UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Specific Vintage Equipment > Vintage Test Gear and Workshop Equipment

Notices

Vintage Test Gear and Workshop Equipment For discussions about vintage test gear and workshop equipment such as coil winders.

Closed Thread
 
Thread Tools
Old 5th Mar 2021, 5:06 pm   #61
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 62
Default Re: Spectrum Analyser TR4172

I was able to check the resistances before this blob popped off. I imagine some of the traces under the blob remained intact, as most of what I measured jibed with what you found.

R1 2.17K
R2 393
R3 9.85K
R4 4.84K
R5 868

R5 is very far from what you saw on yours. Perhaps it is damaged? It does look suspect as if it were blued from too much power. I was able to get on both sides of it, as apparently the actual element is the short strip just above and center of R4. The transistor side of R5's lead is exposed well outside of the epoxy.
73,
Matt
Matt kd4pbs is offline  
Old 5th Mar 2021, 7:05 pm   #62
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Spectrum Analyser TR4172

I think 868 ohms is a bit on the high side as that would reduce the current to about 5mA. This would probably give a slightly lower noise figure (maybe 0.5dB lower?) but the signal handling of the amplifier would be much worse.

I just spotted that you have bought some 2SC5010 parts. I'm not sure if you realise but the 2SC5010 is a modern low noise RF part in a tiny plastic minimold package. It won't be a drop in equivalent in terms of the package size. I used it as a model in Genesys because it was the nearest fit in terms of the RF performance at 2GHz.

I think the original part could an older NEC 2SC2150 in SOT-173X package as mentioned back in post #20. This will have slightly worse noise figure than the 2SC5010 and probably lower gain by about 1dB at 2GHz. The 2SC2150 is used elsewhere in the TR4172 so this was my best guess as to what the actual device is.

I'd expect the 2SC5010 to work in this circuit but it won't fit the artwork neatly. It would be best to fit the emitter leg as close to an RF ground via as possible, There is also a slight risk it might be unstable. It's also only rated to 6V Vce so it would be a bit marginal in this circuit.

Hopefully your RF transistor is OK and it is just the bias board that is faulty.
__________________
Regards, Jeremy G0HZU

Last edited by G0HZU_JMR; 5th Mar 2021 at 7:19 pm.
G0HZU_JMR is offline  
Old 5th Mar 2021, 10:07 pm   #63
SATOLEX
Triode
 
Join Date: Feb 2021
Location: Leiden, Zuid-Holland, The Netherlands
Posts: 26
Default Re: Spectrum Analyser TR4172

Hi Jeremy and Matt,

Sorry, I missed the posts (I did not receive an email when Matt came in?). I repaired the IF1AMP bias circuit. The biasing of the RF transistor seems ok but I’m still missing 20 dB of gain. I discovered a second issue concerning the interconnection between the filter sections. Three of the four have a bad (no) connection (thermal stress probably). I will give an update on the resistor values and voltages tomorrow. About the blob, it popped off after a half hour of scratching with a knife .

Regards, Lex
SATOLEX is offline  
Old 5th Mar 2021, 10:48 pm   #64
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Spectrum Analyser TR4172

Gad to hear you got the bias circuit working Lex!

I can suggest what level of noise floor performance to expect from this analyser once it is repaired if that helps? I've been using my first TR4172 since about 1990 when it was donated to the company I work for so I know it well.

The specs in the manual are a bit conservative for the noise floor. With 0dB attenuation mine would usually deliver a noise floor of about -147dBm/Hz at 0dB attenuation with the preamp off. This would be for frequencies in the range 2-150MHz. The noise floor does gradually degrade a bit above 150MHz but not by much.

To measure this you have to turn on the noise marker to get it to give a corrected noise reading in a 1Hz bandwidth. I rarely used the preamp but it had something like 23dB gain up to about 1GHz.

When tested on its own the IF1 module should have a gain of just under 1dB at 2.05GHz and a noise figure of about 6.5dB. The output intercept point of the IF1 module is about +17dBm according to my old notes. Both of my IF1 modules drew about 16mA from the +15V supply.

For the complete analyser the input IP3 with 0dB attenuation is typically +22dBm across 2-100MHz. Very impressive!

The phase noise for the overall analyser is quite good and across 2-100MHz (from memory) it should achieve something like:

-104dBc/Hz at 1kHz offset
-106dBc/Hz at 5kHz offset
-107dBc/Hz at 10kHz offset
-128dBc/Hz at 100kHz offset (at 100kHz offset this is very low noise!)
-138dBc/Hz at 1MHz offset
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 6th Mar 2021, 1:15 am   #65
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Spectrum Analyser TR4172

I had a closer look at some of my internal photos of my IF1 board and I think the PCB material used for the filter sections must have a very high dielectric constant because the filter dimensions are really small for a 2.05GHz system. I estimate the dielectric constant of the material might be about 10. I also estimate the PCB height to be about 0.8mm.

This would make the 50R microstrip about 0.6mm wide and the square directional filter (diplexer?) at the input will be about 15mm wide. This looks to be about right according to the photos. This means that this is a very exotic PCB material. The nearest I can find to it is Taconic CER-10 at 0.03" height. We use this material at work and it is extremely expensive. However, we buy it as metal backed where the ground plane is bonded to thick metal and I think we buy it with a much thicker layer of CER-10. So this material might not be that expensive when only 0.03" thick with maybe 1oz copper.

I really must emphasise that you should try really hard not to touch the microstrip traces on these little PCBs as they are likely to lift or fall apart because of the corrosive problems caused by the absorber glue.

I put these numbers into Genesys and had a go at designing the square directional filter. Some may prefer to call it a diplexer. The aim was to replicate the dimensions and performance of the real filter.

I found that I had to skinny the traces on one side just like the real filter because this gives the best response at 2.05GHz. See the simulation below. I think the bandwidth of the real circuit will be quite similar. Taconic CER-10 has a loss factor of 0.0035 at 10GHz and I used this in the simulation.
Attached Thumbnails
Click image for larger version

Name:	TR4172DF.jpg
Views:	66
Size:	71.5 KB
ID:	228359  
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 6th Mar 2021, 8:35 am   #66
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,799
Default Re: Spectrum Analyser TR4172

Rogers 3010 perhaps? Er about 10.4

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now  
Old 6th Mar 2021, 1:38 pm   #67
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Spectrum Analyser TR4172

Yes, Rogers 3010 might be another option. Whatever alternatives are considered they ideally need to be available in the correct PCB thickness and I'm not sure what this is yet. It looks to be about 0.8mm which is about 0.03".

Taconic CER-10 is not the prettiest PCB material. It is a dull cardboard brown colour so not that pleasing on the eye. From memory, the original PCB material in the TR4172 IF1 module is a very glossy/glassy grey or off white colour. It looks expensive...

I had a go at guessing the dimensions of the inner IF1 LPF that sits between the square filter and the IF1 amplifier. See the image below where it is circled in red. I guessed it from the photos and used the outer dimensions of the IF1 enclosure to scale it. It looks like this filter is just under 50mm long. I tried drawing the PCB layout into Genesys and then used Sonnet to analyse the PCB layout. This was based on Taconic CER-10 material with a dielectric constant of 10 and a PCB thickness of 0.030".

The results below are surprisingly good. The simulation shows a LPF with a sharp cutoff at about 2050MHz. The passband ripple is very low so it could be that I've managed to get close to the response of the original design.
Attached Thumbnails
Click image for larger version

Name:	TR4172_innerIF1_LPF.jpg
Views:	55
Size:	62.3 KB
ID:	228390   Click image for larger version

Name:	TR4172IF1_LPF.jpg
Views:	72
Size:	57.0 KB
ID:	228392  
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 6th Mar 2021, 2:31 pm   #68
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,799
Default Re: Spectrum Analyser TR4172

Can you see weave in the brown substrate?

3010 is quite plain. Not sure what thickness I used for the GHz transmitters, but 50 Ohms wound up as 22 thou.

It has all the (lack of-)stiffness and breakability of ancient thin vinyl floor tiles.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is online now  
Old 6th Mar 2021, 3:29 pm   #69
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Spectrum Analyser TR4172

The Taconic CER-10 material is quite strange if you look at it close up with a microscope. The only samples I have here are milled PCBs but there is a definite and repeatable pattern in the material surface. It isn't a hard material at all and can be cut into quite easily with a scalpel. It's a very pale brown colour. At work it gets purchased as metal backed. In other words, the ground side of the PCB material is bonded to a thick slab of aluminium several mm thick. The ground vias go through the PCB material and through the aluminium because it is all bonded together as one PCB.

It generally isn't nice to look at or to work with but the RF performance is very good.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 6th Mar 2021, 4:22 pm   #70
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 62
Default Re: Spectrum Analyser TR4172

Quote:
Originally Posted by G0HZU_JMR View Post
I just spotted that you have bought some 2SC5010 parts. I'm not sure if you realise but the 2SC5010 is a modern low noise RF part in a tiny plastic minimold package. It won't be a drop in equivalent in terms of the package size. I used it as a model in Genesys because it was the nearest fit in terms of the RF performance at 2GHz.
Did I write 2SC5010? Sorry - I must have been looking at my notes on this and wrote the wrong part number. It is indeed a couple of 2SC2150s in the ceramic X package I purchased. I think the one I have is dead, but I might very well be wrong on that.
Matt kd4pbs is offline  
Old 6th Mar 2021, 4:29 pm   #71
SATOLEX
Triode
 
Join Date: Feb 2021
Location: Leiden, Zuid-Holland, The Netherlands
Posts: 26
Default Re: Spectrum Analyser TR4172

Thank you all for sharing this information! I have replaced the bias PNP with the BC857B as Jeremy suggested and it is working fine. When the blob was of I have remeasured the resistor values. As expected the broken PNP had effect on the measured resistance value, especially R4. I have measured the following:

R1 = 2k1
R2 = 391
R3 = 9k96
R4 = 4k71
R5 = 323
Vcc = 15.068 V
Icc = 14.2 mA
VTP1 = 5.647 V
VTP2 = 10.225 V
VTP3 = 0.98 V
VTP4 = 10.791 V
VTP5 = 0.831 V

Based on the measured voltages and resistors the current through the RF transistor is about 13.16 mA and is lower than what Jeremy has measured. This current is determined by the resistor values so, what is the truth. In my case the collector voltage (TP1) is 1.5 V higher. Matt indeed your R5 resistor value is not correct. What I have noticed in my case is that because the gold layer was damaged from the emitter of the PNP to the RF transistor I measured a higher resistance. I have added some solder on top of that trace to make the resistance lower. It does not look nice but it was the best I could do at home, see picture.

The MIC substrate which they have used looks like Alumina oxide (Er = 9.8 - 10.1) and the resistors are probably of Nickel chromium (NiCr). A long long time ago I have even designed those kind of circuits on this substrate. I’m more in the chip design now so I have to dig deep in my memory. If the gold layer is damaged or gone than the lossy NiCr layer remains.

My next step is to repair the interconnection in between the filter sections. It looks like that my IF1 module is in a better condition. The gold does not fall of if I touch it but what happens if I solder it? At my work there is one person left of the old MIC department and I will ask him to solder it, good microscope, better light and the right tools. I have to hurry because he will retire in April (the last one). I hope I have more information next week. I have also attached a picture of what I have measured for the 1.8 GHz low pass filter.
Attached Thumbnails
Click image for larger version

Name:	IF1AMP_TR4172_repaired.jpg
Views:	84
Size:	82.7 KB
ID:	228421   Click image for larger version

Name:	IF1_Module_LPF_TR4172.jpg
Views:	72
Size:	89.5 KB
ID:	228422  
SATOLEX is offline  
Old 6th Mar 2021, 4:54 pm   #72
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Spectrum Analyser TR4172

Those readings look good to me now so it looks like the amp section is working again. I really don't know if soldering the gold connections is a bad thing or not. Over time there might be an issue. I felt I had no choice but to solder mine at the output of the IF1 amplifier to the connection to the first BPF section.

I used 60/40 solder and I was very quick with the iron. The biggest fear for me was lifting the microstrip trace at the start of the BPF board due to heat stress. If this happens then I think it would be difficult to repair. From memory I hardly touched the PCB traces with the iron. I think you need to explain to the person that solders it that the copper may lift off if the iron temperature is too hot or if the iron is held there for more than a tiny fraction of a second. If it lifts off and shrivels to nothing then the only option is to use sticky copper tape cut to the same width as the microstrip. At 2GHz this won't really affect the performance as long as it is quite short.

Maybe someone else can suggest a better way to bond the gold strips. Whatever method is used it mustn't involve any upward pull on the microstrip or it could float away if the absorber glue has got to it.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 6th Mar 2021, 5:15 pm   #73
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Spectrum Analyser TR4172

Quote:
The MIC substrate which they have used looks like Alumina oxide (Er = 9.8 - 10.1) and the resistors are probably of Nickel chromium (NiCr). A long long time ago I have even designed those kind of circuits on this substrate. I’m more in the chip design now so I have to dig deep in my memory. If the gold layer is damaged or gone than the lossy NiCr layer remains.
Thanks! I just looked up the specs for this material and it is very impressive with a very low loss factor. I think the metal losses will dominate the substrate material losses in this case but this does look to be a very nice PCB material. I've never worked with it but I will investigate it further as the thermal properties look to be very good too.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 7th Mar 2021, 12:21 am   #74
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 62
Default Re: Spectrum Analyser TR4172

That's great, Lex! Seeing yours, I hold out hope, as long as I don't pull up more traces like I accidentally did already
I ruined the trace from the last capacitor to the output edge of the amp board. I even pulled up the very fine trace running between ground and the two series caps on the output of the transistor.
Check out this ugliness... Ugh!

Click image for larger version

Name:	20210306_175416.jpg
Views:	71
Size:	147.1 KB
ID:	228487

Oddly enough, I had already decided to try the sticky copper route, Jeremy.
I was doing something stupid with solder wick and pulled the whole trace off. I'm not real good at working with this kind of stuff, and it works totally different than any other kind of circuit board I have ever worked with before. The solder (I am using 60/40 as well) seems to even want to stick to the board. Live and learn I guess. Like Lex writes, the remaining stuff under the trace remains conductive. I guess the whole board has the NiCr applied, then the copper, then etched with etchant that removes the copper, then re-masked to create the resistors, then etched with etchant which removes the NiCr? I'm not familiar with this kind of stuff, and I bow to you all who are!
For mine, the traces are actually pretty rugged. I was even able to glue those lifted traces back down on the input thing... I'll call it a tuned narrow band directional coupler just because that sounds kinda cool . The resistors at the ends of the traces even remain connected... I measure ~2K for each.

Click image for larger version

Name:	20210306_160221.jpg
Views:	70
Size:	100.8 KB
ID:	228486

The reason I was soldering on it was due to a high resistance connection between the collector of the NPN amp and the first capacitor after (C3). This was fooling me into thinking the transistor was bad since I was probing on the cap due to not being able to get my probe on the transistor lead (Yes, the transistor appears good thankfully!). I tried just reflowing the solder, but either they used something that makes it impossible to follow up with 60/40 or something with the contamination from the glue creates the problems.

Time to order a PNP transistor and that second capacitor (C4) that I simply could not get soldered down... I measured it at ~ 4pF.

It's times like this that I remember it can't get any more "not working" than it already was. But darned if I ain't doing a good job of making it more broke.
Regards,
Matt
Matt kd4pbs is offline  
Old 7th Mar 2021, 1:00 am   #75
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 62
Default Re: Spectrum Analyser TR4172

Here's S11 for the LO tuned narrow band directional coupler input of the IF interface now. It looks much better.

Click image for larger version

Name:	IF Amp Input PostGlue.gif
Views:	64
Size:	9.5 KB
ID:	228499

Evidently when I checked the response of the RF filter, I used a stored calibration which was only cal'd for reflection. Oops. It looks much better now. This includes two unknown SMA to N adaptors which I did not use in the calibration. EDIT: Oops - looking at the plot, I see it wasn't calibrated at all when I measured it before, even the smith chart measurement was uncorrected. I have no idea what I was thinking that day... I usually don't screw up THAT bad!

Click image for larger version

Name:	LPF-B.gif
Views:	63
Size:	12.6 KB
ID:	228500

The tuning shifted down slightly, but I think it will still perform adequately. Certainly better than before.

Click image for larger version

Name:	VSWR, LO Input Post Glue.gif
Views:	59
Size:	13.6 KB
ID:	228504

Getting there...
Regards,
Matt

Last edited by Matt kd4pbs; 7th Mar 2021 at 1:20 am.
Matt kd4pbs is offline  
Old 7th Mar 2021, 1:55 am   #76
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Spectrum Analyser TR4172

If it's any consolation the same thing happened to the last trace on my IF1 amp board on my first TR4172. However, I don't remember seeing the dark trace stuck underneath. On mine there were also some small RF absorber pads stuck here with glue. The absorber glue was over the microstrip and it completely wrecked that final trace and it fell apart like soft gold leaf or rotted paint. Underneath was just the regular grey PCB material. I had to stick copper tape down in its place. It could be that my first TR4172 used a subtly different PCB material here because the only thing that came away was the metal and I'm sure there was no dark layer under it.

Note that I think the resistor terminations on that square coupler/filter/diplexer should be 50 ohms so I'd check their connections again.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 7th Mar 2021, 4:15 am   #77
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 62
Default Re: Spectrum Analyser TR4172

The odd thing is that the one on the output of that filter that wasn't lifted also reads ~2K. Since it's looking usable on the VNA I will probably leave it alone. Underneath those traces that lifted the NiCr layer was still there as Lex described.
If you look closely, everywhere that the gold traces came off, where it was under the blob and where I screwed it up, has the grey NiCr that comprises the resistors. I even tried to scrape away the NiCr layer with a knife and it would not come off.
Oh, I meant to mention, I think the traces might be a 100% gold alloy. I don't think it's plated copper. It's gold on both sides.

Last edited by Matt kd4pbs; 7th Mar 2021 at 4:25 am.
Matt kd4pbs is offline  
Old 7th Mar 2021, 6:11 pm   #78
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 62
Default Re: Spectrum Analyser TR4172

Thinking more about this, plan B would be to hack in a couple of SMA connectors and use an external 2GHz amplifier if I can't get this one to work.
Matt kd4pbs is offline  
Old 7th Mar 2021, 10:57 pm   #79
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Spectrum Analyser TR4172

It might be worth stopping for a bit until the PNP transistor arrives for the bias board to see if that gets the amp working.

Some of your VNA plots still look a bit strange to me. Your smith chart S11 plot of the IF1 input looks to have a trace containing significant circles within circles and this doesn't look right for the IF1 input port. On a 200MHz span centred on 2050MHz I'd expect to see a single large arc or maybe a complete single circle that resides within a VSWR circle of less than 2:1.

Back in post #22 I posted up a 2 port s-parameter plot of my TR4172 IF1 module and this shows the S11 response with the darker trace. This has a fairly graceful response without the repeating dips and crests seen in your VNA plots.

Could there be something wrong with your VNA cal kit correction tables or maybe there is a strange correction turned on somewhere in the menus? This might cause this issue.

Also, I can't explain why your board shows 2k ohm for both of the directional filter terminations. This filter is a classic design used by other manufacturers in their spectrum analysers. I would expect these terminations to be 50R if the mixer IF port is 50R and the impedance of the input and output microstrip is 50R. Otherwise I can't see how it can meet its design requirement for a wideband termination at the IF port of the mixer.
Attached Thumbnails
Click image for larger version

Name:	IF1_BPF.jpg
Views:	49
Size:	44.1 KB
ID:	228605  
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 7th Mar 2021, 11:14 pm   #80
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: Spectrum Analyser TR4172

Quote:
Originally Posted by Matt kd4pbs View Post
Thinking more about this, plan B would be to hack in a couple of SMA connectors and use an external 2GHz amplifier if I can't get this one to work.
I think it should be fairly easy to get it working again if the main RF transistor is OK. If the little amplifier PCB itself was trashed then I would consider ripping out the little IF1 amp and its bias board and I'd replace it with a new PCB design containing an actively biased ATF-54143 PHEMT. These are sadly obsolete but still available from some vendors.

I did a quick simulation in Genesys and the ATF-54143 gave about 0.7dB noise figure at 2.05GHz when biased at 5V and 20mA. The gain can be set to about 10dB as required and the input and output match can be quite good too. The signal handling performance should be at least as good as the old BJT device. This upgrade would probably lower the noise floor of the analyser by at least 2dB as the datasheet for the 2SC2150 shows a noise figure of about 3.5dB at 2GHz when biased at 15mA.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 12:31 pm.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.