2445B restored (and a question about cal data)


Paul Flinders
 

So, my 2445B is now a much happier oscilloscope :)

Photos at https://groups.io/g/TekScopes/album?id=271502

Once I'd unsoldered everything I could not find any actual trace damage - the pads on the 100n decoupling cap nearest the leaked electrolytics were slightly corroded but everything checked out OK - so I went hunting, especially around the chain of resistors that set the DAC reference current.

Sure enough R2013 (one of the 10k low tempco resistors) was open circuit - I think this has been the downfall of other A5 boards but don't remember who posted the details.

I honestly did not think the corrosion quite extended that far down the board so it's possible these resistors can just fail, but everything is back in place and the 'scope boots with no errors and, on a very quick check, is capable of displaying a trace, cursors etc and all looks good.

There's possibly more to do on this 'scope - certainly check all the functions and sort out the NVRAM; while the 'scope boots with no errors and the calibration looks pretty good the battery in the Dallas chip is 32 years old, give or take.

I have an FRAM waiting to take over but am a bit nervous about unsoldering the NVRAM - my rework skills, well still need a bit of rework to be honest, so I'd really like to get the calibration data out *before* I attack things with the desolder tool.

I understand the calibration data can be read using the diagnostics or over GPIB - does anyone have any pointers to the details (to save me from having to search the whole list archive).

In particular was the problem of transferring the "Exer 02" data to a new NVRAM/FRAM complete with checksums solved. Also can one *write* cal data over GPIB?


Thanks in advance

Paul


Mark Litwack
 

Hi Paul,

There are hidden GPIB commands that can read and write the NVRAM data. Take a look here:

https://www.eevblog.com/forum/testgear/tektronix-2465b-oscilloscope-teardown/msg1904177/#msg1904177

If you wanted to plug the EXER 02 data into the NVRAM with an external programmer, read here:

https://www.eevblog.com/forum/testgear/tektronix-2465b-oscilloscope-teardown/msg1933243/#msg1933243

The checksum verification routine for the 2445 and 2465 was figured out, but unfortunately doesn't work for the A or B series because it includes more that just a simple pass through a section of the NVRAM. Here's a C version of Tek's "spiral-add" checksum for the 2445/2465 if you want to experiment further:

https://www.eevblog.com/forum/testgear/tektronix-2445-2465-cal-settings-earom-er1400/msg927144/#msg927144

My personal opinion is that the FRAM is electrically not a great drop-in replacement for the Dallas part, although most people have success with it. You can search this mailing list's archive for past debate..

-mark

On Mon, Jan 17, 2022 at 05:30 PM, Paul Flinders wrote:

So, my 2445B is now a much happier oscilloscope :)

Photos at https://groups.io/g/TekScopes/album?id=271502

Once I'd unsoldered everything I could not find any actual trace damage
- the pads on the 100n decoupling cap nearest the leaked electrolytics
were slightly corroded but everything checked out OK - so I went
hunting, especially around the chain of resistors that set the DAC
reference current.

Sure enough R2013 (one of the 10k low tempco resistors) was open circuit
- I think this has been the downfall of other A5 boards but don't
remember who posted the details.

I honestly did not think the corrosion quite extended that far down the
board so it's possible these resistors can just fail, but everything is
back in place and the 'scope boots with no errors and, on a very quick
check, is capable of displaying a trace, cursors etc and all looks good.

There's possibly more to do on this 'scope - certainly check all the
functions and sort out the NVRAM; while the 'scope boots with no errors
and the calibration looks pretty good the battery in the Dallas chip is
32 years old, give or take.

I have an FRAM waiting to take over but am a bit nervous about
unsoldering the NVRAM - my rework skills, well still need a bit of
rework to be honest, so I'd really like to get the calibration data out
*before* I attack things with the desolder tool.

I understand the calibration data can be read using the diagnostics or
over GPIB - does anyone have any pointers to the details (to save me
from having to search the whole list archive).

In particular was the problem of transferring the "Exer 02" data to a
new NVRAM/FRAM complete with checksums solved. Also can one *write* cal
data over GPIB?


Thanks in advance

Paul


Paul Flinders
 

On 18/01/2022 16:46, Mark Litwack wrote:
Hi Paul,

...
Thanks for the pointers - will read through later.
My personal opinion is that the FRAM is electrically not a great drop-in replacement for the Dallas part, although most people have success with it. You can search this mailing list's archive for past debate..
Any particular reason why?

I know that the FRAM part does not work if you keep CE low and just use the address lines/WR/OE to drive write or read cycles but I thought that the CPU in the 2465 and related 'scopes toggled drove CE low per access cycle and it was inactive otherwise.

--
Paul


Mark Litwack
 

Hi Paul,

It was a concern about power on/off spurious writes to the FRAM in a circuit designed to support the Dallas part. I really didn't want to clog up everyone's mailbox with the discussion (again). Search for "litwack fram". I'd be happy to clarify any points if you want to send me a direct email.

Thanks,

-mark

On Tue, Jan 18, 2022 at 12:00 PM, Paul Flinders wrote:

On 18/01/2022 16:46, Mark Litwack wrote:
Hi Paul,

...
Thanks for the pointers - will read through later.
My personal opinion is that the FRAM is electrically not a great drop-in
replacement for the Dallas part, although most people have success with it.
You can search this mailing list's archive for past debate..
Any particular reason why?

I know that the FRAM part does not work if you keep CE low and just use
the address lines/WR/OE to drive write or read cycles but I thought that
the CPU in the 2465 and related 'scopes toggled drove CE low per access
cycle and it was inactive otherwise.

--
Paul


Paul Flinders
 

PM sent

What I don't understand is why Tek chose not to socket the NVRAM - with a specified battery lifetime of only 10 years you'd pretty much expect most instruments to need replacement during their lifetime. The fact that the batteries in these NVRAMs have been good for 20 or even 30+ years has been great for hobbyists buying 2nd hand 2445/2455/2465B 'scopes but not something that Tek could have relied on at the time of design/manufacture.

Planned obsolescence perhaps? - given that failure within the warranty period would have been vanishingly rare.

--
Paul

On 18/01/2022 21:20, Mark Litwack wrote:
Hi Paul,

It was a concern about power on/off spurious writes to the FRAM in a circuit designed to support the Dallas part. I really didn't want to clog up everyone's mailbox with the discussion (again). Search for "litwack fram". I'd be happy to clarify any points if you want to send me a direct email.

Thanks,

-mark

On Tue, Jan 18, 2022 at 12:00 PM, Paul Flinders wrote:
On 18/01/2022 16:46, Mark Litwack wrote:
Hi Paul,

...
Thanks for the pointers - will read through later.
My personal opinion is that the FRAM is electrically not a great drop-in
replacement for the Dallas part, although most people have success with it.
You can search this mailing list's archive for past debate..
Any particular reason why?

I know that the FRAM part does not work if you keep CE low and just use
the address lines/WR/OE to drive write or read cycles but I thought that
the CPU in the 2465 and related 'scopes toggled drove CE low per access
cycle and it was inactive otherwise.

--
Paul