Topics

11801 NVRAM

Reginald Beardsley
 

I just got bit by "none of the above". Rather than Dallas, Maxim or ST NVRAM parts, U500 and U511 on the timebase board are NEC D43256C SRAM in a battery powered socket inserted into the PCB socket.

Not having seen this before, I pulled the SRAM from the battery backup on U511, so whatever data might still have been there is now gone.

Is there any documentation of what is stored in U500 and U511? My understanding is that factory "enhanced accuracy" data are stored in one or both.

It's very frustrating not having component level service data as it's all TTL parts and *should* be easy to maintain indefinitely. Though now that I've seen the board, I doubt that the 1.4 ps jitter can be reduced with reasonable effort. It's likely a consequence of the variation in propagation time through a long chain of gates.

Jack2015
 

I did same mistake last week on my CSA803 when replacing NVRAM on the timebase board(selftest error code E5322),
the factory calibration data is gone,now when I measure a 100khz signal it shows me 106Khz,when measure a 1Mhz signal it shows 106Mhz...
I want to know how to fix this problem too.

Dewey Wyatt
 

The way it was explained to me on my 2430a, if you loose the data then you must do a recalibration. I was lucky even though it didn’t pass the self cal due to a low battery the info was still there and I was able to use my tl8662 plus to read and save the data then copy it to the new sram modules. So far so good with it I used a different sram that has a snap cap battery which is replaceable. The data and pinout was the same but it is a surface mounted sram so I had to get boards made and solder the chip to a carrier board then add pins from a header.

Sent from Mail for Windows 10

From: Jack2015
Sent: Sunday, March 24, 2019 8:32 AM
To: TekScopes@groups.io
Subject: Re: [TekScopes] 11801 NVRAM

I did same mistake last week on my CSA803 when replacing NVRAM on the timebase board(selftest error code E5322),
the factory calibration data is gone,now when I measure a 100khz signal it shows me 106Khz,when measure a 1Mhz signal it shows 106Mhz...
I want to know how to fix this problem too.






---
This email has been checked for viruses by AVG.
https://www.avg.com

Jack2015
 

It looks like I have to study the SM and figure out how to do the recalibration by myself,
wondering did any one got same problem and has recalibrated a CSA803x or 11801X before?

Thanks for your information!

Reginald Beardsley
 

I was able to select a DS1230Y NVRAM package on the TNM5000 and read U500. However, it only contained 32 bytes of data. Everything else was alternating 32 bytes of 00 and FF.

I should note that when I selected the NVRAM part, the TNM put up a warning to select the device before inserting the device in the ZIF which is the opposite of what I did.

Does anyone have images of U500 and U511 from a CSA803 or 11801 which has the factory cal data? We'd also need the 200 MHz OXCO measured using a GPSDO referenced counter and the version of the timebase board ROMS. Mine are version 10.1 dated 1991.

Does anyone of the list know who designed the board? It's dated 1988. It they are still alive we might get some help from them.

Is there any information on the recalibration process?

This needs to be documented before all of them fail. Twenty odd years ago there was a detailed description of how to recover a Sun workstation from a failed NVRAM. So if you bought a used system, one of the first things you did was record the serial number, etc for future reference.

Reg

Reginald Beardsley
 

You won't be working by yourself if you send me any information you get. I'll certainly do the same.

This has become a grudge match with Tek management.

--------------------------------------------

On Sun, 3/24/19, Jack2015 <tolkien@...> wrote:

Subject: Re: [TekScopes] 11801 NVRAM
To: TekScopes@groups.io
Date: Sunday, March 24, 2019, 9:24 AM

It looks like I have to study the
SM and figure out how to do the recalibration by myself,
wondering did any one got same problem and has
recalibrated a CSA803x or 11801X before?

Thanks for your information!

Jack2015
 

How about take out the DS1213 and replace it with a pin compatible 256K SRAM(HM52256,CXK58257,D43256)?
It should work if timebase board supply 5V to the SRAM.if it doesn't work then disconnect the VCC pin of SRAM
from IC socket and supply 3v(two 1.5V batteries) to vcc pin of SRAM then try it again(of course the GND of battery must
connected to GND of timebase board).it will make the SRAM work like a battery backed up NVRAM.

Jack2015
 

Forget my idea,the 3V to VCC of SRAM might be too low.too dangerous!
but you can replace the DS1213 with a pin compatible 32k8 SRAM and try it.
if it works you will get battery error but you can exit diagnostic and start to use
the scope.

Reginald Beardsley
 

I have replacements made by ST in transit. As for exiting the Extended Diagnostics, my last post reported that I am not able to do that.

I have tried deleting the failing tests, but cannot.

I have tried exiting diagnostics, but cannot.

I have tried entering debug mode, but cannot.

I have turned off "stop on error" but it has no effect.

I tried changing the jumpers on J712 and J713, but that had no effect.

I have tried installing both of the SD-22 heads, but that had no effect. I have no idea if the SD-22s are any good or not. The channel select LEDs flash on power up, but that is all.

When I compare this instrument to my 485 and 465 it is sad to see how far Tek fell in a very short time.

Reg

bobh@joba.com
 

Re: ROM images, did you check here:

http://www.ko4bb.com/getsimple/index.php?id=manuals

Search ko4bb for 11801.

Bob.

On 3/24/2019 7:26 AM, Reginald Beardsley via Groups.Io wrote:
I was able to select a DS1230Y NVRAM package on the TNM5000 and read U500. However, it only contained 32 bytes of data. Everything else was alternating 32 bytes of 00 and FF.

I should note that when I selected the NVRAM part, the TNM put up a warning to select the device before inserting the device in the ZIF which is the opposite of what I did.

Does anyone have images of U500 and U511 from a CSA803 or 11801 which has the factory cal data? We'd also need the 200 MHz OXCO measured using a GPSDO referenced counter and the version of the timebase board ROMS. Mine are version 10.1 dated 1991.

Does anyone of the list know who designed the board? It's dated 1988. It they are still alive we might get some help from them.

Is there any information on the recalibration process?

This needs to be documented before all of them fail. Twenty odd years ago there was a detailed description of how to recover a Sun workstation from a failed NVRAM. So if you bought a used system, one of the first things you did was record the serial number, etc for future reference.

Reg


Reginald Beardsley
 

Those are *not* 11801 manuals. They are 11801C manuals. I currently have a PDF of the 11801 service manual courtesy of a member of the list and the 11801,11802 & SM11 diagnostics courtesy of TekWiki. I have paper copies of the 11801 user and service manuals in transit.


--------------------------------------------

On Sun, 3/24/19, bobh@... <bobh@...> wrote:

Subject: Re: [TekScopes] 11801 NVRAM
To: TekScopes@groups.io
Date: Sunday, March 24, 2019, 4:44 PM

Re: ROM images, did you check
here:

http://www.ko4bb.com/getsimple/index.php?id=manuals

Search ko4bb for 11801.

Bob.


On 3/24/2019 7:26 AM, Reginald
Beardsley via Groups.Io wrote:
> I was
able to select a DS1230Y NVRAM package on the TNM5000 and
read U500.  However, it only contained 32 bytes of data. 
Everything else was alternating 32 bytes of 00 and FF.
>
> I should note that
when I selected the NVRAM part, the TNM put up a warning to
select the device before inserting the device in the ZIF
which is the opposite of what I did.
>
> Does anyone have images of  U500 and U511
from a CSA803 or 11801 which has the factory cal data? 
We'd also need the 200 MHz OXCO measured using a GPSDO
referenced counter and the version of the timebase board
ROMS.  Mine are version 10.1 dated 1991.
>
> Does anyone of the
list know who designed the board?  It's dated 1988. 
It they are still alive we might get some help from them.
>
> Is there any
information on the recalibration process?
>
> This needs to be
documented before all of them fail. Twenty odd years ago
there was a detailed description of how to recover a Sun
workstation from a failed NVRAM.  So if you bought a used
system, one of the first things you did was record the
serial number, etc for future reference.
>
> Reg
>
>
>
>

bobh@joba.com
 

Yeah, I was referring to the ROM images not manuals.

Bob.

On 3/24/2019 4:18 PM, Reginald Beardsley via Groups.Io wrote:
Those are *not* 11801 manuals. They are 11801C manuals. I currently have a PDF of the 11801 service manual courtesy of a member of the list and the 11801,11802 & SM11 diagnostics courtesy of TekWiki. I have paper copies of the 11801 user and service manuals in transit.


--------------------------------------------
On Sun, 3/24/19, bobh@... <bobh@...> wrote:

Subject: Re: [TekScopes] 11801 NVRAM
To: TekScopes@groups.io
Date: Sunday, March 24, 2019, 4:44 PM
Re: ROM images, did you check
here:
http://www.ko4bb.com/getsimple/index.php?id=manuals
Search ko4bb for 11801.
Bob.
On 3/24/2019 7:26 AM, Reginald
Beardsley via Groups.Io wrote:
> I was
able to select a DS1230Y NVRAM package on the TNM5000 and
read U500.  However, it only contained 32 bytes of data.
Everything else was alternating 32 bytes of 00 and FF.
>
> I should note that
when I selected the NVRAM part, the TNM put up a warning to
select the device before inserting the device in the ZIF
which is the opposite of what I did.
>
> Does anyone have images of  U500 and U511
from a CSA803 or 11801 which has the factory cal data?
We'd also need the 200 MHz OXCO measured using a GPSDO
referenced counter and the version of the timebase board
ROMS.  Mine are version 10.1 dated 1991.
>
> Does anyone of the
list know who designed the board?  It's dated 1988.
It they are still alive we might get some help from them.
>
> Is there any
information on the recalibration process?
>
> This needs to be
documented before all of them fail. Twenty odd years ago
there was a detailed description of how to recover a Sun
workstation from a failed NVRAM.  So if you bought a used
system, one of the first things you did was record the
serial number, etc for future reference.
>
> Reg
>
>
>
>


Reginald Beardsley
 

Things are generally a bit disorganized in the archive. Once I have my replacement NVRAM I'll set about archiving all my ROMs. And writing up a description of the repair process. The only thing I found that was clearly relevant was the EPROMs in some of the sampling heads.

--------------------------------------------

On Mon, 3/25/19, bobh@... <bobh@...> wrote:

Subject: Re: [TekScopes] 11801 NVRAM
To: TekScopes@groups.io
Date: Monday, March 25, 2019, 1:03 AM

Yeah, I was referring to the ROM
images not manuals.

Bob.

be_moulton
 

Noob to the group. I had wondered IF NVRAM data was lost if that actually did remove factory enhanced time interpolation calibration data, I recall vague reference to that in one of the manuals ? I have some equally broken CSA803a and 11801b and haven't taken the plunge to attempt fix yet.

I see the post about the apparent shift in frequency measured changing from known 100KHz to unexpected 106KHz upon losing the NV data. So sounds like it may be true regarding special factory time interpolation cal constants. So, how to restore?

My very limited understanding of the CSA803/11801s is that the 200KHz sampling rate has a tiny time interval added each 200KHz sample so that each subsequent sample taken is shifted very slightly in time to be in a different spot on the repetitive waveform so that an entire cycle of the repetitive waveform can be reassembled from the 512 samples taken. For the measured frequency of the reassembled waveform is to be accurate, then that added time interval to each 200KHz sample must be a very accurate delta time interval as it is dictating the effective sampling rate. So, if the missing factory calibration constants are adjusting that added delta time interval for each subsequent sample, then the DEFAULT "time interpolation" constants following NVram recovery may corrupt the effective time sampling and the calculated frequency of the reassembled effective time sampled waveform. So, if those 32 calibration numbers could be adjusted while monitoring a known precision frequency sinewave, then I wonder if it's possible to see the sampled waveform begin falling into place as the expected sinewave as the cal constants approach a better time interpolation? The challenge would be how to iteratively adjust the NVram constants without removing the NVram chips a thousand times. I don't know of any GPIB way to SET the value of each NVram cal constant, maybe there's some tribal knowledge regarding that ??

Lacking any external wisdom, I may also spend some time bashing my brain against the blue/green mystery machine to attempt restoring a decent calibration. I'm betting somebody knows a trick to do this.

Best of luck.

Albert Otten
 

/
/ The challenge would be how to iteratively adjust the NVram constants without removing the NVram chips a thousand times. I don't know of any GPIB way / to SET the value of each NVram cal constant, maybe there's some tribal knowledge regarding that ??
/

Perhaps you haven't seen this message https://groups.io/g/TekScopes/message/159274 and the previous ones in that topic?
Albert

Reginald Beardsley
 

I'm very interested in how to derive and store the correction terms. There has been discussion of a command to write them, but it appeared not to work when attempted IIRC.

It would be very helpful to know where they are stored in the NVRAM. I've got a programmer, but no idea where to put the data.

My Keysight 33622A is specified as <1 ps jitter. That's less than the 4 ps jitter specified for the 11801. I think I'll play with that a bit and see if I can persuade the 11801 to tell me what the uncalibrated errors are.

Having just built a GPIB-USB adapter using an Uno and AR488 I think I need to read the programming manual for my 11801 to see if I can get digits out via the GPIB port. That would allow removing the NVRAM one time to load the correction values.

Have Fun!
Reg

Reginald Beardsley
 

FWIW I set up my 11801 using the internal trigger and feeding a ~100 kHz ramp from my 33622A. By making slight adjustments to the ramp frequency I get a straight line which intersects the graticule all the way across the screen at 5 ps/div and 2 mV/div using 5120 points and at 1 ps/div with 1024 points. The internal timebase is not as stable as the 33622A which, despite not having the high stability option, has *very* little drift relative to a GPSDO from Leo Bodnar.

I'm obviously severely aliased to get a single cycle image of a 100 kHz ramp in a 10 or 50 ps wide window. That corresponds to 10 femtosecond sampling.

If there is a time base correction that was lost I don't see how I can measure it or what effect it might have. The ramp appears dead straight though it jumps around a lot because of the 4 ps jitter of the 11801 clock. By continually fiddling with the ramp frequency I can stop the display for a single sweep with a diagonal line over the full screen from upper right to lower left when collecting 5120 points.

I'm going to puzzle over this while I fix a rather late lunch. But I can't see how there is a significant issue. I should be able to do better getting digits via the GPIB interface and supplying a trigger signal from the 33622A.

As I removed the NVRAM battery to get it to boot before I replaced the NVRAM chip, there's no question that if there were cal data in the NVRAM chip it was lost.

Have Fun!
Reg

Reginald Beardsley
 

FYI I created a photo album and posted a picture of the 33622A ramp displayed on my 11801 using an SD-26 head at 1 ps/div and 2 mV/div with 1024 samples.

With a trigger signal at 100 kHz, you need a ramp at 100.0001 kHz to test the linearity of the time base. I'm just using the internal timebase, the 33622A set as close as I could and taking the photo at the right moment. It took a couple of tries. I'll try for more rigorous results later.

I'd like to thank be_moulton for suggesting the idea. I had pondered the subject, but lacked the insight needed. It's a great pleasure to have such excellent company join us.

Have Fun!
Reg

Albert Otten
 

Hi Reg,

I think you refer to my message #158778. The GPIB command ccalcorrection does the job. That command was present in the CSA803A but not yet in the CSA803.This suggests that the calibration constants were introduced in the later models. So I suppose this feature is also not yet present in your 11801.

As far as I know there is no GPIB command to read memory contents. The Debug facility in the Extended Diagnostics menu can show memory contents at the scope screen; see also 11801 Diagnostics pdf, section 5.2.7. For viewing the time base NVRAM the Time Base subsystem should be selected in the E.D. menu. With Debug I could not find the GPIB commands in the firmware; probably some memory bank switching is needed to access the proper memory chips. I based my above mentioned conclusion on chip memory dumps obtained from elsewhere.

Albert

On Sun, Sep 15, 2019 at 07:59 PM, Reginald Beardsley wrote:


I'm very interested in how to derive and store the correction terms. There
has been discussion of a command to write them, but it appeared not to work
when attempted IIRC.

It would be very helpful to know where they are stored in the NVRAM. I've got
a programmer, but no idea where to put the data.

My Keysight 33622A is specified as <1 ps jitter. That's less than the 4 ps
jitter specified for the 11801. I think I'll play with that a bit and see if
I can persuade the 11801 to tell me what the uncalibrated errors are.

Having just built a GPIB-USB adapter using an Uno and AR488 I think I need to
read the programming manual for my 11801 to see if I can get digits out via
the GPIB port. That would allow removing the NVRAM one time to load the
correction values.

Have Fun!
Reg

Chuck Harris
 

For what it's worth, the CSA803C, and the 11801C are
the same instrument, +/- two active ports for sampling
plugins, which are substituted for power only ports on
the CSA803C.

They are both covered in the 070-9970-00 Programmer Manual.

By inference, they both have the same command set, which
includes:

CALibrate <alpha> <ui> <link>: <arg>

which auto cal's the command in the argument.

With links,

AUTO: {All | Blowby | LOOpgain | OFFSet | ONUll or TDRAmplitude }

BLOwby:

CSTore: {All | Blowby | LOOpgain | OFFSet | ONUll or TDRAmplitude }

or,

DREcall: {All.....TDRAmplitude}

etc...

So, anything you can do with your CSA803A, you can do and more with
the much later 11801C....

Or, so it appears.... The Programmer Manual is about as clear as mud.

-Chuck Harris





Albert Otten wrote:

Hi Reg,

I think you refer to my message #158778. The GPIB command ccalcorrection does the job. That command was present in the CSA803A but not yet in the CSA803.This suggests that the calibration constants were introduced in the later models. So I suppose this feature is also not yet present in your 11801.

As far as I know there is no GPIB command to read memory contents. The Debug facility in the Extended Diagnostics menu can show memory contents at the scope screen; see also 11801 Diagnostics pdf, section 5.2.7. For viewing the time base NVRAM the Time Base subsystem should be selected in the E.D. menu. With Debug I could not find the GPIB commands in the firmware; probably some memory bank switching is needed to access the proper memory chips. I based my above mentioned conclusion on chip memory dumps obtained from elsewhere.

Albert