Tektronix492 Spectrum analyzer repair / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

Tektronix492 Spectrum analyzer repair

Move Thread

LAN_403

2Spoons

Thu May 17 2012, 06:31AM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

I bought an old TEK492 RF spectrum analyzer off Ebay to help diagnose some interference issues with a GPS project. Seller never mentioned that the 30dB section of the input attentuator was stuffed - probably due to overload.
After seeing a replacement attentuator block listed for $400 (!) I decided to open up my faulty one and see if it could be fixed - i'd only paid $1700 for the thing in the first place so it was worth a crack.

The attenuator has three sections of 10dB, 20dB, and 30dB which are mechanically switched in and out to give 6 steps of 10dB. Its a really fancy looking gold plated transmission line type thing - the analyzer goes to 22GHz, so not surprising . The attenuators sections turned out to be tiny thin film resistor networks on little glass squares, about half the size of a fingernail. Undaunted I pulled out the offending 30dB element and put it under a microscope. The network turned out to consist of half a dozen 'T'sections in series, all laser trimmed. The damage was obvious, and luckily limited to just the first 'resistor'. I decided that bridging the damage with conductive paint would be a reasonable option, as it would only affect the first section out of 6 - a reasonable price to pay for having a fully working attentuator. Applied a dot of silver conductive paint with one hair from a brush, let it dry, and put it all back together.

Quick check using the internal calibration signal and ... exactly 10dB steps all the way! I guess the paint came out to just the right resistance.

foreign1

Wed Sept 05 2012, 02:34AM

Registered Member #940 Joined: Sat Aug 04 2007, 03:43PM
Location:
Posts: 7

That's a right good repair! I have a TEK492, myself, and am seeing some rather disappointing attenuation on the -20dBm CAL signal when I try to calibrate the thing. Looks like I've lost about 30-35dB, and I'm starting to think that it might be the result of a faulty 1st mixer diode. The attenuation does seem to scale properly in 10dB steps, but nowhere does it show that the signal power of the 100MHz signal is where it should be. I've read of people replacing the front-end mixer with manually welded sub-millimeter-sized (!) diodes but I haven't quite worked up the courage yet to try that. Has anyone else here fiddled with this section of the Tek 492?

2Spoons

Wed Sept 05 2012, 03:50AM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

I would have thought that if the mixer diode is dead you would get nothing. Can you see local radio stations if you hook it to an antenna?
Another problem I had with mine: no signals showing, super low noise floor - turned out to be a broken (not blown) fuse. Get a copy of the sevice manual if you haven't already - there is a guy on ebay sells CD copies.

foreign1

Fri Sept 07 2012, 12:49AM

Registered Member #940 Joined: Sat Aug 04 2007, 03:43PM
Location:
Posts: 7

I've read that attenuation can come from either a problematic attenuator (as in your case), or from a troubled mixer diode. I do get signal power from the calibration signal, so I know that it's at least partly working. The attenuation on the signal seems so large, though, that making meaningful measurements seems out of the question.

I do have a copy of the service and operator's manual (shipped with the SA), but I just wanted an idea of what I should start looking at, since I've read that these SAs often develop problems in the front end. Plus, I lack pretty much all of the test equipment used in the service manual for actually diagnosing the issue, so testing things with a power meter, or calibrated source, VNA, etc, is sadly out of the question at the moment. For the fuse, are you referring to one located in the PSU?

Conundrum

Sat Sept 08 2012, 01:15PM

Registered Member #96 Joined: Thu Feb 09 2006, 05:37PM
Location: CI, Earth
Posts: 4061

I have a couple of used tek tunnel's here of possible functionality, PM me if you want them.

Rather than paying me for them, make a donation to 4HV instead, k?
EDIT:- three if you count the one with rusty leads

Usually they work, its very rare to have a failed one unless it has been overloaded.

I also have a few hybrids, etc but mostly pulls and/or marked "BAD" so caveat emptor.
Maybe OK if you have a totally bad chip just to see if it is failed before spending $$$ on a new
old stock one.

-A

2Spoons

Sun Sept 09 2012, 09:54PM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

Yes is was a PSU fuse. The wire in the glass fuses can be a bit fragile and can break with vibration. Certainly that's the first and easiest thing to check: are all the power rails at the right voltages?

foreign1

Wed Sept 12 2012, 05:23AM

Registered Member #940 Joined: Sat Aug 04 2007, 03:43PM
Location:
Posts: 7

I did take a look inside this beast a little while ago, didn't find anything out of the ordinary at the time doing a cursory check of some lines with an o-scope and VOM. I'll have to open it up again and give it another go, a bit more thorough this time. This machine is simply too precious to neglect! I hope I can get it working up to the level it ought to be at ~crosses fingers~

Conundrum

Wed Sept 12 2012, 07:57AM

Registered Member #96 Joined: Thu Feb 09 2006, 05:37PM
Location: CI, Earth
Posts: 4061

Had that happen on a Goauld OS1100, thought I'd fried it. Turns out my one nice probe had a snapped wire inside, as when I wiggled it the noise turned into a nice clean sine wave.

FWIW some probes have a *10/*1 switch which also goes bad, a spray of contact cleaner and rapid switch manipulation followed by ye olde oven trick usually fixes those.

I really must get around to making a ghetto spectrum analyser using a row of TV tuners in parallel driven from a voltage stepper and precision driver.
This is a very effective way to probulate signals well into the low GHz range which is fine for amateur radio use and even for "fixing" stubborn wireless networking problems caused by CheapCCFLs (tm) and their close cousin twice removed the CheapSMPS (tm) which lacks the voodoo known as output T networks.

-A
#include "I *HATE* bad caps.h"

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob