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 > Other Discussions > Homebrew Equipment

Notices

Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members.

Reply
 
Thread Tools
Old 24th Oct 2022, 12:03 am   #181
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: 6-gang FM stereo tuner heads

It might not need the snubber if you stick with the original 47Ω resistor, and you also maintain a good 50Ω termination at the far end of the coax cable. The easiest way to demonstrate the impact of negative resistance is to add an input network that completes a VHF or UHF oscillator. However, this probably won't be easy because the series 0.5pF cap will limit what can realistically be connected at the input to complete the resonator for an oscillator.

The other way to show negative resistance is present would be to sweep through a very narrow UHF bandpass filter response with a 50Ω VNA and then look at the insertion loss on the VNA with a 1dB/div scale.

Something like a single, high Q resonator bandpass filter with very light capacitive coupling to 50Ω ports via (say) 0.5pf caps. If it is a very narrow BPF made with high Q components, you might see a 1.5dB insertion loss on the VNA. If you then probe the centre resonator with a probe that has negative resistance, the signal level seen on the VNA will go up very slightly. The insertion loss would reduce because the probe adds RF energy to the system due to the negative resistance.

The probe would act here like a subtle Q multiplier, but I don't think this is likely to have any significance down at 100MHz. It's probably more likely to be observed up at 400MHz.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 24th Oct 2022, 5:18 pm   #182
regenfreak
Heptode
 
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
Default Re: 6-gang FM stereo tuner heads

Quote:
Originally Posted by G0HZU_JMR View Post
It might not need the snubber if you stick with the original 47Ω resistor, and you also maintain a good 50Ω termination at the far end of the coax cable. The easiest way to demonstrate the impact of negative resistance is to add an input network that completes a VHF or UHF oscillator. However, this probably won't be easy because the series 0.5pF cap will limit what can realistically be connected at the input to complete the resonator for an oscillator.

The other way to show negative resistance is present would be to sweep through a very narrow UHF bandpass filter response with a 50Ω VNA and then look at the insertion loss on the VNA with a 1dB/div scale.

Something like a single, high Q resonator bandpass filter with very light capacitive coupling to 50Ω ports via (say) 0.5pf caps. If it is a very narrow BPF made with high Q components, you might see a 1.5dB insertion loss on the VNA. If you then probe the centre resonator with a probe that has negative resistance, the signal level seen on the VNA will go up very slightly. The insertion loss would reduce because the probe adds RF energy to the system due to the negative resistance.

The probe would act here like a subtle Q multiplier, but I don't think this is likely to have any significance down at 100MHz. It's probably more likely to be observed up at 400MHz.
Thanks. I have a 465MHz 3-pole cavity resonator bandpass filter for such negative resistance test. Based on the experience of other people who built such a probe, the construction with the original 47ohm resistor seems to be give fairly flat S21 plots. The best philosophy is to "keep it simple, stupid". I can't do much now until I receive the pogo test pins.
regenfreak is offline   Reply With Quote
Old 24th Oct 2022, 9:59 pm   #183
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: 6-gang FM stereo tuner heads

I think any negative resistance will be avoided with the 47Ω bias resistor. To try and show this, I had a go at measuring some s2p data of a BF998 in common drain with the g2 pin strapped to the drain as per the probe design.

I then put this s2p data into a model of the probe circuit and compared it to the manufacturer's s2p data for the BF998. The manufacturer's data is really old, and it isn't taken with g2 tied to the drain. It's also taken at 8V. However, the results below were still fairly similar when I look for negative resistance at the input of the probe circuit. In the plots below I compare having 47Ω as the Rbias resistor and also 150Ω as the Rbias resistor when the g1 bias is re-adjusted for 10mA drain current in both cases.

To explore a realistic worst case, I assume the load at the far end of the coax cable can have a VSWR of 1.5:1 so I terminate with 75Ω rather than 50Ω. This will tend to worsen any negative resistance if there is any.

The plots show the 47Ω version looks OK, but the 150Ω version does appear to introduce some negative resistance (as expected).

There isn't much of it though. This amount of negative resistance would be fairly benign although it probably could cause instability if you went looking for it. You can see that the negative resistance is termination sensitive. If I remove the 75Ω load the negative resistance gets worse although it is far worse for the 150Ω bias version. See the third plot below for this.
Attached Thumbnails
Click image for larger version

Name:	Probe_Rs_47Rbias.jpg
Views:	21
Size:	71.4 KB
ID:	266778   Click image for larger version

Name:	Probe_Rs_150Rbias.jpg
Views:	20
Size:	72.3 KB
ID:	266779   Click image for larger version

Name:	Probe_Rs_150RbiasNOLOAD.jpg
Views:	19
Size:	74.7 KB
ID:	266780  
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 24th Oct 2022, 10:57 pm   #184
regenfreak
Heptode
 
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
Default Re: 6-gang FM stereo tuner heads

Quote:
Originally Posted by G0HZU_JMR View Post
I think any negative resistance will be avoided with the 47Ω bias resistor. To try and show this, I had a go at measuring some s2p data of a BF998 in common drain with the g2 pin strapped to the drain as per the probe design.

I then put this s2p data into a model of the probe circuit and compared it to the manufacturer's s2p data for the BF998. The manufacturer's data is really old, and it isn't taken with g2 tied to the drain. It's also taken at 8V. However, the results below were still fairly similar when I look for negative resistance at the input of the probe circuit. In the plots below I compare having 47Ω as the Rbias resistor and also 150Ω as the Rbias resistor when the g1 bias is re-adjusted for 10mA drain current in both cases.

To explore a realistic worst case, I assume the load at the far end of the coax cable can have a VSWR of 1.5:1 so I terminate with 75Ω rather than 50Ω. This will tend to worsen any negative resistance if there is any.

The plots show the 47Ω version looks OK, but the 150Ω version does appear to introduce some negative resistance (as expected).

There isn't much of it though. This amount of negative resistance would be fairly benign although it probably could cause instability if you went looking for it. You can see that the negative resistance is termination sensitive. If I remove the 75Ω load the negative resistance gets worse although it is far worse for the 150Ω bias version. See the third plot below for this.
Thanks! The series input resistance Rs can be negative. So Q becomes negative unless Xs is negative too! I learn something new every day! I can imagine it is difficult to model the parasitic capacitance and inductance at the probe tip using simulation.

I have attached the S22 of the Poorman's FET versus HP 8502A and Philips PM7515 probes from Hirschbuechler's thesis. They just show how well the probe matches the 50 ohm output. None of the S22 look particularly good. As i mentioned before, he added an extra 33 ohm resistor at the 50ohm output.
Attached Thumbnails
Click image for larger version

Name:	FET vs Philips vs HP probes.jpg
Views:	20
Size:	52.1 KB
ID:	266784  
regenfreak is offline   Reply With Quote
Old 25th Oct 2022, 1:12 am   #185
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: 6-gang FM stereo tuner heads

I guess a lot depends on what your requirements or goals are here. A commercial probe design (from HP or Tek?) will have hopefully addressed the negative resistance issue that affects follower circuits be they FET, BJT or valve followers.

The commercial designer doesn't want to manufacture and sell a probe that is inherently unstable. The homebrew designer will gloss over the negative resistance issue. Homebrew probes don't have to adhere to this requirement. The same applies to application note based designs for FET probes or buffers. They probably won't be designed with stability as a mandatory requirement. So, app note circuits and magazine articles need to be treated with caution.

If you were to put a BF998 follower onto a 2 port VNA and casually look at the input reflection coefficient you would see it begin to subtly creep above 1 at a few tens of MHz. When viewed on a smith chart the s11 trace would obviously be outside the smith chart by about 300MHz. So negative resistance would be in plain sight by 300MHz and difficult for a commercial designer to ignore.

There are cheapo homebrew methods that can be adopted to crudely quantify the negative resistance vs frequency. A VNA wouldn't be necessary here, just a coil of wire, a frequency counter and a few chip resistors if the designer is on a tight budget. A 50 TinySA would be nicer and much more reliable to use than the counter though.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 25th Oct 2022, 5:23 pm   #186
regenfreak
Heptode
 
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
Default Re: 6-gang FM stereo tuner heads

Quote:
Originally Posted by G0HZU_JMR View Post
I guess a lot depends on what your requirements or goals are here. A commercial probe design (from HP or Tek?) will have hopefully addressed the negative resistance issue that affects follower circuits be they FET, BJT or valve followers.

The commercial designer doesn't want to manufacture and sell a probe that is inherently unstable. The homebrew designer will gloss over the negative resistance issue. Homebrew probes don't have to adhere to this requirement. The same applies to application note based designs for FET probes or buffers. They probably won't be designed with stability as a mandatory requirement. So, app note circuits and magazine articles need to be treated with caution.

If you were to put a BF998 follower onto a 2 port VNA and casually look at the input reflection coefficient you would see it begin to subtly creep above 1 at a few tens of MHz. When viewed on a smith chart the s11 trace would obviously be outside the smith chart by about 300MHz. So negative resistance would be in plain sight by 300MHz and difficult for a commercial designer to ignore.

There are cheapo homebrew methods that can be adopted to crudely quantify the negative resistance vs frequency. A VNA wouldn't be necessary here, just a coil of wire, a frequency counter and a few chip resistors if the designer is on a tight budget. A 50 TinySA would be nicer and much more reliable to use than the counter though.
My goal is to use my Siglent SSA3021x Plus to do visual sweep alignment in the frequency domain of the RF front-end and IF stages for both valve and solid-state FM broadcast receivers. I rarely bother with time domain measurements. I am not concerned with the perfection in frequency response in the VNA above 108MHz. In the past, I just used a simple HT probe made out of a 100K ohms resistor in series with a 4.7nF high voltage cap for valve FM IF sweep alignment. I simply used a 20MHz Rigol signal generator as a wobbulator and a 200MHz scope. But it would take half an hour to set up. The most challenging part is to align sophisticated RF front ends with multi-stages of double-tuned bandpass filters; e.g. my homebrew valve and dual gate FET 6-gang tuners.

The no-frills Zo probe is limiting for my needs. In a sense, it is similar to what I did with a 100K resistor and a DC blocking cap. Tektronix gave a very good overview the pros and cons of different types of probes in this document:

https://w140.com/tekwiki/images/6/62/062-1146-00.pdf

The visual sweep alignment techniques for solid-state and valve receivers are quite different. The later demands attention to high voltage protection for the expensive spectrum analyzer. So I have made simple SMA adaptors with high voltage 1kV SMT DC blocking capacitors in series with 250ma SMT fast fuses (the smallest rating I can find) in both input and output of the SSA3021X Plus. I have commercial SMA DC blockers but they are only rated 50V, not very reassuring!

Last edited by regenfreak; 25th Oct 2022 at 5:34 pm.
regenfreak is offline   Reply With Quote
Old 25th Oct 2022, 8:18 pm   #187
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: 6-gang FM stereo tuner heads

Yes, it's going to be risky connecting your analyser to valve stuff. As energy stored in the blocking cap is proportional to voltage squared, a lot of energy can be stored in the blocking caps at high HT voltages, and this can produce damaging transients when the energy eventually wants to escape from the cap into a new home. I'd recommend some form of (fast) clamping limiter at the analyser input to minimise this risk.

I managed to find that thesis online, and the technical content isn't that great. Some of it is OK but I'm not impressed by the way the fet probe was developed. The author placed a 5th order LPF at the output and tweaked the values to try and offset some ripple up at UHF. The end result appears to be a network that has smaller cap values than expected. This will make the output network look inductive within the passband of the LPF.

If you place an inductive load at the output of a FET follower, this will effectively lower the input Rp of the FET quite significantly. So, I'm surprised this was done. Perhaps even worse, at higher UHF frequencies I'd expect that network to look capacitive. Putting small capacitive loads at the output of a follower will make the FET input tend towards negative resistance. Up at UHF this may cause the probe to act as a Q multiplier or even become unstable when probing certain loads even with the 47Ω bias resistor present.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 25th Oct 2022, 9:31 pm   #188
regenfreak
Heptode
 
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
Default Re: 6-gang FM stereo tuner heads

Quote:
Originally Posted by G0HZU_JMR View Post
Yes, it's going to be risky connecting your analyser to valve stuff. As energy stored in the blocking cap is proportional to voltage squared, a lot of energy can be stored in the blocking caps at high HT voltages, and this can produce damaging transients when the energy eventually wants to escape from the cap into a new home. I'd recommend some form of (fast) clamping limiter at the analyser input to minimise this risk.

I managed to find that thesis online, and the technical content isn't that great. Some of it is OK but I'm not impressed by the way the fet probe was developed. The author placed a 5th order LPF at the output and tweaked the values to try and offset some ripple up at UHF. The end result appears to be a network that has smaller cap values than expected. This will make the output network look inductive within the passband of the LPF.

If you place an inductive load at the output of a FET follower, this will effectively lower the input Rp of the FET quite significantly. So, I'm surprised this was done. Perhaps even worse, at higher UHF frequencies I'd expect that network to look capacitive. Putting small capacitive loads at the output of a follower will make the FET input tend towards negative resistance. Up at UHF this may cause the probe to act as a Q multiplier or even become unstable when probing certain loads even with the 47Ω bias resistor present.
I didn't bother with his Butterworth LP filter as I am not convinced either. I thought it is not necessary. Funny enough this guy also copied his LP filter idea:

https://electronicprojectsforfun.wor...quency-probes/

Mr Carlson's came up with his spectrum analyzer protector. It was designed for 1M input. I thought his design is over-complicated. He used two strings of BAV99 as voltage clippers. I am not sure if the diodes will add too much stray capacitance at the 50 ohm input and output of the spectrum analyzer with a tracking generator that may cause problem at VHF/ UHF frequencies. His design seems more suitable for relatively low frequency sweep in medium or short wave valve radios.

https://youtu.be/ETKyKC2Zj-M

I will avoid the use of too big value of the blocking capacitor to restrict the stored charge energy. At the moment I use one 1kV 4.7nF blocking cap. I will use three 1KV 10nF capacitors in series giving 3.3nF total and 3kV rating. The 250mV SMT fast fuse has 0.665 ohms dc resistance which will cause some insertion losses but it is probably better be safe than sorry. This guy fired his front end of his Siglent SSA3021x and he repaired it:

https://youtu.be/zQ1lPbTwKaU

At least I have some idea on how to do repair if it ever happens to me. But i rather be very cautious.

Last edited by regenfreak; 25th Oct 2022 at 9:42 pm.
regenfreak is offline   Reply With Quote
Old 25th Oct 2022, 11:13 pm   #189
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: 6-gang FM stereo tuner heads

Quote:
I didn't bother with his Butterworth LP filter as I am not convinced either. I thought it is not necessary. Funny enough this guy also copied his LP filter idea:

https://electronicprojectsforfun.wor...quency-probes/
To be fair, that circuit uses a LPF with a more sensible L to C ratio for a 50Ω design so the LPF will appear fairly transparent up to the cutoff frequency. It looks like a regular 50Ω LPF design. I don't really have an issue with that LPF as it probably won't have much (if any) impact on the input impedance of the probe.

By contrast, the LPF in the thesis has tweaked component values and this will cause the issues I described in my earlier post because it isn't a 50Ω filter anymore.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 26th Oct 2022, 9:28 pm   #190
regenfreak
Heptode
 
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
Default Re: 6-gang FM stereo tuner heads

Quote:
Originally Posted by G0HZU_JMR. Yes, it's going to be risky connecting your analyser to valve stuff. As energy stored in the blocking cap is proportional to voltage squared, a lot of energy can be stored in the blocking caps at high HT voltages, and this can produce damaging transients when the energy eventually wants to escape from the cap into a new home. I'd recommend some form of (fast) clamping limiter at the analyser input to minimise this risk.
It seems that many people popped the two RF switches at the input of the Siglent SSA3021x in spite of the presence of two internal ESD diodes. The SSA3021x has a massive 1 microfarads DC blocker at the input. I guess the large capacitance value helps to improve the low frequency response but it comes at a price of discharging lots of energy like a firecracker in the presence of transient voltage surges. So it is very risky to use it in vintage valve receivers:

http://emccompliance.co.uk/fixing-an...ctrum-analyzer

Commercial microwave transient limiters are very expensive and they seem to use carefully designed Schottky-PIN limiter with built-in attenuator to act as another layer of protection. The limiter consists of a Schottky diode in parallel with a matched PIN diode. The Schottky diode lowers the turn-on threshold voltage and the PIN limiter diode protects the Schottky diode at higher power levels (see attachment). The pair have to be matched and optimized for the best S11 and S21 across broadband frequencies. So the Schottky-PIN limiter are specialized and hard-to-find components.

I have ordered some BAV99 diodes and will try to measure their total capacitance versus frequency in a parallel strings as in Mr Carlson's idea. They work like a bi-directional voltage clipper.

I have looked at some datasheets of SMT ceramic gas discharge tubes rated at 70V (the lowest voltage available) and they have capacitance of 0.5pF. Probably it is rated too high voltage for my need.
Attached Thumbnails
Click image for larger version

Name:	Schottky-pin limiter.jpg
Views:	21
Size:	17.2 KB
ID:	266909  

Last edited by regenfreak; 26th Oct 2022 at 9:41 pm.
regenfreak is offline   Reply With Quote
Old 27th Oct 2022, 9:38 pm   #191
regenfreak
Heptode
 
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
Default Re: 6-gang FM stereo tuner heads

I have done my first ever phase noise measurement using a low-cost, low phase noise 10MHz OCXO signal source.

With -20dbm carrier, the delta marker for the noise power density (dbm per unit Hz) is -119.43-(-21.41) = -98 at 10kHz offset from the 10MHz carrier which is the same as the manufacturer quote of 98dbc/Hz at 1GHz, 10KHz offset.

The capacitance for BAV99 is about 2pF. Now I am also looking at ISS86 Schottsky diode for UHF TV tuner. It has capacitance of 0.85pF. If I put 4-5 of them in series and have back-to-back anti-parallel strings, it may work OK as limiters for FM broadcast band as a protection for transient voltage surges. I am not sure and have to do some measurement first.
Attached Thumbnails
Click image for larger version

Name:	phase noise 10kHz 10M.jpg
Views:	25
Size:	45.0 KB
ID:	266964   Click image for larger version

Name:	phase noise OCXO.jpg
Views:	24
Size:	76.9 KB
ID:	266965  
regenfreak is offline   Reply With Quote
Old 27th Oct 2022, 10:03 pm   #192
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 21,458
Default Re: 6-gang FM stereo tuner heads

A crystal oscillator ought to be a few of orders of magnitude better than that, so you're probably using the crystal oscillator to measure the sum of the phase noise of the local oscillators in your analyser.

You also have to take into account that the -3dB bandwidth of your IF resolution is not quite the same as the equivalent rectangular noise bandwidth. Also there's a subtle one in that many analysers perform logging before averaging and this distorts the pdf of the noise.

If your analyser has a "Noise Marker" function and it's been thoroughly done, it should account for these things for you.

One way of measuring phase noise down to lower levels is to do a mix-down with a known clean signal source, and measure the result with an audio spectrum analyser... they often have reduced phase noise pro rata to the lower frequency LOs they depend on.

A good low phase noise measurement is quite an achievement. There are some good phase noise testers, but the prices are frightening.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline   Reply With Quote
Old 28th Oct 2022, 11:28 am   #193
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: 6-gang FM stereo tuner heads

Looking at the datasheet, the 1SS86 diode looks to be quite weedy as it is only rated to 30mA.

I've got various protection limiters here and unfortunately there is no 'universal' limiter design that I am aware of that can provide good protection at all frequencies and power levels (up to maybe 50W).

Several of my homebrew protection limiters use the classic 1N4148 silicon diode. I typically use 2 pairs of back-to-back 1N4148 diodes in a limiter and these diodes can typically withstand several watts of incident power. The back-to-back 1N4148 diodes should work quite well in a limiter from LF through VHF. Once up at UHF the 1N4148 diodes can lose their limiting action so they won't offer much protection against high power UHF signals. PIN diode limiters generally work well from about 50MHz through UHF. They don't work as well at low frequencies.

Some of my limiters just use four 1N4148 diodes and others include an RF fuse and a DC block ahead of the four 1N4148 diodes. The fuse adds some insertion loss, but the idea is that the fuse will blow before the diodes fail if the limiter sees >5W incident RF power.

The best high-power limiters that can work from VHF up to several GHz tend to use GaAs VPIN technology. Here, you can expect the limiter to cope with 70W to 100W incident power and the small signal insertion loss is typically less than 0.5dB through to 6GHz. When hard limiting at (say) 50W incident power, the leakage level through to the analyser/receiver is about 50mW. That is amazingly good performance. These limiters need careful thermal management. Think in terms of exotic PCB technology and a copper heat spreader in order to keep the limiter from overheating.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 28th Oct 2022, 11:46 am   #194
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: 6-gang FM stereo tuner heads

I haven't tried to use a decent spectrum analyser with valve equipment, but I think I'd avoid this as much as possible.

Instead, I think I would use my little RSP1A SDR with the spectrum analyser software if the aim was to look at narrowband VHF receivers that use valve technology.

I'd probably use RSP1A>>10dB attenuator>>diode limiter>>20dB_attenuator>>probe>>DUT

The diode limiter at the RSP1A starts to limit at 0dBm. The probe would be a cheapo homebrew active probe with its own front end DC block and protection diodes. The RSP1A is quite sensitive so I think the setup above would be OK for most things. I wouldn't be that upset if the RSP1A got damaged as it only cost about 100 new.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 28th Oct 2022, 12:03 pm   #195
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: 6-gang FM stereo tuner heads

See below for the insertion loss and return loss of one of my 1N4148 limiters. It's the circled one in the image below. It uses some really nice bulkhead N connectors and all that is inside it is 4 x 1N4148 diodes and a matching section.

You can see that I soldered the bulkhead connectors direct to thin metal to make up the enclosure. The aim here was to get the insertion loss as low as possible and to achieve 'instrumentation' levels for the return loss/VSWR.

I often use this limiter ahead of a thermocouple power meter and the VSWR needs to be good in order to minimise mismatch uncertainty. This was a serious attempt to optimise the performance as much as possible.

The return loss up to 1GHz is excellent as you can see. This took a few tweaks of the matching section to get this to work the way I wanted it. Note that the amplitude scale for the insertion loss is scaled at 0.1dB/div. The insertion loss is very low because I absorbed the stray capacitance of the diodes into the matching section to achieve a lowpass reponse.
Attached Thumbnails
Click image for larger version

Name:	Limiters4a.jpg
Views:	35
Size:	81.4 KB
ID:	267016   Click image for larger version

Name:	LIM00.PNG
Views:	32
Size:	8.9 KB
ID:	267017   Click image for larger version

Name:	LIM00S11.PNG
Views:	28
Size:	8.7 KB
ID:	267018  
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 28th Oct 2022, 1:33 pm   #196
regenfreak
Heptode
 
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
Default Re: 6-gang FM stereo tuner heads

Quote:
Originally Posted by Radio Wrangler View Post
A crystal oscillator ought to be a few of orders of magnitude better than that, so you're probably using the crystal oscillator to measure the sum of the phase noise of the local oscillators in your analyser.

You also have to take into account that the -3dB bandwidth of your IF resolution is not quite the same as the equivalent rectangular noise bandwidth. Also there's a subtle one in that many analysers perform logging before averaging and this distorts the pdf of the noise.

If your analyser has a "Noise Marker" function and it's been thoroughly done, it should account for these things for you.

One way of measuring phase noise down to lower levels is to do a mix-down with a known clean signal source, and measure the result with an audio spectrum analyser... they often have reduced phase noise pro rata to the lower frequency LOs they depend on.

A good low phase noise measurement is quite an achievement. There are some good phase noise testers, but the prices are frightening.

David
Yes, looking at the datasheet of my cheapo OCXO frequency source. part number OSC5A2B02, the phase noise is -150dbc/Hz at 10kHz offset. I attached the comparison of OCXO with other crystal oscillators as a typical example.

Based on the HP AN 150 series application note: Phase Noise Measurement Seminar, HP used a correction factor of 1.2 to find the equivalent noise bandwidth of the nominal 3db bandwidth of an IF filter.

Yes, Siglent has built-in noise marker, it should take care of the correction.

On a separate issue, I was reading the noise figure measurement in HP AN150-9 application note, they used a noise power correction factor of 1.7 to account for bandwidth, log amp and detector.

Quote:
G0HZU_JMR See below for the insertion loss and return loss of one of my 1N4148 limiters. It's the circled one in the image below. It uses some really nice bulkhead N connectors and all that is inside it is 4 x 1N4148 diodes and a matching section.

You can see that I soldered the bulkhead connectors direct to thin metal to make up the enclosure. The aim here was to get the insertion loss as low as possible and to achieve 'instrumentation' levels for the return loss/VSWR.

I often use this limiter ahead of a thermocouple power meter and the VSWR needs to be good in order to minimise mismatch uncertainty. This was a serious attempt to optimise the performance as much as possible.

The return loss up to 1GHz is excellent as you can see. This took a few tweaks of the matching section to get this to work the way I wanted it. Note that the amplitude scale for the insertion loss is scaled at 0.1dB/div. The insertion loss is very low because I absorbed the stray capacitance of the diodes into the matching section to achieve a lowpass reponse.
Thanks. The S21 and S11 look good.I may try 1N4148 limiters. Those metal enclosures are expensive. I thought the weediness of 1SS86 is a good thing, at least the diode strings blow before the transient voltage surge do any serious damage to the spectrum analyzer. But I may have to keep replacing the diodes...

Last night I ordered two cheap chinese "PIN Diode PIN limiter 10MHz to 6GHz" rated 0dbm to test if they are any good. They come in two versions; one without Faraday cage and the other has very nice CNC milled case. I bought the caseless version and will stick copper foils outside. The CNC cased version is a bit pricy.

Quote:
I haven't tried to use a decent spectrum analyser with valve equipment, but I think I'd avoid this as much as possible.
I did it once aligning an IF strip of an FM stereo tuner with B+ 250V using the Siglent SSA3021x Plus with just 1kv caps and SMT fuses as protection..It was before i realised that SSA3021x has a massive 1 microFard DC blocking cap at the input. I will never do it again. I will use my cheapo 200MHz NWT200 sweeper which costs me 70.
Attached Thumbnails
Click image for larger version

Name:	Phase noise crystall osc.jpg
Views:	24
Size:	55.8 KB
ID:	267026   Click image for larger version

Name:	rectangulat and IF correction.jpg
Views:	22
Size:	37.8 KB
ID:	267027  

Last edited by regenfreak; 28th Oct 2022 at 1:44 pm.
regenfreak is offline   Reply With Quote
Old 29th Oct 2022, 7:37 pm   #197
nemo_07
Triode
 
Join Date: May 2022
Location: Cologne, Germany.
Posts: 28
Default Re: 6-gang FM stereo tuner heads

Quote:
Originally Posted by G0HZU_JMR View Post
One thing to watch out for with any FET based follower circuit is (unwanted) negative resistance at the gate up at VHF. ...
I think that changing the bias resistor from 47Ω to 150Ω is very risky in terms of circuit stability because I expect that it will introduce a fair bit of negative resistance at the gate 1 pin of the BF998. This negative resistance will be generated across a huge bandwidth. Probably spanning VHF and also through a fair bit of the UHF band.
Good that you mentioned it.
The input resistance of source follower loaded with Rs||Cs, at frequency f, given gm, Cgs (ignoring gis) will be

Rp ≃ (gm*Rs+1) / [(6.28*f)*Cgs*Rs(Cgs-Cs*gm*Rs)] (holds well for f up to ca. 100MHz, <300MHz).
Hence the condition for input resistance being positive is

Cgs-Cs*gm*Rs > 0 (*).

Assuming a fixed biasing (for example Id≃10mA for gm=27mS) we still have 2 degrees of freedom to tweak around (i.e. Rs, Cs; we can also modify Cgs by adding a small capacitance):

Rs < Cgs / (Cs*gm) (**) or
Cs < Cgs / (gm*Rs) (***)

Now, with Cs=Cgs=3pF and for Rs = 47Ω||50Ω (perfectly terminated coax) ≃ 24Ω we have gm*Rs≃0.65, and from (*) we get
Cgs-Cs*gm*Rs = 3pF(1-0.65) = 1.05pF > 0 which is Ok, and we get Rp ≃ 87kΩ @100MHz which is also good.

For the same conditions but with Rs = 150Ω||50Ω (coax) ≃ 37.5Ω we have gm*Rs≃1.0125, and from (*) we get
Cgs-Cs*gm*Rs = 3pF(1-1.0125) = -37.5fF < 0 which is not OK and we end up with a negative Rp ≃ -2.5MΩ @100MHz.
At 1GHz the |Rp| will be ca. two orders smaller.
But adding only 0.5pF to Cgs or decreasing Cs will keep things straight.

If we include a DC coupled emitter follower buffer (with series resistors 47Ω on input and output, and Ie≃10mA), we get good separation of source from any parasitics and simultaneously good match to 50Ω coax, with ca. 1Vpp compliance (or 0.5Vpp at terminated coax end), doubled for Ie≃20mA.
For this purpose I would arrange a combined Rs ≃ 75Ω (for AC, but ≃ 200Ω for DC), make Cs = 1pF, and Cgs = 3pF (the total capacitance at G1, Cgtot will still be small, ~2pF).

BTW: The Vdd should be raised by at least 1V for good linearity.
The "Battery reversal protection" with D1 makes me wonder ... Do they still 1:1 copy the original circuit?
nemo_07 is offline   Reply With Quote
Old 30th Oct 2022, 12:08 pm   #198
regenfreak
Heptode
 
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
Default Re: 6-gang FM stereo tuner heads

I received the Tek P6202A and pogo test pins this week. Last night I homebrewed a miniature power plug for P6202A -15V, +15V input socket and powered it up with the Farnell Triple Tops 2D power supply. It is working. I put the ground spring pin from my Rigol scope probe into the tip to eliminate the need of a ground lead.

The gain is very flat when i tested with my short transmission line, however, the results vary depending on whether I terminate the line with a 50-ohm load or not:

gain = -23db with 50 ohm termination of the transmission line
gain = -17.8db with open line

Furthermore, i measured the gain of P6202A with a FY6900 signal gen and a 200Mhz scope at 100MHz (with 50 ohm input). The measured gain is -20db or 10:1 voltage ratio which is the manufacturer's spec. Therefore, the actual gain is somewhere between the open and 50 ohm terminated transmission line measurements. This throws up my long-time question of the measurement of high impedance DUT with large mismatch uncertainty.

I have tried to measure the Rp, Xp of the P6202A (attached) I believe the service manual of the P6202A is incorrect to quote Rp >100k from DC to 500ohms. The Rp should decreases sharply with increasing frequency based whatever small-signal, high frequency model of FETs.

I probed the IF output of FM tuners with P6202A and never get anything like this guy got with the P6202A and HP spectrum analyzers:

http://ham-radio.com/k6sti/swept.htm

I am a bit flabbergasted; he got nice response bandpass curves using the "residue sample at the IF output" while I got flatliner on the spectrum analyzer....Have i missed something?

As expected, the P6292A has greater probe loading than the Poorman's probe due to high input capacitance (about 1.8pF). It loads down the 10.7NHz IF strips greatly, flattening the Qs of the passband filters.


I have attempted to measure the gain of the Poorman probe with a 0.3PF air gap gimmick C1 (.measured at 100kHz with DE-5000). I have used pogo pins for both the probe tip and ground tip. Its performance in term of the flatness of S21 is rather dismal (see attachments). The gain is below -40db (equivalent to 100:1 probe). Note that it has become noisy and difficult to get repeatable S21 measurement even I added a 30db LNA to amplify the signal. The NanaoVNA V2 Plus 4 has a dynamic range is close to 90-100db with averaging.

So I have gone back to Porrman's probe with higher input capacitance (C1 = 0.79p measured at 100kHz with DE-5000), its S21 flatness and gain are more acceptable. Here i have attached the equivalent parallel and series resistance; Rp, Xp, Rs, Cin and ILs. They are crude measurements really but the order of magnitude seems to be OK.
Attached Thumbnails
Click image for larger version

Name:	Tek probe P6202A.jpg
Views:	35
Size:	72.0 KB
ID:	267149   Click image for larger version

Name:	P6202A Rp Xp opwn tran line.jpg
Views:	33
Size:	58.9 KB
ID:	267150   Click image for larger version

Name:	Poorman C1 = 0.79p.jpg
Views:	26
Size:	29.5 KB
ID:	267152   Click image for larger version

Name:	Poorman C1  0.35p gimmick.jpg
Views:	35
Size:	32.1 KB
ID:	267156   Click image for larger version

Name:	poorman's correction.jpg
Views:	27
Size:	61.5 KB
ID:	267163  


Last edited by regenfreak; 30th Oct 2022 at 12:36 pm.
regenfreak is offline   Reply With Quote
Old 5th Nov 2022, 12:31 am   #199
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: 6-gang FM stereo tuner heads

I had a go at modelling both the P6202A probe and the poorman's probe and in both cases the input reflection coefficient is extremely close to 1 all the way up to about 300MHz.

At 150MHz it is in the region of 0.9999 for both probes and this is asking too much of a real world VNA to measure using a classic 1 port s11 measurement. Therefore, whatever the VNA displays, it won't be the correct result. Even my expensive lab VNA will fail here when trying to measure s11.

Both probes are quite similar in terms of the input circuit. The P6202A uses an 8M series tip resistor and a 2M resistor to define a resistive tap. There's also a tiny fraction of 1pF across the 8M tip resistor as part of a compensated capacitive divider. This compensation capacitance is going to be similar to the series cap at the tip of the poor man's probe. Up at RF, the 8M resistor will effectively be invisible.

The P6202A uses a JFET, and this normally has higher input capacitance. However, there is another JFET in the source connection that acts as a current source and the source is only loaded by a common collector BJT. Therefore, the Cgs capacitance of the input JFET will be virtually nulled out.

The P6202A does things this way because the input JFET has to act as a faithful and accurate voltage follower for use with an oscilloscope. The price for this will be lots of negative resistance at the input because the overall circuit at the source pin of the input JFET will mimic a shunt capacitance with a very high shunt resistance. There will also be some Cdg capacitanceat the input of the JFET. The RC snubber circuit attempts to cancel the negative resistance at the cost of slightly increased input capacitance.

I think a fair bit of the input capacitance of the P6202A probe will be in the elongated tip section. If this tip section wasn't there, I think it would have lower input capacitance.

Realistically, I think the best way to measure the Rp and Cp performance of these probes is to measure a very narrow (single resonator) 100MHz BPF with a VNA with an s21 measurement and then see how much the capacitance of the probe pulls the BPF response down in frequency when the probe is connected to the resonator of the BPF. Then see how much the level of the s21 measurement drops as well. This will give an idea for the probe Rp. If the level goes up then Rp is negative. I suppose you could tweak the probe design to get no change in S21 level and this would mean the input Rp of the probe was as high as possible (at the 100MHz test frequency) without actually turning negative.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline   Reply With Quote
Old 5th Nov 2022, 3:30 pm   #200
regenfreak
Heptode
 
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
Default Re: 6-gang FM stereo tuner heads

Quote:
Originally Posted by G0HZU_JMR View Post
I had a go at modelling both the P6202A probe and the poorman's probe and in both cases the input reflection coefficient is extremely close to 1 all the way up to about 300MHz.

At 150MHz it is in the region of 0.9999 for both probes and this is asking too much of a real world VNA to measure using a classic 1 port s11 measurement. Therefore, whatever the VNA displays, it won't be the correct result. Even my expensive lab VNA will fail here when trying to measure s11.

Both probes are quite similar in terms of the input circuit. The P6202A uses an 8M series tip resistor and a 2M resistor to define a resistive tap. There's also a tiny fraction of 1pF across the 8M tip resistor as part of a compensated capacitive divider. This compensation capacitance is going to be similar to the series cap at the tip of the poor man's probe. Up at RF, the 8M resistor will effectively be invisible.

The P6202A uses a JFET, and this normally has higher input capacitance. However, there is another JFET in the source connection that acts as a current source and the source is only loaded by a common collector BJT. Therefore, the Cgs capacitance of the input JFET will be virtually nulled out.

The P6202A does things this way because the input JFET has to act as a faithful and accurate voltage follower for use with an oscilloscope. The price for this will be lots of negative resistance at the input because the overall circuit at the source pin of the input JFET will mimic a shunt capacitance with a very high shunt resistance. There will also be some Cdg capacitanceat the input of the JFET. The RC snubber circuit attempts to cancel the negative resistance at the cost of slightly increased input capacitance.

I think a fair bit of the input capacitance of the P6202A probe will be in the elongated tip section. If this tip section wasn't there, I think it would have lower input capacitance.

Realistically, I think the best way to measure the Rp and Cp performance of these probes is to measure a very narrow (single resonator) 100MHz BPF with a VNA with an s21 measurement and then see how much the capacitance of the probe pulls the BPF response down in frequency when the probe is connected to the resonator of the BPF. Then see how much the level of the s21 measurement drops as well. This will give an idea for the probe Rp. If the level goes up then Rp is negative. I suppose you could tweak the probe design to get no change in S21 level and this would mean the input Rp of the probe was as high as possible (at the 100MHz test frequency) without actually turning negative.

Thanks. Fundamentally, there are intrinsic uncertainties with the methodology of the measurements when the DUT has very large impedance mismatch with the VNA. In my case, I inserted a short 50 ohm transmission line for the measurements of the Poorman's and P6202A probes. The measured S21 results for the open and 50 terminated line are different because the impedance of the probes are somewhere between 50 ohms and infinity. The actual gain (or insertion loss) is somewhere between the open and closed transmission line measurements as verified by the use of the oscilloscope measurement.

I have looked at my 100MHz bandpass filters. By design, they are meant to be maximally flat so the Q are moderate. So I have quickly tried to measure my 465MHz cavity filter with 3 helical resonators(Q=166-250 depending the insertion loss and bandwidth adjustment).

In the 2nd attachment, it was measurement with direct through measurement having the short transmission line included in the calibration and measurement. The S11 is quite poor which suggests that this filter may not be designed for 50-ohm system.

In the 3rd attachment, it is measurement with the Poorman's probe (0.7pF input capacitance) pogo pins pressing against the open transmission line. You can see the bizarre dip at the resonance frequency. I dont know what is going on here.

4th attachment, Poorman's probe with the transmission line terminated by 50 ohm load.


I think the stray capacitance of the P6202A is not due to the length of the tip but it is due to distributed capacitance between the probe tip and the outer ground ring (see attachment 5). My poorman's has much longer probe tip and ground pin but it still has far less input capacitance...my probe tip and ground pin are very far apart..and also I removed most of the ground traces nearby to cut the stray capacitance.

I have made some custom 50-ohm, short, open and through standards in order to measure the Rp, Rs, Cp, Cs of a 10K 1206 resistor, my results look similar to the link below at high frequencies, but they are very different at lower frequencies:

https://www.w0qe.com/Measuring_High_Z_with_VNA.html

PS, correction of my previous post; the P5202A works well with the alignment of 10.7MHz LC IF filters.
Attached Thumbnails
Click image for larger version

Name:	helical resonators Q =166.jpg
Views:	16
Size:	51.6 KB
ID:	267454   Click image for larger version

Name:	helical direct throu poorman.jpg
Views:	18
Size:	35.1 KB
ID:	267455   Click image for larger version

Name:	helical open line poorman.jpg
Views:	18
Size:	35.5 KB
ID:	267456   Click image for larger version

Name:	helical 50 ohm terminated.jpg
Views:	21
Size:	38.3 KB
ID:	267457   Click image for larger version

Name:	P6202A probe tip.jpg
Views:	23
Size:	77.3 KB
ID:	267458  


Last edited by regenfreak; 5th Nov 2022 at 3:39 pm.
regenfreak is offline   Reply With Quote
Reply

Thread Tools



All times are GMT +1. The time now is 2:37 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 - 2023, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.