Advantech R3162 frequency drift & RF test fail

[fsedano]

Hi,

I bought a R3162 (8GHz SA) some months ago and it has been working just fine. However, yesterday it failed, signals show OK in amplitude but with some side bands and 3.7 Mhz offset.

Test fails at 'RF'.

I can't find any service manual for this one, would somebody have any hint? Disassembly pics next...

[fsedano]

Disassembly pics follow - Interestingly I found no CF or something similar on CPU board - Where is FW stored? Backup battery shows full power.

[Fraser]

The R3162 is very similar to the R3132 that I have repaired. I produced a block diagram for it.


The second Local Oscillator is well known for its drifting with age.

Take a look at this R3131 thread as it contains information from me that may prove helpful.

https://www.eevblog.com/forum/testgear/advantest-r3131a-spectrum-analyzer-9khz-3ghz/

Fraser

[fsedano]

Thanks Fraser, indeed great info on that thread, thanks a lot.

I had a look at my 2LO osc by following your document. I didn't remove the blue cable from the feeder before applying 11.3v (which was not a great idea).

Oscillator was almost right, 4MHz off, but very unstable. I adjusted it and then I noticed output power (which was around -10dBm before) was very low (-40dBm).

Reassembled the unit and now it's completely deaf. :-(

I had a look at the bias voltage to ERA-1 and ERA-5 and they're completely off - 3.8volt for ERA-1 (max should be 3.6) and 2.8 for ERA-5 (range is 4.5 to 5.3).

Looking at the actual bias resistors present on the circuit, ERA-1 has 68 Ohm and ERA-5 29 Ohm. That looks to be very low for Vcc=11.3v which I applied

Is the actual Vcc when powered on 11.3? Maybe on 3162 it's supposed to be lower?

It looks at least ERA-5 is toasted, but I think it was fine before my messing around. I hope I didn't break more things by applying power with the blue cable connected to the passthru

Thanks,

[Fraser]

You need to read the whole thread. There appear to be two versions of RF board fitted in the Advantest 3132. not sure about the 3162. One has an additional dropper resistor in line with the supply to the MAR MMIC's. The supply is 11.3V on that version. Those were the RF modules that I worked on. Another forum member discovered a different version of the RF module that did not contain the dropper resistor and that used a lower voltage on the MMIC's. No idea why Advantest made this change but I think my modules were all of the later generation with the dropper resistor.

The good news is that you will not have harmed anything in the PLL drive circuit by leaving the blue wire connected and I doubt you fried the MMIC's either. They are pretty robust little devices. If one has died, the good news is that they are standard Mini-Circuits devices and not select on test or part of a fixed calibration in the Spectrum Analyser.

With regard to the signal level drop...... I have seen that myself when tuning the 2nd LO. It can sometimes be as a result of the Dielectric tuning arm position. I adjusted my 2nd LO for the closest frequency to optimum possible whilst maintaining the nice high level of the signal. I suggest you revisit the adjustment and hopefully when you move the adjustment arm the signal level will return to normal. If not, suspect the buffer MMIC that is on the output of the 2nd LO.

For detail of the lower voltage used to power the MMIC's in the older RF modules, please read the thread as the other forum member measured it.

It is HIGHLY unlikely that you have done any harm to anything but an MMIC in that circuit or anywhere else in the Analyser.

I will go back to that old thread and place a warning on my post that two module versions exist.

Fraser

[fsedano]

Thanks Fraser (and even more thanks because you're replying at 2AM

I gave the adjustment another shot, using now Vcc=7.5v but I can't get more than -40dBm. With the resistors in the circuit (29ohm, recommended value for 7v), bias of ERA-5 is only 1.93v, way below minimum for it(4.5v). So strongly suspect of it being fried.

I ordered a new one, but minimum is 20 units, so it seems I'll have a couple spares. If anybody is on need of one I'll be happy to send it over 

Will report back once the replacement amplifier arrives (it should take a week or so)

Thanks!

[Fraser]

Sorry to hear that the MMIC got fried . I feel responsible.

I would have sent you a replacement MMIC for free as I do have spares.

My apologies for not warning of the different versions of RF module.

Back to your original fault ...... The original 4 MHz 2nd LO error is within the capabilities of the PLL to cope with. That would suggest a problem elsewhere.

Best Wishes

Fraser

[fsedano]

Thanks Fraser, of course you aren't responsible, you're so kind on offering help!. I'll continue troubleshooting once I receive the replacement.

I was checking to see if there's an easy access to that passthru with the cover closed, to check what's the expected voltage there. However, there's no easy access. I could probably place the module with no cover on the chassis and check voltage there. The oscillator will be all over the place, but at least I'll check what's the expected VBias there.

Probably next step would be to check PLL is actually locked, once I receive the MMIC.

Thanks!

[Fraser]

I realised that I had assumed something from one of your posts that may be incorrect.

When you stated that the 2nd LO was off frequency by 4MHz, I assumed you meant with the PL not working and with the module out of the chassis. If this was the case the PLL should pull the oscillator onto the correct frequency.

BUT

If the 4MHz frequency error was measured whilst the module was in the chassis and so the PLL is operating, then that is a classic sign of the 2nd LO needing adjustment as the PLL lock range has been exceeded. Monitoring the PLL control line enables you to see how hard the PLL is working to pull the 2nd LO onto its correct frequency. My document details this I think.

In your case I believe adjustment of the 2nd LO with the PLL control line tied to 0V would be the best option. This is the centre of the PLL range. You will need to fabricate a temporary cover for the VCO with aluminium foil or thin aluminium sheet. Then adjust the 2nd LO and check the frequency with the temporary cover fitted. Repeat until you obtain the normal 3840MHz operating frequency (or close to it) of the 2nd LO. Refit the module covers and insert into the chassis. Check the operating frequency of the 2nd LO to ensure it is on frequency. Now check the PLL control line to see how far away from 0v it has been shifted. If a significant voltage is seen, you should consider repeating the adjustment process.

I suggest this process because I has a fully working later version 2nd LO that I used to take measurements in order to align the off frequency modules. I just checked what frequency it oscillated at when out of the chassis, with covers removed and with the control lighting sitting at its floating bias. As your module is the earlier version it is best to align it with a fixed 0V bias on the varicap diodes. It does take more effort but it is more scientific.

Fraser

[fsedano]

Hello folks,

Some updates on this - Just received my replacement MMICs and yes ERA5 was toasted, due to applying 11v for testing. Replaced it and I'm back at same situation before I started messing around ;-)

I adjusted the LO freq applying 6v, placed it inside the chassis and measured frequency and varicap voltage on normal operation (inside chassis, cover installed).

To my surprise, freq I saw was about 3.834Mhz, instead to 3.840Mhz, and drifts down.  Varicap voltage is around 0.7v, and moves sightly when device warms up towards 0.6v.

Activating autotest brings the varicap voltage to 12v briefly and then goes back to 0.7.

With this result, the theory I have now is the negative rail for the VCO control is gone (or whatever is used to provide <0v to VCO), so PLL can't lock. Next test is sightly manually tunning VCO so PLL can lock with positive voltage. If I manage to get lock, it would confirm this.

Suggestions welcome ;-)

Thanks folks.

[fsedano]

This is with the module inserted on the chassis, V Varicap 0.6v

[fsedano]

Hi,

That did the trick, getting perfect lock now, test passes, calibration is also fine!

Now, wondering if this module is expected to have only positive VCO control? I doubt with a dead rail I'd get perfect measurements...