Marconi TF2370 fault fixing

[PJL]

Not much time today but the 100KHz is OK so it's the complicated 36.4MHz oscillator that is the culprit. The scope images match the repair guide and I followed the realignment instructions that centres the control voltage and it quite happily runs at 36.4MHz with a fixed voltage but then runs faster as soon as the feedback connection is restored.

I will have another run through tomorrow to try to understand whats going on by breaking the loop, varying the control voltage and monitoring the frequency and phase comparator output. It's using another double balanced mixer this time with the 4 diodes on an IC. It converts the edge of the 100KHz clock to a pulse then adds ringing at approximately 36.4MHz and phase locks to this. I guess the ringing might be off frequency?

The 36.4MHz oscillator, is only used for the counter and tracking generator and it could have been wrong for a while. There is still the original fault of low amplitude.

[PJL]

I followed the alignment instructions again and this time adjusted L3 as it was about 40Hz out. After reconnecting the feedback the frequency went low and not high as before. So, I decided to adjust the lock offset when the feedback was connected and found there was a range where it locked so I have left it set to the centre of this range. I would guess the mixer is not quite right but as it works I am going to leave it alone.

Now back to the level issue!

[Radio Wrangler]

That is pretty much the right way to adjust a PLL.

With the loop opened, the phase detector of a good system will bang against one end stop or the other for a phase/frequency detector, or give a sinewave at the unlocked error frequency for a sampling type phase detector (the sinewave may be small if it is high with respect to the loop filter frequency range.

David

[PJL]

Now I am in trouble! I am looking at the 2nd local oscillator buffers (AA7 page 213) that drives the mixer (AZ1 on AC1) that takes off 36.4MHz to make the 200MHz signal for the counter/tracking generator. This is the signal that appears to be low.

The circuit is attached and what I am seeing is the output marked AZ2 on AB9 , also a balanced mixer, is almost twice the amplitude. I can see that the common base TR8 is going to attenuate the signal but I would have thought that any loss would be recovered by TR9 but I guess it would all depend on the loading? I also note that R20 is 270 ohms but it has a mark by it and I wonder if that means adjust to obtain required output level?

So based on this it would seem there is a problem with the TR8/TR9 buffer stages or the loading from the AZ1 balance mixer. I need to search through the manual to see if there is any mention of the amplitude of the AZ2/AB9 output.

[PJL]

I have bought a 1.5GHz low loss probe and a 440MHz high speed log detector. The detector is coming from China but I can still do the sub-150MHz checks in the meantime. Amazing what you can get from ebay for £17 total. At least I won't need to do the dBm conversions then!

[PJL]

I got bored waiting for the detector so decided to replace the mixer diodes seeing as I had bought the SOT ones. Unfortunately, no change to amplitude. I also tried adjusting the 36.4MHz output which is part of the calibration procedure and that does increase the output to the counter and at maximum the counter does change although not reliably.

I am still surprised that the 236MHz output being down just 4dB can have a 10dB impact on the mixer output. Other than the diodes and some tiny transformers there is nothing to the mixer so I guess this is the problem but I will go over the solder connections just in case its a dry joint.

[Diabolical Artificer]

That's disappointing, onwards and upwards. You have your job cut out for you with this, complex beasties, hats off for having a go.

Andy.

[Radio Wrangler]

Too far away, I'm afraid. I've got gear sitting here capable of showing just what's going on. Spectrum analysers to 1800MHz and 22GHz, fast scopes etc.

It's irritating.

Isn't there anyone closer that can help out?

David

[PJL]

Next thing I will do is attach the mixer to the 2022 so I can chart the output for a range of input levels. The target is >+7dBm but the 2022 can only do +6dBm but it should give me enough to determine if the mixer is at fault.

[PJL]

If this were a hard fault then I would have no trouble but it being an amplitude problem makes it extremely difficult. Even if I had the test equipment, it would still be difficult as I have only limited information on the expected amplitudes.

As mentioned in post #24, I looked at the 1st (200-310MHz) and 2nd (236MHz +/-150KHz) VC local oscillators. These are very similar and use the same buffer circuitry, each having an output for the filter then further buffering to another output to the tracking generator circuit. The tracking generator circuitry mixes the 2nd LO with 36.4MHz, followed by a 200MHz IF amp, the output being mixed with the 1st LO to reduce it to the 0-110MHz filter centre frequency.

Both are marked as +10dBm in the manual although >+7dBm is mentioned as the repair threshold for the 2nd LO. It is below this figure and I estimate it is about +3dB.

I then set the 1st LO (Manual scan) so it is fixed at 236.4MHz to match the 2nd LO frequency and compared the 1st and 2nd LO buffer outputs on the scope (PP) using the low-loss probe.

1st LO filter output = 250mV
1st LO generator output = 230mV
2nd LO filter output = 170mV
2nd LO generator output = 115mV

So there appear to be multiple issues
The 2nd LO looks low in general 170<250 and 115<230.
The 2nd LO generator 170>115mV is considerably worse than the 250>230mV.

So there appears to be a general loss of gain. Do BFY90's deteriorate over time?
I also noticed that the +5.5V rail reads +5.25V but I have yet to find out why as the origin of this is a 6V regulated supply.

[Radio Wrangler]

Originally Posted by PJL

Both are marked as +10dBm in the manual although >+7dBm is mentioned as the repair threshold for the 2nd LO. It is below this figure and I estimate it is about +3dB.

That is low enough to create several dB of extra loss in a mixer.

Effectively it gets applied across pairs of Schottky diodes, and once it gets too low to turn diodes on then the loss gets very big in the signal path.

No problems known about the BFY90 It's not a bad transistor

Remember that it's only got a current gain of 5 or 6 by the time you get to 200MHz It might have plenty of Hfe at DC, but Ft is the limiting factor.

David

[PJL]

Done the test with the 2022. It seems the low level from the 2nd LO is a red herring.

Increasing the 236MHz amplitude sees diminishing returns with the increase from +4 to +6dBm having little impact on the amplitude. It starts falling off more rapidly below 0dBm and -6dBm to -8dBm is approximately halving the output.

Even at +6dBm the output is still 10dB down, however, it is difficult to prove that the fault is the mixer as the 200MHz IF will be detuned by the low loss probe (5pF). This is where I could do with the chinese log detector as the intermediate levels are too low for my scope particularly if I use the probe attenuators to lower the input capacitance. The new probe is supposed to be 1pF.

PS: The PP measurements above are readings on the scope but the actuals will be much larger due to the scope bandwidth limitations.

[Radio Wrangler]

If you have a sig gen that covers the frequency range and has a trustable output level, why not use it to shove a signal into the IF at the output of the mixer? Use the analyser to measure the sig gen to see where in the analyser chain it gets things wrong

David

[PJL]

Good plan.

The analyser only goes to 110MHz but I can put 200.2MHz in to the IF and see what level is required to get the -10dB output from the tracking generator. It is supposed to be -20dBm according to page 112. That would help narrow it down.

I don't appear to be getting -20dBm out of the mixer but this could be the probe capacitance. The mixer is driving a 120ohm resistor and the probe 5pF=150ohms @ 200MHz. As mentioned, this is where the tools are causing the problem as I can't use the 10x attenuator as the signal would be too small to see on the scope.

Still no log detector in the post!

[PJL]

-20dBm does get the tracking generator output up close to -10dBm although it seems to be frequency dependent and the counter sort of works but probably needs the threshold adjustment calibrated. At -15dBm everything seems more reliable but this maybe down to panels being removed. In general the whole unit seems a bit marginal (for example the low LO outputs above) and I am wondering if it needs a more general restoration.

I am going to put it on hold till I get the log detector and can do more accurate amplitude measurements. If it still looks like some stages are marginal, I will try rebuilding one with new parts. To get it back to proper working condition may end up being a major project.

[Radio Wrangler]

reopened at PJL's request

David

[PJL]

Dave, thanks for reopening. It has sat in bits on my desk for 2 years so I now either need to fix it or get rid of it.

Recap:
- It used to work but the noise floor was bad.
- It stopped displaying the frequency.
- I established the amplitude of the signal out of the oscillator was too low to trigger the frequency counter.
- I didn't have the means to test amplitude of the high frequency (200+MHz) signals to locate the fault but improvised using a FET probe and a 150MHz scope.
- This gave an indication that the 236.4MHz signal was possibly low.
- Although not documented here, I believe I tried the +6dB from the 2022 for the 236.4MHz signal and it worked but I will retest.
- I ordered a better FET probe and a log detector which arrived (about 2 years ago) and I need to wire them up to resume testing.

I have one question, are the coax cables between the modules likely to have deteriorated? They are only short runs so can I patch in a bit of TV coax for testing?