Marconi TF2370 fault fixing

[PJL]

I have 2 faults:

1. The counter no longer works. It shows 0's all the time.
2. The noise floor is bad. It is not the pre-amplifier as the noise is unaffected by the Vertical Scale (gain) settings.

So far:
Checked the PSU voltages and they are all better than 0.5%.
Started following the fault finding guide User Manual P109 and AG1 pin 9 (timebase) looks about right but appears to have a considerable amount of jitter on negative pulse at slower speeds only. I will investigate later.

The fault finding chart then has me pulling a different board out for each step, but I will do as much as I can on each board starting with AG1 (I don't have an extender card). As it shows all 0's, it is either not getting an input to count or the control logic is not working.

In https://www.vintage-radio.net/forum/...d.php?t=109261 Peter Scott had a similar problem that turned out to be the MC10131 pre-scalar so I'll start there.

[PJL]

Not so easy. The input signal to the counter from the AC1 Tracking Generator is about 1/10th of what it should be. I will check the Tracking Generator output as that comes off the same board and check continuity of the connection between the RF and display units but failing that I will need to open up the RF unit. Looks like I may need some BFY90's as these seem to be their universal RF transistor. The connection is unlikely to be the issue as the amplitude is not frequency dependent.

[PJL]

Definitely the RF unit as the Tracking Generator output is also low, should be -10dB and it's about -26dB. They both come off the same amplifier so I might be lucky but if it's wrong out of the mixer that feeds the amplifier then I am in trouble as I don't have anything that will measure the amplitude of 200MHz+.

Anyone tried one of those cheap chinese AD8310 based power meters?

[PJL]

Could really do with some help now. It looks like the input to the amplifier (+20dBm) is low although measuring -30dBm (20mV PP) with a scope is not so easy.

The circuit for AC1 and AC2 is on p197 of the manual. The two 200+MHz signals that are outside my scope/probe bandwidth are large so they manage to push their way through so maybe I can roughly estimate their level by using the signal generator set to the same frequency and compare levels.

Fixing test equipment always requires yet more exotic test equipment. Options are:
1. I do have a 7904 (500MHz) scope but only 100MHz plugins so I could get a faster vertical amplifier and buy one of those cheap active probes from ebay.
2. Is to buy one of the ebay RF power meters. Do they work?

Any suggestions?

[PJL]

Not had much time to spend on this but what I have done is to get my TK2374 low-loss FET probe (200MHz) working into the 150MHz Tek scope. The response will be far from flat but I have a TF2022 siggen that I can use to create an approximate compensation table.

This should leave me in pretty good shape for fault finding.

[Biggles]

I am not familiar with this particular unit but any Marconi test gear that exhibits low stage gain could point to a failed OM345 chip. This is quite a common problem, but again, I'm not sure if yours uses these. Just a thought.
Alan.

[PJL]

It's not going well! The relevant circuit is attached:

Inputs to 'input balanced mixer':
36.4MHz @-20dBm measures 36.4040 @-20dBm
236.6+150KHz @+10dBm measures 236.597 @+4dBm
The input at the bottom right is
200-310MHz @+10dBm measures 224.485 @+8dBm

The output of this unit then goes to a filter and amplifier feeding the counter and the Tracking Generator Output.
-4dBm into 50 ohms on the RF circuit diagram but I measured it connected to the display unit @-18dBm.

All dBm measurements were made as PP on a scope then compared to the output from a TF2022 set to the same frequency to convert to dBm.

The TK2374 probe is specified as 100Kohms 5pF. The probe is rated to 200MHz and the Tek 465 scope to 150MHz. At 236MHz 290mV PP equated to 4dBm which should have been 1V so attenuation is about -11dB.

The TF2022 is not calibrated and working off a scope trace is not exactly accurate!

It was working when last used a few years ago and has been stored inside the house. Diagnosing an amplitude fault without calibrated test gear is difficult. At the moment it looks like the 236.6 input from the 2nd local oscillator is down by 6dB but it could easily be the probe capacitance.

Another possibility is the input balance mixer is dead. Although a single sided PCB, it is a separate part number and contains 4 diodes that I assume are carefully matched?

[PJL]

For anyone interested in helping diagnose this fault, the manuals can be found on BAMA, http://bama.edebris.com/manuals/marconi/tf2370/

The schematic is p197 in 'Radio Frequency Schematic'
The fault finding guide is p112 in 'User Maintenance Manual'

I have not spent much time on this yet and most of that has been spent on working out how to get reasonable measurements including how to balance all the kit on the bench so I can do those test.

I broke the cardinal rule last evening and read through some of the manual! I had been working off the schematic and need to run through the newly found fault finding guide.

[PJL]

Attached is a picture of the area under investigation. The balanced mixers are in the left hand unit (AC1) top slot (input) and bottom slot (output).

[PJL]

Found the parts list for the balanced mixers and they use a 1N5390. Took a bit of time but I found the Fairchild datasheet for this 'Hot Carrier Diode' (Schottky these days).

[Radio Wrangler]

Right, let's take a look at this analyser overall

RF comes in at 0-110MHz it meets a programmable step attenuator and a lowpass filter before a wideband preamplifier.

It then gets mixed with a local oscillator signal of 200-310 MHz to mix the signal up into a 200MHz fixed IF.

In the 200MHz IF it gets three stages of amplification and filtering before it goes into the second mixer where a 236.6MHz second LO mixes it down to a 36.4 MHz second IF. That 234.6 LO has +/-15-kHz on it, so I guess it gets used for fine tuning and for fine frequency sweeps.

But hello, have I misread anything. Mixing 200 MHz IF with 236.6MHz isn't going to make 36.4 MHz! maybe 200MHz IF is just a name and it really runs a couple of hundred kHz off of 200kHz? Not important at the moment.

Anyway the 36.4 MHz second IF gets mixed down to the 3.6MHz third IF and then mixed down to the 400kHz fourth IF, and it may get mixed even lower to make the narrowest IF bandwidth filters possible.


Meanwhile there is another signal path running in the reverse direction for the tracking generator. But instead of starting with an oscillator at the lowest receiver IF and having a chain of mixers and filters all the way up to the tuned frequency, it looks like they start with a 34.6MHz crystal oscillator somewhere. This is the thing with the +/-50Hz spec on it, I think. Don't worry about that degree of precision, You can twiddle it to match the receiver by peaking the tracker on your narrowest resolution filter. No added test gear needed for the precision adjustment, but you go need to check it is present, close to the right frequency and at the right level

The tracker mixes this 36.4MHz signal with the 236.6MHz +/-150kHz second LO to make a signal at 200 (ish!)MHz which gets a 200MHz tuned 3-stage amp then it hits a mixer where the first LO mixes it down to the receiver's incoming tuned frequency. Job done.

The counter counts the tracker to tell you where you're tuned.


Now, your receiver isn't good with a bad noise floor.
Your counter isn't working properly
and your tracker output is weak

The weak tracker will cause the bad counter and counter waveforms will look jittery. We need to fix the tracker before the counter

The receiver and the tracker don't share any signal path hardware.... but they do share LOs... the 236.6 MHz (+/-150kHz) one and the 200-310MHz one.

Low power on one LO would make the tracker weak AND would make the receiver path noisy.

Ah. now that fits.

It's going to be very difficult working around a lack of gear that covers the frequency range. Frustratingly, I've got analysers to 22GHz, power meters to 40GHz and a 1GHz scope here, but it's a long walk.

David

[PJL]

David, thanks for looking into this. I need to do a few methodical tests but I have a bad feeling about the input balanced mixer. The 36MHz input looks good, the 236MHz input maybe 6dB low but I am seeing very little on the output from the mixer.

I am not sure how badly the 5pF probe capacitance would hit this and the scope can only do 10mV/div at 150MHz otherwise I could use the x10 probe attenuator. Of course it would also help if I wasn't losing 11dB from bandwidth limitations.

How easy would it be to throw together a detector and put the output of the FET probe (50ohm) through the detector and on to a DMM?

I did check the main power rails on the display unit but also need to check local DC conditions as they have been neglected in my 'enthusiasm' to sort out how to measure the RF levels.


Having the right test gear would make it far too easy

[Radio Wrangler]

The throughput of a diode ring mixer drops quickly as the LO drive level is reduced.

The diodes should have been a matched set of schottky diodes. If you need to replace them, go for the 15v rated parts (in the diode ring, each diode can only ever get reverse voltage equal to the drops of the two diodes in the forward path.

Matching isn't terribly important. Parts from the same wafer are usually fine. If you feel bothered, get a dozen diodes and use the diode junction voltage test on a DVM to choose two quads of similar parts and then you have parts for both mixers.

David

[PJL]

I have been avoiding disturbing anything so far, but one option would be to use the TF2022 to serve as the LO. Unfortunately, it is the older model and will only work up to +6dBm into 50ohms so not the +10dBm marked on the schematic.

[PJL]

The output from the input balanced mixer is shown below. Does this look right (photo could be better!)?

A look at the levels at each stage in the 200MHz IF (L1, L2 and L3 taps) and I get the correct gain between stages but levels are all down about 9 to 10dB from those specified in the the repair guide.

I have verified my counter against the TF2022 and if left to stabalise for 30 minutes (it has a crystal oven) it matches the setting on the available 6 digits. This would imply the 2370's 36.4MHz crystal oscillator is out by 400Hz so well outside the +/-50Hz it is supposed to be.

I am not keen on twiddling as this thing is so complicated but it is possible that calibration might fix this fault.

[PJL]

My understanding is this is a double-balanced mixer and is supposed to reject the signal and LO but it clearly isn't rejecting the 36.4MHz so something is wrong? I have 5 schottky diodes on order that are a close match to the original with low leakage and capacitance (surface mount types!).

Can someone confirm this conclusion as I don't want to hack around with the mixer if it's not the problem.

PS: Having thought this through, the trace is misleading due to the scope bandwidth limitations.

[Radio Wrangler]

The scope trace would depend a lot on the harmonic structure of the LO and their relative phasings, so the that and the unmentioned scope scalings make it impossible to say whether it is good or not. I've always probed these things with a spectrum analyser to make sense of them. When I have looked with a scope, the pictures don't give an easily interpreted picture of what is going on.

Double balanced mixers only cancel LO and RF going through to the IF port if they are perfectly balanced AND the LO signal is perfectly balanced. With a purchased mixer, you might get 20-30dB of rejection of LO, with a DIY job like Marconi used you might not see that much. The LO power has to be an awful lot begger than the intended signal power, so even with reasonable balance and therefore rejection, there will be a lot of LO coming out of the mixer (out of BOTH other ports!) Good design is to not reflect it back in, it only gets up to mischief. There's a paper on high dynamic range mixer/receiver design in HP Journal April 1982. Guy Douglas goes int the details of a receiver with a very high dynamic range diode ring mixer, and I go on about the LO needed to drive it.

David

[PJL]

I will fix the 36.4MHz oscillator and move on to the 2nd local oscillator. It is marked as 10dBm and the repair guide suggests a minimum of 7dBm so my 4dBm looks low.

[PJL]

The 36.4 MHz oscillator is derived from:
10MHz (master crystal oscillator) > 80MHz (crystal oscillator phase locked) > 40Mhz (/2) > 4MHz (/10) > 500KHz (/8) > 100KHz (/5) > 36.4MHz (crystal oscillator phase locked)

10MHz reads as 10.0000
3.64MHz reads as 36.4040

Something is not right! I presume one of the crystal oscillators is not locking correctly. I'll fix this first as I am not sure what impact this has.

The 1st LO (200-310MHz) and 2nd LO (236.6+/-150KHz) are used to set the frequency and are phase locked. The sweep is applied to the 1st LO if the range is set to MHz and 2nd LO if set to KHz.

[Radio Wrangler]

Crystals age and exhibit a frequency drift which is asymptotic to a long-term final value.

VCXOs are a right b*gger to design to get enough pulling range.

Pulling range for a VCXO diminishes as the square of its overtone number, so if yo had two VCXOs on say 25 MHz, one using a fundamental mode crystal, the other a third overtone mode crystal, you will find the overtone mode one pulls only about a ninth as far as the fundamental one. This is enough in many cases to make it impossible to phase lock some crystals to the frequency stamped on their can! Same goes for crystal oscillators with ordinary trimmers.

David