FET notes [ J310 ] — 3.5 MHz VFO Frequency Doubler

 Greetings ###

In the past few years, 2 readers emailed asking for a 3.5 MHz doubler for their VFOs. I encouraged these readers to look at EMRFD, radio handbooks & other resources. Today, I built 1 -- and tried to solve some gain concerns with passive doubler networks described by the 2 readers.

The bog standard passive doubler "a.k.a. full wave rectifier" features a trifilar wound transformer, 2 diodes plus an RFC (plus, or minus an additional resistor + bypass cap as shown in EMRFD).  ** see last image **

You'll often get insertion loss of ~8-9 dB. If you connect 10 dBm available power to the input, the 10 dBm @ 3.5 MHz becomes ~2 dBm @ 7 MHz. This is par for the golf course. C'est normal.  

However, when you try and match the doubler output Z to 50 Ω via transformer ratios +/- a parallel tank resistor, you might even drop the signal another 4-5 dB depending on your method. Losses in #43 ferrite wide band transformers might also factor.  Further, after that, perhaps, comes potential filter losses. This is what the 2 readers told me -- and I validated their concerns on my bench.

Above — Today's experimental 3.5 MHz doubler. 

I reckon that like me, many people seeking a 3.5 MHz doubler want to drive a 7.X MHz Direct Conversion receiver LO port with a ~3.5X MHz VFO. The doubler goes in the same container as the mixer/product detector and functions to help reduce 7 MHz leakage getting into the mixer's RF port, the antenna and so forth. 

Leaked 7 MHz LO signal can react with the local environment -- for example suffer phase shifting, 60 cycle modulation and other sequalae -- and then come back into the product detector via the antenna. Among a myriad of other problems, LO leakage may lead to DC on the mixer ports upsetting mixer balance by changing mixer diode bias. The whole goal is to try & contain the 7 MHz LO signal to the mixer LO port &  within the mixer chassis. 

This also involves carefully filtering the DC, however, that is a topic for another day.

To that end -- the first RF amplifier is our familiar common gate job with at least 30 dB of reverse isolation (Measured 33.7 dB). This keeps helps the 7 MHz from leaking down the chassis LO input coaxial connector and coax. I bench matched the input port to 3.5 MHz with a return loss bridge -- and achieved a return loss of 25 dB with the pi match shown.

A standard doubler circuit goes on the FET drain -- I put in some Schottky rectifier diodes that were handy. I didn't match them, as previously when I dynamically matched pairs of these diodes in a home brew single balanced mixer ( 2 diode circuit ) going into a spectrum analyzer, this old bin of diodes seemed remarkably close in characteristics.

I viewed the suppression of the even harmonic LO products
e.g.  --  2LO ± RF, 4LO ± RF ....  as my measure of balance of the diode pair.

I prefer a tuned tank on the diode output -- standard stuff. If you match the tank to 50 Ω with the turns ratio of a secondary winding, typically, a parallel resistor across the primary is required. Doing this, I've measured an output return loss of 18-20 dB at HF in the past. However, the losses make you sad. I want a minimal output amplitude of hopefully 4-5 dBm available power to switch my mixer diodes.

To avoid losing gain, I AC coupled the hot end of the tank to a JFET source follower standing 13 mA drain current to keep the lower , or negative 270 degree AC signal peak from bottoming out when driven with 10 dBm input power.

Above — DSO capture. The raw doubler output before the low pass filter was installed.

Above — DSO capture. The doubler output with the low pass filter.  Driven with 1 V pk-pk or 4 dBm.

Above — DSO capture. Now at 10 dBm available input power. The output calculates to 5.1 dBm. It's OK.  You can still see lots of distortion in the signal.

Above — Spectrum Analyzer capture.  Sadly, when I measured this, I failed to tune the tank after I added the low pass filter to the breadboard. I must have bumped the tank or something. So, the earlier shown DS0 captures feature the correctly peaked RF tank, but this does capture not. Face palm.

Even still, the 3.5X MHz signal is down ~44 .5 dB relative to the 7.0X MHz signal peak.  Not too bad.
I might tweak this circuit some more. I though about reducing the losses incurred with the #43 material trifilar transformer. Because this circuit works at 3.5 MHz, the transformer does need a fair amount of inductive reactance and a #61 material transformer would require more windings... 

I do not want the 3.5 MHz LO to go more than 10 dBm output power to help reduce LO pollution into my radio receiver system that also includes the antenna.   Thank you and Best!

 

** see last image ** EMRFD.  Best book ever written on many RF topics -- including Direct Conversion receivers.

FET notes [ J310 ] - common gate amplifier termination insensitivity

Greetings ###  I'm slowly working on circuits for my next video. I sought a common gate amp with a well-defined, wide-band, output impedance to go on the RF port of a diode ring mixer. I also matched the input to 50 Ω for the entire 40M amateur radio band.

I'll show more hand drawn schematics going forward. I only post a fraction of my experiments -- partly, because I'm lazy; but it takes considerable time to draw schematics with software. Thus, I often don't bother sharing much content nor post that much.  My writing is Ugly style.

 

Above — Common gate amplifier for 7 MHz. This features a TO-92 J310 with a drain current of 12 mA. When raw bench experimenting, I use size 50 toroids -- later if it turns out I only need 15-18 turns of wire, then I may down size to a # 37 toroid etc..

Above — Amplifier measures. Gain = 6.43 dB. From my experience, you'll only get the often mentioned ~10 dB gain when you tune the drain -- plus run an air wound, or a dust-powder toroid like the T50-6 on the output tuned circuit. The #43 ferrite exacts losses, but works OK for wide band transformers -- and in combination with the 2K0 resistor -plus- a drain ferrite bead, or 22-51 Ω drain resistor, behaves itself. (Does not usually oscillate). 

Typically, we keep the FET's gate lead length to the ground plane short [ low inductance] to boost stability as the J310 is a VHF-UHF part.

The common gate or common base topology output impedance proves difficult to well match to 50 Ω using just transformer turns ratios. The usual method is to place a resistor in parallel with the primary coil. This works well & I chose 2K0 Ω for that resistor. 

The turns ratio reflects the Z transform from 2K to 50 Ω. I measured the output return loss from 2 to 13 MHz . The output return loss stayed above 20 dB from 2.1 to 12.8 MHz.

Then I sweep tested from 7.0 to 7.4 MHz. The output port return loss measured 22.8 dB across the whole 40M band. I'm happy. A decent match for my product detector RF port.

To measure return loss on the input port, we place a 50 Ω terminator on the output port. Then to measure the output port return loss, we switch ports-- and place the terminator on the input port.

Above — A wide band BNC 50 Ω terminator. I've got these and SMA versions that go from DC to 18 GHz. You may also make your own with a BNC/SMA connector and resistor(s). We use them to measure the directivity of our return loss bridges -- and calibrate stuff like VNAs.

Note in the data, I performed 1 more measure on the input. I checked the input return loss with no 50 Ω terminator on the output port - open circuit. The return loss across the 40M band = 18.6 dB with an open circuit output. Amazing. Recall, that many RF amplifiers are termination sensitive.

Termination sensitive means that the impedance or resistance seen at the output port may affect the input impedance. In this case, the amplifier input return loss remained ~OK despite the output port going open circuit. 

Further, I measured the reverse isolation of this amp at 32.5 dB. I believe strong reverse isolation is the primary reason that the amp seemed fairly termination insensitive.

This amp seems special -- and it is. The common base (BJT), common gate (FET), and a cascode of 2 FETS, BJTS or a FET + a BJT offer better reverse isolation than standard common emitter/ common source amps.

Above — A sweep in my tracking generator-spectrum analyzer. This device says the gain = 6.65 dB  @7 MHz . The low Q pi network offers some low pass filtration - bonus points.

This RF preamp will go in a HF direct conversion receiver. In the past 25 + years when I've done this -- on 3 occasions, total strangers took the time to email me & wrote that I'm trashing my receiver dynamic range -- that I don't need it -- and to get it out of there! 

Oh no -- its the RF preamp police lol!

I'll say it here -- it's OK to put in a RF preamp in an HF receiver if you want to. Attenuate or boost the signal before the product detector/mixer if you like. You do you. There may be a good case for RF preamps at HF in some circumstances.

Throughout solid state amateur radio history, some rather clever builders placed common gate (or common base)  preamps in front of direct conversion or regenerative receivers to reduce the amount of LO signal at the RF port going out into the antenna, doing bad stuff and ending up back in the mixer. (I'll be covering this in the video).

Perhaps, the listener receives with a subpar antenna? For example, with a small loop antenna -- and needs a little bit of RF gain?  Further, not everyone is an amateur radio contest listener. Many just like to tinker with simple circuits and enjoy making signals bigger.

Above — Another case? The transmit output amp of my old 7 MHz Funster trans-receiver. The receiver is connected to the antenna via a low Q series-tuned network. During transmit and a for a few milliseconds after, the pin diode pair shunts the PA RF output to ground.

On receive, the signal from the antenna passes through the low pass filter, then through a XL = XC = ~400 Ω series tuned network ( 56 pF + 9.1 µH ) to the receiver input. The series resonant circuit + low pass filter exhibits 8.8 dB of losses. A 6-8 dB gain common gate amp might help builders who choose the series connected transmit/receive scheme shown above.

An RF preamp is probably not required in most cases for HF reception. However, there are always exceptions.  A strategically placed amplifier with high reverse isolation might prove helpful and boost your receiver function and enjoyment.  Ultimately, you're in charge. Do whatever you like.

Best to you!

Part Obituaries continued : R.I.P -- dear J310

 Greetings ###

I'm working on a common gate RF amp using the ubiquitous RF home brewer's JFET, the J310. I happened to go on DigiKey and Mouser. Here's what was written for the part I was looking at on DigiKey:

The 10-volt or the higher voltage SMD versions are no longer made!  The TO-92 versions said goodbye a few year ago. I checked some of the substitute parts --- many of them are also no longer made and are just stock.

Yes, people are selling real & bootleg TO-92 versions on EBay etc. -- and also small-scale sellers -- or sellers abroad sell TO-92 J310s.  However, for the average person around the globe, getting these now expensive parts -- and paying shipping costs seems quite a bother. I now have to think carefully when I make a schematic and suggest people use a TO-92 J310 in the 'FET slot' like I casually did 20 years ago.

Fortunately, I saw this coming and actively searched for a "lifetime supply" of TO-92 JFETs around 1998-2002. I purchased hundreds of FETs back in that time slot.

Above — Some of my collection. The leftmost bin also houses 2SC3355s, a 1 GHz NPN part with a low noise figure. I've got BF244 (a 700 MHz nJFET part), 2N5484, MPF102, J310 and a few others in TO-92 for RF. I've also got popular switching JFETs like the J111, J112, J113, & J176 ( p-channel JFET) in my audio parts bins. All these parts are originals and many were in dusty factory packages.

Above — My TO-92 J310 bin is now gold? I bought in minimum quantities of 500 back in the day. Got them for 10-15 cents each when buying in bulk.

 
Above — Top view -- TO-92 J310 bin. About 3/4 of them are Fairchild Semiconductor versions made in 1997. VDS max = 25 volt versions. I also gave away hundreds of J310s plus other parts to some readers of the original QRP HomeBuilder web site for free between 2003 and 2016.  Before shipping got crazy expensive. Also, now to send parts, you have to fill out extensive paperwork.

Above — I have 2 original U310 JFETs. If the J310 is gold, then these are 'platinum'. 

I also collected some ultra low noise through-hole  JFETs for HiFi audio and noise measurement circuitry. These particular FETs now cost a small fortune -- and most affordable versions are no doubt - bootleg.

If you can find a good deal on real TO-92 J310s,  now is the time to grab them. I'm going to probably lean towards using surface mount FETs in future projects for Popcorn. At least people can still get these for now. It's sad to think that such a integral part of home brew electronics is going extinct. 

Analog epilogue -->  Gone extinct -  gone digital,  bye,  I'm gone.  Goodbye old friend.

Addendum:

I've been thinking about mixers a lot. I home brewed 7 different styles to learn more about them.  Just musings. Not to build. All major content for Popcorn will be presented in videos. The audience is not exclusive to Ham or SWL radio -- it's for all people who like straight-forward, accessible, analog electronics. I've got some great ideas to hopefully share. Thanks.

How People Make Circuits without Printed Circuit Boards - Video

 
Addendum added to the text January 13, 2025 - scroll to see -

Resource Files

 

 

 

Addendum January 13, 2025

LOL -- a reader emailed and wants more signal power -- and asked whether they could just swap in a pair of medium power BJTs like the BD139-140 to get more output power. Sadly no. I have a higher power version of this amp in a receiver called Regen 5 plus one other on this Blog. However, there is another way to get more power without greatly increasing the parts count:

Above — My new Popcorn PA version that goes to 700 mW clean signal power without adding a ton more parts. The finals = a TIP 122 &127 pair. 2 more diodes get added to the bias stack to forward bias the 2 pairs of emitter followers. To avoid a big voltage drop, I moved the 22 Ω resistor outside the TIP power followers on the DC positive rail since these BJTs draw much more current than the basic version of this audio PA. Further, a 470 µF AF bypass capacitor gets added to filter the output pair on the positive DC rail.

The TIP 122-127 are low voltage complementary power Darlington transistors in a standard TO-220 package giving a current gain of ~ 1000. They only cost ~ 60 cents per transistor even in Canada with our nearly worthless Canadian dollar.  A BD139-140 transistor usually cost 25-30 cents more.

Above — Schematic and pin out of the TIP 122-127.  

Above —DSO screen capture at about the maximum clean signal power ~ 700 mW.

Above — FFT screen capture with the AF amp driven to 700mW. The harmonic distortion is not as low as the original audio PA shown in the video. However, this is the compromise of using the TIP Darlington transistors + easy-peasy diode biasing scheme with 4 diodes. The better way to bias the power followers is via a transistor like I did in the audio PA Regen 5 back in 2015.  Regardless, the TIP pair are a compromise compared to 2 separate BJTs pairs with optimum circuit design.

Still, to, the harmonic + crossover distortion will still be on par; or better than most of the audio amps you see in many lower power consumer electronics -- or a typical home brew Ham /SWL radio receiver.

Thanks for the email. If you have a question,or request,  please email. 

My readers are the best!   Popcorn Todd