Advantest R3271 spectrum analyzer repair story

[widlokm]

Hello,

Recently I almost finished working on Advantest R3271A 100Hz-26GHz spectrum analyzer. I've got this one cheaply as nonoperational / unknown condition. This is rather old instrument from Japan, with rather nice parameters - especially spectrum coverage. Also all service manuals with schematics are available online.
The unit was dead, when powering it on I could see a led blink on a front panel, but nothing else. It also turned out that someone already worked on my unit, many screws were missing, front panel was not fitted right to the frame and so on. I of course check the power supply - no voltage at any of the outputs.

To get to the power supply almost everything needs to be disassembled (I wanted to this anyway so no a big deal). Power supply itself was not made by Advantest and it was incredibly complicated (2 boards, 8 transformers/inductors, tenths of capacitors). Potential problem could be seen at a first glance - corrosion near capacitors. ABSOLUTELY EVERY electrolytic capacitor in the PS leaked. 

Replacing electrolytic caps fixed the PS and the unit started up to its normal screen, but the displayed spectrum was not right - off by couple of MHz and many additional tones near the main were displayed. The problem was 3.81GHz PLL for the second mixer, that is used in lowest band only. One SMD ceramic capacitor in phase detector become a resistor (or residue under it) and it was dragging output to VCO down. Replacing it fixed the unit. I also fix broken screen cover, and assembly everything right.

Now there is one more thing - selecting frequency span larger then 4GHz does not work - unit shows wrong frequency, and wrong amplitude of the peak. I already know that the circuit driving main YTO is faulty. Advantest generates the frequency sweep by analog circuits, and >4GHz range is the largest. I'm guessing that one of the analog switches in the line is faulty (possibly has a partial short to GND), and when activated in the >4GHz band drags the sweep signal down a bit. That signal is then input to main YTO (unit 4-8GHz LO) driver and we end up with the wrong frequency. I am waiting for the parts now.

Last thing is burned out screen, it is still visible, but poorly. I have a 640x480 LCD panel with the right dimensions, but it is a bare panel. I don't know if creating a driver for it (using STM32F4 with integrated display driver) is a good idea.... It would require taking pixel data from the Advantest at a component level - there is only primitive composite output provided.

I hope that someone will find this description and pictures useful,
Michael Widlok

 

[picburner]

Some time ago I purchased a R3271 in conditions similar to yours, perhaps worse because it lacks the top part of the power supply, so I had to
completely remake a new power supply struggling to make it fit into the small space.

I replaced the CRT with the following LCD-Controller set: LB104SN02-TD01 and M.NT68676, this is the only controller that works with the strange synchronisms generated by the SA.
I didn't want to go crazy creating an 'ad hoc' solution with an FPGA.

[widlokm]

Recreating power supply for this analyzer - GOOD JOB!!!
I think I would go with a 2-module solution, small board with local DC-DC converters to put inside analyzer and separate part doing mains to +/-20V DC or something....

Anyway did You used composite signal generated by the R3271 for display or something else?
Internally it has 4 graphic "planes" with different pixel memories. Outputs of that memories are summed in the end by logic gates, theoretically we could make a 4 color output from it :-). 

[picburner]

Yes, making a new power supply was not really easy, it took me a very long time.
Someone else, if you look in the "Yahoo Groups Advantest Instruments", has done a job similar to mine.

As for the CRT-LCD conversion I tried to take the signals of the various planes to get more colors but I didn't get the desired result.
In the end I used the same signal for the CRT and through a buffer and some resistors I sent it to the RGB inputs of the LCD controller.
The screen is now black and white but by varying the signal levels at the RGB inputs you can get any color.
As I said before not all LCD controllers accept the synchronisms generated by the analyzer: on three cards tested, only one works.

[widlokm]

Many thanks for the info.
With all that problems You mentioned, I think that I will left it with crt for now, I was able to make the screen usable by tuning brightness control on CRT board and by slightly turning the deflection coils (and the tube in opposite direction). Anyway I am working now on a project that will use one of the lcd panels I have, so maybe after having fully working display with my own processor I will came back to the analyzer.

By the way I am still working on the >4GHz frequency span... If that turns out to be some strange passive element problem in a even more strange place I will get angry.

PS. >4GHz span is fixed - it turns out to be an adjustment problem. After checking many DACs, switches and op-amps, I started to analyze currents (easy to measure on scope) in YTO coil and that gave me an idea that adjustment is completely messed up on larger spans. Then I found an option in calibration menu that is not mentioned in service manual... It seems that after earlier service someone adjusted this analyzer using procedure presented in manual and that missed "X10 offset" option in YTO tune. Funny thing is that when performing "SPAN" adjustment later you will not be able to see the problem, so the analyzer was working with wrong large spans since then.

[AlleIZ4DJL]

Dear All,
I have the same unit, R3271A with tracking, working quite nice.
The only issue I see is that amplitude reading is not as stable as it should.
For example, if I connect a high quality attenuator, with proper cabling (+18Ghz qualified),
The amplitude reading is near to be correct but it change continuously +-1dB it is so much as resoution is 0.01dB.
Can anyone give me a suggestion? Im supposing PS is not stable (capacitors leakyng?) but Im quite scared to unassemble the unit without a schematic.
I have service manual but no schematic in it; does anyone can share the schematic?
Thanks for any suggestion. Best regards. Alessandro.

[picburner]

If it is the power supply schematics you are looking for, it isn't in the service manual and it is very difficult to find on the net because it is an OEM product.
You can however check, before disassembling the entire SA and the power supply, if the various voltages have an excessive ripple which can cause the defect found.

[AlleIZ4DJL]

Dear picburner,
thank you for the feedback, I will proceed  in the suggested way.

I'm always disappointed that Advantest does'nt have schematics into service manual like Hp or Tek or others do.

Anyway I'll keep topic informed if there is something to share for repair.

up today im still using analyzer as-is to complete a project; as soon as I will finish project then I will disassemble the analyzer.

As per my experience it is a long way to complete....   

[Tantratron]

Some time ago I purchased a R3271 in conditions similar to yours, perhaps worse because it lacks the top part of the power supply, so I had to
completely remake a new power supply struggling to make it fit into the small space.

I replaced the CRT with the following LCD-Controller set: LB104SN02-TD01 and M.NT68676, this is the only controller that works with the strange synchronisms generated by the SA.
I didn't want to go crazy creating an 'ad hoc' solution with an FPGA.

Do you know if your LCD kit replacement of CRT would work on an Advantest R3261 or R3361 spectrum analyzer serie ?
Thank you, Albert

[picburner]

I am not familiar with the R3261/R3361 SA so I cannot guarantee that it will work.
Correct operation would only be guaranteed if they used the ME-8E1J-TR display assembly.

[nctnico]

IMHO it looks similar enough to try. I used to own an R3261A myself and did a TFT conversion for it (using a CPLD). But it does have a weird screen resolution. Something like 1000x400 ish.

[Tantratron]

IMHO it looks similar enough to try. I used to own an R3261A myself and did a TFT conversion for it (using a CPLD). But it does have a weird screen resolution. Something like 1000x400 ish.

Would you have a dedicated or specific thread on what you did (R3261x or R3361x conversion to LCD-TFT display) ?

[Oldradioguy2]

Nice writeup! Thanks! I bought an R3271A a couple years ago - mine had a vertical hold issue. I'm an old guy and I had a TV repair business back in the late 60's -early 70's so this was an easy fix. For $550 I had a nice spectrum analyzer that passed all self-tests and was spot-on WRT accuracy. For example, a Kustom K-band radar gun with the frequency stated on a boilerplate on the gun agreed exactly with what my "new" R3271A showed on the screen. Pictures of my disassembly and repair work can be found at my w3afc.com web site.

A couple days ago there was a smell like a blown electrolytic, and the unit shut off. It will actually start up but then shut down during boot. It's rainy today and tomorrow so guess what's going on the bench today? Your write up will help.

I also bought a U3641 with the TG option for $650. It was tested and verified fully functional before I bought it. It's 7KHz to 3GHz range is fine for HAM radio, it has a color display, and a button that lets the user freeze what's on the screen yet you can still move the marker around, check frequencies, peaks, etc. And, the 3641 can run on a 12V "car cord". So I use it the most.

Again, thanks for the write-up.

John, w3afc