HP3589A spectrum/network analyzer

[joeqsmith]

I've spent the last week porting over old software for my test equipment to run under Win10 by talking directly with my National Instruments GPIB-ENET controllers.   The last one is the HP3589A which is basically going to require starting over as I want something more generic that supports both the SA and well as the NA features.     The software I had written for it was to make phase noise measurements using the swept SA mode only.   

HP offered both 50 & 75 ohm test sets (35689A & B).   The 3589A uses RS232 to communicate with the test set.   The manuals offer some clues about the protocol.   It uses 9600,8,N,1 and they list a few of the commands. 

1B 2C 49 after power up.
1B 2C 72 4F Forward Reference
1B 2C 72 3F Reverse Reference
1B 2C 72 46 Forward Reflection
1B 2C 72 4A Reverse Reflection
1B 2C 72 36 Reverse Transmission
1B 2C 72 4C Spectrum Port

Sending a ROUTe commands over GPIB will cause the 3589A to attempt to determine the state of the test set.  I was never able to locate where they defined the returning packet from the test set to the 3589A.   

It looks like the last time I tried to sort it out was in 2011.   At that time, I wasn't able to find anyone who had a test set that could sniff the bus.   I had written some software to semi-automate hacking it but was unsuccessful. 

If you have one could you please take the time to sniff the bus.  I would really like to make up a simple test set for it so I can automate some of the measurements. 

My other thought with this particular project, because it's basically starting over and with LabView now being offered for free,  is there any interest in seeing some of the basics of how it works?   


GPIB-ENET
https://www.eevblog.com/forum/testgear/anyone-else-using-the-national-instruments-gpib-enet/msg1016109/#msg1016109

HP3589A test set 
https://www.eevblog.com/forum/testgear/vna-nuts-show-your-vna/msg585482/#msg585482

Labview Community
https://www.eevblog.com/forum/testgear/labview-home-is-no-more-labview-community-edition-is-now-free!/msg3141656/#msg3141656
https://www.eevblog.com/forum/chat/community-version-of-labview-coming/msg2824546/#msg2824546
https://www.eevblog.com/forum/testgear/advantagesdisadvantages-ni-labview-etc/msg2348688/#msg2348688

[joeqsmith]

I spent a couple of hours searching to see if I could find the protocol for the test set but no luck.   

The plan is to make all of the network analyzers have a similar UI and common features which will greatly speed up the development.  LabView was designed for this and makes quick work of it.   

Shown the 3589A measuring a crystal. Also shown is looking at a small extender I had made.   

[joeqsmith]

So far, I am using the 3589A's built-in Short Open cal along with my PCB standards.  At 150MHz, these will be fine.  The unit displays COR suggesting corrections are valid.

Showing S11 of the 50 ohm PCB standard after cal, sweeping 1-150MHz.  I am using a Mini-Circuits ZFDC-20-3 directional coupler.   The original test set may yield better results.

https://www.minicircuits.com/pdfs/ZFDC-20-3.pdf

[joeqsmith]

Looking at the SWR of a 100 ohm resistor.   

The manuals I have don't provide a lot of detail about the internal calibration but after seeing how poor if behaves, I plan to add my own to the software.   

[joeqsmith]

Same loads using software calibration.   

[joeqsmith]

Shown running swept mode, looking at a simple crystal oscillator with peak readout active.  Also looking at the fundamental at a few different RBW settings. 

Good enough for now. 

[joeqsmith]

Shown measuring the 12MHz 10pole filter with roughly the same setup as the NanoVNA in this post:

https://www.eevblog.com/forum/rf-microwave/fun-with-crystal-filters/msg3057094/#msg3057094

Unlike the Nano, the 3589A only has a single receiver and requires an external directional coupler to measure S11.   The other major problem with the 3589A is that it uses GPIB.  In the network modes, it sends up 401 real and imaginary pairs, each IEEE 32-bit format.  Sending that 3.2KBytes takes roughly 3.7seconds.  Overall, its on par with the Nano which would require 4 segments at about a second each.   Both painfully slow.