/repair/2467b-with-a-strange-horizontal-issue-video-link/ , the OP demonstrated that the two over tightened nuts that are securing the U800 were making horizontal problem, sort of stressing out physically the U800 chip. And we're not talking about the thermal cycle that poor chip endures with that two nuts yet. 


Which brings out again, my observation at earlier post at the beginning regarding this securing method on this poor U800.


This U800 shots was the originally assembled, untouched since it was left from Tektronix factory, just look on how they sandwiched the chip's tab with washers. Also its obvious at standard DIP package, when the pins are soldered properly, the bottom of IC's body should not and will never touch the board, even without the washer.




Bottom view of U800, which is clearly the only heat path to dissipate the generated heat.




This is the board where U800 is located, that big exposed pad definitely is not touching the IC's bottom part.


Both last two photos are courtesy of Kirby from -> https://www.eevblog.com/forum/repair/tektronix-2445-repair/

Now its clear, no matter how strongly we fastened those two nuts, it wont improve "significantly" the heat transfer from the chip to the bolt thru that little metal tab since it has washers, let alone transfer it to the pcb big pad right below the chip, which will never happen anyway especially its installed with the two washers at the bottom.

And if the U800 is soldered, I think it doesn't need that two nuts from the beginning right ?

[Satbeginner]

I couldn't help myself, I just bought a second 2465B to repair...

[BravoV]

I couldn't help myself, I just bought a second 2465B to repair...

No ... not that face palm ... infact its good and good news only, no bad news at all. 

Now, visit here for the stress relief amongst brother & sisterhood of -> TEA ...  

[tggzzz]

I couldn't help myself, I just bought a second 2465B to repair...

No ... not that face palm ... infact its good and good news only, no bad news at all. 

Now, visit here for the stress relief amongst brother & sisterhood of -> TEA ...  

That thread relieves stress in the same way that a persistent itch is relieved by a good scratch => better in the short term, but worse in the long term.

[tecman]

Question for all of you restorers.  I am waiting to receive a 2467 I grabbed off ebay.  All appears to be functional, based on the listing info.  I do plan to ra-cap the unit.  It has the same basic HW as the 2465A.  As for the RAM, it has the "Keeper" Li battery.  I plan to replace it with a socket and a CR123 Li cell.  I did the same on my Tek AFG 5101 function generator.  My issue is if the RAM data is still being maintained, I may be able to use an external source to keep the data while I perform the battery surgery.

My question is what happens if you lose RAM data ?  If you do a CAL procedure (steps in the manual), will this refresh the BB RAM and generate a checksum ?

Since there is no way to remove the RAM to read the contents, without losing the data, there must be a method for the scope to "restore" itself from a dead battery.

Any insight is appreciated.

paul

[David Hess]

It has the same basic HW as the 2465A.

1984 - 2445 and 2465
1987 - 2445A, 2455A, 2465A, and 2467
1989 - 2445B, 2455B, 2465B, and 2467B

As for the RAM, it has the "Keeper" Li battery.  I plan to replace it with a socket and a CR123 Li cell.

The A5BT2570 is a "Keeper II" part number LTC-7P which is a 3.6 volt LiSOCl2 cell.  Tektronix must have liked these; they used them in other instruments as well.  Just be sure to replace it with a 3.6 volt cell or the oscilloscope will probably return a low battery warning.  Interestingly enough, the oscilloscope will return a different fault (High battery?  I forget.) if the battery voltage is too high like with 3xAA cells.

My question is what happens if you lose RAM data ?  If you do a CAL procedure (steps in the manual), will this refresh the BB RAM and generate a checksum ?

Yes, doing the external calibration procedure will restore the calibration data and write a new checksum but the calibration procedure is not trivial like on a 2440 series.

Since there is no way to remove the RAM to read the contents, without losing the data, there must be a method for the scope to "restore" itself from a dead battery.

This might be possible through the GPIB interface; I do not remember.  One of the 2465 series experts will know.

[tecman]

David:

Thanks for the info.  I was planning to do a full cal on the unit.

 Based on the photo and s/n it looks like a mid 2467.  The service manual lists 2465a and 2467 as both covered by the same manual.  I looked for a reference site that might cross the s/n to a date (B011907), but no luck.  It does have the PBs instead of the flip switched for mode setting, a'la the 2465a.  When it comes in I will look for date codes.

paul

[MarkL]

Since there is no way to remove the RAM to read the contents, without losing the data, there must be a method for the scope to "restore" itself from a dead battery.

This might be possible through the GPIB interface; I do not remember.  One of the 2465 series experts will know.

I don't know about the 2565A/2467, but the EAROM in the 2456 can dumped via GPIB.  By examining the EPROMs on the GPIB board I was able to figure out the following.  It may be extensible to other models.

First, you have to enable the "special" commands:

  key 0

Then you can dump the EAROM with the following command:

  earom? <address to dump>[,<address to dump>,<...>]

This allows reading out of multiple locations in one command.  For example, to dump the first 10 addresses:

  earom? 0,1,2,3,4,5,6,7,8,9

Then the next 10:

  earom? 10,11,12,13,14,15,16,17,18,19

And so on, until address 199.  The first 100 locations are the EAROM on the main board, and the second 100 is the EAROM on the buffer board.  This is the same as in the exerciser.  The values should match.

The above commands return a line in the form (from the above example 0 to 9):

  EAR 0:84,1:1993,2:10102,3:1893,4:10091,5:1798,6:1494,7:1420,8:9608,9:1431;

The return format is <address>:<value> and everything is in decimal.  You can set an EAROM value by using the same format:

  earom <address>:<value>

Trying to write multiple values in one line didn't work for me and crashed my scope.  YMMV.

It would be interesting if someone could try this on a model other than a 2465(plain).

WARNING!!   If you want to play with any of this, be careful not to overwrite overwrite the EAROM values until you've saved them with a screen capture!  On my 2465 it DOES NOT MATTER if you have the the cal jumper set to off.  These commands are surely hidden for a reason!


If you're even braver, there's another hidden query I've discovered called "BYTE?".  It takes one argument and dumps that number of bytes from somewhere.  I can't figure out what it's dumping, and I can't make it repeat itself except after a reboot.  Eventually after several hundred bytes it causes my 2465 to reboot.  I'm mentioning it because maybe that command will be useful on other models without EAROMs.


If anyone decides to try anything, please post your results.  Obviously this needs a GPIB interface.  I haven't found any equivalent method to write EAROM/NVRAM values from the screen, although now that I found it via GPIB it makes me suspect it's possible.

[Satbeginner]

Hi all,

Today my Mouser shipment with the replacement parts arrived, so i started with the recap of the power supply.

There was the error in the board layout in the manual where the caps C1115 en C1132 were mixed up, but I used the existing population on the board as leading.
I also found this thread where the error in the doc is described: https://www.eevblog.com/forum/testgear/tektronix-2465b-oscilloscope-teardown/475/

I used the 2465B Recap Modified parts list that is floating around (attached with this message) to order at Mouser, and I found 2 points of attention:

The 2 big caps ordered as 330uF 250V need to be axial for proper mounting, and
On my PS there are 3pcs 1uF 50V non polarised caps instead of 2pcs.C1274, C1291 AND C1292!!

Anyway,

The PS part is done, and I really think the minor trembling of the readout is gone now, so, I consider it a succes.

Now I want to improve the cooling of U800 and replace the Dallas chip.
That will be something for tomorrow, I think.
Tomorrow I will also put some pictures of the result of the PS and also my A5 board.

Un saludo,

Satbeginner (Leo)

[Bryan]

Now I want to improve the cooling of U800 and replace the Dallas chip.

I am not convinced the U800 needs to cooled, yes it gets hot, but I am sure Tek took this into account when engineering. I suspect failures are due to a fan that stops working or the cooling vents at the bottom are plugged with lint, or maybe it's used in a environment that is beyond it's ambient temperature specs.

[med6753]

Now I want to improve the cooling of U800 and replace the Dallas chip.

I am not convinced the U800 needs to cooled, yes it gets hot, but I am sure Tek took this into account when engineering. I suspect failures are due to a fan that stops working or the cooling vents at the bottom are plugged with lint, or maybe it's used in a environment that is beyond it's ambient temperature specs.

If it's the Tek manufactured die typically found in the 2465 I would agree. The Maxim die typically found in the 2465A & B has been proven to be more prone to heat failure.

[Bryan]

I thought the Maxim dies were not specific to the U800 but for the other custom chips, U400's maybe

[Satbeginner]

Hi all,

here again some progress on my 2465B revival.

I will share my info on the steps I took so far:

- Recap Power Supply & tests of the old Caps afterwards... ;-)
- Repair/replacement of the Delayed Time Base lever inside the front-panel
- Removal of the Dallas and placement of a socket so it can be replaced when I get my programmer
- Cooling of my Tek U800 and how
- Pictures of the previous replacement of the four caps on the A5 board

Although I am a big fan of "If it ain't broken, don't try to fix it..." I do think this 2465B could be the scope I will use till I die, so I decided to do the recap.
Mainly because when Tek build this thing, they made sure all new components were up to spec, so now, after more than 25 years, I thought it could be useful to renew the old parts that are known to age.

So yesterday I did a full recap of the Power Supply, that job took me a few hours, but all-in-all, not too bad.
I just replaced everything, but I sure was curious about how necessary it actually was, so today I did a quick capacity and ESR test with one of these small all-in-one testers.

As it turned out, more than half of the capacitors were in fact OK!?
Even so, I am happy knowing all parts are now new and "future-proof".

The caps that were bad were:

3,3uF, 350V
10uF, 160V
10uF, 100V
47uF, 25V
4,7uF, 35V
Tantalum 4,7uF

and funny enough, one of the 15 Ohm resistors went up to 4968 Ohm!! without any markings or change of color.

When I got my 2465B, the VAR knob inside the Time Base was missing, and as it turned out, the pull to delayed time base was not working either.
When I looked in to this the little lever that operates the slide switch inside was broken, so that had to be replaced. (I am not a great fan of glue...)
Qservice in Greece had this part, so that was an easy fix.

Next, the Dallas chip.
Almost all the parts have a 90-ties datestamp  on them, so this means the Dallas is -still working- in almost 16 year of extra time!
I choose to go for the original, so today I removed my Dallas and put a socket on the board.
Reading and copying must wait till I receive my programmer, but at least the soldering bit is done today.
I used a simple push-and-release manual tin-sucker that worked OK on all pins but pin 14. (because of the ground-plane inside the board)
I removed the tin from the holes, and then, using a small pointed pair of pliers, I wiggled the pins one by one from left to right, to make sure they were loose inside the holes before pulling on the chip.
To remove the Dallas I applied the soldering iron to pin 14 while carefully pulling the Dallas out.

After that I placed a socket on the board, I choose the double action flat surface type of socket, because they seem to work a bit better with pins of unequal thickness. (tin residue on the used Dallas)
After that, the moment of truth was, will it still have calibration info in it??? Yes!! (I am a happy bunny right now!)

And now the final part: cooling of the U800.
The consensus seem to be, that the TekTronix made chips are "better" in terms of failing on the long term.
Mine is not getting hot, but it may have to do whem touching the chip is is just doing almost nothing?
Maybe when the scope is running on a very high frequency with a high repetition rate, it may get hot due to more capacitive load??

Anyway, I decided to help this little chip a bit.
What I did was, I removed the nut where the metal tab from under the chip is, and made a brass (for thermal conductivity) bushing the same height as the remaining part of the chip.
So basically, I try to guide the heat through this bushing to the top, and there transfer it to a cooling element.
I also put some thermal conductive grease on top of the chip, although this may be of little effect.
The cooling I found at Mouser's, it is a double to-220 part.

I will start adding pictures to show what I did.

Un saudo,

Satbeginner (Leo)