6-gang FM stereo tuner heads

[regenfreak]

Originally Posted by G0HZU_JMR

One thing to be wary of when testing your dual gate mosfet mixer will be the amount of noise on the LO carrier 10.7MHz away. This will be at the tuned RF frequency and the image. Also, there would be wideband noise down as far as 10.7MHz.

I've not examined that many synthesisers that get used for the LO for a dual gate mosfet mixer, but I have seen designs that use a VCO at the fundamental VCO frequency and there is often a tuned LO amp. This helps to filter away the unwanted noise. A diode ring mixer is double balanced so this issue is less of a problem with a diode ring mixer. The port to port isolation of the mixer will be much higher with the diode ring mixer so the noise can't migrate from port to port as easily.


Some lab sig gens will have fairly high noise floor at a 10.7MHz offset from the carrier. By contrast, a well designed VCO based LO could be 30dB cleaner at this offset. In other words, if you were to use a typical lab sig gen for your mixer LO and you didn't filter it with a narrow BPF then you could think that your mixer has a much higher noise figure than it really has. I'd expect to see a noise figure of about 7dB to 10dB with a typical mosfet mixer design assuming there are no carrier noise problems with the LO.

The conversion gain could be anything from 10dB to maybe 16dB depending of the mosfet type used. The output TOI of the mixer could be anywhere from +10dBm to about +17dBm but a lot depends on how much you optimise the LO drive level and the bias point of the mixer. I think it's reasonable to expect a +12dBm TOI at the output of the mosfet mixer.

I would expect significant hurdles to be overcome in the TOI measurement of the dual gate mosfet mixer. I had the intention of using a narrow-band 10.7MHz crystal BPF at the IF port using what I would call "narrow frequency window shifting" technique described by you. It essentially varies the LO frequency to do the IMD measurements in sequential steps to overcome the spectrum analyzer mixer's level limitation. Obviously, the crystal filter would need LC matching networks at both ends. The tricky bit is to design the fairly wideband Pi matching network with a fairly low Q for the input of the dual gate mosfet and the IF transformer with the secondary 50-ohm output impedance for TOI test jig. I haven't had experience with this.

Adding to my predicament, I still lack one VHF signal generator source for such a test. I only have the FY6800 with two channels going up to 100Mhz. I don't have to do any TOI measurements in the VHF frequencies. I will not bother to do it if it gets really time-consuming and expensive. Even I successfully measured the TOI figure and it would give zero impact on my 6-gang FM tuner project. The TOI measurement is for the sake of the measurement itself and nothing else.

The dual gate mosfet mixer has both high input and output impedance, making it easier to implement in a high-performance FM tuner.I would not use a ring-diode mixer for a broadcast wideband FM tuner because the commercially available Mini-Circuits Mixers only work well with low-impedance ports. The impedance matching of the ports is a non-trial problem.

Fair enough if you want to build a QRP receiver or a VHF pre-amp or ring diode mixer with a 50-ohm termination system. There are plenty of circuits to pick and choose from amateur radio communities online. But this is not the direction I want to pursue.

Regarding TOI measurement again, the IF output of a ring diode mixer is very sensitive to the port termination, any significant mismatch that deviates from 50 ohms impedance would result in large degradation of measured TOI. Of course, the addition of an attenuator would "mask" the impedance mismatch, but it would probably be prudent to add a low-Q, 10.7MHz diplexer in front of the 10.7Mhz crystal BPF to ensure 50 ohm termination for the IF port.

[nemo_07]

[QUOTE=G0HZU_JMR;1521589]

... For me the most intuitive way to understand this stuff is to temporarily abandon all equations and s/y/h parameters and just think in terms of how current can lead voltage (and vice-versa) in reactive feedback paths. Then look at the dominant internal feedback capacitance of the device. Assume a purely real source impedance feeding the base, and then trace the feedback path back through the reactance at the emitter and through to the base/source (via the dominant feedback path) and see if this produces a waveform in-phase (can lead to negative resistance at the input) or anti-phase (can lower the input resistance). Putting a capacitive load at the emitter of an emitter follower will usually cause negative resistance at RF frequencies if you trace the signal back to the input. Having inductance at the emitter does the opposite because voltage leads current. ELI the ICE man is one way to remember the phase relationships.

Still the simpler way to demystify the mechanism involved in the creation of "negative input resistance" in emitter/source/cathode follower circuits would be use of vector representation of signal voltages and currents in two complex planes U and I (overlaid for clarity of phase relations). Properly constructed vector diagram based on properly simplified model can show at instant what is going on, how variations of circuit parameters (i.e. gm, Rs, Cs) can modify the input conductance, and make it positive or negative, as desired.
It would also be compatible with and support analytical expressions given in my post few pages back https://www.vintage-radio.net/forum/...&postcount=197

[nemo_07]

Originally Posted by regenfreak

The tricky bit is to design the fairly wideband Pi matching network with a fairly low Q for the input of the dual gate mosfet and the IF transformer with the secondary 50-ohm output impedance for TOI test jig. I haven't had experience with this. ...

The first part you can find at the antenna inputs in some "kitchen-class" FM radios. These are hybrids with four pins, vertical, coated with a sealant mass mostly of brown colour.
Winding an IF filter goes as follows:
Suppose you need a 100pF||2.2uH LC tank for 250kHz 3dB bandwidth, and the AL value of your TOKO cores is unknown.
Wind 20 turns (n=20), assemble the whole, drive the tuning slug to the approximate middle position of tuning range, resonate it with a 100pF and note the resonance frequency. From the well known formula you'll get the inductance value, L_20t. Let it happen to be 2.5uH.
From L = AL*n² we get AL = 2500nH/(20t)² = 6.25 nH/t².
The number required for 2.2uH primary: n1 = sqr(L/AL) = sqr(2200/6.25) = 18.8, we take n1 = 19.
For the required bandwidth we get (loaded) Ql = 10,7/0.25 = 42.8
Given XL = 2*Pi*F*L = 148 Ohm we get the final resonance impedance being Zr = Rp = Ql*Xl = 6.3 kOhm.
From the transformer equation we have
Rp/R2 = n1²/n2² --> n2 = n1*sqr(R2/Rp) = 19*sqr(50/6300) = 1.7, we have to wind something like 1 + 3/4 turns secondary.
Important is to wind it over the "cold" end of primary.

Originally Posted by regenfreak

... I still lack one VHF signal generator source for such a test. I only have the FY6800 with two channels going up to 100Mhz. I don't have to do any TOI measurements in the VHF frequencies. I will not bother to do it if it gets really time-consuming and expensive. Even I successfully measured the TOI figure and it would give zero impact on my 6-gang FM tuner project. The TOI measurement is for the sake of the measurement itself and nothing else. ...

You're right.
BTW: You got another signal generator in the TinySA. It is synthesized so not particularly clean, but for cursory checks should suffice.

[G0HZU_JMR]

I think you need to be wary of the spectral purity of any commercial sig gen that you use for the LO to test a dual gate mosfet mixer. The relatively poor port to port isolation with this type of mixer could cause confusion when you try and measure the mixer noise figure. It usually doesn't really matter how much money you spend on a modern commercial synthesised RF sig gen, the noise floor at about 10MHz away from the carrier is rarely better than -150dBc/Hz. You might find some that can manage -160dBc/Hz but this is rare. Some commercial sig gens can be as poor as -140dBc/Hz.

See below for the noise floor performance of a cheap 134MHz JFET oscillator against a couple of commercial signal generators. The Marconi 2022 and the HP 8648D. You can see that the cheap JFET oscillator is 25-30dB cleaner at a 10MHz offset.

In your case, it will be difficult to do much noise filtering at the LO output because your LO has to tune across about 20MHz. Therefore, if you want to have a decent source to use as a test LO for your mixer development work, I think you will have to make your own oscillator and add a buffer and a booster amp to it. Otherwise, a commercial sig gen is going to be too noisy and it will mask the true noise figure performance of the mosfet mixer.

You can add a narrow BPF at the output of the sig gen to shave off the noise at a 10.7MHz offset, but the BPF would have to be tuned to one part of the FM band only as the FM band is 20MHz wide. I would recommend making a clean VFO to use here. It should be possible to make something with low frequency drift up at 100MHz or so.

[G0HZU_JMR]

Note that the reason the phase noise graph shows a test frequency of 134MHz is because I have a dual gate mosfet mixer eval board here that is designed for the 145MHz amateur band. The LO is ~ 134MHz to convert to a 10.7MHz IF. Most of my dg mosfet experience at work is with the classic old BF981 and also the SMD BF989. These were usually used as low noise RF amplifiers or as buffer amplifiers at work. I've checked my mixer eval board and it has the BF989 mosfet fitted to it at the moment.

[regenfreak]

nemo_07The first part you can find at the antenna inputs in some "kitchen-class" FM radios. These are hybrids with four pins, vertical, coated with a sealant mass mostly of brown colour.
Winding an IF filter goes as follows:
Suppose you need a 100pF||2.2uH LC tank for 250kHz 3dB bandwidth, and the AL value of your TOKO cores is unknown.
Wind 20 turns (n=20), assemble the whole, drive the tuning slug to the approximate middle position of tuning range, resonate it with a 100pF and note the resonance frequency. From the well known formula you'll get the inductance value, L_20t. Let it happen to be 2.5uH.
From L = AL*n² we get AL = 2500nH/(20t)² = 6.25 nH/t².
The number required for 2.2uH primary: n1 = sqr(L/AL) = sqr(2200/6.25) = 18.8, we take n1 = 19.
For the required bandwidth we get (loaded) Ql = 10,7/0.25 = 42.8
Given XL = 2*Pi*F*L = 148 Ohm we get the final resonance impedance being Zr = Rp = Ql*Xl = 6.3 kOhm.
From the transformer equation we have
Rp/R2 = n1²/n2² --> n2 = n1*sqr(R2/Rp) = 19*sqr(50/6300) = 1.7, we have to wind something like 1 + 3/4 turns secondary.
Important is to wind it over the "cold" end of primary.

Cheers. But I was thinking more about the design of the Pi matching network for the RF two-tone at gate 1 of the mosfet 3N201 in the schematic of post #357 from the datsheet.

From the datasheet, it is possible to read off useful parameters ; input admittance |Yfs|, input admittance gis+jbis and output admittance gos+jbos.
For some reason, the reverse transfer admittance is often omitted in datasheets. By working out the complex input impedance Z = 1/Y at the center frequency of the two tones, we can design the Pi network using the Smith chart (or SimSmith) or an online calculator. With a tone spacing of 1MHz and center frequency of 100Mhz, assume the bandwidth is 4Mhz for pi-network,Q = 100/4 = 25. I don't know. It is just a rough quest for Q. If I were to make Q smaller, I may have to connect a parallel resistor across the inductor. The Pi network also works a bit like a low-pass filter for the harmonics from the two-tone signal sources. The Pi network is more flexible than L -network as you can specify the Q. It is basically two back-to-back L networks with virtual resistance at the adjoining nodes. Also the Q of L-network is high for high-impedance circuit.

I think you need to be wary of the spectral purity of any commercial sig gen that you use for the LO to test a dual gate mosfet mixer. The relatively poor port to port isolation with this type of mixer could cause confusion when you try and measure the mixer noise figure. It usually doesn't really matter how much money you spend on a modern commercial synthesised RF sig gen, the noise floor at about 10MHz away from the carrier is rarely better than -150dBc/Hz. You might find some that can manage -160dBc/Hz but this is rare. Some commercial sig gens can be as poor as -140dBc/Hz.

For dual gate mosfet mixer, probably RF-to-LO port isolation is about 20db, LO-to-IF port isolation 30db. Because of the high impedance, the DGM has a very narrow bandwidth and low conversion gain for image noise. Now I think this measurement is a moot exercise purely for learning purpose. I will probably change it to a "thought experiment" without doing all the troublesome practical work. It is a lot of hassle to measure TOI with all the jazz of diplexer, Pi network, If-transformer, crystal filter and frequency source BPFs.

See below for the noise floor performance of a cheap 134MHz JFET oscillator against a couple of commercial signal generators. The Marconi 2022 and the HP 8648D. You can see that the cheap JFET oscillator is 25-30dB cleaner at a 10MHz offset.

Thanks. It is probably good news as I can easily build a BJT oscillator with a buffer amp and air variable capacitor tuning capacitor easily. The air variable capacitor tuned oscillator has higher Q and lower phase noise than varactor-tuned oscillator. With a reduction gear drive for the air capacitor, it should have an acceptable frequency drift. I have several cheapo VHF VCOs from 30Mhz to 1.1 GhZ from China and the Mini circuits POS-200P 95Mhz-120Mhz VCO. The Chinese "toy" VCOs are just for fun and reverse engineering of their circuits.The POS-200P VCO frequency drift is not acceptable on the spectrum analyzer for the critical TOI test even I power it with a regulated battery power source.

[regenfreak]

Attached are examples of the Y21, Y11, Y22, S11 and S21 from BF998 DGM datasheet. It is nice to see both Y and S parameters in a datasheet.

It is very difficult for manufacturers to measure short circuit conditions at high RF frequencies for the two-port network Y-parameters. Probably, Y12 cannot be measured directly. On the other hand, the S-parameters are much easier to measure with a VNA, and they are more intuitive to understand than Y-parameters. So the y-parameters are out of fashion, but most old IEEE papers used y-parameters in their design calculations, and so did many of the dual gate MOSFET datasheets.

[Radio Wrangler]

There are so many different parameter sets to choose from. Each set has its own group of blind spots wher physical conditions are hard to arrange when attempts are made to measure them directly, or where measurement errors are exaggerated.
s-parameters are closely related to the concept of a Zo vector network analyser so their suitability is no surpris. One linking figure was Dick Anderson.

HP did an application note on using a VNA to measure quartz crystal parameters. The real aim was to flog VNAs and blow their trumpet. The industry standard way up to that point was to use a 2-port pi-network with switched loading capacitor. Cathodeon made a beautiful one.

This new 1-port measurement had a fatal flaw. Accuracy of phase measurement around series resonance was crucial. In this region the crystal looks like its ESR. 50 Ohms is well within the bounds for many parts, so you want to know the phase of the nulled reflection....oops.

Unfortunately the folk in our standards labs had attacked our nice, gold-plated, Cathodeon mounts and converted them to mere sockets for the network analyser s11 method. The innards had been junked and were probably in the Craigie landfill. Then we started getting parts failing acceptance tests in incoming inspection, then swapping of test parts to-and-from suppliers showed there was a discrepancy and we were the odd ones out. The lead time on a new Cathodeon mount was painful, but necessary.

Theoretically the network analyser 1-port method is all that's needed. But that theoretically over-simplified and glosses over the finite directivity of the coupler involved. The VNA has multiple receivers, so dump the coupler, use a 2-port mount and the isolation of sepatate paths will easily beat the directivity of any realisable coupler.

The blind spot of the new method was centred right on where measurements were going to come out, and the VNA division had the glimmer of sales twinkling before their eyes when they wrote that app note.

We did replace our old sig gen and vector voltmeter with a shiny new 3577A, but strictly in s21 mode!

Oops!

David

[G0HZU_JMR]

For dual gate mosfet mixer, probably RF-to-LO port isolation is about 20db, LO-to-IF port isolation 30db. Because of the high impedance, the DGM has a very narrow bandwidth and low conversion gain for image noise. Now I think this measurement is a moot exercise purely for learning purpose. I will probably change it to a "thought experiment" without doing all the troublesome practical work. It is a lot of hassle to measure TOI with all the jazz of diplexer, Pi network, If-transformer, crystal filter and frequency source BPFs.

In my previous tests with a 134MHz LO and a high side LO at 156MHz, I used a broadband 4:1 step up transformer to get double the voltage to the G2 pin of the mosfet. My sig gen had to run at about 13dBm output to get good conversion efficiency even with the transformer in place. If the sig gen noise floor at 10.7MHz offset is (say) -145dBc/Hz then the absolute value is 13 - 145 =-132dBm/Hz.

If the mosfet has 20dB L-R isolation then -152dBm/Hz can be assumed to be leaking to the RF port. This is over 20dB higher than thermal noise. This is why a regular commercial sig gen will often spoil any attempt to measure the noise figure of the mixer if the sig gen is used for the LO. I have demonstrated this in the past and some of my sig gens degraded the noise figure of the mixer by over 20dB. Adding a narrow BPF at the sig gen output cured the problem as the filter had well over 20dB rejection at a 10.7MHz offset from the LO frequency. There can also be wideband noise floor issues from the sig gen way down at 10.7MHz and also at the image, but the main problem seems to be noise on the LO at the RF frequency.

[regenfreak]

Originally Posted by Radio Wrangler

There are so many different parameter sets to choose from. Each set has its own group of blind spots wher physical conditions are hard to arrange when attempts are made to measure them directly, or where measurement errors are exaggerated.
s-parameters are closely related to the concept of a Zo vector network analyser so their suitability is no surpris. One linking figure was Dick Anderson.

HP did an application note on using a VNA to measure quartz crystal parameters. The real aim was to flog VNAs and blow their trumpet. The industry standard way up to that point was to use a 2-port pi-network with switched loading capacitor. Cathodeon made a beautiful one.

This new 1-port measurement had a fatal flaw. Accuracy of phase measurement around series resonance was crucial. In this region the crystal looks like its ESR. 50 Ohms is well within the bounds for many parts, so you want to know the phase of the nulled reflection....oops.


Theoretically the network analyser 1-port method is all that's needed. But that theoretically over-simplified and glosses over the finite directivity of the coupler involved. The VNA has multiple receivers, so dump the coupler, use a 2-port mount and the isolation of sepatate paths will easily beat the directivity of any realisable coupler.

The blind spot of the new method was centred right on where measurements were going to come out, and the VNA division had the glimmer of sales twinkling before their eyes when they wrote that app note.

We did replace our old sig gen and vector voltmeter with a shiny new 3577A, but strictly in s21 mode!

Oops!

David

Thanks for a bit of insider's interesting history. I looked at 3577A user manual a few months ago but was confused by why it has 3 terminals and the datasheet claims it can measure S22 and S12, but with the extra 35676A test kit only!

I think the Cathodeon Pi network crystal test fixture described by you is in this link with a photo and schematic:

https://isolalab.com/pinetwork.html

The accuracy of S11, S21, S22 and S12 is affected by many factors. One of the key uncertainties is the interaction between the reflected signal and the directivity leakage of the directional devices inside a VNA. The VNAs use resistive (or hybrid) bridges or directional couplers. The resistive bridge is constructed of Wheatstone bridge. Its balance and directivity heavily rely on the good impedance match at the test port for the DUT. The directivity of the directional coupler is dependent on the spacing and electrical lengths of the transmission lines.

The attached diagrams illustrate the directional couplers inside a two-port two-path VNA. One can see the leakage paths coming from the directional devices. The unwanted directivity signal can add errors to the amplitude and phase angle of the reflected signal vector.

G0HZU_JMR In my previous tests with a 134MHz LO and a high side LO at 156MHz, I used a broadband 4:1 step up transformer to get double the voltage to the G2 pin of the mosfet. My sig gen had to run at about 13dBm output to get good conversion efficiency even with the transformer in place. If the sig gen noise floor at 10.7MHz offset is (say) -145dBc/Hz then the absolute value is 13 - 145 =-132dBm/Hz.

If the mosfet has 20dB L-R isolation then -152dBm/Hz can be assumed to be leaking to the RF port. This is over 20dB higher than thermal noise. This is why a regular commercial sig gen will often spoil any attempt to measure the noise figure of the mixer if the sig gen is used for the LO. I have demonstrated this in the past and some of my sig gens degraded the noise figure of the mixer by over 20dB. Adding a narrow BPF at the sig gen output cured the problem as the filter had well over 20dB rejection at a 10.7MHz offset from the LO frequency. There can also be wideband noise floor issues from the sig gen way down at 10.7MHz and also at the image, but the main problem seems to be noise on the LO at the RF frequency.
Yesterday 8:53 am

Cheers. The BPF at the output of the sig gen can hit two birds wit one stone; cutting down the harmonics and noises.
An active mixer is noisier than an RF amplifier. The mixer duplicates and "folds" the broadband noise as a multiplier of the LO (attached). The attached diagram shows the broad noise of the LO also appears in the image frequency you mentioned.
The operation principle of dual gate mosfet mixer is no less complicated than the ring diode mixer. The DGM mixer can be considered to consist of two transistors in cascode amplifier configuration. The LO drives the upper transistor very hard at gate 2 to the non-linear region that is large enough to modulate the transconductance of the lower transistor with gate 1 to the RF.

The 5th attachment gives a design example of a narrow band DGM mixer using Pi-networks for 50-ohm terminations taken from:

http://www.radiohamtech.com/mixer%20design.pdf

[Radio Wrangler]

Yes, that's the Cathodeon crystal fixture. Too useful to go out of availability with the ending of the Cathodeon company.

In standard form, the nearest piano key in that photo is the switch for a 30pF series load capacitor because these often feature in crystal specifications. However, it can be useful to have a second fixture with a different capacitance load because this allows a different view of the crystal motional parameters and the effects of strays can be better extracted.

There is usually a bit of bent metal used with these things to hold the capacitor key down when needed. Fingers go numb after a while of continuous pressing!


The whole point of doing crystal measurement as a two-port exercise is that the directionality of directional couplers can be taken out of the arithmetic. Yes a very good Wheatstone bridge can be made with purely resistive elements, but its two ports cannot both be ground-related at once. The source and receivers in the VNA are ground related, unbalanced. Consequently a balun is needed. Any unbalance in the balun, unbalances the resistive bridge, and there went the perfect directivity. Of course, signals will leak from one coax cable to another, but a better job can be done than in isolating the two directions on a single cable. So, in practice, the two-port pi-network starts with a quite unfair advantage.

Crystals offer extreme values of Q and so their motional parameters have quite extreme values of C and L, when the frequencies of the resonances are considered. Consequently, seen as a simple s11 measurement, things get pushed into areas where network analysers used alone get rather stressed with degraded accuracy being delivered. Directionality being the biggest factor.

David

[regenfreak]

Originally Posted by Radio Wrangler

The whole point of doing crystal measurement as a two-port exercise is that the directionality of directional couplers can be taken out of the arithmetic. Yes a very good Wheatstone bridge can be made with purely resistive elements, but its two ports cannot both be ground-related at once. The source and receivers in the VNA are ground related, unbalanced. Consequently a balun is needed. Any unbalance in the balun, unbalances the resistive bridge, and there went the perfect directivity. Of course, signals will leak from one coax cable to another, but a better job can be done than in isolating the two directions on a single cable. So, in practice, the two-port pi-network starts with a quite unfair advantage.


David

Cheers. The role of the Balun in a resistive bridge makes perfect sense.

For transmission measurements with a NanoVNA V2 Plus 4, I learnt from the hard way that I should try to use one coaxial cable for port 1 and connect the DUT directly to the port 2 with some form of improvised strain relief for the port connector (e.g. resting blocks to support the weight of the DUT) whenever it is physically possible, particularly in high frequency and broadband measurements. With two coaxial cables, sometimes I got inconsistent, not repeatable or unstable calibrations due to the slight movement of the cables or dodgy cables.

I try to avoid moving or bending the cables at sharp angle during calibration or measurements. Cheap SMA connectors with cables wear out fast and can make me chasing ghosts when the results of the calibration looked funny in the Smith chart. I got some high quality sma cables but they are sometimes too stiff to use.

The cable not only adds insertion loss and phase shift, but also degradation of the directivity. Following the calibration, a large movement of the cables would result in phase shift between the wanted signal and directivity leakage signal. The calibration is supposed to correct the effects of the cable but it is not always fool proof. So it is better to keep the DUT as close to the VNA ports as possible.

[Radio Wrangler]

A lossy ferrite common mode choke core to kill outside currents can be helpful on VNA cables.

Fewer things go bump in the night.

David

[regenfreak]

Originally Posted by Radio Wrangler

A lossy ferrite common mode choke core to kill outside currents can be helpful on VNA cables.

Fewer things go bump in the night.

David

I have seen CM chokes used in balanced dipoles or loop antenna to provide mostly resistive impedance to reduce the common mode current at the outsider surfaces of the shields in a coax cable. Skin effects cause three current paths in a coax. A good HF "wideband CM choke " would have a typical bandwidth of about 10 Mhz, and it may be suitable for the measurements of crystals and filters at HF frequencies with a VNA:

https://gm3sek.files.wordpress.com/2019/01/G3TXQ-RC.pdf

I saw spooky shadows when I tried to push the NanoVNa V2 P4 to the limits in extreme measurements, e.g. measuring the broadband high-frequency characteristics of a 10K 1208 resistor from 1MHz all the way to 4GHz. This is the kind of situation that I would avoid using two coax cables. I can't see a CM choke that can cope with such ultra-wide frequency range sweeping from one MHz to GigaHertz microwave regime.

[Radio Wrangler]

Ah, you don't need fettite which remains reactive over that band. Ferrites go lossy at frequencies above their usefully reactive range, and lossy is also good in this case.

David

[regenfreak]

Originally Posted by Radio Wrangler

Ah, you don't need fettite which remains reactive over that band. Ferrites go lossy at frequencies above their usefully reactive range, and lossy is also good in this case.

David

I see. Up to now, I have only used a big TDK clip-on EMI filter choke for the USB cable connecting the NanoVNA to the PC.

[Radio Wrangler]

Those anti-RFI chokes are principally lossy elements to turn high frequency noise into heat. At lower frequencies they create inductance and reflect the noise. Absorption beats reflection in this case.

If you want to absorb a beam of light, you wouldn't want to try to do it with mirrors. It'll keep bouncing until it escapes.

The usual lossy material is Fair-Rite number 43, or other brand's equivalents. It's very much like cores salvaged from old TV deflection yokes and LOPTs. Efficient for good inductors to a few hundred kHz, good enough for transformers across the HF bands, and good and lossy by 100MHz.

David

[regenfreak]

Originally Posted by Radio Wrangler

Those anti-RFI chokes are principally lossy elements to turn high frequency noise into heat. At lower frequencies they create inductance and reflect the noise. Absorption beats reflection in this case.

If you want to absorb a beam of light, you wouldn't want to try to do it with mirrors. It'll keep bouncing until it escapes.

The usual lossy material is Fair-Rite number 43, or other brand's equivalents. It's very much like cores salvaged from old TV deflection yokes and LOPTs. Efficient for good inductors to a few hundred kHz, good enough for transformers across the HF bands, and good and lossy by 100MHz.

David

Argh Ok I do have a few TV LOPTs lying around...but I used them with ZVS drivers (around 24kHz) for small spark gap tesla coils.

[regenfreak]

I have been thinking about broadband matching networks. Using constant Q circles for broadband matching is like a mole burrowing through networks of tunnels. The mole has been given rules that it has to move either clockwise or anti-clockwise directions along some "magic circles" within a safe zone ( a Q circle) to reach the nest at the centre of the networks. There are forbidden and permissible zones of the circles. Violations of rules or going outside the safe zone Q circle mean whacking in the head by the farmer.

I use a fictitious amplifier as an example. Suppose we need to match a complex impedance of 341.5-j165.5 to 50 ohm centred at 2GHz with Q = 2 and Q = 1.5, respectively, sweeping from 1MHz to 3GHz.

I am not familiar with SimSmith, but I have managed to design the broadband matching networks with:

4-elements with Q = 2 (attachments 1, 2)

6-elements with Q =1.5 (attachments 3,4).

As Q gets smaller, it demands more elements...more constraints for the mole, it has to dig more and shorter networks of tunnels to avoid the farmer.

[Radio Wrangler]

Congratulations! You've just discovered the path which if followed leads you into broadened band, low-Q distributed structures like horns.

An infinite number of infinitesimal value elements, but following a graded density. Sneaking up on them from the lumped-model side.

David