Agilent PNA E835xA with 85092 ECal, Problems

[joeqsmith]

Posted on groups.io as well but thought I may have better luck on EEVBLOG.


The Agilent 85092-60007 3.5mm, 2-port, 30k-9GHz ECal has a USB port marked "USE WITH PNA SERIES".    My Agilent E8367A PNA has the 9GHz option enabled.  When I connect the ECal to the PNA (via USB),  the OS detects the device.  Looking under Device Manager,  it lists "Agilent USB ECal Module".  Driver Details shows e-TEK Labs version 1.2.0.0, Agecalv1.sys   The ECal's Green LED eventually turns on solid.
 
Selecting the calibration wizard and ecal, it displays "An ECal module was not found.  Please be sure the ECal module is connected properly.  (You may need to disconnect and re-connect the USB cable)"
 
I tried two different known good cables, plugging into the front USB port.  I have a powered hub plugged into the rear that has the mouse and keyboard attached.  I tried connecting the ECal to that as well.  In all cases, the OS finds the devices but the PNA fails to locate it.
 
From the small clues Agilent provided, it seemed there was an A,B & C revision.  A was sold by HP and did not appear to support USB.   Looking at photos for the B & C, they do not appear to be marked.   This particular one has no revision markings as well.   The latest manual I have for the PNA calls for the 85092C.  Looking at an older document, it appears they supported the revision B as well:
http://www.av.it.pt/Medidas/Data/Manuais%20&%20Tutoriais/11-%20PNA%20NetWork%20Analyser/CD1/technical_specifications.pdf

Looking on the PNA's disk, there was a directory that was marked Ecal Drivers.   Inside there were two directories for Floppy#1&2.   I had installed the PNA software  using VirtualBox.  I installed these two disks (HP ECal 01.008).  Looking at the help file, it looks like they were used to setup a PC that would sit between the VNA and ECal to control it (see attached).   This was maybe some work around for customers trying to run old ECal units. 

Any ideas on how to determine the revision of the 85092 ECal?  Are both the C&B revisions are compatible with this PNA?  Any thoughts on what it would take to get it working?

[tautech]

Is a FW update available for your E8367A ?

[joeqsmith]

I upgraded the PNA to the latest possible version when I purchased it.   Outside of adding the license to enable the 9GHz option, alignment, replacing the old drive with an SSD and a few repairs, the system is basically factory stock. 


With VirtualBox, I setup a USB device filter.  Device Monitoring Studio looks like it could trace it if Agilent has some way to control it. 

Running the virtual PNA, it see the new device and the OS installs the driver for it.   I then tried the software from the two ECal floppy disks but there is no activity.   I am not aware of any other software outside of the main PNA program that would talk to it over USB.   

[TheSteve]

Being it says use with PNA it is hard to believe it wouldn't be supported. The green LED indicates the unit has warmed up to operating temperature. It also loads the driver which means it can see it.
When you select the calibration wizard what frequency range do you have selected? The full span of the PNA is probably greater than the ECAL range at the low end. There is a slim chance it won't show it if the frequency range you have selected is wider than the ECAL range.

Worst case you could use the parallel interface if you have the adapter box. I used one years ago with the 8753 series.

[joeqsmith]

The ECal is rated for 30k to 9G.  PNA's range is 300k to 9GHz.  I did try 10M to 100M which had no effect. 

The PNA manuals I have make no mention of using the parallel cable, only the USB.  When selecting the wizard/ecal, it shows its attempting to use the USB port.    I wonder if their software would even control a parallel version. 

[joeqsmith]

I reinstalled the software from the second floppy, rebooted and it now it appears to be talking with the ECal.   I had upgraded the firmware when I bought this system to the lastest they supported for the hardware.   I am not sure if this would have had anything to do with it.  The driver appeared to be the same but maybe something else was messed up from the upgrade. 

First thing I notice was while the ECal shows 30kHz, the PNA throws an error when setting the sweep to 300kHz.   It appears to be limited to 50MHz and it reports the N style connectors rather than the 3.5mm.   I would have expected this to be stored in the ECal.   I just let it run through and attached a load and basically it appears to be basically working. 

I'm sure some a lot of user problems now.   Time to do some reading. 

In case someone else runs into this problem and lacks the original floppy that contained the drivers, I have attached it here.   This works with Windows 2k.  I am not sure if it would work with XP.   
 

[Miek]

First thing I notice was while the ECal shows 30kHz, the PNA throws an error when setting the sweep to 300kHz.   It appears to be limited to 50MHz and it reports the N style connectors rather than the 3.5mm.   I would have expected this to be stored in the ECal.

Going by the legacy ECal module reference guide (https://www.keysight.com/gb/en/assets/9018-03493/reference-guides/9018-03493.pdf) and some old Agilent documents (https://testworld.com/wp-content/uploads/configuration-guide-keysight-agilent-hp-8719et-8720et-8722et-8719es-8720es-8722es-network-analyzers.pdf & http://legacy.libertytest.com/assetmanager/uploaded/pdf-2008219-114334-85032e.pdf), the 85092 should have N connectors. The 85093 is the 3.5mm version. I'd be a little worried that a previous owner has done some mix and matching of parts.

[joeqsmith]

Agree.  And looking at photos, for the most part they appear to match with the following:

85091-60004 Sexless 7mm, Sexless 7mm

85092-60007 Female, A Female N
85092-60008 Male, A Female N
85092-60009 Male, A Male N

85093-60007 Female, A Female 3.5
85093-60008 Male, A Female 3.5
85093-60009 Male, A Male 3.5

I find it odd a company like Agilent would have the exact same part number for a 30k or 300kHz part.  Further investigation, it appears the rev A has no support for USB, rev B is rated for 30kHz to 9GHz, and rev C is rated from 300kHz to 9GHz.   So while they kept the same part number, the can be identified by looking at the frequency range and ports available.

While the connectors appear in very good condition.  I have not yet gaged them but over all the unit doesn't appear to have been abused.  I can run through a full 2 port calibration and insert some known devices and the data is reasonable.  So it appears functional.   

I'd be a little worried that a previous owner has done some mix and matching of parts.

I agree with Miek and suspect an outside house (not Agilent or Keysight) swapped the N connectors for the 3.5mm and then overwrote the factory coefficients.  The may not have had a way to change the connector type or model in the devices memory.   I assume who ever did this did not have the proper equipment which is why the minimum reported frequency is 50MHz instead of the rated 30kHz.   

Still, not hopeless.  If Keysight still services these, I may be able to have them calibrate it for the full range with the 3.5mm connectors.   This would be ideal but I'll have to check with them.   The other option would be to borrow a set of standards and characterize it myself .   

***
I had borrowed a set of Agilent 85033E standards that are in cal.   These are rated for 9GHz.  Using these to create a user setup seems like one option.

Sadly, it doesn't appear the PNA's software supports this.  Looks like Keysight would be the only option and with it being so old, that may be a dead end.   I'll write them. 

***
Add details to identify the hardware revision

[joeqsmith]

I noticed where a few people have attempted to hack these old ECal units:
https://www.eevblog.com/forum/rf-microwave/legacy-hpagilent-ecals-with-open-source-control-to-cal-any*-vna/msg5506081/#msg5506081

Interesting is the above link, they are using it with a 50MHz VNA.   I wonder if someone bought a bunch of damaged ones and attempting to repair them (swapping connectors and alignment) using these hacks to reprogram them.   Not a bad idea if the sellers are clear about that is what is happening and they disclose the details.   I'm surprised they would have ran it on a better VNA to get the full working range and reset the part number so at least the PNA would report everything correctly. 

If you're in the market for an old ECal, learn from my mistakes!    I'll report back once I hear from Keysight.   

[joeqsmith]

It's been over a week now since I contacted Keysight.  While they initially responded quickly, asking for further details and then elevating the case, that was the end of it.  So it appears this unit can not be recovered. 

Let's assume that Agilent/Keysight did not modify this unit, suggesting there are tools available to realign one.  I've done a fair amount of searching and came up with very few clues.   I found one document that mentions undocumented commands for the PNA to control the ECal.  Then we have the group attempting to hack the A version through the parallel port.   I am not sure how the added USB board effects that.  I also didn't see where they had a way to realign the ECal and write the data back into the device.

I also tried to find details on how to program the ECal directly through the USB port but again, found nothing.  I thought I would see what it would take to write my own tools.   The first problem is going to be getting some sort of driver for Windows 10.   Using Labview's driver wizard, when running a test, it complains the ECal is not VISA compatible.

[joeqsmith]

To work around this, we can try and use the RAW mode.   "NI-VISA USB RAW mode supports Interrupt, Bulk and Control data transfers. Isochronous transfers are not supported. " Still using the wizard but selecting other rather than the identified connection, then manually entering the VID/PID.

This will create the INF file, which you can then just copy into the Windows/INF directory.  Windows 10 requires all the drivers to be signed.  They have Inf2Cat & signtool which could be used to but MS provided the advanced boot menu what allows you to disable driver signature enforcement.  After installing our driver and rebooting, the system is back to normal.

[joeqsmith]

Windows reports the new driver and running the Device Monitor, it now reports the ECal.

[joeqsmith]

I then wrote a simple program to try and get the ECal to respond with anything.   This was a total bust.   I am not aware of any tools outside of the PNA that will communicate with the ECal.   I can't seem to run the Agilent software in a virtual box without the actual PNA hardware attached.  Normally, would just install a sniffer on the target but with this thing being so old,  it would mean finding a compatible sniffer that actually works..

Maybe something like this could be used:
https://github.com/ataradov/usb-sniffer

[TheSteve]

Have you tried the secure erase tool running on a PC? I've been able to backup(and I believe restore) the entire writable flash region with that.
The tool is called ECalWipe.

[joeqsmith]

Have you tried the secure erase tool running on a PC? I've been able to backup(and I believe restore) the entire writable flash region with that.
The tool is called ECalWipe.

Nice find.  Installing their drivers under Windows 10, and running the program, it does talk with the device.    I tried the ALL data dump and searched for the end.  Seems there is no user data (all 0xFF), which is what the PNA reported.     

Are you aware if anyone has decoded their dat file format? 

[joeqsmith]

It appears that if you run through a calibration using the mechanical standards, then use one of the User calibrations, you can change the frequency range.   If you set the range outside of the ECal and then attempt to run though the User cal, it will not let you.  So order is important.   It does appear to work correctly so no custom software or hacking is needed.   

Next step is to borrow a set of mechanical standards again.