Advantest R3131A Spectrum Analyzer 9KHz-3GHz

[Fraser]

Having slept on this I do not want to walk away from you on this problem. It is very hard doing this sort of diagnostics remotely though.

I have decided to continue offering advice, at least until I run out of ideas.

I have studied the Type 1 and type 2 pictures and believe that there is a significant difference that we must confirm before proceeding.

From what I can see, the Type 2 module uses 390 ,Ohm voltage dropper resistor in series with the 11.3V supply to the VCO. This is supported by the fact that only 34 Ohms is present in the MMIC bias supply lines. 34 Ohms is way too low for 11V operation. I would expect at least 300 Ohms.

The Type 1 module has no visible 390 Ohm resistor. I originally thought it was under the PCB but that is an assumption on my part.

If we assume for a minute that the Type 1 module uses a lower supply voltage, that would explain the excess current draw at 11.3V. The question is, what should its supply voltage bevwhen bench testing the VCO ?

My units had a 200mA current draw so you could set the current  limit on your PSU for 200mA, set the voltage to 3V, connect to the VCO power rail, and slowly increase the voltage. The voltage at which the current limiter kicks in will be very close to the normal operating voltage. Very rough but if it is near 5V then that will very likely be the supply voltage used.

As a second and important confirmation of this. Connect a long wire to the power supply pin of the VCO where you have been connecting your PSU. Reinsert the module into the chassis without its lid. Switch on the analyzer and measure the voltage between the wire and the shell of the module. That will provide the operating voltage that we must use.

OK, with regard to your latest weird voltage readings. This is unlikely to be Varicap failure. Look at my schematic for a type 1 module. The mid point of the varicap is the supply rail. One outer point is connected to the PLL control line directly but the other is connected to the PLL control line via those 3 components that loop around the pair of varicaps. There could be an intermittent connection at work here.

Before we make any assumptions we need to establish some facts and be logical in our approach. The correct supply voltage for tests is crucial so let's sort that out first. More tests to follow.

Fraser

[jc81382]

Fraser,

Just when you think you are done... I drag you back in... (haha, I really do appreciate it thought)

Its funny you mention connecting a "long power wire" to the supply pin and leaving the cover off, because that is exactly what I did last night. The voltage  was 6.13 V being supplied to the module while inserted and powered on.

You are also correct that at 6.13 V is draws ~ .2 A on the PS, its actually a little higher than that but it is in the 2's. I do have a picture I took from a few days back and at 5.1 V if was drawing 2.14 so it is not too much more than that, I can get an exact answer tonight.

Do you want me to re-attempt the adjustment procedure at 6.13 v?  I'm pretty sure I did and was still unable to get 3805, but I am happy to try again.

Is it possible that the dielectric sweeper arm's characteristics can change over time? I'm only asking because I noticed that I tend to get my highest freq output when it is removed completely. However if all it is, is blank piece of PCB board I can try to make a replacement as I have access to a wood cutting CNC that may cut out a replacement piece. (just a thought)

[Fraser]

Hiya,

I hate to admit defeat if a little more effort could bring success !

The dielectric trimmer including the PCB on which the VCO is mounted can change its characteristics over time but we have to work with it as replacement is not an option.

Remember the picture I posted of a type 1 module that was without its dielectric trimmer ? Your experience may be the reason for such an extreme measure. Let us not go down that route yet though.

OK, this where we are at......

1. We now know that the VCO in a type q module runs at 6V and not 11.3 V
2. The VCO supply also applies the Varicap diodes so they will have only 6V on them rather than 11V which is significant as it effects the control voltage range
3. We do not know the control voltage range of the PLL as it will be different to the Type 2 module.
4. The Varicap diodes would normally have around 20 to 30V maximum p.d. for a decent tuning range if all of that range is to be available. The type 2 module has a maximum p.d. Across the diodes or around 24V (-12V to +11.3V)
5. If we for a moment assume the same 24V maximum p.d. Across the the varicaps used in the type 1 module, the PLL control voltage range will be -18V to +6V with a mid point of -6V.

That gives us the following

-18V to +6V provides a 24V p.d across the Varicap diodes
+6V to +6V provides a p.d. Of zero across the Varicap diodes.

The PLL adjusts the control voltage around a mid point in its range which would be -6V as opposed to the 0V that I used on the type 2 module. This is different because of the lower supply voltage on the Varicap diodes. Everything has shifted down by 6V.

Hope you are following me on this.

6. It is possible that the type 1 module has a much smaller VCO tuning range and only uses part of the Varicap diode available capacitance range. That would potentially mean that a type 1 module has a 6V supply to the varicaps and its control voltage is symmetrical around 0V. Such would mean that the control voltage has a range from -6V to +6V with a mid point of 0V as with the type 2 module. This would mean the VCO control range is only half that of the type 2 as maximum p.d. Across the varicaps is only 12V instead of 24V.

7. The best way to determine the control voltage range of the PLL is to trace the CV line back to the PLL PCB and the driver chip using simple continuity checks. Once the driver chip has been found the voltages on its power pins may be measured in the same way that the VCO supply voltage was checked....a long test wire. If we see +6V and -18V we know that we have a 24V range and if we see +6V and -6V we know we have a 12V tuning range. Sadly, if your u it has a failed negative voltage rail, we may not see the expected negative voltage reading. We will cross that bridge if we come to it.

8. The frequency to which the VCO is aligned will very much depend upon the tuning range of the VCO and p.d across the Varicap diodes at the time it is set up. We shall deal with that when we have the facts regarding the tuning range.

9. It is a possibility that there is an intermittent open circuit in the connection between the extreme ends of the Varicap diodes. The connection between these two points is made using the three unidentifiec components mounted in an arc around the two diodes. This connection delivers the 6V to the leftmost diode so if that connection fails, that diode can no longer be biased and that will ruin the tuning range by halving it and also causing an error during alignment. The identity of the three small components would be useful. I shall detail how to test them and the continuity of the supply chain in a separate message. This will follow later.

Fraser

[jc81382]

Fraser,

Sorry for the delay, I don't know if it was the case for you but for my the site was down most of the day yesterday so I was unable to post anything. As a result I haven't started on your last post either.

I assume that the most important thing at this point is to jump down to #7 as I think we need to know the actual voltage we are working with correct?

The way I was able to confirm the 6V was to solder to that pin with the blue wire come off of it in the area of the LO2. Do you think I will be making this measurement off of a similar pin or off of a IC ?

It is around lunch time here on the East Coast so if you have a chance to replay before I get home I will proceed accordingly.

Also, regarding #9 it seems i did have a bit of a bad solder joint across that "unspecified" component. I did a quick solder job and i am now getting voltage on both ends of the vericaps and in the middle again.

[jc81382]

Fraser,

Not sure if this is good news or bad news for my situation...

I decided to go down at start fresh again to make sure I wasn't missing anything obvious.

I first connected a lead to the "pin" to confirm a 6.2 V supply when the unit is in the chassis, this is confirmed.

I then removed it and applied 6.2 V form my supply and although I was unable to lock in the 3805 when the sweeper installed (I could get as close of 3795), as soon as I removed it I immediately go 3805 (even powered on and off a few time and it always came right back).

I figured I would put it back in the chassis and power it to up see what freq was coming out of the set port when it was installed. These reading were all over the place from ~4000 down to ~2000. However as soon as I tried to run the self cal it would go right to 3840, every time.

Unless I missed something I believe these are the 2 freq we should be expecting in this situation, correct?

Can you think of anything stupid/simple i might be missing at this point.

[Fraser]

Hi,

Sorry I have been occupied on some other work but having just seen your post things are looking up. The most important thing is to have 3840 MHz when the Analyser is running. We do need to check the voltage on the PLL control line to establish what voltage is present. That will give some idea of the PLL state.

With the frequency counter showing 3840 MHz does the Spectrum Analyser still indicate a fault when self test is run, if so, is it the same as previously ?

Fraser

[jc81382]

***I was going to post to you in response to your question, however when I went to confirm the measurements I spent some more time messing around so this post is  a 2 parter. Hopefully some of it is useful***

PART 1:

The voltage on the PLL control is pretty consistent at 6.2V while installed in the chassis, +/- .02V (I am measuring this via a wire I soldered to the post).

While in the chassis I was not getting 3840 Mhz exactly all the time, however it was pretty close it only fluctuated by 20Mhz or so . Every time I initiate the "Self Cal", at that moment I hear a "click" noise, the unit would lock on to 3840.

Yes, even though I am seeing 3840 I still get Fail on the Self Test and also the 807 error on the Self Cal.

PART 2: (some other things I did)

I decided while I was down there I would take the lead wire off and put the cover back on to see what happened. I even took the opposite side of the module apart to see if there was a spot to supply power through the passthrough and there is not.

When the unit is installed in the chassis and the cover is on I no longer got the 3840 it was more like 3856, plus when I ran the Self Cal that Freq would not change like it did with the cover off. For some reason it would just stay at 3856 and not change. (Now I cant think of a reason why with would happen with the cover off and not on)

So just for the fun of it I removed it again, put the sweeper back on and adjusted an offset of 16 Mhz so that when I reinstalled it into the chassis I was getting 3840. (I had to set the "Adjusted Freq" to 3793 by the way)

However even when I was able to dial in the 3840 in this manner it still didn't work.

[Fraser]

Two comments.

The PLL control voltage is not the supply to the VCO. I am. Nor certain that you are monitoring the correct point for the PLL control voltage. You can actually monitor the control voltage on the outer pins of either of the Varicap diodes. The point where they are connected together will always be 6V. The problem with monitoring the voltage on the Varicap diode is that you will adversely effect the tuning of the VCO. That is why we really need to find the point where the PLL control line leaves the metal chassis and passes up to the PLL PCB. Monitoring at that point does not adversely effect the VCO tuning.

The PLL control line has a 15K Ohm resistor in series with it at the Varicap diode so any continuity checks between the Varicap diode ends and the PLL PCB control line output will read 15K when you have continuity. I will have to draw the test points on the VCO diagram for you and that will be a point that avoids the 15K resistor.

All your symptoms suggest to me that the Phase Locked Loop is not controlling the VCO. We are back to the likelihood that the PLL VCO control line output is open circuit, missing a supply rail or the driver chip is faulty.

Sadly whilst the PLL is not locked, you will never get a calibration pass.

I am a bit busy this evening so will write again tomorrow.

Fraser

[Fraser]

While you await further comment from me you could assist a great deal by providing the following photographs.

1. Picture of the meal chassis on the opposite side to the VCO and with the PLL board removed. I wan t to see if I can track down the PLL control line visually. if there is a Connector that plugs into the PLL board and is connected to several places on the chassis below, please provide photographs of all these connection points...... One will be the PLL control line !

2. Detailed pictures (several please) of the PLL PCB, both sides including overall shots and close up, say breaking the PCB into three sections. I may be able to identify the PLL and its driver if the pictures are good enough.

3. If able, please also photograph the VCO side of the chassis, with all covers removed so that I can look for then PLL control path.

If a particular picture is of interest I will provide my email address so that you can send me a higher resolution image. Standard 640x480 is OK for initial viewing on this forum.

Best Wishes

Fraser

[jc81382]

Fraser,

I hope these pics are what you are looking for. Even though I resized them I could not fit more than 3 on a post so I put them on Postimage to share them. The link is here  [url]http://postimg.org/gallery/22m0qrrpq/806c6b82//[url]

I took a single shot of each board, then 3 closeups, then a panorama.

I dismantled and took picks of all boards/covers on the opposite side (opposite side from where we have been working) as well. There were a total of 3 covers that I removed on that side.

Also as a side note is seems that all the pass-throughs from the side with the adjustment are coming throughs, but soldered directly to the circuit boards.

Please let me know if I missed something.

[jc81382]

Fraser,

(Please know I have zero intention of rushing you along)

I just wanted to confirm you were able to see the pictures I put on Postimage.

If you have a chance please confirm.

Thanks.

[Fraser]

Hi, yes no problem. Pictures can be seen.

I am a bit buried in work at the moment but will come back on this as soon as possible.

Fraser

[jc81382]

Please, no rush at all.

I just wanted to confirm.

Thank you.

[Fraser]

Be assured, I have not forgotten you. Just a lot on my plate at the moment.

I will be back !

Fraser

[jc81382]

No worries, I'm in a holding pattern awaiting you instructions.

[jc81382]

Fraser,

Any feedback on how much access you think I will need to the circuit boards?

I still have the unit apart on my bench and I want to put as much back together as possible so I don't lose or accidentally break anything.

Thanks

[lelandford]

I have an Advantest R3131 that has the tracking generator hardware installed but does not seem to realize it.

I believe the issue is that the lithium battery which was backing up the realtime clock and a couple of static memory chips failed. 

I replaced the battery and now the fact that it had the TG option is no longer remembered.

I don't know how to restored the option setting in the static memory.

Anyone have any experience with this?

Microlease wants $4K to fix this and thinks it may have to be sent to Japan which is out of the question.

I've looked everywhere for clues as to how to restore the option but no success so far. I've tried all suggestions that I've been given but as I said...no success. There is a Shift 5 and a Shift 9 function that may help but each is asking for a password that I have not been able to crack yet.

Hoping one of you guys holds the key.

Thanks,
Leland
K9SDR

[TheSteve]

I haven't seen anyone mention they have figured out a way to re-enable the tracking gen after the battery dies - the question has been asked several times. Eventually the right person will be determined enough to disassemble the firmware and figure it out, when they do I hope they share it with everyone.

[nctnico]

I posted this before: there is a small button on the CPU board near the rear of the spectrum analyser (you have to remove the rear cover). I had some problems with my R3131 after I replaced the battery. Pressing the button seemed to re-initialise the data in the battery backed-up RAM. It wouldn't surprise me if that procedure also gets the tracking generator working again.

[lelandford]

nctnico,

I have tried the reset button on the cpu board. Unfortunately it has no effect for re-enabling the tracking generator.

I appreciate the suggestion though. I'm open to any other suggestions.

I believe there are option enabling bytes that must be set in the static ram that is backed up by the battery. Unfortunately without a utility running on the R3131 I have no definite idea where those memory bytes reside in the address space and what the code is for enabling it. I'm feeling like the passwords for shift-5 and shift-9 might open up a utility but I have not yet cracked the password for either of those functions. I have tried several published passwords for other Advantest analyzers but none of those seem to work either. Sure wish I could figure it out. I would love to be able to share the answer with everyone else.

Leland
K9SDR

[Wuerstchenhund]

Silly question: have you asked Advantest if they can help? I don't have much experience with them but sometimes manufacturers are friendly enough to hobbyists to help them (re-)enable options for old gear.

[nctnico]

nctnico,

I have tried the reset button on the cpu board. Unfortunately it has no effect for re-enabling the tracking generator.

Did you press it long or short? Try pressing it for a couple of seconds AND keep it pressed in during power on. Maybe it takes a longer press or it is only detected during power on.
Your calibration data is likely to dissapear as well. You can check by enabling the 30MHz calibration signal. If it is off then the calibration data has been erased.

[RadioAct]

Hello Guys/Gals,

I am not here to hijack this thread but I do think I may be able to offer some bits of information. Background is that I have recently aquired a R3132 SA and upon checking it out it seemed like there was a Oscillator/PLL issue. I connected and measured the reference output frequency and it was off by about 3MHz (27Mhz ish) and had a all around instability about it, power seemed in the ballpark but not perfect.

Jump on google and start reading and I find out about the RF VCO aging issue. Sounds likely. I stumbled on this thread and read about jc81382 R3132 and his and Fraser's battle with it. Oddly enough the RF board I have in my R3132 appears identical to jc81382's photos and not like the one in the pictures of Fraser's alignment guide. Since I was going through the same procedure I decided to make a few measurements and post my results here hopefully they will be of some value. I will get some pictures up asap as well.

The first thing I did was open up the RF deck and attach some 28Ga spy wires to the power supply feed-through to the 2nd LO MMIC and another to the PLL feedback (TP501) running to the varicap structure. If you take out one screw holding the RF probe power there is a nice hole to get small wires out. I put everything back together and fired it up. Everything came up just fine and I made these measurements:

MMIC Power: 6.53
PLL Feedback: -9.58

So it appears that this RF board uses 6.5V for the MMIC power supply to the 2nd LO and I am going to guess the tuning range is +/- 9.5V. Yup my VCO needs some adjusting. As a confidence check I did a measurement on the reference signal with a Tek MDO4000 scope with the SA option and I found the R3132 reference to be bang on 30Mhz at -20dBm. I have booked some time next week to go and visit a friend who has a whole room full of SA, sig gens, power meters, and the little bits to connect them all together. He has an Agilent SA that does 6.7Ghz that should help get the VCO tuned up again. I would like to try and get the VCO aligned again so the PLL will lock and then feed in some reference signals and see how the unit is in accuracy department.

I will report back after I give it a go.

Ty

[Fraser]

Hi RadioAct,

Many thanks for this useful information. Your readings are very helpful.

Sadly I have been well and truly buried in other work so had not made any progress on tracing the PLL control line path to the Varicaps. You have now detailed a way to monitor said control voltage quite easily.

The challenge with setting up the VCO is knowing what frequency to set it to without PLL control active and with the cover off. It might appear to be as simple as 'with 0V on the varicaps the centre frequency may be set' . Sadly the PLL control line does not rest at 0V on the uninstalled module so that would need to be addressed. Also, with the cover removed, the VCO may oscillate at a different frequency to when the cover is on. You can sometimes use aluminium foil to simulate the cover during adjustments but it is fiddly with the dielectric trimmer design.

Tuning the VCO can take a lot of trial and error. I was fortunate enough to have a working module to use as a reference so I knew what the VCO should be set to with the PLL inactive and cover off. Without that I would have likely tied the Varicap diode control line to 0v and set the VCO to near its correct operating frequency. I would then have checked how many MHz the VCO shifted with the cover fitted. Re-adjusted the VCO trimmer to compensate and repeated the whole operation, again and again until the VCO was pretty much set to its most accurate frequency possible with the trimmer. I must warn that the dielectric trimmer is a PITA to set, as small movements can cause large shifts in frequency and the adjustment is not linear.

I may have been too fussy about getting the VCO frequency very accurate as the PLL can pull it onto operational frequency within reasonable limits.

Good luck with your adjustments.

Fraser

[lelandford]

Wuerstchenhund...I have contacted Microlease who is the designated Advantest service organization for the US. As I mentioned in a previous post, they are the ones who quoted me $4k to fix it. I can buy another analyzer for that. I responded to their quote by telling them my situation and that I am using this analyzer for hobby purposes only and that I would appreciate some information on how I could fix it myself. They did not respond back. I took that to mean they didn't want to bother with me if they could not get $4k out of me. I have not tried to contact Advantest directly in Japan, so maybe I'll try that next.

nctnico...Yeah I have tried holding it down all during the power up process and also a short press. Both seem to simply reset the processor and unfortunately neither fixes it.

Thanks,
Leland
K9SDR