Spectrum Analyser TR4172

[Radio Wrangler]

Originally Posted by Matt kd4pbs

Lex,
I was in the process of chasing down gain at each stage in my 4172, and thought of you... does my "Point L" look like your "Point L"? I bet it does.
Looks like there might have been a common issue with these, or perhaps this is part of what TR/Advantest did to calibrate these?
Attachment 229614

I rather like the DIY diode ring mixer in the photo. Doing the transformer for one end with trifilar wire in a 2-hole bead and then anchoring it all with a toko style can is neat. HP used toroids on top of little 6-legged bases and glue.

David

[SATOLEX]

Originally Posted by Matt kd4pbs

Lex,
I was in the process of chasing down gain at each stage in my 4172, and thought of you... does my "Point L" look like your "Point L"? I bet it does.
Looks like there might have been a common issue with these, or perhaps this is part of what TR/Advantest did to calibrate these?
Attachment 229614

Yes, my Point L looks exactly the same (indeed an TR/Advantest TP). I had the Idea to cut it here to perform a measurement on the second RF down conversion part or make a RF probe with some SMD components (56 Ohms and 1K). Unfortunately I have some other things to do at this moment so this has to wait. Please keep me informed.
Regards, Lex

[Matt kd4pbs]

It appears that they have cut the trace at the factory and re-soldered a jumper back in on mine. I was unable to get any work done on this over the weekend due to other projects. So far, I've tracked the gain/loss in each stage from the output of the 1st mixer to the input of the frequency compensation/gain adjusting amp stage, and each stage so far is within +/- 1dB of what the service manual indicates. I am hoping to find the issue within there.
David, as you know, these were some interesting times for the Japanese engineers. Late 70s/early 80s was when Japan was finally becoming a major player in the electronics game, and "Made In Japan" was finally removing the stigma it held since WWII. Some of their work was truly innovative, like this transformer. The ring mixer design is pretty decent considering the lower IF frequencies used here. Other parts of this, like the inadvertent use of highly corrosive adhesives in very sensitive areas, and also like the huge 1K resistor tacked in the output of my first mixer are a bit, as we say in the states, "Ghetto".

[Matt kd4pbs]

Something is quite fishy about this amplifier section after the 3rd mixer. With the -20dBm test signal feeding the analyzer input 1, I followed the signal through the output of the first mixer up to the input of this amp. I see about -14.6dBm on the input to Q19, the first transistor in this amplifier with my high impedance RF millivoltmeter probe. This works out pretty much inline with the gain/loss up to this point. According to the manual, I should see 14dB gain after this amp, which appears to be a frequency compensation amplifier as well as being a broadband amp. At least that's what I'm assuming the manual means by, "30MHz f. Gain collection Amp.". Regardless of what exactly it is, I see very little gain, if any, between the input of this amplifier and point M on the schematic, the input to either of the two RBW BPFs. I don't know if my gain issue is due to the output being loaded by a faulty component in the BPF switching network or due to a bad transistor. I do know that Q20 runs warmer than Q19, but looking at the voltages on the transistors, it appears that at least at DC, neither Q19 nor Q20 are bad. I will mock up the circuit in Multisim today to see what I get at DC levels.
Making progress...

[Matt kd4pbs]

I'm a bit befuddled here now.
Has anyone ever experienced a transistor failure mode in which it's DC operating characteristics look absolutely normal, but exhibits a loss of gain at RF levels in which it was designed to operate? I modeled the whole Q19/Q20 circuit with the opamp and Q21 section (seems to be operating as an AGC) and my DC measurements are within a few millivolts of what I see on Multisim. I see virtually zero gain, if not perhaps a bit of loss while probing with my RF millivoltmeter's high impedance probe, but in Multisim I see at least twice the voltage going out as going in, so it should be about 6dB of gain at least. This is odd as well, considering the manual states that this amp should provide 14.2dB of gain. Normally, I'd just wholesale replace the transistors, but since this board is such a pain to remove to replace them, I'd rather be a little more conservative in my method.
Thoughts?
Regards,
Matt

[SATOLEX]

Hi Matt, what I understand is that if you measure at the 51 Ohm resistor at the input (R219) and the output point M than you have no gain. The DC bias point looks ok and is comparable with my measurement #137. I assumed that this is the controllable gain amplifier to calibrate the RF chain with the cal. screw on the front (the 5 dB). In my case the LEVEL CON voltage changes from 2.22 to 6.12 if I turn the screw on the front (according schematic it should be 5.5 V ± 3.5 V). This voltage is going to the OPAMP input. Maybe it is the (wrong) loading of the switchable BPF but I assume that it is a 50 Ohms filter if it is switched correctly. I hope I have some time left this weekend to do some measurements. In my case I think there is also something wrong with LO2 but I can look to that at a later moment. Did you already check the gain from the input of the amplifier (R219, 51 Ohm) to IF OUT. In your case this should be 9 dB less. It is best to measure IF OUT with an analyser because the LO4 feedthrough is high for what I have seen.

Regards, Lex

[G0HZU_JMR]

Hi Matt, I've added a wider view of the schematic below. I think the purpose of the complicated opamp circuit on the right may be to try and linearise the operation of the gain control to make it nicer for the operator to use. I think diode D57 is there to act as a non-linear resistor and its resistance is set by the current though it and this current is contributed to by the opamp+Q21.

The resistance of D57 will affect the gain of the amplifier Q19+Q20.

Normally, the resistance characteristic of D57 would be very abrupt but I think the idea is to control this better.

If the gain is low then maybe there isn't much current passing through D57 or maybe D57 is faulty.

I'm not sure what type of diode it is but if it's a silicon diode it will probably have a resistance of 26/Ibias so if it had 0.5mA through it the resistance might be 52ohms and this would give your circuit about 12dB gain.

I think the minimum current will be =(15V supply minus a diode drop)/27k =0.5mA.

The opamp probably just tops this up with a tiny amount more current through D57 to get the expected 14dB gain. So I would check out the health of D57 and maybe check the DC voltage at R154. it looks like there should be about 14.3V here.

If D57 dies open circuit the gain will drop to less than unity.

[Radio Wrangler]

Q21 with the diode networks in its emitter is an amalgam of a "piecewise-linear" shaping network and a current source. The current is pulled from the 15v supply via D57 which is most probably a PIN diode and this is its DC bias feed.

D57 handles the emitter RF current of Q20 via DC block C317 and away to ground via C320

D58 probably gives diode drop compensation for Q21 Vbe

David

[G0HZU_JMR]

Yes, I just had a look and the manual states D57 is a 1SV35 silicon diode and it's likely to be a PIN diode based on a quick google. I couldn't find a datasheet though.

A cheap (sub $1) and available equivalent might be the Vishay BA479G PIN diode. This has an ideality factor close to 1 so it should have a resistance of 26/Ibias(mA).

Mouser have loads of them in stock at about 35p each here in the UK and the USA.

https://www.mouser.co.uk/datasheet/2...9g-1767857.pdf

As R154 (27k) appears to bias D57 with a minimum current of 0.5mA then the resistance using a BA479G will be 52 ohms. See the BA479G datasheet graph below.

If this gives too much gain on the lowest setting of the gain control then I suppose the 27k resistor could be increased to 33k and then retry the circuit. This assumes D57 has failed open circuit and needs to be replaced.

I don't think the op amp can quite reverse bias D57 if something has gone wrong with the opamp section and it slews to the positive supply rail. However, just to make sure I'd check the opamp/Q21 just to make sure it isn't slewing high enough to start to rob D57 of current.

[Matt kd4pbs]

Thank you kind sirs!
Yes, Lex - that's correct. I do have a little bit of gain, but only just a dB or so. It can vary via using the R151 control on the RF board, and will also vary using the front panel gain control. The gain appears to be generated with Q19, as if I probe the collector of Q19 / base of Q20 junction, I see slightly more level coming out than going in, variable with R151. Once I probe the collector of Q20/point M though, it really gives little gain if any, variable with the front panel level control over a range of 5 dB or so. I verified what I saw on the RF millivoltmeter with a lesser but still capable SpecAn on my Motorola R2032. The 30MHz signal is indeed not being amplified very greatly. Since I don't have a high impedance probe for the R2032, I simply utilized a 100MHz 10x scope probe since I was not worried so much about an accurate value. just a comparison between input and output. I also utilized the procedure from the calibration section of the manual to adjust the two RBW filters to examine gain through these two transistors, and the results were the same; nowhere near the 14dB or so gain that the manual states it should be capable of. I did not check the gain from R219 to the IF out.
Josh, I concur that the IC1/Q21 circuit serves to linearize the level control, like David wrote, by applying a variable slope depending on the current through Q21, the shape of which is set by the three diode/voltage dividers on it's emitter. I'm just confounded by why they would use an IC to drive Q21 instead of a simple voltage divider though; this is why I assumed it to have some sort of AGC function as well.
Thank you, gentlemen. I will interrogate this circuit more closely in the next few days and report back with my findings.
73,
Matt

[Matt kd4pbs]

For the record, the 1S35 is indeed a PIN diode. Some internet sources incorrectly cross it to a 1N4148/1N41450/1N914.

https://drive.google.com/file/d/1e21...ew?usp=sharing

[Radio Wrangler]

The impedance presented by the base of the transistor will vary as the different diode-switched networks come into play, and the circuit's a bit easier to design with the base driven from a low impedance source.

The signal amplifier needs to be able to produce some gain. RFery is usually in a relatively low impedance environment so that PIN diode in the emitter needs to run down to fairly low resistance values, so the bias current does need to reach a good few milliamps, so the emitter network of the piecewise-linear transistor is lowish impedance, exacerbating the base loading on the control voltage, and opamps were starting to become cheap commodities.

There's my take on an adjustable piecewise shaper that can both increase and decrease slope in the HP Journal April 1982. I didn't think of it at the time, but I should have patented it. In this case it was linearising a swept VCO and needed to be quite accurate.

David

[G0HZU_JMR]

I suppose if the gain control is actually doing something sensible in terms of adjustment range then maybe D57 has to be OK? The other thing that could kill the gain would be a cap in the amplifier stage that has gone high impedance. There could a dry solder joint at one of the caps for example.

[Radio Wrangler]

Might be time for trying a bit of freezer spray?

Yes, transistors can have their RF gain and noise ruined, with seemingly lesser effect on DC conditions. It's one of the symptoms of something that has had vbe reverse biassed into avalanche, but not with enough current to completely kill it.

A decoupler/blocker gone higher resistance than normal in this circuit could spoil the gain.
So Q20 is running 36 mA and Q19 8.4mA accorging to the emitter voltages and resistor values. 1.8v drop on 51 Ohms says 35.29mA which fits.

The circuit looks a bit naive. There seem to be no anti-parasitic oscillation components which I'd expect if some lively devices were used with plenty of available gain at RF. However, this is a relatively early Japanese instrument, and the Japanese were evolving a design/manufacturing style of carefully optimising component choice to minimise component count.
Around that time I was working at HP and we had issues with finding some, er, excessively familiar circuitry in competing instruments from a Japanese firm, though not T-R and in an instrument a lot more specialised than spectrum analysers.... particularly in some areas which were the work of another forum member. I think at the time he said he felt 'sincerely flattered'. Did I say he has a dry sense of humour?

Anyway, it was a time when Japanese test gear manufacturers were just moving into more complex instruments that basic sig gens, and they were feeling their way around and learning.

If you swap Q19 or Q20, you may have to add some parasitic stoppers.

David

[G0HZU_JMR]

I had a look at the IF amplifier circuit in a bit more detail and I'd be nervous to change the transistors Q19 and Q20. I think they are both 2SC1426 BJTs.

Although this circuit only runs at 30MHz it looks like they have chosen to use a decent BJT device in terms of signal handling and noise figure. The 2SC1426 is suitable for wideband RF amps for example.

There's no feedback so I think the design partly relies on the quality of the 2SC1426 to minimise IMD distortion. The diode D57 needs to behave as a resistor to preserve linearity too. The BA479G diode would probably be OK here.

The amp sits just after a diode ring mixer and it is probably a high level mixer. The OIP3 of the mixer is probably +11dBm. I simulated the amplifier using BFR106 devices although the BFR106 is just a SOT23 device. The amp gave an OIP3 of +28dBm when I adjusted the gain to +14dB.

With the 51R termination resistor and the series 33R at the input to the amplifier the simulator predicts a noise figure of over 11dB. This seems a bit poor for an IF amplifier stage in a spectrum analyser. It looks like the designer wanted to provide a very well defined resistive termination at the input and the output and this seems to be at the expense of noise figure. I suppose these resistors will help with stability as well.

[Radio Wrangler]

The 51R resistor will keep the mixer happy, broadband, just so long as the following transistor Q19 doesn't load it much.

This means that the majority of the power from the mixer winds up in the resistor, and that the power gain demanded of the two transistors is rather large. This is not a good move in terms of local noise figure, but OK so long as the gain in earlier stages makes it relatively unimportant.

Using a high-Z probe may show that there isn't much voltage gain going on in Q19, Q20, but the input impedance arrangement may mean that there is plenty of power gain just to keep the voltage similar while the impedance level changes significantly.

David

[Matt kd4pbs]

From what I see, David gets the beer on this one. It's lots of current gain, and my RF millivoltmeter isn't seeing this fact. The PIN diode seems fine. I can adjust the gain through this amp using the gain controls as expected. This amp section was a distraction. Time to move on down the line.
That being said, thank you guys! I am WAY out of my league here, but learning tons from some of the most intelligent minds on the subject. I am VERY grateful every time something like that has happened in my life. Like my knowledge of all of the high power triode, tetrode, klystron, and IOT amplifiers I have maintained over the past 33 years, while minuscule compared to those who I helped at the time, eventually rubbed off, and I thank you gentlemen for smearing a bit of that knowledge on my synapses!
73,
Matt

[G0HZU_JMR]

If it helps, the manual states in figure 10.10 that this amp typically has a gain of 14.2dB at 30MHz in a healthy TR4172. It is designed as a 50R and 50R out amplifier stage and I think it is possible to demonstrate this with a very simple analysis.

The input impedance is set by the 51R shunt resistor as the input impedance of Q19 will be much higher than 51R. So there will be about unity voltage gain at the base of Q19.

Q19 has 560R at the collector and maybe 56R RF resistance in the emitter so this might deliver x10 voltage gain at 30MHz at the collector of Q19. Q20 may have about 50R RF resistance in the emitter assuming 0.5mA in D57 and the collector load is 51R || 50R = 25R.

So there might be a halving of voltage at the collector of Q20. So a combination of x10 and x0.5 means a voltage gain of 5 which is about 14dB. In reality, I think the circuit will need some more current in D57 because I think it will struggle to give a flat gain of 14dB out to 30MHz.

[SATOLEX]

Hi, I have found the loss problem in the RF part . After measuring with a HP RF probe at several points in the RF chain I discovered that I had a lot off loss after the 206 MHz band-pass filter, point L in the schematic. At first I did not find anything but after a closer look I discovered that two of the four inductors did not have a good ground connection anymore (broken), see picture. So Matt, check your 206 MHz BPF filter! I have also measured the 30 MHz VGA but I also concluded that the impedances are not equal and probably also not 50 Ohms so you can not measure it correctly with a high impedance probe. For the gain of the 30 MHz amplifier I have measured about +2 dB but because the gain control is working correctly I assume that this stage is ok.

Next, I will solder the inductors and have to trim the filter again. It will be critical because the bandwidth is only 7 MHz. Probably they have bent the inductors to tune the filter because the shape is not exact a spiral anymore? Maybe that is the reason why it is broken. Even one inductor is touching the side wall? It looks like that somebody has already tried to repair this in the past. For me it is clear why TR made the L test point as is. I do not have a lot of time at this moment but I will keep you informed.

Regards, Lex

[SATOLEX]

Sorry, forgot picture.
73,
Lex