Agilent probe interface ( AutoProbe and Probe Id)

[KE5FX]

I found some more info and was able to make a working Autoprobe with EEPROM:

https://electronicprojectsforfun.wordpress.com/using-the-keysight-autoprobe-interface-in-your-own-projects/hacking-keysight-autoprobe-eeproms/

Some very nice work there, Wolfgang, especially on the 3D printing job.   

Also intrigued by your PythonScreenShot app, but it doesn't get very far, at least under Windows:



Is there an easy way to install all the (seemingly countless) dependencies at once?   Anaconda, maybe?  If this can be made accessible to people who aren't Python gurus, I'd like to add a link to it in my GPIB Toolkit readme file. (PM me if you like, so we don't hijack this unrelated thread.)

[2N3055]

That is some nice work!!

I simply love the little red handle on the BNC !!

[Miek]

I've done quite a bit of work reverse engineering the firmware on my MSO6034A scope, so I might be able to fill in some blanks.

It definitely does check the model number against a list in the firmware (see attached image).
For older probes, it seems to have an ID number instead of the model string.

I went through the code that does the probe calibration and it seems to get the default offset & gain values from floats at 0x53 and 0x57 in the EEPROM for the newer named probes. Older probes may have them at 0x40 & 0x44 instead. I haven't tested changing them, but I think you're better setup to try it out

[Wolfgang]

... seems you need to install the pyvisa library.
Then it should work.
Should you miss other libraries, have  a glance at the beginning of my code where the libs are included.
I use PIP. It resolves dependencies automatically.

Much success !

Regards
  Wolfgang DL1DWG

[Wolfgang]

I've done quite a bit of work reverse engineering the firmware on my MSO6034A scope, so I might be able to fill in some blanks.

It definitely does check the model number against a list in the firmware (see attached image).
For older probes, it seems to have an ID number instead of the model string.

I went through the code that does the probe calibration and it seems to get the default offset & gain values from floats at 0x53 and 0x57 in the EEPROM for the newer named probes. Older probes may have them at 0x40 & 0x44 instead. I haven't tested changing them, but I think you're better setup to try it out

Hi Miek,

agrees with my observation that a "database" (you just proved its dumb hardcompiled code) of probe types is kept in the scope.

I also agree that there ARE probes that write back calibration information. An example is my (very complex) N7026 high-sensivity current probe. When you insert this one, I sniffed that not just probe ID and serial number are retrieved, but also a lot of other information at several locations. An I2C scan also reveals several EEPROM locations, not just 0x51. When you calibrate, quite some info is written back to the probe EEPROM.

Older (and simpler) probes like the N2795A or N2796A dont read anything else than serial number and probe type. This was verified by sniffing the bus during a calibration operation.

The expected outcome of my modest probe attempts is to supply a calibratable and detectable interface to probes that Keysight does not supply. At the moment, I have three types:
1) wide voltage range 1:10 FET probes with minimal capacitance (10-200Mz). This works now with AutoProbe, I'll write a webpage in a few
     weeks about this
2) 1.-1.5GHz AC Fet probes. Works, but for the moment without Autoprobe. Could also be run under N2795A.
3) current transformer probes to measure small AC currents in the frequency range from 2MHz to 200MHz (I needed those for my PhD work).
    Candidate for current probe missing.

Something I would also like to have is a differential probe with wide input range. That should not be too difficult, but I lack a similar Keysight model to
start from.

For type 3 I am currently searching for a close enough Keysight probe that can be used to make the hack possible.

As said, Keysight does not want to disclose the interface (I guess to avoid competition for their probes), so some creativity will be needed.
Everybody having similar issues and/or other Keysight probes of interest (I just have N2795A and N7026A). is invited to join efforts by suppling EEPROM images.

Regards
  Wolfgang DL1DWG

[Wolfgang]

Hi,

I just added the N2818A 200MHz 1:10 differential probe to my list.
Still looking for a decent robust diffamp that can be made with uncritical components.

[KE5FX]

Hi,

I just added the N2818A 200MHz 1:10 differential probe to my list.
Still looking for a decent robust diffamp that can be made with uncritical components.

Depending on what you mean by 'wide input range,' the LMH5401-based probe that nctnico sells might be of interest.  It works very well with the 6000 series.

[Cerebus]

As said, Keysight does not want to disclose the interface (I guess to avoid competition for their probes), so some creativity will be needed.
Everybody having similar issues and/or other Keysight probes of interest (I just have N2795A and N7026A). is invited to join efforts by suppling EEPROM images.

Interesting choice of verb, 'disclose' as it's exactly what a patent does. Like this one: US Patent US5939875A - Universal probe interface - Assignee: Agilent, expired

The patent unfortunately doesn't go into the EEPROM format, but the 'example' table in it of resistor encoded probes is in fact 'actual' rather than 'exemplary' as far as I've been able to determine by comparing it with a few real examples. It's not the "golden goose" of all the possible EEPROM encodings, but it's better than a kick in the teeth.

[Wolfgang]

As said, Keysight does not want to disclose the interface (I guess to avoid competition for their probes), so some creativity will be needed.
Everybody having similar issues and/or other Keysight probes of interest (I just have N2795A and N7026A). is invited to join efforts by suppling EEPROM images.

Interesting choice of verb, 'disclose' as it's exactly what a patent does. Like this one: US Patent US5939875A - Universal probe interface - Assignee: Agilent, expired

The patent unfortunately doesn't go into the EEPROM format, but the 'example' table in it of resistor encoded probes is in fact 'actual' rather than 'exemplary' as far as I've been able to determine by comparing it with a few real examples. It's not the "golden goose" of all the possible EEPROM encodings, but it's better than a kick in the teeth.

Hi,

by "disclose" I meant that they do not allow you to define your own probe. Now we now why: They hardcoded the probe IDs into their software, and if they find a smart probe (RId = 46.6K) and its not in their code they reject it. Somebody on EEVBLOG found the code by reverse-engineering a scope and extracted the probe types they support. There is another expired patent explaining the EEPROM content.

Working with the resistor Id is fine (I tried all values published and verified a table of working values and probe types), but leaves you without advanced support. For what I need a few FET probes with modest frequency ranges, but extra low capacitances are OK by now,
and the calibration feature of EEPROM enabled probes is practical so I use it.

Another probe of interest would be a very small current transformer probe (like Tek), but I am still looking for a Keysight equivalent.

[artag]

I have an 1153A probe which I use with a DSO7014A. It's not supposed to work (I think it requires the 6V supply that's not available) but I removed pin 5 so it can't reject it and it seems to work fine.

Happy to make measurements on that if it adds to the sum of knowledge.