UK Vintage Radio Repair and Restoration Discussion Forum

UK Vintage Radio Repair and Restoration Discussion Forum (
-   Vintage Amateur and Military Radio (
-   -   Franklin VFO ? (

Granitehill 8th May 2019 7:23 pm

Franklin VFO ?
I've been using an HP3325 synthesizer to drive my Racal Squadcall but it's overkill and I'd sooner use an external VFO driver that's more in keeping with the 1960s feeling.

I'm sort of pulled in the direction of a Franklin VFO, but this circuit seems to be out of fashion for no particular reason.

The basic spec will be :
11-16 MHz (possibly bandswitched or plug-in tuned circuits)
300 mV RMS output (via buffer amp, of course)
Supply voltage 12 or 18V DC
Oscillator circuit to use a pair of Jfets
Rigid construction and temperature stable box are a given.
Running constantly - no keying

My guess on the benefits :
The Franklin would be the most stable - less frequency determining components than the competition.
Reasonably constant level over quite a wide band.

The major unknown would be the noise level. Since the oscillator is running at
low level, would'nt this imply a highish close-in noise spectrum ?
Has anyone got any ideas on this, rather than the more usual Clapp, Seiler, Vackar etc ?

G6Tanuki 8th May 2019 8:42 pm

Re: Franklin VFO ?
The Franklin, or its related cousin the 'cathode coupled' oscillator, both work well; the advantage of the Franklin is that the active devices are only very loosely coupled to the L/C so helping to reduce any drift caused by the characteristics of the active devices changing with supply-voltage or junction-temperature..

I used a Franklin [with a 12AT7 valve] many years ago to bring back to life an old CR100 receiver which had suffered burnout of the anode-coupling windings on a couple of the LO coils, this being easier than exhuming the LO coils from the coil-box and rewinding the coupling windings. It worked just fine!

G3PIJpeter 9th May 2019 7:39 pm

Re: Franklin VFO ?
Have a look at the G3PDM Vackar plus buffer - included in the article at

- Peter

Granitehill 9th May 2019 8:51 pm

Re: Franklin VFO ?
Thanks Peter - I'd considered the G3PDM circuit, and its got a lot to recommend it. However, there's still quite a lot of RF current carrying components directly attached to the tuned circuit. On that basis, I'm guessing that a Franklin would still be a better choice for ultimate stability, having only the parallel tuned LC components as potential drift sources, and given the very loose coupling to the other circuit elements.

I've still got a gut feeling that the Franklin is a leading candidate, and I'm struggling to see why it seemed to disappear from view - is there a good reason other than being seen as "Old fashioned" ?

On the subject of stabilising older valve VFOs, I had the same sort of drift on a Heathkit HG10 VFO many years ago. The eventual fix was to stabilise the heater voltage via a 3 terminal regulator (converted to DC heater as well). The G3PDM solution would have been rather better, but I wasn't aware of his work at that time.

G6Tanuki 9th May 2019 9:31 pm

Re: Franklin VFO ?

Originally Posted by Granitehill (Post 1144081)
I've still got a gut feeling that the Franklin is a leading candidate, and I'm struggling to see why it seemed to disappear from view - is there a good reason other than being seen as "Old fashioned" ?

Digital Frequency Synthesis!

Professionally, nowt's been controlled using free-running oscillators for half a century. And amateur stuff's been much the same since the 80s.

These days you can buy a spectacularly-stable DDS module for about half the price of the reduction-drive you'd need to fit on the capacitor of a free-running-VFO tuning-capacitor.

G3PIJpeter 9th May 2019 10:41 pm

Re: Franklin VFO ?

Originally Posted by Granitehill (Post 1144081)
I'd considered the G3PDM circuit, and its got a lot to recommend it. However, there's still quite a lot of RF current carrying components directly attached to the tuned circuit. .

True - but I made a 'PDM Vackar for 3.5 Mc/s and multiplied up in the LG300 to 21 Mc/s and still had a signal sufficiently stable for CW contacts. It is absolutely rock steady so long as all the frequency-determining components are good-quality and are firmly bolted or Araldited to the VFO box.
Peter G3PIJ

Granitehill 9th May 2019 11:56 pm

Re: Franklin VFO ?
Thanks for the advice - the logical answer would be to go for digital synthesis, of course. The idea to go for a free running VFO is a consciously retro thing. Analogue nostalgia !

I'm off to ponder the options....

Andrew2 10th May 2019 8:00 am

Re: Franklin VFO ?
What about a Vackar or Franklin (or even a good old Colpitts) with a Huff-Puff stabiliser? I've used HP on most of my VFO's for many years and not only do they hold the osc rock steady, but they are fascinating things to play with.
Dangle a cheap frequency counter (one with offsets) on the output and you have a stable rig with digital readout. I set the lock-intervals at around 50 Hz, which seems fine for SSB.

Bazz4CQJ 10th May 2019 8:07 am

Re: Franklin VFO ?
Back in the '70's, I spent quite a lot of time playing with a VFO running at ~11MHz, and then multiplied by 12 to act as the LO for the receiver in a 2m Pye Vanguard. I think I copied a circuit (very conventional) published in SWM for that purpose, which used a '3819 FET and two isolation transistors, but I don't think it was a Vacker.

I installed it inside the Vanguard, a situation where the temperature must have cycled a lot, increasing markedly if the Tx was kept on any length of time. In those days, it was possible to find small tubular ceramic caps of various temperature coefficients. After putting some effort in to it, the outcome was "adequate" for the purpose. I suspect that finding caps with a range of tempcos these days is far less easy. In hindsight, I cannot imagine why I didn't mount it externally ???.


G0HZU_JMR 11th May 2019 1:13 pm

Re: Franklin VFO ?

I've still got a gut feeling that the Franklin is a leading candidate, and I'm struggling to see why it seemed to disappear from view - is there a good reason other than being seen as "Old fashioned" ?
I had a look in my old Radcom Handbook and there is a Franklin oscillator design on P6.16 but this uses valves.

The Franklin circuit does look like it favours valves rather than semiconductors because of the voltages and impedances that are involved in the design. Do you have a proposed circuit of one that uses modern semiconductors?

Radio Wrangler 11th May 2019 7:07 pm

Re: Franklin VFO ?
If you don't want to go retro then there are many choices

DDS give low phase noise but have spurs around -70dBc

There are 1-chip fractional-N synths with everything on board and do almost DC to daylight in small steps, but the phase noise pedestal sticks out over hundreds of kHz... truly 'orrible.

If you want to go retro the G3PDM Vackar is a good example of its kind and he made the point that the mechanics of construction are more important than the exact choice of oscillator circuit. The problem is the unavailability of the Oxley Tempatrimmer, but I put a solution to that in the ARRL handbook oscillator chapter.

How retro do you want to go? Valves? it can be done. The Hartley in the RA17 was reasonably stable and resettable. Ot all comes down to mechanics again.

The Japanese big three went ga-ga over synthesisers because they saw two advantages: Saving the cost of good gearboxes and strong construction, and providing memories easily as a bullet point in the brag-sheets.

First decide on your wanted frequency range, then decide which era you want to build something from. Only with those decided can you start to make progress towards designing something.

The Franklin does not have light coupling to the tank implicit in its topology. It all depends on the choice of values. You have to choose appropriate values to make it lightly coupled. Exactly the same can be said of the Vackar.


G3PIJpeter 11th May 2019 7:30 pm

Re: Franklin VFO ?

Originally Posted by Radio Wrangler (Post 1144506)
The Franklin does not have light coupling to the tank implicit in its topology. It all depends on the choice of values. You have to choose appropriate values to make it lightly coupled. Exactly the same can be said of the Vackar.

In the Vackar schematic given in the link at Post No. 3 above, the coupling capacitor C2 can be turned down to the point where the circuit is just oscillating. I assume this is the point at which the loading is lightest. Not a lot is coupling the tank to the FET at that point.

Granitehill 11th May 2019 8:20 pm

Re: Franklin VFO ?
There's an article in "Rob's Web" which set me on this path. See :
This refers to a piece in "Ham Radio" magazine November 1989 written by VE3RF. This uses a pair of MPF102 JFETs and is probably where I'd start off. (I'm not looking for a design to copy as I want to use the project as a learning experience).
I'm not aiming to be too retro - no valves - it's to drive an early solid state Racal manpack (Squadcal) which was crystal switched rather than synthesized. I want to keep to the spirit of the time, so no digital synthesis. That's what I have at the moment - I'm using an HP3325A which works great, but is a sledgehammer cracking a nut in this context !

G0HZU_JMR 11th May 2019 8:54 pm

Re: Franklin VFO ?
1 Attachment(s)
OK thanks. I had a look at your link and it shows the JFET circuit below.
This doesn't look that great to me if you want a low phase noise oscillator. At first glance it looks like a scaled down version of the valve version but I think there are a couple of potential issues with the JFET version below.

Normally you would want lots of voltage swing in the resonator if you want to explore designs with very low phase noise. But the design below is limited by the circuit design. You can gets lots of voltage swing with a valve version but that is because it runs from large DC voltages. With the circuit below you might get 10Vpkpk here but that is just a guess and it depends on the components used in the design.

I'd expect to see a lot more than 10Vpkpk in a valve version. Even a regular solid state oscillator running from just 5V can deliver >50Vpkpk in a resonator. But I think the circuit below will be nowhere near this.

The other issue is that I think the 1kΩ in the drain of the JFET on the left is very low in value and I think it is going to (slightly) mess up the gain and phase response through the resonator. You would normally design for zero phase (around the loop) at the same frequency where there is maximum group delay in the resonator. This is good for stability and phase noise. But this design seems to mess this up a bit I think. Adding some shunt capacitance somewhere in the loop of the circuit below might improve things (by introducing some phase shift) but that is a bit of a band aid fix. The valve version looks better because it has a much higher resistance at the valve anode. Therefore, I'd expect the gain/phase/group delay response to be better in the valve version. But it depends how much you care about getting really low phase noise.

As drawn below I'm fairly sure it will still oscillate but maybe not with the full potential of the circuit. The part of the design I like the least is that 1kΩ resistance in the drain of the JFET on the left.

Granitehill 11th May 2019 10:32 pm

Re: Franklin VFO ?
Useful comments - thanks. I am also a bit suspicious about the noise performance. It seems that there is a trade-off of noise against stability inherent in the design, as the Franklin seems to feature a low level of oscillation, and relies on external buffers to get something useful out.

I'm quite optimistic over it's high stability potential, but the it remains to be seen whether or not the noise level is a problem in practice. The specific use is to drive a very modest rig, so I think some experiments will be in order. (I've got a good baseline standard from the present HP3325 synthesiser - which has a pretty good phase noise performance from what I've heard)

I feel a winter project coming on...

G0HZU_JMR 11th May 2019 10:47 pm

Re: Franklin VFO ?
If it helps, I quickly simulated it in open loop and also as a negative resistance oscillator and the loop response appears better aligned if there is about 70pF shunt capacitance at the drain of the JFET on the right.

This seems to correct the misalignment between the frequency of the peak in group delay and the frequency where there is unity phase shift around the loop. It also increases the negative resistance and I think the Vpkpk in the resonator would improve with this capacitor added because I'd expect the oscillator to 'park itself' closer to the centre of the resonance frequency of the LC resonator in the centre of the circuit. Without this external capacitance I think the oscillator won't be running at the peak of the response of the Hi Q LC resonator. It will be a bit off to one side I think. So this would be bad for stability and phase noise.

But adding what is effectively an RC time constant into the loop (to redefine the phase shift around the loop to correct for the misalignment) would be a bit unconventional.

Radio Wrangler 11th May 2019 11:06 pm

Re: Franklin VFO ?
The phase noise performance of the 3325A is fair, but not in the good league. It's a single-loop fractional-N synthesiser, but it is the 'Analog Phase Interpolation' type rather than the noise-shaping sigma-delta type. If it's been carefully adjusted (it has *lots* of adjustments) the fractional frequency spurs can be got below -70dBc. I know this beastie well, I had the job of redesigning its synthesiser around a new chipset when the old one went obsolete in the 1990s.

The 3335A is a much cleaner synthesiser, multiloop, but with API frac-N in the less significant loop.

The 8662A and 8663A sig gens were the low phase noise standards, if your ears could tolerate the fans.


G0HZU_JMR 12th May 2019 1:40 pm

Re: Franklin VFO ?
1 Attachment(s)
I'd expect a free running VFO at ~10MHz to have really good phase noise but I guess it depends on what the ultimate design goal is in terms of phase noise and stability. A Franklin oscillator should have a really high loaded Q because of the light coupling to the resonator and this should be good for phase noise and stability. I think my simulation gave an optimum loaded Q of about 50 but this was simulated down at 6MHz and it assumes that the gain and phase response is optimised at the peak in group delay in the resonator.

If there is a 0dBm power level and I assume a low device NF and a lowish flicker corner frequency then the phase noise at 1kHz offset would be better than -120dBc/Hz for this oscillator according to Leeson's equation. At 10kHz offset the simulation/equation predicts about -150dBc/Hz. If you look at the classic old ARRL phase noise graph from the 1980s this level of performance would be slightly better than the 'good' curve in the image below.

If the oscillator is redesigned to run up at 15MHz it will still be within a few dB of this and could be better if the design was tweaked a bit.

Most (synthesised general coverage) ham transceivers from the 1980s had really poor phase noise in comparison to this level of performance. They had to run the first LO up at maybe 70MHz which doesn't help, but the overall design of the LO in these early synthesised radios wasn't great.

G0HZU_JMR 14th May 2019 12:08 am

Re: Franklin VFO ?
1 Attachment(s)
If it helps, I tweaked the resonator and the tap caps to suit 11MHz operation and the simulator predicted a loaded Q of 50 for my changes. I think it's possible to go higher than this but this is a reasonable benchmark. With the gain/phase response aligned it looks like 25-30Vpkpk is possible in the resonator.

This is several mW and if I pump the numbers into Leeson's equation in an excel spreadsheet I get a phase noise prediction as below. Therefore, I'd expect that this is roughly what a real circuit will perform like assuming it is built correctly. In the real world, the phase noise response below about a 100Hz offset will be worse than this because of microphony and PSU noise/pickup but I think the response from 300Hz out to 100kHz should be about what you can expect from this circuit. It would be possible to improve on this with more tweaking but the phase noise below represents a basic/initial attempt at this circuit I think. So you might get slightly better results or slightly worse although I doubt it will be more than a few dB worse than this basic attempt unless something is wrong in the circuit.

The spreadsheet used for the plot below is normally used for VCO design so there is a contribution curve for noise from the tuning/steering resistance in the Vt line. But in this case this has been deactivated as there is no voltage control used in your circuit. So ignore the 'steering resistor noise' legend on the right of the plot.

One thing to watch out for is the sheer size of the voltage waveform at the second JFET if a high loaded Q is designed for. I think it would be possible to get close to the reverse breakdown voltage for the JFET here unless the circuit was modified to prevent this happening.

One niggle with this circuit is that it isn't immediately obvious how to (efficiently) tap off energy from it with a buffer amplifier. This is because of the very large voltage waveforms and the high impedances involved.

Radio Wrangler 14th May 2019 6:41 am

Re: Franklin VFO ?
There is a capacitive potentiometer at the gate of the second FET.... essentially the coupling capacitor as the series element, the gate-channel capacitance as the shunt element. This circuit may benefit from having a fairly small coupling C, or maybe added shunt C from gate to ground.

It is possible that gate breakdown could act as the amplitude limiting mechanism, with gruesome effects on the operating Q and noise behaviour.


All times are GMT +1. The time now is 10:58 pm.

Powered by vBulletin®
Copyright ©2000 - 2019, vBulletin Solutions, Inc.
Copyright ©2002 - 2019, Paul Stenning.