6-gang FM stereo tuner heads

[regenfreak]

another series resistor to the IF out hot side, as required (value depending on IF transformer ratio; ~50-150 Ohm should do).

No, 50-150 resistor would heavily load down the IF amp LC tank giving completely wrong response curve. Min value would be 100k to 330K plus HV DC isolation cap. Valve IF amp has very high I/O impedance.

[nemo_07]

Originally Posted by regenfreak

I am not interested in making sweeps from antenna input to IF out because my FM IF strips use complicated staggered tuning (in contract to synchronous tuning in mono FM or AM IF strips) to max out the stereo fidelity. The IF responses will mask out the misalignment issues in the individual single and double tuned RF front end stages. It is easy to do frequency sweep of IF stages but it is hard to do visual sweep on the FM front ends...

I didn't mean the output of the IF strip, but the output of the IF transformer of front end, and assumed it having a smaller bandwidth than that of preceding stages. You could get a meaningful picture, improving the resolution of the sweep by adding another IF filter (of high Q) past this IF transformer.
And sorry, I had a typical transistor-oriented IF secondary impedance ~100-200 Ohm in my head.

[nemo_07]

Originally Posted by regenfreak

No, 50-150 resistor would heavily load down the IF amp LC tank giving completely wrong response curve. Min value would be 100k to 330K plus HV DC isolation cap. Valve IF amp has very high I/O impedance.

And again a correction is due: the narrow IF bandwidth, as mentioned, is good for inspection of distortions and spurs, but for gain and selectivity sweeps the opposite is true, i.e. the IF transformer bandwidth should be made considerably wider than the expected front bandwidth.

[regenfreak]

I didn't mean the output of the IF strip, but the output of the IF transformer of front end, and assumed it having a smaller bandwidth than that of preceding stages. You could get a meaningful picture, improving the resolution of the sweep by adding another IF filter (of high Q) past this IF transformer.
And sorry, I had a typical transistor-oriented IF secondary impedance ~100-200 Ohm in my head.

As I said before, once you take the output of the first IF stage, you mask out the critical information about the tracking and sweep response of the variable double tuned stages across 88-108MHz. The IF stage has its own response shape. The sweep response from the output of the 1st IF stage does not show the true overall response of the single-double-double tuned stages of the RF front end. I dont have the proper equipment to measure distortion.

I managed to do RF sweep of RF frond end of a 4 gang FM double gate FET tuners with high Z probe but have not been successful with valve FM tuners. Great caution must be taken with valve tuners.

I am starting to build the dual gate FET version of the 6-gang tuner. It should be easier to test and align without worrying about the high voltage or frying the measurement equipment. But it will be still tricky to align those double tuned amp stages because we still have high impedance in the I/O of the filters.

the IF transformer bandwidth should be made considerably wider than the expected front bandwidth

The opposite is true. The IF stages provides lots of the selectivity. Typical FM IF bandwidth is about 250KHz for wide IF high fidelity stereo. The RF front end has quite wide bandwidth (see the RCA paper example in my previous post number 20 and 25 of this thread. With double-single tuned stages, it is about 700-800kHz). This is because it is very difficult to align multiple tuned gangs (over 3 gangs) because of the tracking errors and imperfect filter response shapes. The bandpass shape and bandwidth of the double tune filter changes with the frequencies across the band. 2nd order Butterworth bandpass filter has very sharp tuning; a small misalignment would cause massive insertion losses. On the other hand, a single tuned LC gang has a relatively low loaded Q, small misalignment wont hurt much.

[nemo_07]

Originally Posted by regenfreak

I managed to do RF sweep of RF frond end of a 4 gang FM double gate FET tuners with high Z probe but have not been successful with valve FM tuners. Great caution must be taken with valve tuners.

I am starting to build the dual gate FET version of the 6-gang tuner. It should be easier to test and align without worrying about the high voltage or frying the measurement equipment. But it will be still tricky to align those double tuned amp stages because we still have high impedance in the I/O of the filters.

The opposite is true. The IF stages provides lots of the selectivity. Typical FM IF bandwidth is about 250KHz for wide IF high fidelity stereo. The RF front end has quite wide bandwidth (see the RCA paper example in my previous post number 20 and 25 of this thread. With double-single tuned stages, it is about 700-800kHz). This is because it is very difficult to align multiple tuned gangs (over 3 gangs) because of the tracking errors and imperfect filter response shapes. The bandpass shape and bandwidth of the double tune filter changes with the frequencies across the band. 2nd order Butterworth bandpass filter has very sharp tuning; a small misalignment would cause massive insertion losses. On the other hand, a single tuned LC gang has a relatively low loaded Q, small misalignment wont hurt much.

Again, you will need a narrow bandwidth preselection to visualize and locate spurs.
But to visualize the bandwidth of the front end you will need a "panorama" bandwidth at the detector (or to kill the IF transformer selectivity). A simplest way would be to replace the IF transformer at the mixers plate by ~1k Ohm resistor and to use a cheapest version of Hi-Z 1:1000 probe consisting of series connected 1pF/500V mica and 1000pF (plus a Si diode for additional DC leak protection, if you insist). This, of course, would require few orders higher generator levels-

[Radio Wrangler]

Regenfreak, It would help if you didn't cut quoted text so closely and left in the bit that says who you're quoting. It makes the thread a lot esier to follow. Where this bit is missing we sometimes get other preople thinking it was something they said and the confusion just explodes.

David

[regenfreak]

Originally Posted by nemo_07

Again, you will need a narrow bandwidth preselection to visualize and locate spurs.
But to visualize the bandwidth of the front end you will need a "panorama" bandwidth at the detector (or to kill the IF transformer selectivity). A simplest way would be to replace the IF transformer at the mixers plate by ~1k Ohm resistor and to use a cheapest version of Hi-Z 1:1000 probe consisting of series connected 1pF/500V mica and 1000pF (plus a Si diode for additional DC leak protection, if you insist). This, of course, would require few orders higher generator levels-

Interesting idea. I can give it a try. i have tried all kind of tricks on the book you can imagine. Unfortunately, i dont have a VHF signal generator at the moment. The TinySA may have
not enough power level in its RF generation output.

[regenfreak]

Originally Posted by Radio Wrangler

Regenfreak, It would help if you didn't cut quoted text so closely and left in the bit that says who you're quoting. It makes the thread a lot esier to follow. Where this bit is missing we sometimes get other preople thinking it was something they said and the confusion just explodes.

David

Oops.sorry. My last few replies were for Nemo 07.

[nemo_07]

Originally Posted by regenfreak

... Interesting idea. I can give it a try. ... The TinySA may have not enough power level in its RF generation output.

Wait a minute. It is variant one of three.
If you use it with a 1m long piece of high quality coax cable, it will show ultra-low loading, pretty good amplitude flatness and essentially constant, nearly zero phase shift, from ~10MHz(-3dB low corner) well into GHz range, provided C2 (1000pF) is a leadless (chip) type and the diode is a small signal low capacitance type (no Schottky).
With its ~1pF (C1) input it would detune a bit IF tanks, but completely the front end. With its insertion loss of ~60dB and your TinySA (noise floor -102dBm @ 30kHz RBW, max sin output -6dBm) the max. available span for selectivity check would be ~36dB for a cold tank, increased by a gain (if any) in the hot state.

Removing the C2, you get a "narrow band" (a sort of high pass amplitude response with a slope ~20dB/decade) variant 2, with IL~50dB @10MHz and ~30dB @100MHz, and varying phase shift (not suitable for VNA measurements, but still usable for narrow band scalar sweeps). The possible span will improve by ~10dB @10MHz to ~30dB @100MHz (for cold unit). Loading @100MHz will be noticeable.

Now, adding a 0.1pF/500V (C01) in series to C1 makes up the variant 3. The IL will jump to ~70dB @10MHz to ~50dB @100MHz, but the span for hot unit with some gain would be usable. And now you'll get virtually no loading @10-100MHz band, and max. detuning @100MHz less than -200kHz.

Would it be a crack enough?

[regenfreak]

Originally Posted by nemo_07

Wait a minute. It is variant one of three.
If you use it with a 1m long piece of high quality coax cable, it will show ultra-low loading, pretty good amplitude flatness and essentially constant, nearly zero phase shift, from ~10MHz(-3dB low corner) well into GHz range, provided C2 (1000pF) is a leadless (chip) type and the diode is a small signal low capacitance type (no Schottky).
With its ~1pF (C1) input it would detune a bit IF tanks, but completely the front end. With its insertion loss of ~60dB and your TinySA (noise floor -102dBm @ 30kHz RBW, max sin output -6dBm) the max. available span for selectivity check would be ~36dB for a cold tank, increased by a gain (if any) in the hot state.

Removing the C2, you get a "narrow band" (a sort of high pass amplitude response with a slope ~20dB/decade) variant 2, with IL~50dB @10MHz and ~30dB @100MHz, and varying phase shift (not suitable for VNA measurements, but still usable for narrow band scalar sweeps). The possible span will improve by ~10dB @10MHz to ~30dB @100MHz (for cold unit). Loading @100MHz will be noticeable.

Now, adding a 0.1pF/500V (C01) in series to C1 makes up the variant 3. The IL will jump to ~70dB @10MHz to ~50dB @100MHz, but the span for hot unit with some gain would be usable. And now you'll get virtually no loading @10-100MHz band, and max. detuning @100MHz less than -200kHz.

Would it be a crack enough?

I don't quite follow the logic of your three ideas. Why do you need such large IL as big as -70-50db for the LC tank to qualify to be "lightly loaded"? My NWT200 has only the dynamic range of 70db. The TinySA has the dynamic range probably of about 100db (it states 120db for its detector).

I have 0.1pF smd capacitors but if you put two wires a few mm apart, its stray capacitance would be much bigger!

In the measurement of unload Q of a single tuned LC resonator with capacitive coupling wires (like in a cavity resonator) for I/O using a spectrum analyzer with tracking generator ( I use NWT200), all you need is to introduce about -20 to -30db IL for the capacitive coupling wires for the measured Q to be unload (Qo). Please see the equations and graph of Qu/QL versus attenuation in db (red line).

[nemo_07]

Originally Posted by regenfreak

I don't quite follow the logic of your three ideas. Why do you need such large IL as big as -70-50db for the LC tank to qualify to be "lightly loaded"? My NWT200 has only the dynamic range of 70db. The TinySA has the dynamic range probably of about 100db (it states 120db for its detector).

You get it wrong way.
If you put a probe at hot end of 100MHz tank you get excessive detuning, unless the tip capacitance is ~0.1pF or less. The high IL is the consequence, not prerequisite.
You wanted Hi-Z probe with Lo-Z output, but found nowhere. So, here you've got a proposition. Take it or leave it.
Anyway, you are free to experiment with values of your choice.

Originally Posted by regenfreak

I have 0.1pF smd capacitors but if you put two wires a few mm apart, its stray capacitance would be much bigger!

It would be higher, it could be smaller. Look at the basics.
Or am I to believe that fabricating 0.1pF SMD capacitors is a pure nonsense?
BTW. It (C01) doesn't have to be SMD type. If you find and get a leaded 0.1pF one, the apparent paradox of wires will ... disappear. Right?

[regenfreak]

Going back to the idea of having 50-ohm I/O bandpass filters, I have acquired some MMIC Gali-51+, MAR-1, Mar-2 Mar-4, 6 and 11. Theses have Darlington configuration internally matched to 50ohms. The advantages of 50ohm I/O impedance is easy to align with VNA or sweep generator. I thought I could explore the use of passive ring diode mixer like SBL-1+ or active double-balanced Gilbert cell mixer/oscillator NE602.


In order to test the feasibility of the ideas, I built the fixed tuned MMIC cascade amp with Gali-51+ with double-tune capacitive coupled bandpass and inductive coupled bandpass. It is working ok.

Furthermore I made a variable tuned MMIC amp using a 7-gang air gang variable capacitor. It is too lossy and difficult to tune to my liking. After all, it seems not a good idea. When it comes to variable tuned MMIC cascade amplifier, it is impossible to obtain the end coupling caps matching 50ohm I/O impedance across the whole FM band. The Hi-Z to 50-ohm transformation is frequency dependent. To add to the complication, the nodal frequencies are much higher than the filter center frequency in 50 ohms matched version, this makes it impossible to obtain an exact analytical solution that simultaneously satisfies all the constraints: acceptable bandwidth, impedance matching and tracking correctly across the FM band. So I am going back to the high Z double dual gate FET project.

Attachment 3-4 is the testing of SBL-1+ with my new poorman's 100MhZ FY6900 signal gen.

Attachment 6 is my 300 ohms to 50 ohm balun. I tested with 300ohm resistor, I got SWR of 1.4 at FM band. I am working on a new version that will give SWR close to 1 at VHF.

I am hoping to get a combiner and will measure the IP3 two tune measurement of an amplifier.

[nemo_07]

Originally Posted by regenfreak

... I made a variable tuned MMIC amp using a 7-gang air gang variable capacitor. It is too lossy and difficult to tune to my liking. After all, it seems not a good idea. When it comes to variable tuned MMIC cascade amplifier, it is impossible to obtain the end coupling caps matching 50ohm I/O impedance across the whole FM band. ...

Attachment 6 is my 300 ohms to 50 ohm balun. I tested with 300ohm resistor, I got SWR of 1.4 at FM band. I am working on a new version that will give SWR close to 1 at VHF.

There is not really a problem to match Hi-Z resonant circuits to any arbitrary low impedance. Look at any application note for any VHF/UHF dual gate MOSFET. A mismatch along FM band (around 3dB or so) is not a problem also (easily solved for purpose of any measurement).
The problems with MMICs, like for any BJTs, are intermodulations and spurs. Diode mixers are not much better in this respect.
... But you might put a MMIC or two in use to boost the dynamic range of the Hi-Z probe we've talked about.

This balun doesn't look good. Low insertion loss requires tight coupling, what means good intimacy of wires involved over the whole length of winding. Think of them as of a micro-transmission line. To wind a good 300 ohms to 50 ohm (6:1) balun is not a trivial task. You would better take a 4:1 one and use ordinary resistive matching.
If you need isolated primary to secondary, 300 Ohm to 75 Ohm balun (or 200 Ohm to 50 Ohm), you would start with 3 pieces of thin enameled copper wire, ~10 cm long, as convenient. Twist them together (about 1 full turn/cm length), then wind through the core few turns (say 6), cut the ends to the required length. Now, the ends of one of the wires will be the 75 ohm side. The other two will be connected in series (take care of phases), common point serving as mid-tap, both other ends as 300 Ohm balanced ends.
You may look at the Figure 2 here: https://www.minicircuits.com/pdfs/MPD_Transformers.pdf
This is a 4:1 balun wound on a toroidal core.
Note, how the separation of primary versus secondary terminals was solved, whereby all 3 wires pass exactly the same path around the core.
NB. For sake of core losses its volume should be kept as small as possible. High permeability material is advantageous.

[regenfreak]

Originally Posted by nemo_07

There is not really a problem to match Hi-Z resonant circuits to any arbitrary low impedance. Look at any application note for any VHF/UHF dual gate MOSFET. A mismatch along FM band (around 3dB or so) is not a problem also (easily solved for purpose of any measurement).
The problems with MMICs, like for any BJTs, are intermodulations and spurs. Diode mixers are not much better in this respect.
... But you might put a MMIC or two in use to boost the dynamic range of the Hi-Z probe we've talked about.

This balun doesn't look good. Low insertion loss requires tight coupling, what means good intimacy of wires involved over the whole length of winding. Think of them as of a micro-transmission line. To wind a good 300 ohms to 50 ohm (6:1) balun is not a trivial task. You would better take a 4:1 one and use ordinary resistive matching.
If you need isolated primary to secondary, 300 Ohm to 75 Ohm balun (or 200 Ohm to 50 Ohm), you would start with 3 pieces of thin enameled copper wire, ~10 cm long, as convenient. Twist them together (about 1 full turn/cm length), then wind through the core few turns (say 6), cut the ends to the required length. Now, the ends of one of the wires will be the 75 ohm side. The other two will be connected in series (take care of phases), common point serving as mid-tap, both other ends as 300 Ohm balanced ends.
You may look at the Figure 2 here: https://www.minicircuits.com/pdfs/MPD_Transformers.pdf
This is a 4:1 balun wound on a toroidal core.
Note, how the separation of primary versus secondary terminals was solved, whereby all 3 wires pass exactly the same path around the core.
NB. For sake of core losses its volume should be kept as small as possible. High permeability material is advantageous.

My current 1: 6 balun is based on this design:

https://pa1ejo.wordpress.com/2020/05...50-to-300-ohm/

His one has SWR of 1.33 at 10MHz and approx. 2.0 at 100MHz. Mine has SWR = 1.4 at 100MHz which is not bad. I use BN61-1502 binocular core.

I am going to build the 5 wire version based on a web article. Annoyingly his web site is down now:
https://vk6ysf.com/balun_6-1.htm

I am waiting for the multi-color ribbon cable to arrive on the post. I bought a T130-17 for this.

I dont need to make 1:4 balun as I have some ready-made micro smd 1:4 ones lying around. Previously I purchased them to make a crystal tester for VNA.

PS I have ordered a RF power splitter with -35db isolation so the two ports do not see each in the IP3 measurement.

[regenfreak]

Actually the SWR (attached )of my 6:1 balun looks better than the one in the article.

[Radio Wrangler]

Be careful not to judge baluns entirely on VSWR. Loss can be important.

A 50 ohm resistor or 75 ohm has wonderful VSWR!

David

[regenfreak]

Originally Posted by Radio Wrangler

Be careful not to judge baluns entirely on VSWR. Loss can be important.

A 50 ohm resistor or 75 ohm has wonderful VSWR!

David

That's true.

vk6ysf web site is still down. His design his 5 windings Ruthroff voltage transformer and is relatively easy to wind. It has a measured VSWR of 1.

This is an alternative 5-windings 6:1 design which shows a VSWR of about 1. It is more complicated to wind than vk6ysf version.

http://www.spirat.com.au/vk5zvs/pic32.htm

[regenfreak]

Clive Ruthroff 1959 paper:

http://sezador.radioscanner.ru/artic...f_Aug-1959.pdf

[Radio Wrangler]

Thanks, one I left in my collection at HP.

David

[regenfreak]

At last I have completed the dual gate FET version of the 6-gang tuner ( 2 x 3SK45 plus BPJ mixer and oscillator). After simple alignments of the oscillator and the antenna coil, I have powered it for the first time. Similar to the valve version, it is tracking perfectly and working brilliantly with excellent sensitivity and selectivity. It sounds amazing even I have not adjusted the two bandpass stages. I simply connected to it to the IF input of my DIY valve stereo FM tuner. I can get full stereo MPX for all the weak and strong stations across the FM band,. It is one of the rare occasions a DIY project that worked the first time in the first power-up, it is very satisfying! So the project has been a success.

The unload Qu of the front end tank is measured to be 130. In the design calculations, I chose lower I/O impedance (approx. 5.3K ohms) for the solid state variable tuned bandpass filters on purpose. I have used coil tappings for all the input and output bandpass stages to reduce loading effect of the dual gate FETs on the LC tanks. The positions of the tappings are just rough guesses....It is very complicated to calculate the tapping points of the coils. I have read the below IEEE paper by Klein and I can struggle to understand the design calculations involving I/O conductance, admittance, impedance and stability criteria. But the author showed how the coil tappings can be determined in his example calculations:

Mosfet tuner design by Klein, 1970, Taxes Instrument:

https://ieeexplore.ieee.org/document/4079818

I have used 0-1pF silver plated piston trimmers for the bandpass coupling capacitors.