Note: At around 3:00 I mentioned "5 turns less" - I meant to say "turns divided by 5". So say 15 turns on the resonator side would be 3 turns on the input side.
Informative demonstration of BPF coupling. Those T37-0 cores must have mostly filler in the mix. Re link coupling into and out of BPFs at HF, I've always used 3T and have never seen anyone do anything to optimise the impedance match to 50r of these windings.
The 'Mix 0' is just a phenolic form, isn't it? This project sounds like fun. I really like such experimental builds. BTW, I've used 'gimmicks' occasionally over the years. They always seemed like a shaky idea, but they always worked well and maintained their capacitance. Of course, that depends on the wire type used.
Thats what I thought too John but this www.amidoncorp.com/t37-0/ says its got an iron powder core. It does seem like its just phenolic - I don't have enough to break one and find out! Thanks for the comment!
@@na5y, I actually just called them and they verified that the zero material was indeed iron powder. BUT, unfortunately, Micrometals says the zero mix is phenolic. RF Microwave lists the Mix 0 as follows: Iron powder toroid core T50-0, non magnetic MIX 0 - KEY FEATURES - Non-magnetic, non-conductive phenolic material I've been around the industry long enough to know that even corporations get things wrong, (like 5,000 diodes with the band painted on the anode end.) I think the reality of the situation is that these manufacturers provide phenolic toroids for Mix 0, but list them with their iron powder cores because the 'Mix 0' doesn't fit into any other category, and so some of the documentation, (such as the above,) ends up calling it powdered iron when it's actually phenolic. Perhaps a magnet can verify that.
Your are right it a good idea to know effect of some parts. Il you want a simulation closer to reality always add a low series resistor with your inductor. You will see the effect on quality factor. Yes at 144Mhz with toroid, the effect is lower...This is the first parameters we read when we choose inductor in industry design. At 144Mhz you can also use proximity coupling with air coil : more difficult to calculate but you can adjust both inductor value and coupling for experiment. At this frequency we can simulate accurate S21 design with : inductors S parameters, smd parts, ideal capacitor in simulation, add PCB capactive effect : about 1pF/parts. Easier if you replace L transformer with capacitive transformer. At the end the true result is very close to measurement. More difficult at 440Mhz. For transformer ratio calculation and impedance LC I suggest read SA605 SA602 philips application note.
Oleg, this is a direct conversion I/Q receiver - the output of the mixing process will produce an I and Q audio signal. That Audio signal can then be subsequently digitized and pretty much any demodulation can be applied (SSB, AM, FM, etc). The input to the mixing process is the RF itself + an LO which will come from an si5351 frequency synth. There is however no digitizing of signal on this board. I hope this answers your question and thank you for the comment!
@@na5y Thanks, but I know how it works because I made it myself (ruclips.net/video/nGn-KkACT7o/видео.html). I'm wondering how such a high frequency will work?
@@R1CBU Its all analog on this board though - the FST3253 is the mixer and it can switch up to GHz frequencies. The si5351 can output at least into 200 MHz. Once the mixing is done its all at audio frequency after that.
The maximum frequency of operation of Tayloe Detector is limited by FST3253, particularly by the time of channel switching. It can be from 1 to 6 ns according to the datasheet, which gives us the maximum **guaranteed** receiving frequency of 41 MHz. In practice it may work up to 75 MHz or so but definitely not at 150 Mhz. You are going to need a downcoverter for this.
Yeah I had read about the downconverter need elsewhere. Still I will try... Just walk me through your math though. With a quadrature signal at 2,14 of the FST3253 that will result in transition through 00, 01, 11, 10 at 4x the LO frequency. Going from a 0 to 1 on say, pin 2 will take between 0-6ns. Worst case that will take 6ns - and because I need to transition 4 times per cycle that will take 24ns - which is 41 MHz. Is that your thinking?
@@na5y Yes. Also, I actually checked that it doesn't work on such high frequencies. The instance I tested worked up to 75 Mhz, so it had 1000*1000*1000/75/1000/1000/4 = 3.3 ns switching time.
@@R2AUK Good to know - like I said I will proceed with the experiment and see where it lands me! Its interesting that there is such a wide range of switching times in the datasheet. The TI datasheet says 1.5ns as minimum which would be approx 160Mhz. Like you said thats not even close to a guarantee. Interesting also in Dan Tayloe's original paper he talks about the solution potentially working into the GHz range. I wonder if he was referring to a theoretical switch rather than the FST3253. Thanks for the comment Alex!
If you make this work it will be a bloody miracle 😊 seriously use a single conversion Rx with an IF around 9mhz. This is what my current project is about. Single conversion hf transceiver using some of the ideas from your channel. No analogue audio switching, route i2s internally.
I have my doubts - we'll see. The FST3253 based Tayloe detector can go up to GHz, and the si5351 can certainly do 150MHz quadrature. The traces on the board may be a problem of course.
I like miracles. I like to push circuits to their limits, then do work-arounds to push them a little further. To hand-match components. To come up with tricks and innovations. To 'make something work'. The 'gimmick' is a perfect example. I've used them for years, and the hams around me just look at them and shake their heads. They think you have to buy a specific component with a specific value stamped on it from a manufacturer in order to make a circuit work. But you don't. Not if you know what you're doing. Electronics is fun. :)
I'm not knocking your try at this, but it has real difficulties. I've done this stuff in the past, a long time ago in the land of Oz. I live in nl these days. My current RF knowledge is a bit rusty these days but if you want to bounce some ideas around I would be happy to pitch in.73s. a Long time ex vk6
Sure Anthony - happy to chat. Should be straightforward to find out whether or not it is going to at least do the mixing or not. Sensitivity is a whole other thing. More to come!
@@na5y The problem with the 3253 is best case the S inputs can run at 1GHz, no typical figures and worst case of around 200mHz. You need then to allow for prop max of 0.25ns, so to get an output best base your operation could no more than 750Mhz approx. That wouldn't provide enough time to charge the integrating cap. The other problem is gain. If you check the sensivity of this detector at 10M compare to 20M you will find the gain has dropped, not sure how much, 3db at a guess. There are no specs for this, but is only going to be a lot worse at 150mHz.
@@anthonyclark7290 Your arguments make sense - I am definitely interested in whether this will work practically. We'll see - I haven't even run the si5351 at 150Mhz in quadrature - thats the first open question for me. Thank you for the comment!
Note: At around 3:00 I mentioned "5 turns less" - I meant to say "turns divided by 5". So say 15 turns on the resonator side would be 3 turns on the input side.
Informative demonstration of BPF coupling. Those T37-0 cores must have mostly filler in the mix. Re link coupling into and out of BPFs at HF, I've always used 3T and have never seen anyone do anything to optimise the impedance match to 50r of these windings.
Thank you Paul. The impedance matching is interesting as I've never seen a writeup on it.
Very interesting diversion from your transceiver build :)
Thank you Andre
The 'Mix 0' is just a phenolic form, isn't it? This project sounds like fun. I really like such experimental builds. BTW, I've used 'gimmicks' occasionally over the years. They always seemed like a shaky idea, but they always worked well and maintained their capacitance. Of course, that depends on the wire type used.
Thats what I thought too John but this www.amidoncorp.com/t37-0/ says its got an iron powder core. It does seem like its just phenolic - I don't have enough to break one and find out! Thanks for the comment!
@@na5y, I actually just called them and they verified that the zero material was indeed iron powder. BUT, unfortunately, Micrometals says the zero mix is phenolic. RF Microwave lists the Mix 0 as follows:
Iron powder toroid core T50-0, non magnetic
MIX 0 - KEY FEATURES
- Non-magnetic, non-conductive phenolic material
I've been around the industry long enough to know that even corporations get things wrong, (like 5,000 diodes with the band painted on the anode end.) I think the reality of the situation is that these manufacturers provide phenolic toroids for Mix 0, but list them with their iron powder cores because the 'Mix 0' doesn't fit into any other category, and so some of the documentation, (such as the above,) ends up calling it powdered iron when it's actually phenolic. Perhaps a magnet can verify that.
@@johnwest7993 Interesting - I might have to sacrifice one to find out. Thanks for the update!
@@na5y, the phenolic toroids I have weigh next to nothing compared to iron power cores. That would be a pretty good hint, that and a good magnet.
Your are right it a good idea to know effect of some parts.
Il you want a simulation closer to reality always add a low series resistor with your inductor. You will see the effect on quality factor. Yes at 144Mhz with toroid, the effect is lower...This is the first parameters we read when we choose inductor in industry design.
At 144Mhz you can also use proximity coupling with air coil : more difficult to calculate but you can adjust both inductor value and coupling for experiment.
At this frequency we can simulate accurate S21 design with : inductors S parameters, smd parts, ideal capacitor in simulation, add PCB capactive effect : about 1pF/parts. Easier if you replace L transformer with capacitive transformer. At the end the true result is very close to measurement. More difficult at 440Mhz.
For transformer ratio calculation and impedance LC I suggest read SA605 SA602 philips application note.
Thank you Remi - thats some great advice and tips!
.. and I'd agree - not sure I'd attempt this at 440MHz
Will it work up to the Nyquist frequency? Or above her?
Oleg, this is a direct conversion I/Q receiver - the output of the mixing process will produce an I and Q audio signal. That Audio signal can then be subsequently digitized and pretty much any demodulation can be applied (SSB, AM, FM, etc). The input to the mixing process is the RF itself + an LO which will come from an si5351 frequency synth. There is however no digitizing of signal on this board. I hope this answers your question and thank you for the comment!
@@na5y Thanks, but I know how it works because I made it myself (ruclips.net/video/nGn-KkACT7o/видео.html). I'm wondering how such a high frequency will work?
@@R1CBU Its all analog on this board though - the FST3253 is the mixer and it can switch up to GHz frequencies. The si5351 can output at least into 200 MHz. Once the mixing is done its all at audio frequency after that.
@@na5y Thank's! I did not expect such speed from this multiplexer!
Well that's what I read - we'll see how well it works in practice ;) - Thank you for the comments!
Interesting that you didn't tweak the two variable caps at all? That could have been all that was needed to get up to the 140-150 MHz range, no?
Apologies - I really should have showed the tweaking of the trimcaps part. I had already adjusted the caps in both cases earlier.
The maximum frequency of operation of Tayloe Detector is limited by FST3253, particularly by the time of channel switching. It can be from 1 to 6 ns according to the datasheet, which gives us the maximum **guaranteed** receiving frequency of 41 MHz. In practice it may work up to 75 MHz or so but definitely not at 150 Mhz. You are going to need a downcoverter for this.
Yeah I had read about the downconverter need elsewhere. Still I will try...
Just walk me through your math though.
With a quadrature signal at 2,14 of the FST3253 that will result in transition through 00, 01, 11, 10 at 4x the LO frequency. Going from a 0 to 1 on say, pin 2 will take between 0-6ns. Worst case that will take 6ns - and because I need to transition 4 times per cycle that will take 24ns - which is 41 MHz. Is that your thinking?
@@na5y Yes. Also, I actually checked that it doesn't work on such high frequencies. The instance I tested worked up to 75 Mhz, so it had 1000*1000*1000/75/1000/1000/4 = 3.3 ns switching time.
@@R2AUK Good to know - like I said I will proceed with the experiment and see where it lands me! Its interesting that there is such a wide range of switching times in the datasheet. The TI datasheet says 1.5ns as minimum which would be approx 160Mhz. Like you said thats not even close to a guarantee. Interesting also in Dan Tayloe's original paper he talks about the solution potentially working into the GHz range. I wonder if he was referring to a theoretical switch rather than the FST3253. Thanks for the comment Alex!
I was curious about how well it worked on 6 meters. I'm thinking of adding that band to my SDX experimental rig. It sounds like it may work OK.
@@johnwest7993 I will include a test at 50MHz and you'll see the result
Gimmicks caps make sense for scenarios but nowadays with the range of ceramic capacitors available, they create more issues than they solve.
If you make this work it will be a bloody miracle 😊 seriously use a single conversion Rx with an IF around 9mhz. This is what my current project is about. Single conversion hf transceiver using some of the ideas from your channel. No analogue audio switching, route i2s internally.
I have my doubts - we'll see. The FST3253 based Tayloe detector can go up to GHz, and the si5351 can certainly do 150MHz quadrature. The traces on the board may be a problem of course.
I like miracles. I like to push circuits to their limits, then do work-arounds to push them a little further. To hand-match components. To come up with tricks and innovations. To 'make something work'. The 'gimmick' is a perfect example. I've used them for years, and the hams around me just look at them and shake their heads. They think you have to buy a specific component with a specific value stamped on it from a manufacturer in order to make a circuit work. But you don't. Not if you know what you're doing. Electronics is fun. :)
I'm not knocking your try at this, but it has real difficulties. I've done this stuff in the past, a long time ago in the land of Oz. I live in nl these days.
My current RF knowledge is a bit rusty these days but if you want to bounce some ideas around I would be happy to pitch in.73s. a Long time ex vk6
Sure Anthony - happy to chat. Should be straightforward to find out whether or not it is going to at least do the mixing or not. Sensitivity is a whole other thing. More to come!
@@na5y The problem with the 3253 is best case the S inputs can run at 1GHz, no typical figures and worst case of around 200mHz. You need then to allow for prop max of 0.25ns, so to get an output best base your operation could no more than 750Mhz approx. That wouldn't provide enough time to charge the integrating cap. The other problem is gain. If you check the sensivity of this detector at 10M compare to 20M you will find the gain has dropped, not sure how much, 3db at a guess. There are no specs for this, but is only going to be a lot worse at 150mHz.
@@anthonyclark7290 Your arguments make sense - I am definitely interested in whether this will work practically. We'll see - I haven't even run the si5351 at 150Mhz in quadrature - thats the first open question for me. Thank you for the comment!