Author Topic: Advantest R3131A Spectrum Analyzer 9KHz-3GHz (Read 128334 times)

jc81382 · « **Reply #125 on:** February 08, 2016, 07:18:44 pm »

Fraser,

Also, any idea what type of current the circuit will draw during adjustment? I wonder if my power supply is underrated.

Fraser · « **Reply #126 on:** February 08, 2016, 07:55:50 pm »

If you have 11.3v at the input to the PCB your PSU is not current limiting and fine for the task. I do not recall how much current is drawn but it wasn't much.

Fraser

jc81382 · « **Reply #127 on:** February 08, 2016, 11:27:51 pm »

Fraser,

So I hooked everything back up to give the adjustment another shot and same this as before, I can't seem to hit that 3805 I can get to 3767.

I then decided to see how much current it is pulling off the power supply when feeding the 11.3 v and I am maxing out the power supply to .75 amps.

Should I try a more powerful supply?

Thoughts?

jc81382 · « **Reply #128 on:** February 08, 2016, 11:44:09 pm »

Just for fun I hooked up to the 2 output in my PS.

This time I was only drawing around .5 amps but as soon as I hooked the RF board up my voltage dropped to like 6 v (on my Dmm that I am using to monitor) I then disconnected the Dmm from the PS and took a reading at the varicap and I was getting 9.4v and then 9.7 on the PS.

I'm starting to think I have a PS problem.

Fraser · « **Reply #129 on:** February 09, 2016, 12:03:17 am »

By applying power to the 2nd VCO supply rail you will be providing power to other circuits and components on that same supply rail, even though they are not actually doing anything useful. 0.75A is not an unreasonable current draw.

If you cannot get the VCO to 3805MHz then there is a problem with its tuning range that needs investigation. The dielectric tuning 'plate' needs to be very carefully moved through its full arc of travel whilst watching the frequency on the counter. There should be ore than one part of the arc where it approaches the correct 3805MHz. It can be pretty fiddly to get right. The greatest error in the set frequency that I would accept would be 3804MHz to 3806MHz. Anything further away from 3805MHz is forcing the PLL too far off its centre setting and thermal drift could cause issues later.

Sorry that I missed the large error in the frequency of the VCO earlier. Too many jobs on the go at once and late nights

Fraser

jc81382 · « **Reply #130 on:** February 09, 2016, 12:55:23 am »

no worries.

Yeah I do notice that as I twist it the freq will cycle to a low freq back to a high and at about ~180deg it will get close to the same reading but never as high as the 3805. I think when I took the pics at 3797 that was the closest I ever got to 3805.

I still cant figure out why on the first PS output I max out the current then on the second I only draw half the current but the voltage drops.

Is it reasonable to assume that if I am not supplying the proper voltage and required amperage that I would never achieve the 3805?

Either way I am trying to get my hands on another power supply to try tomorrow, I will keep you posted.

Fraser · « **Reply #131 on:** February 09, 2016, 12:06:22 pm »

If you can apply 11.3V the PSU is OK. Sounds like your second PSU channel is sick.

It is a problem that you cannot achieve 3805MHz. Such would suggest an issue in the VCO. You could have a failed Varicap diode or other tuning component. To prove this I would isolate the Varicap diode control line and bench test the VCO's tuning range. It's tuning range should be approximately 40MHz. Sadly IIRC this is not such a simple test to configure as the Varicap diodes sit across a split supply with the control line moving plus and minus around 0V. It would need some careful configuration to avoid potential damage to other circuits.

This is becoming a more complex job I am sorry to say. A problem in the VCO is challenging due to its 3GHz operating frequency. With the greatest of regret I do not have the time to write a full test sequence and draw the schematic so that we can isolate the cause of the problem. If you were in the UK I would suggest you send the module to me and I would repair it for you. Sadly postage costs from your location would make such an expensive option.

Sorry that I cannot be of greater assistance to you on this problem. Your colleague who has worked on SA repair before may be able to help you isolate the problem further. I recommend that the schematic for the VCO be drawn so that all potential causes of the frequency error may be identified. The Varicap diodes are the first place I would suspect until I proved that they presented the correct capacitance range to the VCO. There is also the matter of the VCO control line from the PLL appearing to be sat at exactly 0V. That is not right at all. If the VCO is not producing 3840 MHz the control line should be a long way off 0V as it attempts to pull the VCO onto the correct frequency.

Whilst writing this message, I have had another thought and relatively easy test for you. ...........

It would be worth checking that the positive and negative supply rails to the Varicap diodes are present. Working from memory, there should be +12v on one end of the pair and -12V on the other end. It is possible, just possible you have lost the -12V rail or connection so the PLL can only drive the control line to 0V and not negative, as required. I say this because I think the PLL should cope with an initial setting of 3797Mhz.

You need to do some continuity tests to establish where the dual supply rails to the Varicap diodes comes from. Hopefully you can find a convenient point to attach some wires so that the module may be inserted into the chassis and switched on. All power supply rails will then be provided by the SA chassis for you. Check that the voltages are present. I will need to check my notes on the power supply rails used on the 2nd VCO it was either +-12V or +-20V. Whichever, it should be symmetrical around 0V.

Fraser

Fraser · « **Reply #132 on:** February 09, 2016, 12:12:15 pm »

My alignment guide contains all the details. The maximum p.d. across the Varicap diodes is 24V. This is across split power rails of -12V and +12V with the PLL moving the control line between these two limits. You could well have lost you -12V rail at the PLL VCO Varicap diode driver. Definitely worth checking.

My memory also failed me on the VCO spec. It can be swung plus and minus 40MHz around its centre frequency so has a range of 80MHz.
With this range, the PLL should definitely be able to cope with your 3797MHz centre frequency. All this will mean is that the PLL will have less tuning range available to it in one direction. The more I think about this, the more I think you have lost your -12V rail at the VCO control line driver.

Fraser

Fraser · « **Reply #133 on:** February 09, 2016, 01:03:46 pm »

I feel like I am missing something here. Working from memory is never a great idea. I will spend some time looking at the VCO pictures I took whilst working on these modules to determine the exact topology. Once I have an understanding of that, I can advise on how it is supposed to operate. The problem at the moment for me is that I designed my alignment process around just supplying 11.3V to the VCO. The Varicap diode control line settles at 10.3V naturally so I worked with that to select the required frequency setting for the dielectric trimmer. It's a pretty nasty alignment technique but it's easy to do with the cover removed and it works well. The KISS principle at work.

However, when a fault is present we need a better understanding of what is happening in the VCO as the +12V and -12V rails are not present during my alignment procedure.

This is something I will have to study in order to make helpful comment.

You can however proceed with checking the voltages that are being applied to either end of the Varicap diode pair and also the Supply rails to the PLL driver chip that drives the Varicap diode control line.

Fraser

jc81382 · « **Reply #134 on:** February 09, 2016, 01:08:27 pm »

I will do that today and let you know what happens. Thank you again for all your help.

I was able to get my hands on a Tek 2520 Programmable Power Supply so I am interested to see if that makes any difference.

Fraser · « **Reply #135 on:** February 09, 2016, 01:30:46 pm »

Yep, as I suspected, My memory is failing me. Having looked at the picture you provided of your VCO tuning, it is not as I thought. Please await an update from me and tests that need to be done.

Fraser

Fraser · « **Reply #136 on:** February 09, 2016, 02:09:55 pm »

OK I am having a feeling of Deja-vu when looking at the VCO. I must have reverse engineered it previously. I shall look for any notes that I still have and if necessary I will redraw the schematic. I have little time so the result will be hand drawn and functional rather than pretty

More comment as soon as I am done looking at this circuit.

Fraser

jc81382 · « **Reply #137 on:** February 09, 2016, 02:19:25 pm »

Not a problem, let me know if any additional pictures on my end might help.

Are you thinking that the adjustment procedure does not apply to this particular revision or are you trying to advise on additional troubleshooting procedures?

Fraser · « **Reply #138 on:** February 09, 2016, 02:24:42 pm »

No, the adjustment procedure still works fine. It is the cause of the failure to lock a at 3840MHz that I am investigating. My line of thinking is that the 0v you are seeing on the control line is because the PLL has lost half its tuning voltage range. I suspect the -12V rail has gone but need to prove it to myself.

Fraser

jc81382 · « **Reply #139 on:** February 09, 2016, 02:30:46 pm »

OK.

Also, just as a side note (in case you don't have pics of the board you worked on)... I only saw 1 of those pass through pins come through to the other side that would allow access to take a measurements on while the cover is on the board and it is installed in the chassis.

Fraser · « **Reply #140 on:** February 09, 2016, 03:55:10 pm »

No worries,

I am drawing the VCO right now. It s a bit different to how I remembered it but I am already suspicious that the -12V rail or missing from the LM324 that drives the Varicap diodes. Either that or the LM324 has failed.

It should be clearer once I post the schematic

Fraser

Fraser · « **Reply #141 on:** February 09, 2016, 04:56:32 pm »

Just for info if anyone else works on one of these R3132 VCO's ....... the design was changed in older and newer versions so the schematics, though similar, are crucially different in one particular area. Namely the injection point of the PLL control voltage ! The older model had a design flaw and when they fixed that they moved the PLL control voltage connection to the varicap diodes and reversed the diodes on the PCB. This is why I have previously visited this VCO.... it obviously had me scratching my head last time I worked on one of these. The good news is that I have a photographic record of both versions from which to work.

All good fun.

I will post the simplified schematics for both to aid others in fault finding. The VCO is very simple in DC terms so we should be able to sort this issue out.

I am off to the VETs now, so will complete this later. (The vets isn't for me

....the cats need their annual inoculations)

Fraser

Fraser · « **Reply #142 on:** February 09, 2016, 10:31:20 pm »

OK all done

Attached are the simple style schematics for the two versions of 2nd Local Oscillator found in the R3132. I include the very much simplified DC model to the right of the schematic. Your module is a Type 1 early version.

As can be seen, changes have been made around the area of the varicap diodes in the Type 2 version. The changes have improved the design but the simple alignment procedure that I wrote applies to both versions in all respects except where you measure the control voltage on the varicap diodes..

In both versions, the Varicap diodes are in series with the VCO resonant circuit and this is how they influence the oscillation frequency of the VCO. In purely DC terms however, the two diodes are in parallel and the potential difference across them is created by tying one end of the parallel pair to +11.3V whilst the other end is connected to the Control Voltage (CV) line coming from the Phase Locked Loop (PLL) PCB. The PLL produces a voltage at its output in the range -12V to +12V. As such it is capable of providing ~0V to ~23V across the varicap diode pair.

If CV -12V then the p.d. to +11.3V on the other end of the varicap diodes is 23.3V
If CV +12V then the p.d. to +11.3V on the other end of the varicap diodes is 0.7V (but forward bias on varicap)

In my alignment procedure the CV pin settles at +10.3V and the VCO produces 3805MHz. When the PLL is active, the CV line drops to around 0V, which is its mid point of output capability, and the VCO rises in frequency to its specified 3840MHz. In the alignment phase, the p.d across the two Varicap diodes is 1V. When the module is reinserted into the chassis and the PLL is active, the p.d across the varicap diodes increases to approximately 11.3V.

Now if we consider the loss of the -12V rail to the PLL output LM324 chip we have the following potential scenario........

As stated, when the module is aligned in the manner that I detailed it has no PLL control. When the PLL is active it applies approximately 0V to the CV pin, the VCO should be pretty much on frequency at 3840MHz. If however the alignment phase has proven that the VCO cannot be set at 3805MHz, due to component tolerances in the VCO circuit, we may have to settle for something near to that frequency.

The PLL can swing the VCO through a range of -40MHz to +40MHz around its set centre frequency. 3805MHz is the ideal alignment frequency but less than 10MHz either way will not be significant in most cases. That would provide an acceptable alignment frequency range of 3795MHz to 3815MHz. It may even be acceptable to use a frequency as much as 20MHz away from ideal, but this does limit the PLL's lock capability in one frequency drift direction as it is already compensating for the alignment error. This is already explained in my alignment guide.

We know that an alignment frequency of 3805MHz is optimal as when the PLL is active it only needs to apply approximately 0V to the CV line to gain lock at 3840MHz. It is therefore pretty much in the centre of its control range. If however we could only achieve 3790MHz during the alignment, the frequency will be too low and the PLL will compensate by applying the required voltage to the CV line. In this example it would actually apply a negative voltage on the CV line in order to generate greater potential difference across the two varicap diodes. If the -12V power rail is not present at the LM324 on the PLL output, a negative voltage cannot be applied to the CV line. The result will be a PLL trying to pull the VCO onto frequency and failing to do so because it cannot create the required p.d across the varicap diodes. It will do its best though and try to drive the LM324 for negative output. The output will sit at 0V however as the LM324 cannot supply what it does not have, i.e a negative potential.

Well this is my present theory on what has gone wrong but as I do not have the module in front of me, you will need to do some testing to prove whether I am right or wrong about this. I am very tired right now so regret I cannot help further for the moment.

Fraser

UPDATE : PLEASE READ

It has become apparent to me that the type 1 module may differ to the type 2 module in ways that I cannot capture as I do not have a type 1 module available to test. The simple schematics that I have drawn come from my type 2 module repair notes, and photographic records of both module types. Please treat these schematics with caution and use them as guidance only as their content IS NOT guaranteed accurate !

Caution should also be exercised when using my simple alignment procedure on type 1 modules as there Are differences in the type 1 module PCB. The principal remains valid however.

Thank You

Fraser

Fraser · « **Reply #143 on:** February 10, 2016, 12:32:13 am »

Some pictures of the second Local Oscillator VCO and associated PLL PCB.

There are two different examples of the Type 2 VCO and one of the type 1 VCO.

Note that the Type 1 VCO has been mauled at some time by the service agent. The dielectric trimmer 'wiper' has been removed and they have used an SMD resistor to act as a stripline tuning element on the striplines. Nasty workmanship. From this evidence it would appear that Type 1 VCO's may be hard to get within centre frequency spec using just the dielectric trimmer wiper. I checked the VCO alignment and it was acceptable. The module was working fine and that SA had a corrupted Firmware CF card. Out of five units, this was the only one that was without its dielectric wiper.

I have previously taken a picture of me probing an output pin on an LM324 located on the PLL PCB. I can only assume that this was the CV line going to the Varicap diodes. This is a later type PLL and may not be identical to the earlier version. LM324 pinout attached. It is a single or dual supply rail device. In the PLL it should be a dual supply rail configuration in order to swing the output from -12V to +12V under PLL control.

Hope this helps.

Fraser · « **Reply #144 on:** February 10, 2016, 12:52:39 am »

The PLL PCB is well labelled and has test points for both the +12V and -12V rails.

I attach a close up picture.

TP10 is +12.6V
TP11 is -12.6V

The presence of these voltages at these test points does not prove that the voltage is reaching the PLL CV line output driver. You can check continuity between the test point and its associated LM324 power supply pin with the module on the bench. Then solder a temporary thin test wire to each test point and measure the voltage with respect to chassis whilst the module is in the chassis and powered.

There remains the possibility that the LM324 is faulty so the presence of the correct voltage at its supply pins does not mean that it is working correctly. In the attached LM324 op-amp equivalent circuit, Q13 failing open-circuit would cause the loss of the negative voltage swing at the output. We will consider next moves after you have tested the supply rails.

I almost forgot, my modules had a current draw of approximately 200mA on the 11.3V supply when doing the alignment procedure with the module out of the chassis. Yours appears to be drawing significantly more current. That may need further investigation.

Fraser

Fraser · « **Reply #145 on:** February 10, 2016, 01:22:25 am »

One final comment before bed.....

On my simple schematics I have shown the three components that are across the Varicap diodes in your Type 1 module as inductors, but in truth I have never tested them in order identify whether they are inductors or possibly even diodes. The link is essential to connect the CV line to both varicap diodes. Inductors would normally make a mess of the tuning. They may be diodes. Three in series drops voltage but lowers their total capacitance. The diode test mode on a multimeter may resolve this unknown but I doubt it has anything to do with your problem.

Fraser

jc81382 · « **Reply #146 on:** February 10, 2016, 03:10:58 am »

Fraser,

I can only hope that all your efforts will end up helping someone other than me some day...

I have to be honest that without a procedure similar your adjustment procedure I worry that this is over my head. Granted as I said before, at this point I have nothing to loose at this point because either way I don't have a working unit.

As I mentioned in an earlier post I was able to get my hands on another power supply today and I wanted to let you know what happened with that...

As it turns out when 11.3 V are applied to the point specified in your adjustment procedure the unit was drawing .7 amps, so just outside the max of the unit I was using previously.

Unfortunately with this power supply I had the same results of not being able to achieve the 3805.

At this point i did my best to take some of the measurements you were referring to in your previous posts, and this is what I ended up with...

with 11.3 V applied

Fraser · « **Reply #147 on:** February 10, 2016, 01:47:23 pm »

Hiya,

With great regret I have to stop this remote diagnosis here. This is not because I do not want to help, but rather my concern that I do not have the older Type1 module in front of me and it could have significant differences in its design to those that I have worked on. The increase in current from what is expected, to an abnormal draw could be a failed component or it could be because of a difference in module design. This makes proceeding without the ability to take reference measurements on a good SA too risky.

I am sorry that this little journey into your R3132 must come to an end here but it would be better to find someone local who can repair the unit for you. It is likely a simple fault and worth the effort.

For other readers..... Please consider the schematic for the Type 1 module suspect, especially the voltages as I did not have those on record from when I worked on an older generation of module. The newer modules have an additional 390 Ohm power resistor which may be significant. Without further investigation I can not be certain of its function. Sadly neither my time nor my energy levels permit such an investigation.

Fraser

Fraser · « **Reply #148 on:** February 10, 2016, 05:59:41 pm »

Having given this situation some thought today, I am concerned that the type 1 module has some significant differences to the type 2 that I have worked on in greater detail. Though I do not recall any power supply differences I believe that the type 2 module contains a 390 Ohm power resistor as a dropper for the MMIC's. The type 1 module has no such resistor and this may be the reason for the unusually high 700mA current draw when the VCO is powered from 11.3V. Whilst the MMICs have survived, this is not a good idea.

I do not have a type 1 module to carry out tests on so cannot risk providing further comment on the VCO circuit or its power supply arrangement.

I have added a note to my simple schematics posting. These schematics were produced from repair notes produced when I repaired 5 R3132 analysers. I also had detailed photographs of the two module types. My notes suggest both have the VCO running off of 11.3V but I have no way to prove this.

This has been an interesting investigation and I have not minded spending some time on it, drawing schematics etc, but it also highlights the risks involved in remote diagnostics and equipment version differences.

I sincerely hope that this sick R3132 is eventually repaired. It is likely to be a simple fault and they are worth the effort. I paid GBP600 for mine so they are not cheap.

Fraser

jc81382 · « **Reply #149 on:** February 11, 2016, 02:34:09 am »

Fraser,

Quick question for you, hopefully.

Do you have any idea how I could go about getting replacement vericap? I assume the have specific sizes and ratings.

For some reason today I decided to remove the dielectric swiper to see what happened and when I did the unit put out a freq of 3805-3804 consistently. I actually powered down the supply a few time and each time it settled in (this was while supplying 11.3V).

I wondered if that might have solved my issue but when I put it back together it hadn't.

So I tried to confirm the voltage on the vericaps and I was getting the expected voltage between the two vericaps and on one end, but the other end had no voltage. I figured it might be worth trying to replace them.

Thanks


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

Author Topic: Advantest R3131A Spectrum Analyzer 9KHz-3GHz (Read 128334 times)

Share me