Spectrum Analyser TR4172

[G0HZU_JMR]

Looking back at my repair notes I may have actually measured the conversion loss in my mixer at some point because I have a gain loss spreadsheet in the folder. I think the conversion loss is frequency dependent because there is an RC pre-emphasis network at the input to the mixer. It looks like I saw 13dB conversion loss here but I was measuring at an RF input frequency of 50MHz where the RC network will be lossy. My RC network shows a 68R resistor and not the 39R in the manual and this is the case for both of my TR4172 analysers. This will degrade the conversion loss at LF.

I looked at my old repair images and I think my 1st mixer device 'must' have been replaced at some point because the gold legs on the quad package still look shiny and clean. Also the soldering around it looks a bit clumsy.

Looking at my old images there are some scary 'scabs' of corrosion all around the insides of this mixer module so I assume this module has absorber stuck in the lid (can't remember). Therefore, it could be that one of the connection legs of your first mixer has finally crumbled and broken possibly due to the simple act of setting the module down which might be enough to finally kill it off. How much corrosion is there on the four legs of the diode mixer?

I suspect that my first TR4172 succumbed to the dreaded absorber glue in the first mixer many years ago so maybe the corrosion happens within the first few years? I think I would remember if it had ever failed at work so it may have had the mixer repair back in 1989/90 before it arrived at the company.

[Matt kd4pbs]

I will do that tomorrow Lex.
I was able to fix my keyboard issue. It was what I call an ID-10-T error. I have also been noticing that the analyzer will simply lock up with no reason, causing me to need to do a master reset to recover. I find that this is most prevalent when narrowing the bandwidth. I also notice that on very narrow RBW settings, the signal of question, be it an external oscillator or the internal 50MHz source, it will jump around in position quite a bit. Sometimes, on wider bandwidths, just pressing the CENT FREQ button will cause it to shift as well. In order to attempt to overcome this, I re-seated all of the boards in the display section. Well, evidently I also accidentally switched one of the DIP switches on the GPIB/IO board. These switches turn on or off different IRQs to the CPU, and I had accidentally turned off the one for the keyboard. Live and learn.
It's good to know that the issue is related to incorrect levels somewhere. Even though I have lied to the analyzer to force it to measure correct levels by raising the 1st IF output level, it still skips and locks up, so if it is skipping due to incorrect level somewhere, then it is likely a low LO level someplace else. Perhaps you and I have the exact same issue. Well, other than the ID-10-T issues I have to deal with

[Matt kd4pbs]

Incidentally, after checking some date codes on some components, I did not see anything newer than '83. So for the record, this is a Takeda Riken branded TR4172 likely made in '83. It finally succumbed to the lead rot in the 1st mixer in 2021, just as Jeremy described (see post #35 for the ugliness). I was investigating the major loss in sensitivity a few weeks ago and this is exactly what I found. Thankfully I was able to perform major surgery on the mixer and repair the problem. It seems mechanically solid now, and is certainly working electronically, but I don't know how long that will last. Hopefully at least another 38 years. My guess is that it isn't tuned as well as it was originally though... but I am not in need of perfection.
73,
Matt

[Matt kd4pbs]

Jeremy,
I was perusing the manual on this to see if I could gain some insight as to where the loss in gain could be and stumbled upon this. It is with respect to your mentioning that the trace is way too bright when used in VNA mode...

(7) Other features
Each operation of the []T (CONTRAST) key increments only the
intensity of the displayed impedance response trace or the
graticule. The character information readouts remain at the
same intensity. Operation of this key first increases the trace
intensity in four levels; if the []T (BRIGHT) key is pressed
a fifth time, the intensity returns to the original level.
Next, operation of this key increases the graticule brightness.
This trace and graticule intensifying feature is convenient for
highlighting the impedance response trace for photographing, or
other occasions.
Operation of the []H (HELP) key provides a listing of the
special key functions used in the impedance measurement mode on
the display (see Figure 9-25.)

Regards...
Matt

[G0HZU_JMR]

Thanks! It's a few years since I last tried to use the TR4172 in impedance mode with the smith chart but at the end of the calibration the display shows the expected trace dot at the centre of the smith chart. However, this tiny dot at (50 + j0) is so insanely bright anyone would be scrambling for the intensity button and praying the CRT wasn't already burned. Luckily I was quick enough to avoid this problem.

[SATOLEX]

I have measured the mixer separately and it worked fine. It has a conversion loss of about 8 dB if I apply a 800 MHz signal to the RF input. I have also remeasured the IF1 and observed the same interconnection issue. Outside the analyser it was working fine. I have opened the mixer to see what the condition is. It has the same issue as the mixer of Matt, lead rot. What do you think, do not touch it and wait until it falls of or clean it? I saw a note on the cover that the diode was replaced in 1989. It would be wise to buy a spare diode but what Jeremy mentioned is that it is hard to get one. The 39 Ohm series resistance looks strange, what do you think?

Regards, Lex

[Matt kd4pbs]

If it's working, I think leaving it alone is a good idea. Mine suddenly went to about 40+dB deaf, and that was from the lead finally giving away. I had to literally scrape away some of the black epoxy (NOT the same stuff as what's on the 1st IF Amp!) to expose enough metal to be able to solder to it. It was NOT at all a fun job!
This mixer does indeed seem to be unobtanium. I made an inquiry with a company in the states about getting one from them and after I answered all their questions to satisfy them I wasn't building a thermonuclear detonator, they ghosted me.
Looking through my 1st mixer input attenuator, the two shunts measure 147.44Ω and 147.9Ω, and the series reads 16.325Ω. As Jeremy pointed out, that's about right for a 3dB pi attenuator. Odd that yours is much higher. I wonder if it got toasted?
I am digging for the 1.84G post-filter connection with my NRP18T at the ready... Hopefully I will be able to find a convenient spot to ready this.

[G0HZU_JMR]

Yes, probably best to leave it if it works and try and find a replacement in slow time. See below for a zoomed in image of the mixer in my first TR4172. Almost certainly this has been replaced at some point because it looks so clean.

The soldering around it looks poor so I'm not sure this was replaced by someone skilled in soldering! This image is in a folder dated "2015 repair" so this is how it looked 5 years ago. I'm not sure I ever looked at the mixer in the other TR4172. The chances are it will look a corroded mess if it is still the original mixer.

[Matt kd4pbs]

Well, that explains that...
The 1.85G filter is on the PCB on the RF block without connections into and out of the filter. It just uses the other branch of the Wilkinson splitter and connects directly to the filter with a stripline.

Edit: I don't see where to measure the output of this filter. Is it buried down deep?

[Matt kd4pbs]

I think I need to locate some extender cards to use in calibrating this guy. Even though I still have to find the ultimate reason why it's about 9 dB deaf, there is still a need to calibrate. Anyone have any great ideas here? I have a bunch of extender cards... somewhere. In case I can't find them, or can't find what is needed, as anyone investigated what is required? I know there's a need for two 28x2 cards and I think a 22x2 card. Any other ideas on this?
Regards,
Matt

[SATOLEX]

Matt, I have measured the other branch which is going to the T.G. to get a feeling about the power level out of the 1.84 GHz generator. It is most convenient to measure it at the output of the external 1.84 GHz BPF which is going to the tracking generator, see picture. The power out of this filter was about 0 dBm. This seems somewhat low based on the figures mentioned in the schematic. I have also added the response of the external filter. It is hard to measure inside the RF module. In mine I have seen that somebody already tried to measure the second mixer and the rest of the RF down conversion. They have cut the line at L, see schematic.

I saw that you already have your RF part open so it is easy to measure the RF gain adjustment voltage range (screw on front), see schematic. What is your gain and voltage adjustment range? I forgot to write the gain down but I think it was about 5 dB and 2.22 - 6.12 V.

Regards, Lex

[Matt kd4pbs]

Sorry, I see what I did there... That's why I couldn't find it. I thought I remember seeing it before, but now I have some stuff in the way since I ditched the amp and filters in the IF interface...

I'll get back in there later today. I'll also check the gain adjustment.

[SATOLEX]

I have checked the RF gain cal. adjustment again and in my case it is indeed 5 dB. I also checked the RF gain again and I’m sure that the gain loss of the RF part is 14 dB, more than yours. Actually it is even more because the cal. screw is at max. This is based on the level diagram in the manual. The RF part is pretty narrow-band so the input frequency has to be precise. This gives in my case an IF out max at 3.34 MHz. The place where they have cut the RF circuit was not clear in the previous picture, see attachment. Before I break the circuit I want to be sure that the LO levels are correct and the gain adjustment is work as it should be.

Regards, Lex

[Matt kd4pbs]

Here's what I found:
Output of 1.85GHz 2nd LO at input to filter: 11.99dBm
Output of 1.85GHz 2nd LO BPF:8.9dBm
Filter S21:


My filter response is a bit different than yours, I see a bit less loss, and it looks like my 2nd LO is considerably higher than what you see. Since the 2nd LO should not be affected by the 1st LO level, I'm sure my excess gain I made there has nothing to do with the higher level.

My front panel CAL control can bring the -20dBm 50MHz signal from -16dBm (full CCW) to -21.1dBm (full CW) at 1MHz span, 10KHz RBW.

Hope this helps!
73,
Matt

[6AL5W-Martin]

i have a service manual for 4122, if that can help let me know.
Martin

[Matt kd4pbs]

After much weeping and gnashing of teeth I think I found an extender card solution. I was able to locate a 22-position x 2 row (44 total) extender card on that auction site for a very good price. In fact, if one wishes to buy a new one, Vector still makes them.
That left only the double 28 position x 2 row (56 total) extender card problem. I would have been happy with just two separate ones even. I could find .15" and .1" pitch 56 position extenders, but was coming up short with finding two identical .156" cards. After some google-fu, I discovered something called a JAMMA connector, which is a universal Japanese standard for video game boards starting ~'85. The big auction site gave me sales for plenty of pre-made JAMMA extenders, with the caveat that they are all wired, not circuit board based. I figure I can just mount it on a piece of plexiglass (perspex) or something similar. Best of all, they are not too terribly expensive. I should be getting these in a couple of days, so I'll report on their effectiveness. From what I can tell, there are many adjustments that are likely needed on our old 4172s that require an extender card.

[SATOLEX]

Thank you Matt, this is very helpful to me. The cal. range is the same and it is clear that I first have to dig into the second LO generator. I did not yet open that part. It is clear that at your side the second LO driver is ok although Q16 can still be inactive. To find the RF gain loss problem flowchart 66 can be helpful. My paper version is slightly different, see picture. I have measured some DC voltages at amplifier stages and calculated the current based on the resistance values in the schematic. Maybe it is helpful to share them.

Q16, Ie = 12.8 mA
Q17, Ie = 25.5 mA
Q18, Ie = 17 mA
Q19, Ic = 8.1 mA
Q20, Ic = 34.5 mA
Q22, Ic = 21.4 mA
Q25, Ie = 22.8 mA

Thanks Martin but I already have found some manuals of the TR4172.

Regards, Lex

[Matt kd4pbs]

For the manual I have, chart 66 is the RF block.
Thankfully I have an oldie but goodie Boonton RF millivolt meter good to 1.2GHz with high impedance probe.
Alt least you're getting it narrowed down. Good job!

[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?

[Matt kd4pbs]

Wow - seven pages on a necrothread.
I want to take time out to give HUGE kudos, thanks, and props to the forum owner... THANK YOU! There is literally no other place, including Dave's EEV board on the internets, save for the manuals on Repeater Builder and BAMA that have much of any info on this old gal.
Hopefully these rantings help someone in the future looking to fix one of these.
Regards,
Matt