HP 4195A network /spectrum analyer

[ppppenguin]

I have owned this instrument from new (c1990) and it has always worked flawlessly. Until now. It's stuck in some kind of boot loop. It says "memory test" on screen for a few seconds, then boots to a normal display and is responsive to controls for about a second or two. Then goes back to the memory test. Rinse and repeat. I don't recall ever seeing the memory test screen before.

I have the original user and maintenance manuals as supplied. They don't contain schematics, not even a decent block diagram. There is a list of modules, their functions and a few photos showing their locations. Unlike many Tektronix manuals, there is no faultfinding guide.

Looking at this video: https://www.youtube.com/watch?v=0VhyDC1DTCk&t=36s  it looks like there's a lithium thionyl chloride battery in there which is probably now dead. The manual mentions EEPROMs so it's not clear if calibration data is stored in battery backed RAM or not.

Any ideas? I'm off to the Broadcast Engineering Museum for a few days so won't be looking a the 4195A immediately.

[Mike Watterson]

If it's too awkward or expensive to fix, there is the amazing TinySA Basic and TinySA Ultra.

I did have an HP141T with four RF plug-ins. It died (a third time), so I got a 1.5 GHz Rigol with Tracking generator (works as signal generator if span= 0). I did recently get a TinySA Basic just to use as radio waterfall spectrum display (via buffer just in front of IF). It's actually pretty good to align and test from 100 kHz to GHz, but best at 1 MHz to 350 MHz. Signal Generator mode, but you need an external 30dB attenuater for radio alignment. Can't use Generator as a Tracking Generator without an external mixer and 433.92 MHz LO (which happens to be what the cheap 434 MHz gadgets use!).
There is a TinySA Groups.io and the designer answers questions not in the Wiki.

[Mike Watterson]

Well, you can try a Lithium coin cell or two Alkaline AA and reset etc. If that fails, then unless you are lucky, it's doomed. But you know that already!

[ppppenguin]

For network analysis there's the Nano VNA: https://www.tindie.com/products/hcxqsgro...-v2-plus4/

I took the top cover off the 415A a moment ago. There's an inner cover held on by about 20 screws before you can get at the PCBs. Another time. First port of call is obviously the battery. On that video it looks like a lithium thionyl chloride but it could be anything. Without schematics I won't know until I've opened it up.

I don't care too much about loss of calibration data. For most of my likely uses it will be better than I need, even uncalibrated.

All I was trying to do today was look at a UN1/163C 60kHz antenna. This BBC design is built in length of plastic drainpipe with end caps. https://www.bbceng.info/ti/eqpt/UN1_163.pdf This is intended for use with a BBC-designed MSF receiver which in turn sends time signals to a GNAT: https://www.bbceng.info/Designs/designs_...trbbc.html I've got all the kit temporarily until I take it to Richard Russell who designed the GNAT.

We have another GNAT at the Broadcast Engineering Museum but not the MSF kit. It will keep time (not very well) on its own xtal and while switched on it uses sync from our main SPG which is now GPS locked. This means it ticks at the right speed but actual time has to be set manually.

[Mike Watterson]

I have a Nano VNA too. Great for aerial design / checking, or making real 1/4 wave stubs.

[Nick]

Here's the service manual

[ppppenguin]

Many thanks.

[ppppenguin]

I opened it up today and found the board with the backup battery. A 2.4V 90mA NiCad and very dead. About 0.2V. There are some rough equivalents available at a price but it's going to get a pair of AAA NiMH cells for now and probably forever. Soldering to ordinary cells isn't very good for them but it's OK if done swiftly.

Calibration will be lost but within reason I don't care.

The service manual that Nick found (and is on the Agilent website) is not the right one for my instrument. Mine has a serial number starting 2904J while the manual is for 2738J. The board I'm working on (CPU and memory) looks totally different to the drawings in the manual. The CPU is a 68000 in a huge DIL package.

[Nick]

Had to replace an unobtainium Renata lithium cell in a bit of Tek kit a while back:


https://photos.app.goo.gl/uHYpdPiZpsfoCksMA

[ppppenguin]

If fitted the the pair of AAA cells, a bit fiddly but no problems. The 4195A now boots OK and seems to work on everything I've tried. I'm sure that calibration is adrift in all sorts of ways but I doubt it will matter for the jobs I use it for. When doing 2 port transmission measurements you always calibrate against your test cables to zero them out. For 1 port return loss measurements you always calibrate against at least an open circuit and a 75R termination.