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.

Reply
 
Thread Tools
Old 7th Mar 2021, 11:18 pm   #81
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 59
Default Re: Spectrum Analyser TR4172

I thought I had seen a spec from an NEC datasheet stating the 2SC2150 had a ~5dB nf. Of course that was a max rating.

I think you're right on the input plot. Most likely this is because I do not have a SMA cal kit with polynomials known. Then again, it is quite possible that at least calibrating it with what I do have would give better response than calibrating to the N connectors and using the N-SMA transitions I have on hand, but it's from one of those $1.98 mini VNAs that I just had to play with. The cost of a genuine HP SMA cal kit is way more than I paid for this spec an.
I guess I could try tacking on a resistor to the ends of those traces to make 50 ohms, and go from there, but as you can see in the pictures, no resistor was on there and unless the substrate resistors changed that greatly, they most definitely are not 50 ohms.
I agree that something isn't quite right. I'll certainly have something to do until the regulator transistor comes in.
Thanks bunches, Jeremy!
Regards,
Matt

Last edited by Matt kd4pbs; 7th Mar 2021 at 11:26 pm.
Matt kd4pbs is offline   Reply With Quote
Old 8th Mar 2021, 12:07 am   #82
G0HZU_JMR
Nonode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 2,435
Default Re: Spectrum Analyser TR4172

I wouldn't try adding any resistors just yet because you ideally need to find a very good quality 50R termination resistor to fit to the directional filter if the termination resistors really are faulty. It may be the case that a pair of stacked 100R 0402 resistors will give the best results at 2GHz for example.
However, it's best to leave it for now because it looks like you did a great job of gluing down the microstrip on that filter. It would be a shame to spoil it if the soldering iron trashes it all again.

Maybe best to wait for the PNP BJT and fix the bias board. When I had my IF1 module open I was really scared to touch any of it because the PCB traces were so fragile because of the absorber glue effects. I was really happy to get it closed up without disturbing any more PCB traces.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 8th Mar 2021, 12:19 am   #83
G0HZU_JMR
Nonode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 2,435
Default Re: Spectrum Analyser TR4172

Quote:
I thought I had seen a spec from an NEC datasheet stating the 2SC2150 had a ~5dB nf. Of course that was a max rating.
I've still got my classic and probably quite rare 1989 NEC databook for high frequency devices and I was quite surprised that the 2SC2148, 2SC2149 and 2SC2150 BJTs were included in it.

The datasheet graphs show about 3.5dB noise figure at 1.85GHz at 15mA bias current for the 2SC2150. My old notes for the IF1 module showed that I measured about 1dB gain and just over 6dB noise figure for the complete IF1 module after I repaired it. If there is 2dB loss in the directional filter and maybe 0.8dB in the LPF and the 2SC2150 manages a 3.5dB noise figure then this does stack up to be just over 6dB NF for the IF1 module at 2.05GHz.

This old NEC databook really is a very good book with all kinds of application notes in the back for BJT and MOSFET LNA design. It also has app notes for diode ring mixer design and TV tuner design from the 1980s. It has datasheets and lengthy circuit app notes for all the classic low noise devices like the 3SK88(A), 3SK74, the wonderful 2SC3355/6/7 BJT and others like the 2SC2026. It even has about 10 pages devoted to homebrew diode ring mixers using various Schottky diodes.

It is one of my most valued databooks. NEC have added a very strange front cover to this databook showing a drawing of a teddybear sat on a beach ball. I've always called it my teddybear book...

I haven't seen it online anywhere but I keep looking and hoping.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 8th Mar 2021, 12:19 pm   #84
G0HZU_JMR
Nonode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 2,435
Default Re: Spectrum Analyser TR4172

I had a rummage online to try and find some info on directional filter design (for spectrum analysers) and the best I could come up with is in the old HP journal from August 1979. This covers the classic HP8566A spectrum analyser and it uses a directional filter on range 1 up to 2.5GHz.

https://www.hpl.hp.com/hpjournal/pdf...Fs/1979-08.pdf

Have a look at page 12 where there is a decent picture of a directional filter and also a brief technical description.

Also, your recent VNA plots of the IF1 input show a reasonably low VSWR across a 200MHz span and this suggests that the 50R resistors might be intact so this is confusing me a little. If the resistors were toasted to 2k ohm then I'd expect to see the VSWR skyrocket at the edges of that 200MHz span.

Probably best to sort out the correction table for your VNA cal kit or make an SMA cal kit using good quality end launch connectors. A carefully built cal kit like this is good to about 3GHz even with basic corrections in the cal kit definition file.
Attached Thumbnails
Click image for larger version

Name:	HP8566ADF.jpg
Views:	20
Size:	32.5 KB
ID:	228641  
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 8th Mar 2021, 2:33 pm   #85
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 17,645
Default Re: Spectrum Analyser TR4172

Notice also, on page 7 who was the project manager for the 8566A - Siegfried Linkwitz.

He's also the designer of Craig Sawyer's favourite loudspeakers. He was also involved in a surround sound system for cinemas in competition with Dolby.

I wound up working with some of that gang about 12 years after that HPJ article. Very impressive people.

I've always viewed those directional filters as rat-races where you also use the frequency dependency.

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   Reply With Quote
Old 8th Mar 2021, 5:21 pm   #86
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 59
Default Re: Spectrum Analyser TR4172

You all are very kind helping me learn this. Thank you so much for the patience!
I will certainly attempt to calibrate to the ends of the SMA connectors and re-visit this issue.
In thinking about this filter coupler and the termination issue, I could experiment with simply placing a termination resistor on the end to ground without soldering, but pressing down on it with a non-metallic tool and see what that does. I guess another possibility is to just place a circulator at the input, and I just happen to have a couple of decent 1.9-2.1GHz circulators.
As for the 2sc2150's noise figure, here's what I found in a 1995 NEC semiconductor selection guide...
Click image for larger version

Name:	2sc2150.png
Views:	9
Size:	34.7 KB
ID:	228670

Thanks for the lunchtime reading on the HP Journal - I love browsing those!
Matt kd4pbs is offline   Reply With Quote
Old 8th Mar 2021, 5:59 pm   #87
G0HZU_JMR
Nonode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 2,435
Default Re: Spectrum Analyser TR4172

To test the input 50R termination of the directional filter I'd be tempted to just use a DMM to measure the resistance across the input SMA connector at the internal soldered post (PCB) side of the SMA connector. I wouldn't recommend touching any part of the directional filter itself with the probe tips. Then ground the other end of the DMM to the metal body of the SMA connector. Hopefully it will show 50 ohms. If not then there is almost certainly a broken connection somewhere.

Do you have a 6GHz HP8753D VNA? Looking at your colour plots I guess it is the full 2 port 8753D. This is a very nice full 2 port VNA. It really is worth trying to set yourself up with a fully corrected SMA cal kit if it is a full 2 port 8753D.

This needn't cost much. My first SMA cal kit (for a full 2 port VNA) cost very little because I made it from salvaged SMA connectors but even if you bought new SMA connectors it would only cost about $50 including some test fixture SMA connections that would get used for some PCB dev work.
__________________
Regards, Jeremy G0HZU

Last edited by G0HZU_JMR; 8th Mar 2021 at 6:04 pm.
G0HZU_JMR is offline   Reply With Quote
Old 8th Mar 2021, 6:55 pm   #88
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 59
Default Re: Spectrum Analyser TR4172

That's how I determined the terminators were ~2K as they are... checking the input SMA and the output at the inter-board connection wire.
This is indeed a 6GHz 8753D. She's an oldie but goodie. It was gifted to me by a very dear friend and fellow broadcast engineer. Shortly after, it developed the typical HP problem on these series of VNA; a dead YTO. Thankfully I was able to locate one on that auction site, so all said, I have ~$500 into it. I've got no excuse for not playing around on the 5.8GHz and below microwave bands I guess... especially since I have lots of surplus 7GHz stuff just laying around that can probably be coerced operate down there.
I can certainly make an SMA cal kit, but I have no way to measure the polynomial corrections for it, so it would be less than ideal. Then again, it's likely that it would be far more ideal than slapping these unknown N to SMA adaptors on the ends and working from there. Again, I've got no real excuse... especially since the cables I currently use are on the order of 8 meters a piece of RG-400. Not ideal for microwave use at all, and very much impractical for benchtop use. It is another one of those "round tuit" things I need to work on. I know; I'm a mess - I'm sure it explains some of those crappy readouts you see from that Matt guy

David, I was doing some light reading on your old co-worker Siegfried. Amazing guy, he was. Interestingly, there is a local (for me) connection to him. The company he and some other of those HP guys created back in the 90s is right down the road from where I work, about 8 miles, half-way to our transmitter site. Small world!
https://www.stereophile.com/interviews/503/index.html
Matt kd4pbs is offline   Reply With Quote
Old 8th Mar 2021, 8:34 pm   #89
SATOLEX
Triode
 
Join Date: Feb 2021
Location: Leiden, Zuid-Holland, The Netherlands
Posts: 26
Default Re: Spectrum Analyser TR4172

Hi, I agree that the resistor has to be 50 Ohms. The reason that you measure 2K is because the gold has no connection to the end anymore with the resistive layer. Probably because it is glued. You can even calculate the resistance value roughly. The square resistance of the resistive layer is about 50 Ohms because it is almost a square. The length of the line from SMA to resistor is about 20 mm and the average width about 0.5 mm. So, 20 mm/0.5 mm is 40 squares which means that the resistance is about 40*50 Ohm = 2000 Ohm. What do you think?

Regards, Lex
SATOLEX is offline   Reply With Quote
Old 8th Mar 2021, 9:23 pm   #90
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 59
Default Re: Spectrum Analyser TR4172

I'm guessing you're spot on, Lex. Even the trace which did not come loose measures almost the same resistance to ground, and I think this is what was confusing. Not only that, but the resistance reads the same both before and after I glued it. I think quite simply, I need to add resistance to the ends to make 50 ohms.
Matt kd4pbs is offline   Reply With Quote
Old 8th Mar 2021, 9:45 pm   #91
G0HZU_JMR
Nonode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 2,435
Default Re: Spectrum Analyser TR4172

It's worth investigating a few contenders for the resistors first. You ideally only want to be soldering this once.

I don't have any 50R or 100R resistors in 0402 and I'm guessing that 0603 will be a bit big to fit. If 0603 does fit I have some 50R (49.9R) resistors rated to 6GHz I could test but I suspect they won't be that great at 2GHz in terms of VSWR. A pair of stacked 100R resistors will probably be better and this might have to be in the 0402 package to fit neatly?

My homemade SMA cal kit load was made with stacked 100R resistors and it compares really well when measured against an Agilent Ecal cal kit. It shows about 40dB return loss at 2GHz for example. I did spend quite a bit of time trying to find the very best 0603 SMD 100R resistors from various manufacturers though.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 8th Mar 2021, 9:49 pm   #92
G0HZU_JMR
Nonode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 2,435
Default Re: Spectrum Analyser TR4172

Quote:
Originally Posted by SATOLEX View Post
Hi, I agree that the resistor has to be 50 Ohms. The reason that you measure 2K is because the gold has no connection to the end anymore with the resistive layer. Probably because it is glued. You can even calculate the resistance value roughly. The square resistance of the resistive layer is about 50 Ohms because it is almost a square. The length of the line from SMA to resistor is about 20 mm and the average width about 0.5 mm. So, 20 mm/0.5 mm is 40 squares which means that the resistance is about 40*50 Ohm = 2000 Ohm. What do you think?

Regards, Lex
Neat analysis! Could this explain some of the subtle differences we see in the bias board? What about the blue marks in the resistance sections seen on the bias board. Is this some sort of thermal stress or could it be stress from the horrible absorber glue? At one point I thought the blue marks might be a by product of trimming the resistance at the factory but I'm not sure the resistances need to be very accurate.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 8th Mar 2021, 9:59 pm   #93
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 17,645
Default Re: Spectrum Analyser TR4172

Laser trim gives a sharp line cut, usually L-shaped

Air abrasive trim is much blurrier to look at.

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   Reply With Quote
Old 9th Mar 2021, 1:25 am   #94
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 59
Default Re: Spectrum Analyser TR4172

I only have 0603 size on hand, but I can likely salvage something smaller out of some old gear. Whatever I use, for sure I'll only solder once but test on the VNA by just pressing it on there first.
Matt kd4pbs is offline   Reply With Quote
Old 9th Mar 2021, 1:30 am   #95
G0HZU_JMR
Nonode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 2,435
Default Re: Spectrum Analyser TR4172

This is not something I have any experience of so I can only guess what causes the blueing on the resistors.

I did dig out my box of special resistors and managed to find some very snazzy FC series 0603 50 ohm resistors from Vishay. These are part of the 40GHz series and are available at Mouser in the UK here:

https://www.mouser.co.uk/ProductDeta...LEn2ejug%3D%3D

The datasheet indicates that the VSWR of the 50 ohm 0402 version is extremely low. Easily good enough for this application.

https://www.mouser.co.uk/ProductDeta...LEn2ejug%3D%3D

These cost about 1.50 each here in the UK.

I'm not sure if 0402 will be too small or if 0603 will be too big. However, these FC resistors from Vishay would be a good contender in my opinion. I can try and test the 0603 version tomorrow on my VNA as I have a few of them here on SMD tape.

It might be possible to use ordinary 0603 resistors as the termination quality might not be that critical for the directional filter. I'm not sure how good the original printed resistors would be in terms of VSWR at 2GHz anyway.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 9th Mar 2021, 2:27 am   #96
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 59
Default Re: Spectrum Analyser TR4172

Hopefully I'll get a chance in the next couple of days to possibly put a lid on this thing. I just realized I have some of these on hand:
https://html.alldatasheet.com/html-p.../1/SS8550.html
They appear to be a close enough match to the 2SA812-M6 PNP in this application.
If I can get it working right, I can imagine the 4172 will work better than it ever has for me, even with sub-optimal terminators.
Matt kd4pbs is offline   Reply With Quote
Old 9th Mar 2021, 8:28 pm   #97
SATOLEX
Triode
 
Join Date: Feb 2021
Location: Leiden, Zuid-Holland, The Netherlands
Posts: 26
Default Re: Spectrum Analyser TR4172

Quote:
Originally Posted by G0HZU_JMR View Post
Neat analysis! Could this explain some of the subtle differences we see in the bias board? What about the blue marks in the resistance sections seen on the bias board. Is this some sort of thermal stress or could it be stress from the horrible absorber glue? At one point I thought the blue marks might be a by product of trimming the resistance at the factory but I'm not sure the resistances need to be very accurate.
In my case it can explain that R5 is a bit higher than yours. Some gold is missing in between R5 and the PNP emitter. The resistors R5 and R2 are most important to set the RF transistor current and collector voltage if you assume that the divide ratio of R3 and R4 is correct. I think that the resistors are not trimmed. It is a pity that the documentation is missing of this IF1 block. It is also possible that there is some difference between series of the TR4172. How can I find out which year mine is?

Regards, Lex
SATOLEX is offline   Reply With Quote
Old 9th Mar 2021, 9:26 pm   #98
G0HZU_JMR
Nonode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 2,435
Default Re: Spectrum Analyser TR4172

I'm not sure how to find out how old it is other than to look at the date codes on the main chips in the digital sections. I did this on mine many years ago and the youngest date codes I could find were from late 1986. The analyser was donated to the company some time around 1990/1 and I was the first to use it.

It didn't look new at the time so mine was probably made in 1987. I bought it from the company well over 15 years ago and for many years it was my main spectrum analyser here at home.

I'm not sure about Advantest serial numbers. It might be possible to decode the date from the serial number but I don't know how. I think the TR4172 dates back to 1981 but I'm not sure when Advantest stopped making them.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 10th Mar 2021, 12:21 am   #99
Matt kd4pbs
Tetrode
 
Join Date: Mar 2021
Location: Raleigh, North Carolina, USA.
Posts: 59
Default Re: Spectrum Analyser TR4172

Made up some sane-length RG400 N to SMA cables today - about a meter each. Perfect for the bench. This certainly answers the question as to whether an unknown SMA cal kit is better than a known N cal kit with unknown N to SMA adaptors at the end of an eight meter mediocre-at-microwave-frequencies cable. I'm almost ashamed to admit that it is merely the little calibration standards that came with this $1.98 NanoVNA I got to play with.
This is the IF input before adding terminators.

Click image for larger version

Name:	SMA IF Input 2K Terms.gif
Views:	21
Size:	8.6 KB
ID:	228786

Much more sane than the same measurement from post #46 and #75.

Oh, just an FYI, you are evil, Jeremy. You've got me eyeing a real SMA Cal Kit from Dr. Kirby. It would cost about as much as I already have in the VNA.
Can I please tell my wife that YOU made me do it if I end up purchasing it?

Lex, that reminds me that I still have to look at some component date codes on mine. Being that it's a Takeda Riken branded TR4172, it wouldn't surprise me if it was one of the first ones to roll off of the assembly lines.

73,
Matt
Matt kd4pbs is offline   Reply With Quote
Old 10th Mar 2021, 2:04 am   #100
G0HZU_JMR
Nonode
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 2,435
Default Re: Spectrum Analyser TR4172

I'm not sure I'd go for a Kirkby kit but lots of people seem happy with them.

I'm still confused by your VNA plot of s11 because now it looks about right for a healthy directional filter. Could it be that despite the broken connection there is still sufficient coupling capacitance there to make the broken connection act as a DC block at 2GHz? This would mean the resistors are still there at 2GHz. If so, adding the piggy 50R resistors might make it worse...
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Reply

Thread Tools



All times are GMT. The time now is 7:51 am.


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 - 2021, vBulletin Solutions, Inc.
Copyright ©2002 - 2021, Paul Stenning.