Advantest R3265A Spectrum Analyzer Repair Help : Input Attenuator

[ARF]

Hi guys,
I bought this spectrum analyzer in the "does not turn on" condition, after receiving it turned out that it turns on and even shows a picture, it was just that the brightness control was not at maximum.
The image on the screen was distorted (CRT board problem). I just checked that it reacts on the signal at the input and turned it off. I was afraid of leakage of capacitors inside. Before buying, I looked at the repair reports for such a device and I was sure that the power supply was broken there due to leaked capacitors.

However, my device was assembled in 1996 and almost entirely on Nichicon capacitors. They turned out to be in good condition, the small capacitors on the boards turned out to be worn out (not all). Elna capacitors were used in the CRT unit and they completely degraded. Because of this, there was a distortion of the image.
I changed all the capacitors (except the input block) and assembled the device to test it.

I connected a signal generator and tuned it to 25MHz / +10dBm. On the screen, I saw the level of -17dBm.

-10dBm level of "CAL OUT" displayed as -32dBm

I tried to calibrate the device. I have connected CAL OUT (on the front panel) with the input.
If I select "CAL ALL", then the error "Calibration error of INPUT ATT" was displayed;
If I select "TOTAL GAIN", then the error "Calibration signal not detected" was displayed;

I don’t think I was mistaken in the assembly, but anything is possible.
Maybe I'm doing something wrong?

[ARF]

The problem is dependent on RBW.
With RBW 1 MHz, the level is measured correctly.
There is a slight drop at 300KHz.
At 100KHz, there is a strong drop in the signal swing and it is displayed incorrectly on the screen.
I found a place in the amplifier, after which the signal swing falls.

I don't quite understand how this unit works, especially the part with the L14 coil.

[memogara]

Dear ARF,
Have you fixed this problem? I bought an R3265A and it has similar problems like yours. The unit works with RBW>= 300kHz but the signal drops significantly when I decrease RBW. When I do the calibration, I get the same message like you. I'd appreciate if you have an update.
Thanks

[ARF]

Yes, I fixed it. As the RBW decreases, the band of the filter in the picture above becomes narrower and narrower. It turned out that the resonance was shifted away from the required 21.42 MHz.
After a detailed inspection, it turned out that the adjusting screw of the L14 coil was broken off - someone tried to adjust the coil and the plastic screw broke off. There are several similar coils in the device, avoid adjusting them, the plastic screws have become very fragile. If you still have to adjust, first carefully dissolve the fixing glue on top of the coil.

In my case, it turned out that the L13 coil was broken and because of this, the resonant frequency of this block shifted. Someone tried to configure it using the L14 adjustment and it is possible, but not correct. Start by checking L13 and replacing it if necessary. It is important to comply with the DCR for L13. On this board in another place there is the same coil and it seems that the DCR was 20 ohms, at first I bought a coil where the DCR was 7 ohms - it is impossible to perform auto-calibration with it.

To check, you need to connect a network analyzer to this board, you can find this in the instructions in the calibration section. You need to get the result below:
(before and fixed - the center must be 21.42MHz)

[memogara]

Thank you very much. I managed to download R3265A service manual vol 1 to 3. I found the schematic you posted (vol 2, p127). As per manual, L13 is 330uH and L14 is custom in the Log amp section. However, I couldn't find the VNA connection and calibration section. Could you be more specific? By the way I haven't opened the unit yet. I wonder how complicated it's to locate the Logamp and disconnect it.

[ARF]

This is a very complex device with a huge number of components. However, all boards can be opened like a book for ease of repair and maintenance. You need the second top board under the processor. The connection is described in the calibration section, look carefully.

[memogara]

I found it on page 218, section 5.3.2. It says it can be done by adjusting L14 and R239 (only if something else is not broken like in your case).
I don't have R3361B but I guess I can connect a VNA similarly (port 1 to J8 and port 2 to J9). Span shouldn't matter as you set yours to 42MHz, not like the procedure calling for 5MHz. I don't know how accurate my old VNA will be providing -10dBm. But I don't understand how R3265A picks up this VNA signal. VNA will be sweeping at that point. How do I catch the signal on the SA as described in steps 5 to 11? And finally, I will only carry out the adjustment on this p218 right, not do the MAG AMP adjustment? Many thanks.

[ARF]

I did not understand your question. I used VNA and you can use any VNA. You connect the input/output to the LOG-MAG block via adapters, configure the RBW and see where the filter resonance is. You need to understand that; there is some reason, most likely a breakdown or degradation of a component.

If the resonance is shifted, then you need to first check all the components around it, and only then try to adjust L14. Because I was able to get a positive result without the L13 coil using the L14 adjustment - the device passed self-calibration. However, there is a reason that this L13 is installed there.

[memogara]

Sorry for confusion. I understand the filter adjustment using the VNA. But I was trying to establish how R3265A is being set at the same time. Please have a look at the attachment. It  asks for the same tuning whilst doing some key presses to store values into the SA.

[ARF]

You should turn-on R3265 and use its keyboard)

[memogara]

Thank you. I managed to set it all up tonight. It looks like L14 is already set and glued. It seems to be centred at 21.35MHz. I didn't want to disturb L14 with all the silicon glue so I think it should be fine. when I set the sweep time to 2s on the VNA, I get the SA screen on photo3 (Span 50KHz, RBW 300Hz). So far so good. I can increase span and get a slower scan (photo 4). When I now reduce RBW to 100Hz, it all disappears. In order to get a fast scan and see better resolution, I need to reduce both span and RBW but whilst doing so, at some point I lose the signal. I am not sure what combination I should keep them at.
One thing you mentioned: Filter gets narrower as you reduce RBW. Mine changes from green to orange on photo 2 when RBW goes down to some KHz and seems to stay constant on VNA even if I reduce it 10Hz. I donit know why it doesn't change as you said.
It needs a huge understanding of how it works in order to fix these devices.

[ARF]

There are 2 options narrower and wider (the instructions contain explanations of how this block works). The center of the yellow graph should be at 21.42, but you have it at 21.35, I think this is the reason for the problem. I see that someone has tuned L14...the glue looks damaged.

[memogara]

I managed to tune the filter to the exact centre of 21.42MHz. But the problem didn't go away.
When I reduce RBW from 300KHz down to 100KHz, the signal drops about 40dB and it sits hardly on screen (like a tiny kneedle).

[ARF]

I downloaded the documentation and remembered a little how it was...
You should have the following set of documentation:

Advantest_R3265A_R3271A_Spectrum_Analyzer_Operator_Manual.pdf
Advantest_R3265A_R3271A_Spectrum_Analyzer_Vol_I_Service_Manual.pdf - theory of operation with block diagrams (they are different than in Vol II/III) and calibration procedures
Advantest_R3265A_R3271A_Spectrum_Analyzer_Vol_II_Service_Manual.pdf - electrical circuits and useful block diagrams part 1
Advantest_R3265A_R3271A_Spectrum_Analyzer_Vol_III_Service_Manual.pdf - electrical circuits and useful block diagrams part 2

Your next stop is the IF block. This is a block of filters that switches depending on the video filter settings. This unit also has fairly clear instructions for diagnostics and calibration. Additionally, you will need a high-impedance probe for your network analyzer, a signal generator and a mixer.

You can start by simply confirming (with scope) that the signal level drops after a certain stage of the IF block / at the output of the IF block.

So...
Depending on the RBW the signal passes through different filters on the IF board.
- for RBW 300k-3M the path is almost straight, without filters.
- for RBW  100k and below, filters appear on the path and additionally the intermediate frequency is reduced to 3.58 MHz, then at the output it rises again to 21.42 MHz.

I would start by applying 25 MHz from the built-in calibrator to the input of the spectrum analyzer, then disconnect the output cable from the IF and switching the RBW between 100K-300K and confirm that the level decreases STRONGLY at 100K and there should be no drop in level at the input signal.

And check 25M clock on the IF block.