A Quad-1625 Modulator or Speech Amplifier, straight out of the junkbox!

[w8khk]

After seeing Steve's excellent post to a reference describing improved modulator grid drive technique, I was prompted to share my latest project, completed a week ago.  On a slightly smaller scale, my project shows practical application of an almost identical driver circuit, but with no driver transformer.  The entire project was completed in about five days of free time.

Link to Steve's post:  
http://amfone.net/Amforum/index.php?topic=46695.msg333980;topicseen#msg333980

I had been toying with the idea of building a moderately-powered audio amplifier, to use as a modulator for my Viking II and/or a speech amplifier for the 810 modulators in my P-P 250-TH rig.  There are many different ways to improve upon 50s technology with 30s and 40s tubes and other components, and I had several ideas I wished to try.  The results were so good, that I decided the new amplifier would be used for the center channel in my home theatre system, and I will build two more stereo versions for the front and surround channels.

I will post schematics and some photos of the process of recycling rusty junk into something for a shelf in the living room.  (It can still be used in the ham shack, of course!)

I have built and studied more push-pull audio amplifiers than I can count.  Looking a the pros and cons of many designs, I devised a hybrid design, based on the classic Williamson, melded with features of the UTC MLF (Multiple Loop Feedback) design.  I added to these designs a regulated screen supply, and cathode follower drive for the push-pull parallel output tubes.  (While the circuit includes 47 ohm grid stoppers on the output tubes, these were found to be unnecessary.)

Recycled, junk-box parts were used, with the exception of the 1625 tube sockets, output connectors, resistors and small capacitors.  Even the 10 by 12 inch chassis was reused, and I repurposed as many of the holes as possible.  The transformers were dirty and rusty, and the wires were clipped short.  I removed the end bells, cleaned and painted the bells and laminations, then attached new leads, leaving them long as I soldered them into the new circuit.  

All the tubes are used.  The 1625s are from a pair of ARC-5 transmitters that my father rack-mounted and used as 75 and 40 meter exciters for his P-P 304-TL rig.  (I have plenty of 1625s, these just happened to be handy.  I will not scrap or part out the ARC-5s.)  All the other tubes are from my boxes and drawers of old tube stock, and all worked fine without any issues.

The main power transformer is a Stancor C8412, providing 800 volts center tapped at 200 mA, a 5 volt 3 amp, and a 6.3 volt 5 amp winding.  A 10 henry, 200 mA choke is included, and the supply is set up such that a jumper can select either choke or capacitor input.  It is interesting that the transformer is rated for 117 volts input, but with 125 volts input the voltage, under load, of the 5 and 6.3 volt windings in series is 12.5 volts, making the 1625s happy.  A bridge rectifier and filter provides 12.6 volts DC, a 6SN7 in series with a pair of parallel 12AU7s in the preamplifier stages have less than 15 millivolts ripple on their heaters.  A separate Stancor PS-8416 provides 6.3 volts at 1 ampere for the 6L6 and 6AU6 in the regulator section, as well as 250 volts center tapped, with a full-wave rectifier provides the negative bias of 160 volts DC.

I chose a quad of output tubes in push-pull parallel because the Thordarson T-22S78 60 watt output transformer from the junk box has a primary impedance just over 3000 ohms.

The cathode follower 6SN7 has the bias set individually for each grid, and this bias is coupled directly to the 1625s, allowing the bias to be balanced between the output tubes.  About 20 volts negative bias is required on the 1625 grids for around 30 milliamperes cathode current for each output tube, assuming a regulated screen voltage of around 240 volts DC.  Each 1625 cathode is tied to ground via a ten ohm resistor, and a test jack is provided to measure the current - 1 volt indicates 100 milliamperes.  This is much simpler than adding closed-circuit headphone jacks.  The resistors do not provide any significant cathode bias, but they will serve as inexpensive fuses, should one of the 1625s fail.  Each power transformer is fused separately for the required primary current.

I tested with and without the feedback circuits, but found that, for the power levels I want to run (maximum 50 watts) the feedback is not really necessary, even for concert level sound in the living room.

After the first post I will add photos showing the build process, from scrap to finished product.  I hope this will provide ideas for others that wish to build class AB2 tetrode audio power amplifiers for their station.

[w8khk]

Assembly Progress Photos

[w8khk]

Transformer Restoration

[w8khk]

Project completed, on the test bench, showing with the HP distortion analyzer and Tek oscilloscope that no inverse feedback is needed.  It would be added if I decide to drive the 810 grids in the 250-TH rig.

[K1JJ]

Hi Rick -

Nice wiring job underneath!!   Neat and solid as a rock.  That's the way I like to see them turn out.

T

[w8khk]

Hi Rick -

Nice wiring job underneath!!   Neat and solid as a rock.  That's the way I like to see them turn out.

T

Tom, thanks for the kind words.  

I think the most important wiring detail is the grounding method.  In order to minimize or eliminate hum in audio circuits, especially low-level stages, it is necessary to make sure the grounding method does not introduce hum.  If the chassis is used as a ground bus, and there is any electromagnetic disturbance in the area, say a power transformer or filter choke, then the AC current induced in the chassis will be adding hum to the signal.

I chose to use a number-12 ground bus wire to return all circuits to ground, and it is connected to the chassis at only one point.  The best place to connect the ground to the chassis is at the lowest level, highest impedance input connector.  From there, no other circuit connections are made to the chassis.  The grid return resistors for each stage are grounded to the isolated bus wire.  The decoupling bypass capacitors for each stage are grounded to the bus as close as possible to the grid return, ensuring there is no difference in potential between the power supply ground for that stage and the signal return ground.  The actual power supply ground and the AC safety ground attach to the same bus at the FAR end of the bus wire.  This way, each stage has extremely clean B+ source, and any hum currents in the chassis are not introduced into the audio path because nothing is connected to the chassis proper.

All AC-powered filaments (tube heaters) receive power through a twisted pair of wires that do not return to chassis ground at any point in the circuit.  The twisting of the wires cancels the electromagnetic hum radiation from the wires.  In low-level audio, the chassis should NEVER be used as a filament ground return.

The low-level amplifier stages receive filtered DC from a bridge rectifier circuit, with a series resistor added to reduce the voltage to the nominal rated voltage required for each tube.  When the 12.6 volts is rectified and filtered, the resulting voltage is somewhere between RMS from the transformer and peak (RMS * 1.414) and it is necessary to reduce this voltage via a dropping resistor such that tubes run at their rated filament voltage.  The ripple on the filament voltage on the 12AU7 input and phase inverter stages is less than 10 millivolts with a 3300 uF electrolytic at the rectifier, and another at the 6.3 volts on these two tubes in parallel.  The drop across the cathode follower 6SN7 adds additional filtering to reduce the ripple on the input stages.  (See how this is done in the power supply schematic, which may be downloaded from the link in the first post of this thread.)

Another step that can be taken to reduce hum and noise in low level amplifiers is to elevate the filaments to a DC potential above the cathodes of the tubes.  If this is not done, there is a potential for the filament and cathode to function as a diode, whenever the cathode is positive with respect to the filament.  By elevating the filament to a higher positive potential than the cathode, there is no current flow in the filament/cathode diode structure to modulate the cathode bias, thus removing another potential source of hum and noise.  All high-quality tube amplifiers employ this technique.

[K1JJ]

Rick,

That's a very interesting discussion about using an isolated-from-chassis ground bus and other common point connections - and DC fils.   I never used such a comprehensive set of techniques aimed at hum reduction for audio, but then again, I've had my share of hum battles too.    Would you say that it is OK to use the chassis as a ground plane  when building a set of gain stages in an RF amp, but to use these bus isolation techniques for audio?

When listening on a good receiver like an SDR and well filtered supplies, I can usually hear some 60 or 120 Hz hum on at least 40% of the older tube AM transmitters I hear on the air, maybe even more. I wonder how many of the older 40's-60's rig designs followed your grounding precautions.  Probably not many.

I use all low level MOSFET solid state audio amplifiers for my TX audio drivers and RX amplifiers, so have no problems. And maybe the high level tube plate modulators, being of low gain do not need the bus isolation. Or should I consider this for the high level chassis in the 813s, 4X1 X 4X1s plate modulated rig, etc?


This seems something to think about when building up low level tube audio amplifiers for both ham TX and RX -  and music entertainment use.

T

[w8khk]

Yes, grounding requirements for low-level audio are much different than those needed for RF.  It is no problem using the chassis as a ground plane for RF power amplifiers, and the use of the chassis for the filament return also presents no problem.  This applies to indirectly-heated cathodes, where the cathode current is not part and parcel to the AC filament current.  Directly heated cathodes, those tubes with no separate cathode, are of course susceptible to hum if the cathode current is NOT in the path of the center-tap of the transformer feeding the directly heated filament.

The higher-level audio (power) stages, with separate cathode (indirectly heated cathode) such as 807s need no special care, as the signal level impressed on the grid of these stages is so much larger than any hum from an unbalanced, or grounded filament return.  For the 813, 4-1000, etc, it is of course necessary to provide a center-tapped cathode return path to minimize hum from the filament.  Other than this, I would not suggest any changes to the wiring in the higher-power stages, whether it be RF or audio.

In the low level stages, all the things I listed in earlier posts contribute to minimizing the hum and noise between the cathode and grid of the lowest level stages. My father taught me most of these techniques when I was around 6 or 7.  They have served me well.  By the way, HP also promotes most of these techniques, to reduce PARD, their acronym for Periodic and Random Disturbances, or noise.  These techniques are apparent in the assembly of many of their instruments.  Almost all of their tube instruments that deal with low-level audio have circuitry to elevate the filament to a potential positive with respect to the cathode, to eliminate hum from filament-cathode diode current.

When I finished building the amplifier last week, I found the need to adjust one resistor value in order to properly set the bias point of the first two cascaded stages of amplification.   No other changes were required of the completed amplifier.  When turning up the input gain control to maximum, with no input device connected, the output is completely silent.  I connected a Hewlett Packard 400-D AC VTVM to the 16 ohm output terminals, and set the meter range to one millivolt full scale, and the meter did not move; the output noise level was below the test instrument noise threshold.  The only thing that could have caused any hum would have been a short or leakage between filament and cathode in the 60 year old tubes, but even that did not happen.  I did use two-watt metal film resistors for the majority of the circuit.

In other threads on this forum, there have been debates and discussions as to whether resistors cause noise in audio amplifier circuits. I can say from experience that they definitely can and do create noise, but it can be minimized.  Years ago, when building a preamplifier for a stereo tape recorder, amplifying the output from the tape head dealt with signal levels even lower than studio microphone or turntable magnetic cartridge levels.  The hiss, pops, and shot noise caused by current through resistors in the circuit were more objectionable than the noise created in the first stage active device, whether it was a tube or transistor.  Discussion with some audio engineers at the RCA labs revealed that carbon composition resistors were the culprit.  I switched to metal film resistors, and the noise disappeared almost completely.  Using two watt metal film, instead of quarter or half watt resistors, maximized the improvement, because resistor temperature changes exacerbate the problem.  This does not apply to all resistors in the circuit, but only a few.  Most important is the plate or collector output resistor in the first stage, and second is the grid return at the input to the first tube, and the base biasing resistors at the input transistor.  Irregular current flow in the carbon comp resistor is noise that is amplified in the later stages, and is minimized with metal film resistors.  This is absolutely not snake oil or audio-phoolery, it is clearly measurable with meter, oscilloscope, and the human ear.  It is akin to selecting a lower-noise tube or transistor for the lowest level circuits that generate noise which is amplified in the later stages.  Think of this in a similar light to having a low IMD exciter followed by a linear amplifier.   Garbage in, MORE Garbage out.

I know for certain that many of the earlier AM transmitters ground one side of the filament circuit, using the chassis as the return wire.  Examples are the Collins 32V series, the Vikings and the Dixie 100s.  In the 32V2 and the Johnson Viking II, I have measured the hum, then re-wired the filament circuit with isolated twisted pair, and the subsequent hum measured was greatly reduced.  If I had it to do over again, I would simply disable the low-level mic stages, and feed audio into a later stage at line level, leaving the rig in its stock form.