The AM Forum

This evening on the whatz fah dinnah net, we got into a discussion about W3DUQ's old upside down tube modulator / transmitter circuit. Bear was especially interested in it. I remembered that I had a jpeg of the original skizmatic of it. It is basically a high-level balanced modulator.

If anyone has questions about it, I may be able to answer some, but I have never built one.
                         
                                                           The Slab Bacon

I'm trying to understand it, so I re-drew it slightly more conventional.

Is the upper tube energized by the negative modulation polarity?
No? - because the clamp tube kills that?

I am missing something. When does the upper tube come into play? It seems like it can only work if the modulating voltage goes below zero, but there is the clamp.

Also it looks like the top tube receives only half the amount of drive as the bottom tube. Both drives are in phase, so I am not sure what is balalced as referenced by X in the original diagram.

Did I draw it wrong?

also I could not tell if the resistor from the lower tube's screen to mod B+ is 40K or another value.

on Frank's drawing IMMSMR R1 is a variable resistor which sets the point where the upside down tube begins to conduct audio as the class C amp approaches 100% negative. On the input side there is a parallel inductor variable cap network that I remember using. Pull the GD clamp tube and just watch the drive.

Patrick,
           Per our discussion with Bill last night, (and on previous discussions) You can either tie both control grids together and use a push-pull output tank, or feed the control grids push-pull (as shown) and tie the plates into a single ended pi output.

Like Dave said toss the clamp tube. It only gets in the way.

One tube conducts on the positive and the other tube conducts on the negative half cycle of the modulating audio voltage. Thus producing double sideband without a carrier. (note on the original skiz one end of the mod tranny secondary says B+ or ground . The amount of B+ that is applied to the other end of the mod tranny will produce enough carrier (DSBRC) to make it receivable with an AM detector and not need a BFO. With dramatic audio modulation peaks.

The whole thing basically runs on the AC audio modulating power. Bill has said that he used to run a pair of 4CX300s to modulate it without overworking the finals severely.

the input RF drive in this circuit is fed through a balun to provide somewhat equal drive to both tubes. I guess the whole trick to this circuit is jockeying the component values around to keep it balanced.

                                                               the Slab Bacon 

Is this the same as the old Globe DSB-100 transmitter?   It was a high level balanced modulator in the final amps.

My guess is the clamp tube's location in the circuit is a mistake.  The clamp tube should be AFTER the screen resistor, to pull the screen voltage to 0 when there is no grid current.

But, one question I *do* have is the need for the push-pull RF input (the grids are out of phase).  I can't see any good reason for this.  The "peak" tube is cut off anyway, and the plates are in parallel.  There is no "cancellation" of the carrier happening, because only one tube at a time is functioning.  The reduced carrier aspect of the circuit appears to be a function of a large amount of audio applied to the RF amplifier, with respect to the carrier DC.

Am I missing something?  At no time are both tubes conducting at the same time, so I fail to see the necessity for the balanced drive.

The input circuit transformer connections also appear to be a bit funky.  I bet in actuality, there was no connection between the primary and the secondaries, save for the ground.

I certainly remember Bill DUQ running the circuit.  On a good detector, it was fine - heaven help you if your detector wasn't up to par  :D

Regards,

Steve

maybe I'm missing something but I don't see this circuit operating any where near symetrically ... the upside down tube does not start to conduct until the negative swing of the modulating voltage completely offsets the B+ .... only then does the cathode go negative with respect to the plate ... what I'm not sure about is the effect of the rf + going swing on the dc grounded plate ... quite unorthodox but interesting .... 73 ...John

hi Steve .... just saw yer post .... This reminds me of a high level DSB generator from WCHB (Orr) with 2 4-250 's operating at 4kV ... circuit still resembles a push - push freq. multiplier ... javascript:void(0);

just had a chance to consider what Bacon had written and to further consider this interesting circuit ....

the application of B+ is to offset the modulation from symetrical ....if no B+ were applied then each tube would conduct some roughly equal amount set by class C bias during the + then - modulating voltage excursions and all pa power wpold be provided by the modulator with a limit of 100 percent before distortion ...  so by adding some B+ this modulation would tend to assymetry % in direct ratio of dc to pp mod voltage ....very clever ...73 ...John

8:30 pm  some more headscratching .... as Bacon and others have said, 0 V on B+ gives dsbsc ... to operate am some imbalancing is required .... this is the trick to figure since the B+ on the plate tube will dissipate more heat quickly than the dc ground on the plate tube .... I think metering both currents would be in order .... I am still trying to understand why the input transformer .... each grid could be driven from same phase signal .... maybe unbalancing the drive amplitude is what was going on to try to rebalance the load ....I wonder about using a 3rd tube with unmod B+ similarly just to generate carrier ...the best thing about this is the balanced loading on the modulator...could get interesting ...whatcha think ?

6:30 am april 8 2009 ... this 3 tube topology has captured my thinking .... I propose to call it a bird amplifier (a carrier tube flanked by 2 small mod tubes)... this seems like a direct synthesis technique .... since the carrier tube is acting as a single delta function and is slightly separate in frequency from the dsb generator spectra (separated by lowest modulating frequency), I don't believe that a combiner / ? will be required ... just combine the outputs as in the two tube (upside down or bi phase driven) designs ... in addition, since the mod xfmr would be carrying only balanced audio currents, this would allow use of commonly available push pull audio output xfmrs reverse driven by a solid state audio pa .... with this many outputs tied together will need to mind Miller effects and impedance

This circuit posted above (c. 1967) isn't the same 6DQ5 circuit that Bill/DUQ sent me in 1972, but is somewhat similar.  Maybe he made some improvements. Sadly, the schematic he sent me is long gone.

After some playing, mine worked really FB.  The only problem, as usual, is that the high audio peaks above 150% caused distortion in most receivers. I finally decided to go back to conventional modulation.  But it does perform well when the design is correct.

T

Tom- "somewhere" in my files I have an actual blueprint schematic of the 'DUQ upside down tube rig, and the "unilevel" tube compressor he used on it too. I never built the upside down tube PA but did copy the mixer and tone control section of the unilevel. it was my first decent audio chain (1973) and I still have it. it was in service at Mamaroneck until just a few years ago.

Chris,

It should be framed for the OB archives... :-)   Yep I remember WA2OMH in Mamaroneck.  The guys used to say,  "I can't believe Chris is only 16"  ....  :-)



The differences from the posted schematic are:

1)  In 1972 Bill used a standard push-pull L/C input coil to drive the 180 degree out of phase tubes. No toroids in the newer version.

2) No clamp tube.



I was thinking of building up a triode upside down tube rig. Not sure what yet.  The only "inefficiency" I see is having to use the SAME upside down tube as the final. The upside down (peak) tube draws very little current when it conducts as compared to the main final. It's such a waste of tube. But I understand that it's needed - to match the impedances of the two tubes in the circuit.

Maybe something like an 833A final and an 811A upside down tube would work if the 811A was biased so that it's impedance was artificially made to look lower, I dunno.

One thing that does need to be set up correctly is the little "wavelet" -  the conduction of this peak tube. The mini waveform must look exactly like the main waveform when going over 100%  (but smaller) or it will generate crud that sounds like IMD.  I think that peak tube needs to have a separate biasing arrangement and independent way of loading it to fine tune this wavelet.   The wavelet looks exactly like a balanced modulator waveform of any ssb rig running on AM.   The advantage of the class C upside down tube bal mod is simply for class C power efficiency, that's it.  The modulators, being class B, take a hit just like any plate modulated rig.


In 1972 I built the 6DQ5 upside down tube rig.  In 1988 I did it with a pair of 4CX-250's to be later changed to a pair 4-400A's..  A few years later I did it with a pair of 4-1000A's. Now THAT was a cool rig.


I might build another for the fun of it all, but an NE-602 with a linear will do the job even better at a slight reduction in overall efficiency.

T

Why bother with all that complication when you can just run a bigger final tube or more of them in normal mode to get more monkey swing.

You mean conventional plate modulation or a solid state bal mod using a linear?


I think the issue come down to resolving the following:

I'd be curious of opinions comparing a 180% modulated AM waveform using hard negative peak limiting in a conventional plate modulated rig vs: a balanced modulator which actually lets the negative peak through via the little negative peak "wavelet" between positive peaks.

The two waveforms look substantially different, but which one is "truer?"    Since ssb runs a balanced modulator with literally infinite modulation I would think that is the better way to represent the waveform when AM is modulated 180%+ rather than limiting/clipping the negative peak off.

Does the diode detector even "hear" this negative peak wavelet or are the positive peaks all that are needed?    (The negative wavelet is an exact duplicate of the bigger positive peaks.)


Basically, to prevent splatter, the difference is either keeping the negative baseline "fat" so it never sees negative 100% -  or fill it in with a wavelet. Pobably it makes no difference how it's done.

Opinions?

with apologies in advance to some who would rather not consider this....

Had an interesting qso with WA1HLR yesterday on 20 mtr (for about 3 hrs ) about this and other methods of:

(1) high level DSB/AM/SSBwRC generation ( in a glow in the dark fashion )
(2) doing this without audio transformers
(3) doing this with low distortion
(4) doing this without temperamental hard to adjust circuitry ( a goal not yet discussed)

since this was a technical discussion, please don't move the thread ...

The high level balanced modulator (or push-push topology, if you wish ) should be capable of meeting these criteria with a little extra work and circuit additions:

(1) the Upside down circuit
(2) the Bill Orr circuit

I have plans to modify a Globe DSB 100 to experiment with these and other variations ...73...John

John said:
The high level balanced modulator (or push-push topology, if you wish ) should be capable of meeting these criteria with a little extra work and circuit additions:

(1) the Upside down circuit
(2) the Bill Orr circuit


Reply by Jim, WD5JKO: Ok, so what is the Bill Orr circuit? I can see now the turmoil if we get DSB w/RC on 20M!  :D That sure would be fun though..

_______________________________________________________________________________

Tom said:
Does the diode detector even "hear" this negative peak wavelet or are the positive peaks all that are needed?    (The negative wavelet is an exact duplicate of the bigger positive peaks.)

Basically, to prevent splatter, the difference is either keeping the negative baseline "fat" so it never sees negative 100% -  or fill it in with a wavelet. Pobably it makes no difference how it's done.


Reply by Jim, WD5JKO: Tom, one weakness (of many) with the central electronics 20A on AM is that the negative peaks become "wavelet's" as you call them. The positive peaks are compressed, so in order to get a high percentage of modulation on AM, you have to put up with this issue. On a receiver with a diode detector, you hear a series of raspy clicks when this happens; the sound is just awful unless you run the gain down. Switch to SSB receiver, or AM synchronous detector, and it sounds fine.

A few years ago I went through a 20A conversion (to higher power using a EL-34 RF final tube). In the process I dealt with this issue you mention with a circuit that I call PNCL (progressive negative cycle loading). By attenuating the negative modulation peak in order to eliminate the "wavelet", the perceived quality on receive with a diode AM detector is much improved, and louder. Sure this does create distortion, but it is mostly 2nd harmonic so it gives you that wonderful tube sound.  ;D

I have a write-up of this rig with some scope traces with and without PNCL turned on:

open the link below and click on "20AQRO3.doc"
http://pages.prodigy.net/jcandela/CE20AQRO/

Fun stuff!
Jim
WD5JKO

If the circuit is just used to accommodate occasional negative overmodulation peaks, there is no problem using an ordinary AM receiver.  You probably never would hear  the  difference on an envelope detector.  But if you run up the modulation to increase sideband power, it sounds like an AM signal with selective fading  distortion or like slopbucket with the BFO off, depending on the sideband/carrier ratio.

With synchronous detection, DSBRC with up to 15 dB of carrier reduction should sound identical to full carrier AM.  Reception should be easy, with the availability of the SE-3, or the SDR synch detectors some people are now using.  The Sherwood performs excellently, but I'm sure the cost keeps a lot of people away.  The SDR version works almost as well, but then you are tethered to a computer.

The overall efficiency would be better using a low level exciter and linear amplifier.  With high level modulation, the efficiency is already reduced to 50-65% at the high power modulator stage, and further reduced to 60-70% of that, with the class-C amplifier, giving a total efficiency of about 30-45%, and even that is an optimistic figure.  Using the low level exciter, you get about the same efficiency as a leenyar running slopbucket, 50-65%.  It would be easier to run a linear in this service, since you are not contending with the limitations imposed by the full carrier.

Perhaps the efficiency of high level modulation could be greatly improved using class D or E, with a PDM modulator.  Wonder if any of the class E guys have ever considered the possibility of running a class-E final in balanced modulator configuration?

ok .... if you are interested in the Orr circuit either obtain a WCHB ed 15 or 16 OR go to BAMA and download a WRL DSB 100 schizmatic .... either one shows basically a push push topology tetrode multiplier with push pull audio on the screens ....works class C ...classic high level DSB generator

about efficiency .... I think high efficiency is what attracts me to this method and possible variants .... what I see as possible is independent control of carrier and sideband power and system distortion being a receiver detector issue rather than a generation necessity

I fully agree that synch detection with usb/lsb/dsb modes as described by Tim WA1HLR in our qso as being highly desirable.  Tim mentioned the ITU/IARU (?) standard for ssb inserted carrier as 10 db down as I recall ... listen to CHU on 7845 (i'm not sure of this freq ) to set your watch and hear this method ( I need verification of this ! )

I tried to work some ssb tonight .... how did I ever tune those poeple in ?  ???  ???

A double balanced mixer to generate DSB followed by a combiner so you and insert any carrier level you want is a cool set up I have seen.

GFZ ...sorry, I can't remember your name ....I saw Francis on QRZ database .... I am interested how


this was done ...

I was considering adding a 3rd tube just to generate carrier and controlling its screen voltage to vary power ...

Tim told me of a TMC transmitter that was capable many modes ...

I would appreciate your posting this    ... 73 ....John
 

I have an old Traynor Bass guitar amp (YBA-3A , (4) 6CA7 tubes - middle version) and its mounted upside down in the head...first wondered maybe some guy did this for a reason but then found pics of others that were the same. Always wondered why that Canadian company did that..

73, Tim K8WBL

I'll post it if I can find it. I sent Jay a copy W1VD a couple months ago. I think it is on my work computer but I'm finally hobe so it will need to wait till Monday

A couple months back I threw together a balanced modulator using a Mini Circuits diode ring mixer. +7 dBm to the LO port from a sig gen (VFO). Std line level audio (attenuated) into the i-f port. Output from the rf port drove an MRI amplifier chain. A potentiometer sent  DC into the i-f port which allowed adjustment of the carrier level over about a 20 dB range . Got sidetracked with other projects and never made detailed measurements. It looked promising.     

You can do the same thing with a couple of splitter/combiners, some attenuators and a DBM. I think it would be easier just to unbalance the DBM for whatever carrier level you want, unless this induces undue distortion.

When you offset a DBM it becomes temperature sensitive so carrier level will change with diode temperature. It also reduces the dynamic range of the mixer. So just use it to generate DSB then add a combiner after it to inject the carrier. A step attenuator would work fine. Then you just need a splitter to provide carrier to the DBM LO and the carrier inject path. The DC offset method will work fine though.

My notes indicate that the carrier level was somewhat 'touchy' to adjust using the dc offset...the combiner method is probably the better approach especially considering temperature effects.   

For precision, repeatable measurements, the DBM, splitter combiner method with step attenuators is the way to go. I did that back in the 80's for a work project. I laughed to myself while doing it, thinking, this is the most expensive AM rig I've ever run (it was all HP stuff, including several sig gens, 8663s, 8660s and the like).

With the upside down tube circuit, the carrier level is rock stable, since you set it by adjusting DC plate voltage.  Zero DC voltage, you have suppressed carrier  down to -30 dB or so.  Full plate voltage, you have regular plate modulated AM.  Anywhere in between, you have a reduced carrier which is a function of the DC plate voltage.

The screen modulated balanced modulator like the DSB100 has considerable distortion, since screen grid modulation is inherently non-linear, especially at the extremes of the SG voltage swing.  With regular AM, you can minimise distortion by limiting modulation to 85% or so.  But with the DSB100 circuit, you are driving the modulation of each tube beyond 100% in both directions.

I have a little UTC LS series modulation transformer rated for about 20 watts of audio.  I could use that to modulate a small class-C upside-down tube final, then amplify that with a big leen-yar.  Or I could use the modified Eico 730 modulator that is presently in my 10m rig.

I remember the old phasing type slopbucket exciters back in the early 60's, like the Central Electronics series,  Heapshit SB-10 and Johnson Pacemaker, in which the carrier null drifted badly and had to be fiddled with constantly during the course of a QSO.

Hi Don ... thanks for sharing your thinking and experience ... following the discussion with Tim, decided the first thing I want to do to the DSB 100 is to replace the 6aq5/transformer with 2 cathode followers .... I think a vert osc/driver like a 6ea7 would do the job nicely, one per 6dq6  ... I have that circuit designed and it is relatively simple ... can adjust each 6dq6 q point for balance/unbalance experiments .... will use negative offset voltage (like dx60 screem mod) to enhance linearity .... keep those cards and letters comin'   73   John

ps how big of an upside down tx have you worked with or is it a conceptual analysis?

Indeed.  I have one those, a CE 100V.  It produces surprisingly good sounding AM with the audio limiter and audio filter bypassed.  And the carrier null does drift.  So it occurs to me in DSB with the carrier not very nulled it seems like I'd have reduced carrier AM.  I think I'll run the experiment with my buddies who have various types synchronous detectors to see how well this works.

I'll hook up the peak reading bird Bird and feel legit if it doesn't bounce past 1500 very often.