Pullen Mixer

[N4LTA]

I have decided to use a Pullen Mixer in an  am  receiver that I am building.

I will use a 6C4 local oscillator. I have seen some comments that say a cathode follower buffer might be necessary and some say that the Pullen only lightly loads the LO. Do I need a CF?

I am in the process of punching holes and trying to get this thing layed out.

Pat
N4LTA

[WU2D]

The second section of the Pullen is in effect, a high Z input, so it takes care of the loading situation.

Mike WU2D

[K1ZJH]

Anyone have any suggestions regarding the terminating impedance for a Pullen mixer?  I assume one wants to avoid the high off resonance impedances when driving a xtal filter; so what would be a good starting point for the grid resistor on a cathode follower? 15K or so?

Pete K1ZJH

[KM1H]

If you research the subject you will find that Keats Pullen and follow up published info suggest a buffer.

This may not be required with a crystal oscillator.

Carl
KM1H

[N4LTA]

Yep - That is why I asked the question - Pullen said a buffer might be necessary - yet the input impedance is very high and others have not used one and said it is not necessary.

So - has anyone built one and used or not used a buffer?

Pat
N4LTA

[K1ZJH]

Impedance of the LO or Output port?  I'd think you'd want the output feeding a fairly well defined, broadbanded termination, lest the voltage gain soar when the crystal sees signals that are off resonance and the filter goes to a very high impedance.

Pete

[N4LTA]

There will be no crystal filter. I am asking about the LO port.

[k4kyv]

The problem with the buffer stage is deterioration of the performance of the mixer.  It inevitably has some nonlinearity and internal noise.  You want as few active stages as possible between the antenna and main selectivity filter.  This includes rf amplifiers, mixers, i.f. amplifiers and buffer stages of any kind.  This is the reason the dynamic range of multi-conversion receivers is inferior to single conversion ones.

Ideally, a receiver would have no tuned rf stage at all, but instead a passive tuned network between the antenna and mixer, and the mixer would feed directly into the crystal, mechanical, L-C or whatever selectivity filter is used.

You need a mixer with enough gain that the signal coming in directly from the antenna is able to overcome internal noise in the mixer and deliver enough signal through the filter that the weakest signal received will overcome the noise floor of the i.f. amplifier and stages that follow.

One of the better mixer circuits uses a beam deflection tube, such as the 7360 or similar.

Diode mixers are extremely linear, but have insertion loss instead of gain.  An extremely good receiving antenna that collects plenty of signal would be needed in order to use one of these without an rf stage to directly feed the selectivity filter, and still be able to pick up the weakest signals all the way down to the atmospheric noise floor.

It would be interesting to compare the performance of the Pullen vs beam deflection mixer circuits.

[KM1H]

Ive built quite a few Pullens Pat while using a S-40B as a test bed. Tubes tested have been the 6SL7, 6SN7, 7F8, 12AT7, 6ES8 and 6J6. In that radio pulling was noticed on the 2 higher bands.

Its also been apparent on the highest HF band of a NC-183, SX-28, a pair of HQ-140X's, and HRO-5. However since my primary use for vintage radios is AM with a bit of low band CW its not a real issue.

As a product detector driven by an unbuffered BFO Ive experienced no pulling in any radio and Ive used up a lot of my NOS surplus 6J6's that way.

On a project that will be used for SSB/CW Id suggest buffering a free running first LO that doesnt have a high degree of built in isolation. Also be sure there is sufficient LO voltage.

I havent modified a xtal controlled LO set yet but a R-388 and 75A2 are on the project list where I doubt there will be any pulling.

A cathode follower at the IF output of a Pullen would be a good idea for isolation and feeding any low impedance circuit or filter.  Any noise contribution can be simply dismissed as inconsequential as the front end has already established the system noise figure which should be below band noise even on 10M. I use a HP-8970A NF meter to verify performance. In a multi conversion design I wouldnt use circuits that would degrade the benefits established by the Pullen. This means dont use a 6BE6/6SA7!

I have not spent much time on the W2DXH Pullen revision. The single attempt using it as a first mixer led to instability. Maybe revisit when I have time but for now the basic Pullen is fine for my needs.


Carl
KM1H

[N4LTA]

Carl,

This is a receiver that I plan to use on 80 and possibly 40 for AM reception. I have left some room for a BFO - But initially it won't have one.

My plan was to use it to test some ideas and make a very decent receiver for monitoring the local AM net. It will be a table model radio as such. It will have no RF amp (not needed at these frequencies) will have a 6C4 LO that will be as stable as I can make it, a 6ES8 Pullen Mixer  - The 6ES8 is probably overkill but I have some, 2 stages of  455 Khz IF, a detector that I am still pondering - (6AL5 biased diode or an infinite Z detector and I probably will have a space for Steve's (QIX's) solid state detector/AVC circuit. I made a PC board  for it and have not built it)

I thought seriously about a 7360 or the TV tube you suggested (I have both of these) but the problems of the input port not liking complex impedances made me a little gun-shy. I don't need or want a RF amp ahead of the mixer.

It is more or less a receiver experiment for fun but I do plan to use it. So far  - I have the power supply working and the 6BQ5 output tube working fine.



I got around to wiring the 6C4 LO tonite. Using a hand wound inductor on an old 3/8" ceramic form (tapped for feedback) I got it working at around 3.9 mhz first time with a single 270 pf NPO cap. That's in the ballpark of what I was looking for (4300 Khz)so that is working OK. I'll figure out what size variable I need and add 2 or 3 NPO's and maybe a polystyrene to get the LO as stable as I can.

 I get about 7-8 volts Pk to PK on the feedback tap.

Pullen says that for some tubes a LO injection voltage of 2-5 volts Pk to Pk works fine. Has anyone used a 6ES8 and remember what the injection level was?

Thanks for any additional info. I hope to get the mixer wired tomarrow night.

[KM1H]

I answered your PM already with the injection info Pat.

Let me know how the 6ES8 handles with no RF stage. Thats something I dont believe in even for 160 for any radio.

Carl
KM1H

[N4LTA]

What would you suggest as a RF amp ahead of the mixer for 80 and 40?

Pat
N4LTA

[N2DTS]

I am with you Don!
I use a very high Q tuned circuit with a link coupled antenna input right into the mixer, then out of the mixer is the filter.

You need no rf amp before the mixer below 14 MHz if you use the transmitting antenna.
Many radios were designed for use with little short antenna's and even with the RF amp work poorly.

My homebrews have more gain than is needed, and I run the IF gain at 3/4.

When does mixer noise become important?
I just use a crappy setup, LO in one grid, ANT in the other, and with the antenna input shorted its dead quiet.

With the antenna port hooked up to the station control, but no antenna selected, or a different receiver selected, I can hear stations, that is with no connection to the antenna, a 1/4 gap on the rotory switch.....

When I had the R390a, and the 756 pro3, the homebrew was always better copy of very weak signals.
On the R390a and others, the signal was always down in the hiss...

Brett
 

The problem with the buffer stage is deterioration of the performance of the mixer.  It inevitably has some nonlinearity and internal noise.  You want as few active stages as possible between the antenna and main selectivity filter.  This includes rf amplifiers, mixers, i.f. amplifiers and buffer stages of any kind.  This is the reason the dynamic range of multi-conversion receivers is inferior to single conversion ones.

Ideally, a receiver would have no tuned rf stage at all, but instead a passive tuned network between the antenna and mixer, and the mixer would feed directly into the crystal, mechanical, L-C or whatever selectivity filter is used.

You need a mixer with enough gain that the signal coming in directly from the antenna is able to overcome internal noise in the mixer and deliver enough signal through the filter that the weakest signal received will overcome the noise floor of the i.f. amplifier and stages that follow.

One of the better mixer circuits uses a beam deflection tube, such as the 7360 or similar.

Diode mixers are extremely linear, but have insertion loss instead of gain.  An extremely good receiving antenna that collects plenty of signal would be needed in order to use one of these without an rf stage to directly feed the selectivity filter, and still be able to pick up the weakest signals all the way down to the atmospheric noise floor.

It would be interesting to compare the performance of the Pullen vs beam deflection mixer circuits.

[N4LTA]

I think Don misunderstood  what I was talking about in using a buffer.

I had not and wouldn't put a buffer in the RF path.

The buffer that I was asking about was in response to Keats Pullen's comment that the Pullen mixer showed a tendency to "pull" the LO at higher frequencies. He suggested  putting a cathode follower to buffer the LO. Carls tests showed the same LO pulling at higher frequencies. Since this receiver is not going to be used above 40 meters, I am not going to install it.

Conventional wisdom is that you have as little signal altering ahead of the mixer as practical. That is why I am not putting a RF stage ahead of it at the low frequencies that this receiver is designed for.

Carl disagrees and I would like to hear his opinion.

The 7360 mixer is supposed to be a very good performer but has been reported to have bad tendencies if the RF port is not matched to the design (usually 50 ohms). The usual way to correct this is to add a RF amp stage with a known constant output impedance. That somewhat defeats the purpose - in that the 7360 requires an active stage ahead of the mixer - or is rumored to require one.

Wire antennas rarely supply a constant impedance to a receiver.

This receiver is somewhat of a test bed and I want to try to observe some of the state of the art  tube technology. I have already build many simple and not so simple solid state receivers and a few superhet tube receivers. The ability to hear weak signals is not my goal - that can easily be done with a simple tube regen receiver - I find them to be about as sensitive as you can get - and also nearly unusable for communication.

Dynamic Range - IMD and Blocking are much more important to me than MDS

The best receiver that I have ever used is my K3 - hands down - I have not used a R390 but have heard most everything else including a 74A4 Howard Mills receiver I owned a few years ago..

The K3 Handles itself with close in blasting signals and is a joy to listen to. I have an ICOM 718 that I bough to play with a few years ago and after listening to the K3 - I can't stand to listen to the 718 with the grating background noise.

That's my opinion only.

[KM1H]

Any AGC controlable tube will work Pat, even a 2.5V 58 setup for unity gain.  The RF stage is used to present a reasonably constant load and match to the mixer and a single LC stage to the antenna doesnt cut it. A 7360 directly to an antenna is a known disaster with some antennas and Ive no experience running a Pullen that way. Reading thru Pullens notes and analyzing the circuit leads me to believe it isnt a good idea. An RF stage with gain wont faze it and you have some place else to use AGC control for an all around better performer. I doubt that even a double tuned passive preselector will provide sufficient isolation so why take a chance?

A 6L7 or 6K8/6SA7/6BE6 noise generator will be covered by a decent antennas noise on 160-40M and several basic radios do well that way, and if all you are looking for is absolute basics get a S-38, SW-54, or NC-60 and be done with it. But what you are going for is signal handling with the Pullen which is the whole reason for this exercise.

Once you get it all working it might be interesting to see what happens with the RF tube replaced with a 100pf cap.

Carl

[N4LTA]

I have room for the RF amp - I am building this thing piecemeal - so I will try it with and without.

The article that got me thinking about this was in November 2008 Electric Radio for a high performance front end using an Amperex frame grid tube (I don't remember the number) with a double tuned LC preselector in front of it  and a single tuned LC circuit after it all on a 3 gang variable cap.

Would you recommend a tube for the front end - reasonably high performance - in case I won't to upgrade this animal later.

Thanks,

pat
N4LTA

[KM1H]

I use the 6GM6 almost exclusively for swaps. Lowest noise figure of the 7 pin tubes on 10M, and darn near overload proof. For 9 pin the 6EH7 and 6EJ7 are hard to beat.  The 7788 is hot as a pistol with obscene Gm but short lived.

The 6DC6 beloved by Collins can be a good overload performer but there had been some supplier problems resulting in inconsistency. Also RCA had been caught rebranding 6CB6's. Ive replaced the 6DC6 in my 75A4, SX-101A and SX-115 with noticable improvement. I havent touched the R-390A.

Carl
KM1H

[N4LTA]

I ordered several this afternoon (6GM6) after some searches showed some of you older posts.

I suppose a tuned circuit before it ( I like to use toroids) with a  link coupling to the antenna and a tuned circuit after it using a dual gang cap would be suitable?

Thanks again for your help.

Pat
N4LTA

[KM1H]

Im a big dual winding toroid fan Pat. Ive installed several in boatanchors for a substantial SNR performance without tube mods which are then the next step. Having a HP Q Meter helps evaluate and select the best performers.

I get more enjoyment squeezing high performance out of vintage radios than I do using them every day for hours on end!  Kinda sick, huh

Carl
KM1H

[N4LTA]

I'm more of a builder than operator myself. I am trying to do more operating.

Don't have a Q meter - Would like to pick one up - I do have some decent test eq. - HP SA - HP RF Gen  400 Mhz scope - HP Pwr meter and some decent HP power supplies.

I have had some pretty good results using toroids in tube transmitters - I have a 40 watt CW transmitter using an 807 with a T130-6 in the pi-net. About 28 watts out and the toroid runs cool.

I had a 160 meter 4-1000A linear with a T400A-2 as the pi net inductor. Easy 2000 watts out and never had the slightest problem with the toroid.

[N2DTS]

As far as Q goes, what do you think has a higher Q, a toroid or some big B+W coil stock?

I dont have a Q meter, but had much better results with the B+W coil stock and an air variable cap than with any sort of cores.

Brett

[N4LTA]

I'd guess the B and W. It would depend on the type of core material. The ferrite material Qs are lower.

The Q on a piece of wire with an air core is going to be about as high as you can get. There is also a toroid core with a permeability the same as air - to allow the self shielding of the toroid form and high Q of an air core inductor.

I'd never use any type of core material on an VFO inductor toroid or otherwise - only air.  Some say the type 6 and 7 material is OK but it all has variable permeability with temperature.


The problems that I have with the B and W stuff for low level inductors is that it takes up space and is not self shielding.


BTW - Thanks for the info you sent me. It has been very helpful.

pat
N4LTA

[N2DTS]

What I did was mount the B+W coil for the input on an 8 pin octal plug, then installed a socket for it on the chassis. That way I could experiment with different setups.

I tried tunable ceramic forms, untuned ceramic forms, and the B+W coil stock.
I also tried various multi coil and cap setups, but what worked best was one B+W coil with link antenna input, and a 365pf cap.
If you get the coil right, it can cover 160 to 40 meters and is has very sharp tuning.
The Q must be high enough that I don't seem to have any image problems at all.

The first receiver used slug tuned forms for the LO, and along with the octal tubes drifts a lot on startup.
Its good enough for AM, plus the frequency readout is digital so shows the actual frequency (no calibration needed), I just have to retune for the first 2 minutes or so.

The 2nd receiver LO used the B+W coil stock in a small metal box, and it drifts very little, at startup or after.

I read all sorts of stuff in old QST's about guys building uber stable VFO's using huge coils in even bigger boxes, like transmitting coils!

Brett

[KM1H]

A mix 6 or 7 powdered iron toroid makes for an extremely stable VFO for a FET. Id suspect that it would be more than adequate in a Vackar 12AT7 type circuit also but I havent gone that route yet. Never use ferrite cores in a VFO. Even slug tuning requires playing forever with TC circuits.

A lot of my HP and other high end test equipment came from various internal company auctions where I worked or publicized belly up ones. The HP 4342A Q meter shows up at times at ham prices. A real sleeper is the military version TS-617D/U. The TS-617C/U is a superb late 70's tube Q Meter that doesnt have the inherent problems of the Boonton which are also available under earlier TS-617 letter designations.

Ive had no problem using various size 2 Mix powdered iron cores to add 160M to commercial amps up to the 1500W level. Alpha and others have used toroids since the 70's at QRO levels yet hams generally seem to shun them.

In vintage receivers the coils used are pretty lossy on the higher frequencies with Q's measured in the 100-125 range. Some 30's sets with less developed insulator technology have soaked up enough moisture and pollution that Q's are much lower. It makes me wonder what will happen to the uber cheap Heathkit and other basically paper cores as they age? A 2 or 6 mix of at least a T80 size will measure around 225-250 and offer a 5-15 dB improvement in images just as the antenna coil. Tracking is straightforward by tweaking turns spacing for the low end and the original trimmer at the high end. I havent used them in the following stages.

My HB project is using all toroids with relay switching. Nothing destroys the Q of an airwound coil faster than placing it inside the chassis with other coils, parts, and wiring. Its not much better with slug tuned coils and all the associated switching and haywire associated in the average boatanchor. The last time I used plug in coils in a receiver project was around 1956 .  However they do the job in the FB7-XA and various HRO's in use here.

Carl
KM1H

[w3jn]

It's *always* best to use a buffer stage between the osc and next stage - be it mixer, mulitplier, or whatever.

Instead of using a 6C4, why not use a 6J6.  Takes the same room.  Cathode couple (ie don't completely bypass the common cathode lead).  Grid leak of 100K or so.  Take the signal off the plate of the second section.   That's the way I did it in my HB and I didn't notice any pulling at all.