EMRFD Message Archive 6390

Message Date From Subject
6390 2011-06-16 13:45:02 Mike Dinolfo 1n914 with jfet vfo
I've been recently working to wire up a jfet VFO. It's a solid state
conversion of a Heathkit (remember them?) VF-1, using J310 JFET's in
place of the tube. My ultimate intent is to use the VF-1 to drive a
homebrew 40M-10M CW transmitter in the 20-50 watt output range. But
I've encountered some "wierd" behaviour that I'm trying to understand.
The conversion is (very loosely) based on the QST Dec 1972 article on
solid-stating the VF-1. (The original circuit of the VF-1 is a series
tuned LC 6au6 pentode oscillator.)

The VF-1 has several bandswitched ranges. Range 1 generates a 160M
fundamental, Range 2 generates a 40M fundamental, and Range 3 generates
a fundamental at ??? (but the Range 3 switch position is marked "11M" so
it's presumably a sub-harmonic of the CB band; probably about 6.8 mhz.)
In modding the VF-1 and looking at the output on a Tek 454 (200 mhz
bandwidth) scope, I found out that the Range 2 and 3 outputs were
"squegging" (I think that's the correct term) with an almost
instantaneous buildup of VFO "fundamental" signal (around 7 mhz), then
decaying in amplitude rapidly to zero, then building up again. The
squegging frequency was about 700 khz, and it was intermittent. If I
tried turning off the power supply to the VFO, and then turning it back
on, the squegging would go away. But if I turned the Range switch (from
Range 2 to Range 1, and back to Range 2) the squegging would occur. I
tried various things to eliminate the squegging- adding bypass caps at
various points, shunting the 1 mH choke in the JFET source circuit with
a 220 ohm resistor, etc. But the squegging remained.

Then I tried adding a 1n914 diode from JFET gate to ground. Bingo! The
squegging consistently disappeared. However, now I notice a new
phenomenon- the output signal (at ~ 7mhz) has a harmonic (I think it's
the 24th harmonic of the fundamental, if my 60+ year old eyesight is
telling me correctly) with a maximum voltage amplitude about one-tenth
that of the fundamental. Also, the amplitude of the (24th?) harmonic
varies with an amplitude variation that tracks the 7 mhz fundamental.
It looks like the 24th harmonic amplitude is relatively constant over
the entire cycle of the 7 mhz waveform except when the 7 mhz waveform is
at about a maximum positive (or maybe it's the negative) peak; at that
point the 24th harmonic amplitude drops down nearly to zero, and then
builds back up over 3 or 4 cycles of the 24th harmonic frequency to a
constant amplitude.

So my question is: Do I leave the 1n914 diode in place, or remove it?
My gut feeling is: leave it in place, and don't worry about the 24th
harmonic signal; it's going to get stripped away by tuned circuits in
followup amplifier circuits. But also- does the the 1N914 have any
adverse impact? I have heard that it degrades phase noise, but maybe I
don't need to be concerned with that issue in this application (although
I'm also planning to eventually subvert the VF-1 output for use with a
direct conversion receiver- initially a KK7B Micro-R1 "improved"; maybe
eventually a MicroR2). And one other question that I have: When I was
previously getting a "non-squegging" signal with the original circuit
(without the 1n914, but turning on the power supply with the Range
switch already in position 2), the 7 mhz output signal was absolutely
clean (no evidence of a 24th harmonic.) So why would the presence of
the 1N914 in the circuit create the 24th harmonic signal? And one last
question: any suggestions on how I might eliminate the "squegging" in
the original circuit (without the 1n914)? I would prefer that any
harmonics that I need are intentionally generated in followup (probably
class C) amplifier stages rather than at the VFO source.

Thanks (in advance) for any thoughts that you guys and gals can offer.

Mike Dinolfo n4mwp
6392 2011-06-16 16:02:55 Ed - K9EW Re: 1n914 with jfet vfo
Interesting project, Mike.

I wonder if you're getting too much feedback which is swamping the FET
oscillator, creating the squegging.

This would also be consistent with the harmonics generated by the non-linear
1N914 diode on the input to the oscillator.

Perhaps some negative feedback would tame it down.

Keep us posted.

73,
ed - k9ew


6393 2011-06-16 16:12:33 Wes Re: 1n914 with jfet vfo
Hello Mike,

I suspect that the problem that is leading to the squeeging action is that 1.1 mH RF Choke that is in the source of the FET. I recall many years ago seeing a similar behavior when trying to convert a vacuum tube VFO to solid state. Like the Heath VF-1, it was also a Clapp topology with a big RFC in the cathode lead. One possible way to tame the beast is to add about 500 or 1K of resistance in series with the RFC. That will kill the Q at low frequency to prevent the potential for oscillation.

It could be that the same problem occurred with the QST design, but the fellow didn't have the instruments to see it.

I am confused when you talk about the 24th harmonic of the signal. If you are operating the circuit at 7 MHz, the 24th harmonic is going to be at 168 MHz. While you could see that with a 454 scope, you would not unless the 7 MHz was not present. If the 7 was there, it would probably dominate the display.

There is more discussion in EMRFD about the operati
6395 2011-06-16 19:51:40 Mike Dinolfo 1n914 with jfet vfo: Solved!
All:

In response to Wes w7zoi and Ed k9ew: Thanks for your thoughts, and it
looks like you were both correct. I tried inserting a series resistor
(actually a 2Kohm miniature trimpot) in series with the millihenry
choke, and I was able to "fine tune" the resistance value to (a)
eliminate the squegging, and (b) eliminate the 24th harmonic. And
without that pesky 1N914. I tried to take a photo of the scope trace
with the original circuit to demonstrate what the original signal looked
like, but didn't have much success. I suspect that this solution
(adding a series resistor) is one way of implementing Ed k9ew's
suggestiion to add some "negative feedback".

Doing a project like this offers perhaps the best of both old and new-
with a wide open chassis like the VF-1 provides, with miniature
components it's real convenient to implement "ugly construction,"
particularly since the chassis is already copper plated and I can solder
components directly to the chassis (with a big enough soldering
iron/gun, of course!) Life is good.

Thanks again for your help!

Mike Dinolfo
n4mwp

6404 2011-06-19 12:40:34 be4july86k Re: 1n914 with jfet vfo: Solved!
Mike

Many years ago built a Heath GR-78. It's a band switching, SS, general
coverage receiver. The upper bands never worked correctly. Finally
obtained a good scope and saw the local oscillator was squeeging on
the bad bands. EMRFD discusses squeeging on page 4-4. The LO uses an
MPF105 FET. The gate coupling capacitor was originally 100 pF.
Changing the capacitor to 5 pF stopped the squeeging and output was
good on all bands. Also added the diode from gate to ground.

I did notice a much higher frequency signal riding
6406 2011-06-19 16:15:02 Mike Dinolfo Re: 1n914 with jfet vfo: Solved!
Buck:

Thanks for your suggestion. So far, I have it working OK (nice clean
output when viewed on a wideband scope, at least for the higher
frequency output). There isn't any series gate coupling cap; just a 510
pf cap from gate to source and another 510 pf cap from source to ground;
these two caps provide RF feedback. Then there's a series tuned
(capacitor tuned) LC circuit from gate to ground, a 1 meg resistor from
gate to ground, and a series connection from source to ground of 1.1 mH
and about 1000 ohms.

6407 2011-06-19 20:10:03 davidpnewkirk Re: 1n914 with jfet vfo
> I've been recently working to wire up a jfet VFO. It's a solid state
> conversion of a Heathkit (remember them?) VF-1, using J310 JFET's in
> place of the tube. My ultimate intent is to use the VF-1 to drive a
> homebrew 40M-10M CW transmitter in the 20-50 watt output range. But
> I've encountered some "wierd" behaviour that I'm trying to understand.

and

> bandwidth) scope, I found out that the Range 2 and 3 outputs were
> "squegging" (I think that's the correct term) with an almost
> instantaneous buildup of VFO "fundamental" signal (around 7 mhz), then
> decaying in amplitude rapidly to zero, then building up again. The
> squegging frequency was about 700 khz, and it was intermittent. If I
> tried turning off the power supply to the VFO, and then turning it back
> on, the squegging would go away. But if I turned the Range switch (from
> Range 2 to Range 1, and back to Range 2) the squegging would occur. I
> tried various things to eliminate the squegging- adding bypass caps at
> various points, shunting the 1 mH choke in the JFET source circuit with
> a 220 ohm resistor, etc. But the squegging remained.
>
> Then I tried adding a 1n914 diode from JFET gate to ground. Bingo! The
> squegging consistently disappeared.

I'm going to suggest something a little different about eliminating the squegging problem you're encountering (and which quite a few experimenters with JFET oscillators have experienced at one time or another: I'm going to suggest that you use two JFETs, or maybe a JFET and a BJT, to synthesize the 6AU6 tube you seek to replace.

I've just had a look at that December 1972 *QST* article ("New Life for the Heath VF-1 VFO" by Robert M. Glorioso, W1EBW, pages 18-19). In the original VF-1 circuit, a 6AU6 pentode serves as the VFO's only active device. In the conversion, the author replaces the 6AU6 with a a JFET (2N3823, but this is not critical) and then bolts on a second JFET as a common-source amplifier (with its 560-ohm source resistor unbypassed) ac-coupled to the first JFET's hot-for-RF source.

The resulting circuit is not a direct equivalent of the circuit he started with in this key sense: A vacuum-tube pentode (or tetrode) is equivalent to two triodes in cascode, but the circuit he ended up with uses two separate triodes that are *not* in cascode. Instead, one serves as a common-drain triode oscillator and the other serves as a common-source triode amplifier with a tuned anode. The oscillator JFET is likely squegging on its own as a result of two-terminal negative resistance in its gate or drain, but because the amplifier JFET is ac-coupled to the oscillator tuned circuit *and* its drain is tuned, it may well be getting into the oscillatory act as well, and/or may do so if the conditions in the actual oscillator device are right.

Squegging is really superregeneration, and vice versa. Superregeneration happens when an RF oscillator simultaneously oscillates at AF or low RF such that its wanted RF oscillation starts and stops with each cycle of the unwanted lower-frequency oscillation. A practical superregenerator does this *quenching* at a supersonic rate so the interruptions are inaudible to the user; blocked oscillators, used in some RF ID systems, may quench at a subaudible rate.

Users of regenerative receivers based on vacuum-tube triodes experienced this same effect as *fringe howl*, a fixed-frequency AF oscillation that occurs right around the point at which a regenerative detector goes into and out of oscillation--at the *fringe* of oscillation.

And why are triodes the bad actors? Because a triode's anode current--at least in vacuum tubes and JFETs--is greatly dependent on its anode voltage. The net effect seems to be that setting up the conditions for three-terminal RF oscillation in these devices--and it's worse with JFETs--also sets up conditions of *two*-terminal negative resistance that can result in simultaneous RC- or LC-based oscillation at a frequency set by time constant(s) in the device grid/gate and/or plate/drain. Inserting enough positive resistance in series with such a negative resistance is one way to stop such oscillations; if you're lucky, the wanted oscillation is still going when the unwanted one is killed.

For users of vacuum-tube regenerative detectors, the advent of the tetrode largely did away with fringe howl as triode users had known it. A vacuum-tube tetrode inserts a second grid between grid 1 and the anode; this second grid, operated at a positive voltage significantly lower than that on the anode, makes the tube's anode current relatively independent of the anode voltage. Fringe howl based on anode negative resistance interacting with a reactive output load--say, the interstage transformers commonly use to couple triode detectors to the first AF amplifier--just can't get going as a result.

So what I suggest is that you replace your 6AU6 with a solid-state tetrode. A dual-gate MOSFET is the first choice; a BF998 from Digi-Key would be great, if surface mountable. (:-)) If you don't have one or a leaded equivalent, two JFETs in cascode, or a hybrid cascode (BJT [upper/output triode] in cascode with a JFET [lower/input triode]) will serve. The two-cascode-JFET mixer of EMRFD would be a good starting point. Any JFET from the MPF102 through the J310 will perk.

The output amplifier of the December 1972 article should not be necessary and may well contribute to instability because it's pretty much set up as a wannabe tuned-gate, tuned-drain oscillator by virtue of its ac-coupling to the oscillator resonator/cathode network. (That the author did not bypass the source of that output amplifier makes me wonder about exactly what behaviors he encountered.)

Just connect your two cascoded devices in place of the 6AU6 as follows:

upper device drain/collector: 6AU6 plate
upper device gate/base: 6AU6 screen (grid 2)
upper device source/emitter: to lower-device drain
lower device gate: 6AU6 grid 1
lower device source: 6AU6 cathode.

And power the shebang from 9 V max (with a BF998) or 12 V with two JFETS or a hybrid cascode. Set up the cascode upper-control-element (gate or base) biasing as per the JFET-cascode-mixer treatment in EMRFD or the hybrid-cascode technique as described by Hayward and Damm.

Wes's point about de-Qing the oscillator cathode choke with series resistance is a good one. To do that and pretty much make unnecessary a gate-clamping diode on the lower device, add, say, 470 to 1000 ohms, *unbypassed*, in series with the choke--as much resistance as will allow the stage to reliably start oscillating while giving sufficient output.

Finally, don't get your output by coupling to the lower-device source--get it from the upper device drain/collector, just as in the original 6AU6 circuit. If you must add more amplification, ac-couple it from there; the buffering by the upper cascode device of that point from the oscillator tuned circuit is a major feature of the cascode arrangement.

Best regards,

Dave
amateur radio W9VES