Author Topic: Advantest R3131A Spectrum Analyzer 9KHz-3GHz (Read 128305 times)

jc81382 · « **Reply #100 on:** February 06, 2016, 05:40:59 pm »

So the plot thickens...

It seems that was confused because my unit does not have that pass thought.

I took off the top circuit board to get a good look/pic of the unit and as you can see the only pass thought on the entire this was the one i was trying to use in the first place. (top right area)

I also added a pic of how I connected and how I was trying to take the measurement.

Do you think i should be expecting a voltage at this point?

Was there anything else I had to do other than turning the unit on to get a reading?

Fraser · « **Reply #101 on:** February 06, 2016, 05:56:43 pm »

Hmmmm how odd. I have not seen a unit without that feedthru before. I will need to check my notes on this SA to see where I was monitoring the PLL voltage. Sorry I do not recall the purpose of the feedthru that you are monitoring.

Fraser

jc81382 · « **Reply #102 on:** February 06, 2016, 05:59:40 pm »

sounds good.

There was another plate that was under the circuit board and I took that off to see if the pass throughs came up from there and they did not. it was just another circuit board that has some RF Cable going thought the sidewall that connected to the circuit board that usually rests on the top.

jc81382 · « **Reply #103 on:** February 06, 2016, 06:03:19 pm »

Another question.

How sensitive are the cables that connect the RF board to the board above it?

I noticed when I was connecting things back up that one of the cable was pretty crimped (it was the one that crosses over the copper SMT.

I was just wondering since that connects the 2 boards together could that be causing the RF Fail?

Fraser · « **Reply #104 on:** February 06, 2016, 06:38:12 pm »

Ok I just checked the paper files that I had knocking around for the 3132. My memory has failed me. I have met your exact model of module before. I have a couple of printed pictures showing the difference ! Attached.

The aim of the measurement exercise is to get onto the control voltage line of the 2nd L.O. varicap diodes whilst the covers are on and the module in the chassis. The feedthru that I specified is a DC connection from the varicap diodes to the PLL control voltage driver. I could have chosen a point on the PLL board but that is just making life difficult for no good reason

In the case of the module that did not have the convenient monitoring point where I expected it, I would have carried out DC continuity tests between the varicap diodes common point (between the two diodes) and the feedthru's that emanated from the modules metal shell. One of them would be the control line from the PLL and provide continuity. The blue wired feedthru could well be the right one but you need to prove it.

I have bee thinking about your previous reading of 0V. If you are monitoring the PLL control voltage, it could be that you have a fault shorting the control line to 0V. The fault could be the driver chip on the PLL board or a short to chassis somewhere along the control line. Such a short would still bring the VCO almosyt onto its correct frequency as the nominal control voltage is approx 0V, but the PLL would have lost control of that VCO which can only be bad !

With regard to your crimped solid coax cabled. The copper coax cables are vulnerable to damage and if distorted too badly they can either present a poor impedance match, and standing wave issues, or even a short of the centre conductor to the outer copper screen. The insulation is PTFE which is quite soft and easily crushed to the point of a short, inner conductor to outer. To do so the coax would need to be pretty seriously bent to a right angle though. Minor outer copper sleeve distortion can be tolerated. Much depends upon the signal being carried by the cable and its frequency.

Keep us advised on progress and I will try to help but my repairs were an awful long time ago and the old memory is failing a bit. I am not in a position to dismantle either of my R3132's at the moment so regret direct comparisons are not possible.

Fraser

jc81382 · « **Reply #105 on:** February 06, 2016, 07:42:58 pm »

Ok... so here is what i got...

I took the board out again and took the cover off.

I put my DMM at the 2K ohm (continuity check setting) to that passthrough that was giving me 0V and checked the common point of the vericaps and got nothing (OL on the DMM).

I found the other side of that passthrough and immediately got 0 Ohms and beep (as expected).

But then as i touched the chassis and other passthroughts (like the one that I was applying voltage to when adjusting the Freq) and i kept getting reading of .3XX.

Do you think this means a short?

I wouldn't expect that point to be giving me a reading when touching the chassis and the there "blue" passhtroughs.

Fraser · « **Reply #106 on:** February 06, 2016, 08:01:28 pm »

This test was only to find low ohmage continuity in order to trace the control line. You cannot deduce much from your other readings as the control line is coming from a driver IC and we do not know what effect such has. However 0.3k ohms is not low enough to be a shorted control line.

You need to check the 2nd LO frequency whilst the SA is operational. No way of avoiding it. I know it's a pain with regard to access, but is is necessary.

Fraser

jc81382 · « **Reply #107 on:** February 06, 2016, 08:09:54 pm »

So basically find a way to hook up the counter to that test port on the side that was monitoring the Freq before?

Since I don't have any options that take up space on the right hand side should I just try to drill or remove a spot on the chassis where that port is so I can plug into it?

Fraser · « **Reply #108 on:** February 06, 2016, 08:25:04 pm »

EEEEEK drill the chassis....... Nooooooo. That is a recipe for a swarf disaster

Take a piece of solid core insulated copper wire with a core approx 1mm diameter. Strip off approx 5mm of insulation and bend the bare wire through 90 degrees. Insert the bare wire into the SMA connector. Run the wire along the side of the module about 50mm and hold it in place with tape. The end of this wire should be positioned such that it can be accessed by your frequency counter probe. Use the module case for the ground. Keep the wire as short as practical. This is bad RF practice on many levels but as the port is connected to a directional coupler monitor port and we are just checking frequency, it is acceptable.

Fraser

jc81382 · « **Reply #109 on:** February 06, 2016, 08:52:19 pm »

Hahaha....

I will give it a shot later today (just got to a birthday party) and let you know what happens.

jc81382 · « **Reply #110 on:** February 07, 2016, 01:06:01 am »

Fraiser,

Here is what I have.

I took a piece of CAT5 cable and removed 2 of the conductors.

I put one into the center of the port like you said and took the other and mounted it to one of the screws next to the port.

I then taped the wires down to the side of the unit and slid it back in place, then attached all the connections.

When I tried to get a reading off the Freq Counter it was all over the place. The pics attached were all taken within 15 seconds of each other.

I will wait till you have a chance to review and let me know your thoughts.

Thanks

Fraser · « **Reply #111 on:** February 07, 2016, 01:59:24 am »

I see where your problem lies.

The unscreened CAT5 wire from the port is way too long. It needs to be kept as short as possible. You need a proper coaxial test lead to go from the frequency counter to the short CAT 5 cable. A 3 GHz signal is not going to like travelling far on open wires.

If you can arrange that the CAT 5 wire just extends the ports centre pin to a location that you can get at inside the chassis, that should be a nice short length. I think a chassis side rail obscures the port so you just need to place your wire so that it gets around that obstruction. The croc clips of the screened frequency counter cable just connects to the extended centre connection of the port and the metal module case.

Time for bed now as its 02:00 here in the UK.

Best Wishes

Fraser

jc81382 · « **Reply #112 on:** February 07, 2016, 02:03:28 am »

2:00 AM... I wasn't expecting to hear from you till tomorrow.

I will run out tomorrow morning and try to find a BNC to alagator. I know we have some at work but I won't be able to grab those till Monday.

Stay tuned

Fraser · « **Reply #113 on:** February 07, 2016, 02:07:52 pm »

I just took another look at your pictures. You need to use the 500Mhz to 20 GHz input on your frequency counter and not the BNC input that only goes up to 500MHz maximum.

The signal you are monitoring is over 3GHz.

Your test lead will need the correct connector for the GHz input on your counter.

Fraser

jc81382 · « **Reply #114 on:** February 07, 2016, 02:09:26 pm »

Fraiser,

I was not getting any signal at that range, so I decided to try to 500 range and that is where I was getting the readings.

Fraser · « **Reply #115 on:** February 07, 2016, 02:13:26 pm »

Likely due to more sensitivity on the lower range and picking up random RF from the environment. I see the high range is fitted with a BNC connector as well so you can buy a BNC to croc clip or BNC to grabber coaxial lead. Always a useful lead to have and it can be used with your 3132 when you get it working.

Fraser

jc81382 · « **Reply #116 on:** February 07, 2016, 02:16:31 pm »

Ive been looking online for some place local to grab one today but my local electronics store in town is closed on Sunday and Radio Shack doesn't seem to have them.

Im gonna try to grad one off a co-worker and if not I'll grab one from work.

jc81382 · « **Reply #117 on:** February 07, 2016, 06:07:31 pm »

Fraser,

With the lack of BNG to Alligator, I decided to get creative.

As I looked inside the chassis it seems that I could remove the "divider" from the inside, therefore giving me access to that port that we are trying to monitor.

After taking off the 6 screws on the bottom, then the face plate to reveal 6 more I was able to take it out and hook right up to it (just barely because of the length of my BNC).

Either way it worked and I am getting a Freq of around 3.8 Ghz.

Please see attached pics and let me know your thoughts.

jc81382 · « **Reply #118 on:** February 07, 2016, 06:31:52 pm »

For some reason when I upload the 2nd pic it always shows ends up upside down.. Sorry

Fraser · « **Reply #119 on:** February 07, 2016, 07:58:15 pm »

Sadly that is not good news. The frequency should be exactly 3840MHz. You frequency is almost 2MHZ out. Which is significant in the 2nd stage of down conversion in the 3132.

I am suspicious of the 0V reading on the PLL 2nd LO control line. It should be trying to bring the frequency to exactly 3840MHz and definitely not 0V.

I believe you have a problem on the 2nd LO PLL or its control line drive to the associated VCO. The reason it is sat almost on frequency is that my simple alignment procedure sets the frequency with 10.3V on the control point of the Varicap diodes. The VCO should then be on the correct frequency when the control line drops to around 0V, which is the PLL control voltage centre point. If you just force the varicap diode control input to 0V it will still be pretty close to core t but it is not under PLL control.

OK, you read 10.3V on the varicap control point so the wiring from the Varicap diodes to the PLL is not shorted. It could be a fault in the control line driver. It would be interesting to monitor the frequency on the monitoring port to see if it drifts as the VCO warms up. If it does not, it would appear the PLL is active but holding the wrong frequency. I would expect you to see frequency drift though, as I think your PLL is at fault here.

Diagnosing the fault in the PLL really needs a good understanding of its chips and a schematic to identify appropriate test points. The alternative is to reverse engineer it and create a schematic for that part of the circuit. I regret I cannot help on the schematic front.

You may be facing quite a challenge here. How about your colleague who has worked on SA repair helping you ?

Fraser

jc81382 · « **Reply #120 on:** February 07, 2016, 08:42:45 pm »

So I'm not confident in my 0v reading as I was unable to take a reading where your procedure called for.

I was thinking of trying to retake that reading now that I don't have the divider in the way by leaving the cover off of the RF board and connecting directly to the connector that I was applying the 11.3v to (I assume that is the point that you want the reading from correct)?

I just don't know how much error I will introduce to the circuit with the cover off. And I also don't know of any other location to try to take that reading properly.

What are your thought on that?

Also I was never able to dial in to the 3805 using your procedure. Do you think that might be why I am off by 2Ghz?

Fraser · « **Reply #121 on:** February 07, 2016, 09:09:08 pm »

Ok, to answer your question in order

1. The point that you are measuring for the 10.3 V reading during alignment is the control line from the PLL. The 11.3V supply point is not the same and would tell you only that the correct supply was present on the VCO.

2. If you cam confirm zero ohms between the Varicap diodes control voltage point and any point on the route back to the PLL board, that will give you a place to measure the control voltage with the covers on.

3. Do not operate the module with its cover off. It will cause all manner of issues including detuning the VCO.

4. Your present reading is off by 2MHz and not 2GHz. You should aim to be within plus or minus 1MHz of 3805MHz in my procedure, but that is just to get the PLL running with its full positive and negative capability. IIRC the PLL canoe ally pull the VCO plus or minus 20MHz. If you set the frequency to 3810MHz, the LL will still bring the VCO onto frequency but with less further adjustment available to compensate for thermal drift. It will only have 15MHz of PLL adjustment left.

Fraser

jc81382 · « **Reply #122 on:** February 08, 2016, 02:36:19 pm »

Fraser,

I'm not sure if this is what you are referring to in you #4 statement but...

What if I take another reading as the unit sits in the chassis and make a note of the Freq.

Remove the unit, and perform your procedure, but instead of trying to lock in the 3805 I adjust the "trim" to offset the reading that I am getting with the module installed?

For example say I still get a reading of 3838 with the unit installed in the chassis. But when I remove it and start your adjustment procedure I am getting 3767. Should I just try to adjust the module up another 2Mhz to 3769 and assume that it will correlate to another 2Mhz once it is install again? Is it worth a show?

Also should I be getting 3805 or 3840 when the units is in the chassis? I'm confused because I was adjusting for 3805 but you were expecting to see 3840 in the last post. Maybe that is correct but I just wanted to make sure.

Am I making sense?

Also, going back a few posts, I left the unit running and went back to it a few times over 45 min and it did not seem to drift from that last point of 3838.

Fraser · « **Reply #123 on:** February 08, 2016, 06:37:49 pm »

OK let's see if I can explain.

The alignment procedure I produced sets the VCO dielectric trimmer with the PLL inoperative. The frequency is set to 3805MHz as the Varicap diodes settle with 10.3v on the VCO control line.

Once the covers are fitted and the module is back in the SA, the PLL changes the voltage on the control line to approx 0V. At this voltage the VCO is running at its correct frequency of 3840MHz and is trimmed to ex actly that frequency by the PLL.

Do not set up the VCO to another frequency as that will not help your situation. The PLL can normally work perfectly well when the VCO is aligned as I have detailed. I spent a lot of time testing the alignment procedure as I had several modules to repair. The lack of a cover on the module does effect the frequency of VCO oscillation so my procedure takes account of this for you.

An important question..... How sure are you that the frequency counter is accurate ? If it is not calibrated or checked against a known good microwave frequency, it cannot be trusted.

Fraser

jc81382 · « **Reply #124 on:** February 08, 2016, 06:59:34 pm »

Ok... I think I get it now... assuming I am close to that 3805 when applying 11.3V to the circuit (using your adjustment procedure), it should automatically adjust the voltage to lock in the Freq when it is all put back together. Is that right?

Keep in mind I was never able to get that 3805, and I don't know if based on what you saw I was able to get close enough.

With that said the one thing that I did notice when monitoring the DC power supply with my DMM (when I was applying the 11.3V called out in your adjustment procedure) is that the Voltage seemed to drift. So I wonder if that might be causing an issue and if i should let things sit and settle for a while before attempting anything.

I work for a calibration lab so I know Freq Counter (HP 5343A) is calibrated.

The last calibration was on 9/17/15 and using our GPS Reference Standard it read exactly 10Ghz. The tollerance is +/-.58 Hz for this particular.

The Power Supply (HP 6205B) was last calibrated on 10/6/15 and was also in spec.

I was using just a cheep Ideal DMM so it is possible that is where I was getting the drift. I will try to confirm this when I get home tonight.

(again sorry for going around in circles but I just want to make sure I am not missing something simple)

If I am applying 11.3V to the circuit and I am able to tune it to 3805Mhz then i should be getting a reading 10.3 between the Vericaps? Do I have that correct?


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: Advantest R3131A Spectrum Analyzer 9KHz-3GHz (Read 128305 times)

Share me