Tektronix 2465B oscilloscope teardown

[malch]

This one is for BravoV. I have the same vertical geometry problem you had [the words are too low to read]
You said that you adjusted the  vertical readout trimpot.
 I cant find this "vertical readout trimpot" in the schematics or troubleshooting. I cant even find those words in a 2445B or 2465B service manual.
Could you tell me what R number you adjusted?

[BravoV]

This one is for BravoV. I have the same vertical geometry problem you had [the words are too low to read]
You said that you adjusted the  vertical readout trimpot.
 I cant find this "vertical readout trimpot" in the schematics or troubleshooting. I cant even find those words in a 2445B or 2465B service manual.
Could you tell me what R number you adjusted?

Malch, remember, I'm NOT the expert on this problem, and "maybe" my method is not the right way to fix this problem. So proceed with your own risk.

My suggestion, read the service manual, and understand what these 2 pots do. Also use a sharpie/marker to mark the current position at the pot, so if you suspect you screwed up  while adjusting these pots, you can turn it back to the previous position.

The attached photo shots below should speak for it self where to find the pot, sorry, I forgot which one  , but definitely one of these two, again, read and understand what are their purpose 1st before adjusting it.

[malch]

I just received my 2445B, Its like new. The serial number is B064xxx , so I will have to change the caps in A5 and check the rest.
The Dallas has a 1990 date. It was last calibrated in Oct 2001.
I have the downloaded manual . I looked for many hours and could not find that fault mentioned.

So thank you very much for the info. I would not have found it as there are 8 possible trim pots. 

P.S. Why does everybody hide their serial numbers?
 Got the failure prone .068uf caps.
Found the manual page for screen centering. page 5-22 CAL 07

[linux-works]

OT: but suppose someone sees your serial and makes a fake report to some police dept that your unit was stolen.

you just posted your unit with your serial number.  its now up to you to prove you did not steal it.

why even get into that mess?  why allow bad people to try to screw you over?

so, many of us avoid it by not posting the serial's.  no reason to give serials on gear away, anyway.  what gain (to anyone) would there be?  I've already listed at least one danger.

[MarkL]

Wondering if anyone has attempted to hand-copy cal constants using the "Calibration RAM Examine" thingie in the exercisers (or has at least verified that they are stored as shown in the RAM)...

I did this with my 2465, but I took a more lazy approach.  I took a video of the screen while flipping through all the memory locations.

The 2465 has an EAROM which hasn't gone bad (maybe yet), but now I have all the cal values if it does.

[tymm]

Cool -- though do you know whether the constants can just be copied into a new IC and have it work?

I've copied down the data from mine - and think there's a good chance it would work just straightforward... though I can see lots of ways that it could be more complicated (cal data stored at an offset address in the NVRAM, secondary checksums, etc).

Looking forward to hearing from sparkybg; will probably just go ahead with mine and test whether things match if there's no confirmation by the time I'm ready to go.

[MarkL]

Cool -- though do you know whether the constants can just be copied into a new IC and have it work?

No, I don't know.   I'm assuming it's accurately reporting the word at each location.  I'll have to test reprogramming in battle.

Living on the edge?  Maybe.  But I haven't heard any tales yet of the EAROMs going bad, and I don't use the scope enough to invest the time to fix what might not be a problem.

Another thing I've done on other equipment is to use an exercise routine to cycle through NVRAM and clip on a logic analyzer to capture the bytes.  Then you *know* whether or not you have all the cal data.

[malch]

Watch out you dont squash or bend the two little stiff wires going to the middle of the CRT
 while trying to get the psu cards out.

[sparkybg]

Done:
- Full recap of power supply - the immediate result is sharper display, better triggering, and (although I am not completely sure) faster "boot-up" after the scope is powered up. I've used Panasonic FR, low ESR Nichicon and low ESR Rubicon caps.
- Full refurbishment of A5 board. I replaced the DS1225Y NVRAM with FM18W08 FRAM. No problems whatsoever. I was able to read the contents of my original NVRAM and put them in the FRAM without any problem. The scope boots and works normally.

To do:
- Full recap of main board.
- Replacement of the only electrolytic cap in the CRT supply.
- Full calibration, if I find a place with all the equipment needed, or at least vertical calibration. All other is OK, but vertical is with 2-3%  lower amplitude than cursors say.

The calibration constants from exercise 2 are located at addresses 1E00 - 1FFF in the NVRAM. I don't know if they can be used with another dump. Maybe I will try just out of curiosity. Maybe I will able to tell which constants are for vertical calibration when I calibrate the vertical.

I will post some pictures of the process when all the tasks are done.

Update: I have just tried a 2465B S/N B05xxxx dump on my 2445B S/N B06xxxx with my original calibration constants. It works flawlessly! So, it is a good idea to backup your constats from exercise 2 before you do anything with the NVRAM chip. It can save you a calibration if something goes wrong when you try to read your NVRAM after desoldering.

[BravoV]

The calibration constants from exercise 2 are located at addresses 1E00 - 1FFF in the NVRAM. I don't know if they can be used with another dump. Maybe I will try just out of curiosity. Maybe I will able to tell which constants are for vertical calibration when I calibrate the vertical.

Great info !  Please, keep us updated on your quest for calibration constants if you decided to decode it.

I will post some pictures of the process when all the tasks are done.

 

[tymm]

Update: I have just tried a 2465B S/N B05xxxx dump on my 2445B S/N B06xxxx with my original calibration constants. It works flawlessly! So, it is a good idea to backup your constats from exercise 2 before you do anything with the NVRAM chip. It can save you a calibration if something goes wrong when you try to read your NVRAM after desoldering.

Sweet -- thanks so much for testing it out and following up.  Really great to know that it works!

[sparkybg]

Just two pictures of FRAM mod for now. Enjoy.

[sparkybg]

Malch, remember, I'm NOT the expert on this problem, and "maybe" my method is not the right way to fix this problem. So proceed with your own risk.

My suggestion, read the service manual, and understand what these 2 pots do. Also use a sharpie/marker to mark the current position at the pot, so if you suspect you screwed up  while adjusting these pots, you can turn it back to the previous position.

Use these pots without any marking or so. It is pointless - they are only to center the readout vertically. They have nothing to do with the calibration, or cursors. Try them - you will see for yourself that only the readout (top and bottom text fields on the crt) will move. One of the few things in this scope that does not interfere with something else. Maybe only horizontal or vertical readout jitter will have to be readjusted, but again - you can do this also without affecting anything else.

[malch]

Thanks sparkyb. I am going to get at pin 7 today to add a battery to my Dallas. I,m thinking of staying with the Dallas with a big battery.

[sparkybg]

The choice is yours of course.

But, from my point of view, the FRAM is much better solution. FM16W08 costs around 10EUR in Mouser Europe, than you need a SOIC28 to DIP28 PCB (maybe a dollar or so, from eBay for example), and you are done with this problem forever.

On the other side, no one has confirmed (nor can confirm) how long the external battery will last, considering it has a 20 years old battery connected in parallel with it. The desoldering of the old chip is the most hard work to do, and once it is done, you have completed 90% of the work anyway.

...well, I bought a ZD215 desoldering station recently, and desoldering of the old chip was a piece of cake with it. It was done in under 2 minutes with perfect result. The recapping of the PSU was also a pretty easy task using such instrument.

P.S.: Bear in mind that you absolutely HAVE to change the caps in the power supply, especially the polyester/polypropylene X/Y type ones. These are a ticking bomb waiting to explode. I had an older 2445 scope before and two of them exploded. Also, some of the electrolytic ones had 15 times bigger ESR compared to the the caps I replaced them with. You can leave the A1 board without recap, but the power supply and the A5 board are absolutely "must-do's" if you are planning to use your scope for some years.

[Howardlong]

My programmer wouldn't deal with the DS1225Y (a TL866A, it's listed, but it doesn't work), so I built my own.

It's been tested with both the DS1225Y and the RAMTRON FM18W08-SG (although only the bottom 8k bytes of the RAMTRON is used).

I built it with parts that I had in the lab, it's based on a PIC18F4550, and I breadboarded it. It's controlled from a TTL 3.3v level serial port. I used an FTDI TTL-232R-3V3, but a 5V one should work just as well. The device must be powered from 5V, it will detect a brown out and reset if the voltage falls below about 4.3V or so, this is to make sure that unexpected writes don't occur during power up/power down (control signals go Hi-Z and there are also pullups on /WE and /CE).

The NVRAM data is stored in the PIC's flash program area, so it remains in place after power down.



Schematic attached (Eagle and PDF), and source code and .hex file for use MPLAB 8 IDE (compiler is XC8), although it should import into MPLAB X if that's what you prefer.

Here is how to use it:

Insert source NVRAM and power up...

Code: [Select]

1E: reset status

--- DS1225Y NVRAM copy programmer ---
Select:
  [R]ead NVRAM into PIC flash
  [V]erify NVRAM with PIC flash
  [W]rite PIC flash to NVRAM
  erase PIC [F]lash
  erase [N]VRAM
  [B]lank check PIC flash
  fill NVRAM with ran[D]om data
  fill PIC flash with rand[O]m data
  [P]rint NVRAM contents
  print PIC f[L]ash contents
  get [C]hecksums
> R

Erasing flash...
Reading 1F00
Read completed.
Verifying 1F00
Verify completed: success.

--- DS1225Y NVRAM copy programmer ---
Select:
  [R]ead NVRAM into PIC flash
  [V]erify NVRAM with PIC flash
  [W]rite PIC flash to NVRAM
  erase PIC [F]lash
  erase [N]VRAM
  [B]lank check PIC flash
  fill NVRAM with ran[D]om data
  fill PIC flash with rand[O]m data
  [P]rint NVRAM contents
  print PIC f[L]ash contents
  get [C]hecksums
> C
NVRAM checksum=0xF9AB
Flash checksum=0xF9AB

--- DS1225Y NVRAM copy programmer ---
Select:
  [R]ead NVRAM into PIC flash
  [V]erify NVRAM with PIC flash
  [W]rite PIC flash to NVRAM
  erase PIC [F]lash
  erase [N]VRAM
  [B]lank check PIC flash
  fill NVRAM with ran[D]om data
  fill PIC flash with rand[O]m data
  [P]rint NVRAM contents
  print PIC f[L]ash contents
  get [C]hecksums
>

Power down and insert new NVRAM chip, then power back up...

Code: [Select]

1E: reset status

--- DS1225Y NVRAM copy programmer ---
Select:
  [R]ead NVRAM into PIC flash
  [V]erify NVRAM with PIC flash
  [W]rite PIC flash to NVRAM
  erase PIC [F]lash
  erase [N]VRAM
  [B]lank check PIC flash
  fill NVRAM with ran[D]om data
  fill PIC flash with rand[O]m data
  [P]rint NVRAM contents
  print PIC f[L]ash contents
  get [C]hecksums
> W
Writing 1F00
Write completed.
Verifying 1F00
Verify completed: success.

--- DS1225Y NVRAM copy programmer ---
Select:
  [R]ead NVRAM into PIC flash
  [V]erify NVRAM with PIC flash
  [W]rite PIC flash to NVRAM
  erase PIC [F]lash
  erase [N]VRAM
  [B]lank check PIC flash
  fill NVRAM with ran[D]om data
  fill PIC flash with rand[O]m data
  [P]rint NVRAM contents
  print PIC f[L]ash contents
  get [C]hecksums
> C
NVRAM checksum=0xF9AB
Flash checksum=0xF9AB

--- DS1225Y NVRAM copy programmer ---
Select:
  [R]ead NVRAM into PIC flash
  [V]erify NVRAM with PIC flash
  [W]rite PIC flash to NVRAM
  erase PIC [F]lash
  erase [N]VRAM
  [B]lank check PIC flash
  fill NVRAM with ran[D]om data
  fill PIC flash with rand[O]m data
  [P]rint NVRAM contents
  print PIC f[L]ash contents
  get [C]hecksums
>

That's it.

[malch]

I already have all the fram pieces in my watch at ebay.
I also found an earom on ebay [new] and a earom burner.$$

Removing the Dallas and soldering the socket has been done.
I dug in the epoxy and checked the Dallas battery voltage=3.45v.
So now I will copy the Dallas to my hard drive before anything else.

My idea is to add a laptop battery [connector and all] and disconnect the original battery.
If that fails or is impractable I still have the bin to burn to a new chip-Dallas or Fram.

Mouser is sending all the caps for A5 and the LVPS cards.
Two points.
 If I had to do all over again I would only buy a 24x5b with a lower serial number than 50,000 to get rid of the SMDs.
 I am replacing 3 of them, but there are still 20 to 40 of them left in the scope.[replaced the 4 axial ones too]

It is a pleasure working on this scope mostly because of the people on this thread.

[sparkybg]

Make sure you get "W" FRAM. FM18W08 or FM16W08. These lasts 100 times longer (10E14 cycles) then devices without "W" (10E12 cycles).

I have not seen "W" devices on eBay. And, I prefer to buy from well known reseller (Farnell, Digikey, Mouser) than eBay.

[malch]

This pic is for anybody interested in the battery connection  under the epoxy. A strap is spot welded to a watch battery with one end going straight to pin 14.
I drilled two holes, one on each side so I could get my side cutters in to make the fatal cut.
I used a 1.2mm carbide router bit in a Dremel to do the cut.

[grumpydoc]

Make sure you get "W" FRAM. FM18W08 or FM16W08. These lasts 100 times longer (10E14 cycles) then devices without "W" (10E12 cycles).

I have not seen "W" devices on eBay. And, I prefer to buy from well known reseller (Farnell, Digikey, Mouser) than eBay.

Are you sure - the cypress datasheets list the "W" types as "Wide voltage" - i.e 2.7 to 4.5V max. The Non "W" 5.5V max devices are rated for 4.5 to 5.5V Vcc max.

Both are rated for 10¹⁴ operations.

Or, in other words if you sequentially accessed every location in the device at maximum speed (130ns cycle time) it would take 130ns x 10¹⁴ x 32768/8 = 16k years to wear the device out.

[BravoV]

@Howardlong

Thanks for sharing that ! 

.... 16k years to wear the device out.

Hmmm ... give it a really-really generous de-rating, say only 1% of 16K years = 160 years , ok, I'm content now. 

[sparkybg]

Are you sure...

Yes, I am.

Both are rated for 10¹⁴ operations.

No, they aren't. Look again.

Here's FM1608 datasheet:
http://www.farnell.com/datasheets/82469.pdf

And here is FM16W08 datasheet:
http://www.cypress.com/?docID=48242

[grumpydoc]

No, they aren't. Look again.

Here's FM1608 datasheet:
http://www.farnell.com/datasheets/82469.pdf

And here is FM16W08 datasheet:
http://www.cypress.com/?docID=48242

Ah, I see your confusion

EDIT: I see mine as well - it's the newer FM1608B which is currently on the Cypress site - guess I missed that small part number change 

You have compared the data sheet for a Ramtron part with the data sheet for a slightly different Cypress part and concluded that the difference in endurance is due to the slight change in part number - but failing to spot that there is at least one other difference between these parts.

The supply voltage on the FM1608 is 4.5 to 5.5V
The supply voltage on the FM16W08 is 2.7 to 5.5V

In fact if you look over all of the Cypress parts http://www.cypress.com/?id=4986&addcols=&parametric=html&filter_184=32Kb+x+8#parametric you will see that they are all specified as 10¹⁴ cycles, both the "W" and non "W" part numbers.

So in your example it was the manufacturer that made the difference between 10¹² and 10¹⁴, however I suspect it is really process because the Ramtron data sheet is from 2007 and the Cypress one from 2014.

In fact there is a Ramtron FM16W08 which is wide supply voltage and 10¹⁴ cycles but the data sheet for that is quite recent as well (2012).

It's all moot anyway because Cypress acquired Ramtron last year.

So, yes, there are 32kx8 FRAMs with 10¹² endurance and 32kx8 FRAMs with 10¹⁴ endurance, and buying the wide supply voltage FM16W08 should get you a part with the better endurance but the "W" in that part number does not denote the endurance, it denotes the extended supply voltage tolerance.

[sparkybg]

Now, go to eBay, seek for "FM1608", and all the parts you will get will be Ramtron ones.

Then, do the same with "FM16W08", and you will again get some Ramtron devices.

Then, on the Cypress page, look for "FM1608", and you will get nothing. 16W08 is there.

So, in practice, buying FM1608 from eBay, you cannot know what exactly you are buying - a device with 10E14 or a device with 10E12 cycles. Buying 16W08 you are sure it is 10E14.

P.S: I would like to see that datasheet for 1608 with 10E14 rating. I cannot find it anywhere. Can you please give a link or upload it here?

[malch]

On page 3 BravoV posted this
 "Sad news, a failed to attempt to find the better replacement for the pesky Dallas DS1225Y BBSRAM, cause the F-RAM Ramtron FM1608-120 is not compatible. It was an impulse purchase thru a friend visiting abroad without checking the detail 1st, and also I was mis-leaded by info gathered from the Tektronix's Yahoo mailing list"
How come his chips didnt work but yours does?