High Power Vacuum Tube Amp Using Push Pull 3CX300A1 Ceramic Triodes

Preliminary Release - 6/6/97

since Last Update - 9/30/98

Here is the schematic for the amplifier section of the "Beast":
NOTE: Use "Landscape" mode to print.


This project was inspired by a visit with Eric Barbour, Applications Engineer for Svetlana , in his Portola Valley, CA office. I was in Silicon Valley on other business and decided to pop in to meet face to face. Eric had, in the past year or so, been very supportive of my efforts with my Single Ended SV811-10 Amplifier.

During our conversation I mentioned that I tend to keep out of the mainstream when it comes to audio design. This is what originally lead me to try the SV811 and SV572 types over the currently popular tube choices.
With that he handed me a pair of 3CX300A1 and said "Whadaya think you can do with these?!"

I must admit, at first I was not too thrilled by these little hockey pucks".
I thought to myself "What's the point of using tubes if ya can't see 'em glow?!" However, I took the pair and the data sheet and decided that I would accept the challenge...

Back at my hotel that night I studied the 3CX's data sheet. I worked out a few possible Single Ended operating points. Some looked more promising than others. And if I had the output iron on hand I would definitely try some of them. However, it became clear to me that where this tube really shines is in a high power push-pull application.
With a 300W plate dissipation rating, and a plate voltage rating well over 1000V,you could easily get over 300W out of a single pair in Class AB.
Tim Allen would be proud...MORE POWER! Uh, uh, uhg!

A First Step....

This incarnation of the "Beast" provides only a "modest" 100W or so. Although 100W is nothing to sneeze at, this value is well below its ultimate capability. It is limited only by the 660V power supply that I currently have available.
For example: 200W is certainly achievable from a B+ of 850V (tranny load = 1.9K). ...280W will require around 1000V.
Q: Why consider 280W and not 300W or more?
A: The biggest, commonly available output transformer I am aware of is the Hammond 280W, 1.9K job. And that sucker weighs in at 28 lbs. Good lord! Pass the spinach, Popeye...

A proper power supply is in the works. I hope to boost the output power to at least the 200W level before I'm done. Then again, 280W is only another 150V more of B+..........

Design Details

Input/Driver Stage

The input stage is comprised of a differential amplifier made from the triode sections of two Svetlana 6BM8. A 3mA constant current diode is used in the common cathode circuit. This diode is returned to a negative supply to provide some bias headroom. (The actual grid bias will be between 0.7V and 1.0V, but the pinch-off voltage for this constant current diode is around 2.5V. So a negative supply is required.)

The input stage is direct coupled to the driver stage.
The driver stage is yet another differential amp made out of the Power Pentode sections of the 6BM8. The each pentode section is triode connected for better linearity, but also for simplicity. (I'd rather not have to mess with another screen supply.)

Another novelty of this driver (beside the DC coupling) is the "Choke Loading" of the plates. In general, choke loading offers several advantages over resistive plate loading. A choke load presents a practically flat ac load line (high impedance) while only dropping a small amount of DC. Among other things, this results in a very linear response and a large signal swing from a relatively low Ebb supply. This allowed me to keep both the driver and input sections on the same supply. A resistive load would have pushed the Ebb for the driver to 500V or more, just to get the 80V peak swing that I was aiming for.
The fact is: the choke loaded driver presented here is capable of almost 200V peak!

The cathode resistor is sized to pull about 10mADC thru the cathode follower. This value was chosen mainly to overcome the >100pF input capacitance of the 3CX300's. Grid current is meant to be avoided in this design.

Output stage

As mentioned before, the Push-Pull pair of 3CX300A1 's are currently seeing a plate voltage of about 660V. Adjustable Fixed bias (how's that for an oxymoron?!) can set the bias current from 50mA to over 200mA. Given the 1.9K plate-plate load, 100mA per plate works out well for a Class AB operating point. Going to the published plate curves, this arrangement promises about 120W on paper. A little over 100W was measured into a real speaker load.

Another option that presented itself was to load the tubes with 3.8K instead of 1.9K. This was done by connecting the 8 ohm nominal load to the 4 ohm tap of the output transformer. This arrangement drops the theroetical power to just a hair over 100W. One significant benefit of this approach is that the output stage is now operating in Class A. In fact, the bias can be taken down to about 80mA per tube. (This seemed hardly necessary as far as the tubes were concerned! It was more of an advantage for the sub-optimal power supply which would now only have to provide a relatively constant 160mA rather than 200mA to 400mA + at max. signal.)

Power Supply

The Power Supply was built in a separate enclosure so it could be swapped out for a more suitable supply down the road. The Power Supply Chassis houses the Main B+ Supply (from which the driver/input stage's Ebb supply is derived). It also contains a -115VDC bias supply, a +16VDC supply for the (24V) cooling fans, and a -16VDC supply for the input stage's current sink. There are also two 6.3VAC filament supplies. One of these is for the two output tubes (@ 2 X 2.65A), and the other is for the two 6BM8 input tubes (@ 2 x 0.78A). The two supplies were separated because the input filaments needed to be biased up to about 80 - 100V to satisfy the +/-100V heater-filament rating. (Remember, the direct coupling places the pentode-section cathodes at 100 to 150V.)

The Main Power supply is designed around a surplus 980VCT/250mA tranny. Using series silicon diodes and stacked 220uf/450V cap input I end up with about +690VDC under load at the first cap. The first LC filter (6H/80 ohm, 220uF/2) provides a relatively clean B+ (approx. 660VDC) for the output stage. The second RC filter (10K, 220uF/2) cleans the supply further for the more sensitive input gain stages. Ebb is approximately +410V.

The rest of the supplies are nothing fancy.

Construction Details

NEW Pictures of the BEAST:

Photo 1 Shows the amp section during power and distortion measurements. Scope trace is more visible in Photo 2.
Pretty compact for a 120W tube amp.
Volume control and input jack on front. Notice feedback cable connected to 1/8 in. jack on top surface to left of the input tubes.

Photo 2 shows the whole mess ... The power supply chassis and output transformer are simply laid on the workbench next to the amp and scope. Notice the X/Y trace on the scope. Not too shabby, eh?! This is at 100W output !!!
Ultramini 6j6/6aq5 guitar amp and Eimac 250TH in background.

Photo 3 Shows the guts of the amp section.
Not a whole heck of a lot in there!
Notice small muffin fans under each output tube.

Photo 4 Better quality Pic

Photo 4 'nother Better quality Pic

The "Beast" is only a MONO amp, for now. It is currently in three pieces:
The AMP section, the Power Supply, and the Output Transformer.

The AMP section

The amplifier is quite compact for its power handling capability. I was able to fit this mono unit into a 10" (L) X 8" (W) X 2.5" (H) aluminum chassis.
For now, the bottom and one 8" X 2.5" end are open to provide good air flow to cool the 3CX300A1 output tubes. A small 2.5" square muffin fan is mounted on stand-offs beneath each output tube. A glass chimney is used on each tube as well. I basically followed the mounting instructions given HERE.

Power Supply

The power supply is built in/on a 17" (L) X 8" (W) X 3" aluminum chassis. The Power Tranny, Filter Choke, and the bank of six computer grade caps are mounted on top. The smaller stuff is kept inside. A 12 position barrier strip is mounted inside to provide connections to the outside world. A bundled harness comes into the chassis thru an access hole.

Output Transformer

This thing is so damn big that I decided not to bother mounting it to anything. I left just enough room on the power supply chassis, except I'm afraid it will bend in half if i picked it up by the middle!
When I have to move the whole thing out of the way I simply place it on a section of 2 X 12 (decking lumber).


Calculated Performance

The following performance was calculated using the plate curves provided by Svetlana:

For Rload= 1.9k plate-plate, Ebo=660V, & Ibo=100mA:

Pout 123W
Plate Dissipation 90W/tube max. signal (Class AB)
Power Consumption 300W

For Rload= 3.8k plate-plate, Ebo=660V, & Ibo=82mA:

Pout 100W
Plate Dissipation 54W/tube quiescent (Class A)
Power Consumption200W

Measured Performance:

For Rload= 1.9k plate-plate, Ebo=660V, & Ibo=100mA:

Pout 110W (90Vpp into 9.2ohms resistive at clipping)
Overall Gain approx 40dB (100X)
Driver Gain approx 46dB (200X)
Input sensitivity approx 0.1Vrms (0.3V peak)

For Rload= 3.8k plate-plate, Ebo=660V, & Ibo=82mA:

Pout 87W (80Vpp into 9.2ohms resistive at clipping)
Overall Gain approx 37dB (70X)
Driver Gain approx 46dB (200X)
Input sensitivity approx 0.1Vrms (0.3V peak)

More Test data will be provided as it is generated. Frequency Response and possibly Distortion measurements are planned. So far, a quick measure of the circuit's linearity was obtained via an "X-Y" plot on my scope. Looks pretty flat (even better with the 3.8K load) with a nice soft turnover into clipping. This was done with real music input and not just with a sine wave.

Listening Impressions

I have listened to this amp extensively over the last two weeks. This sucker ROCKS!!!
I couldn't manage to get past 50% on the volume control before the walls started shaking. Priceless tubes were ready to fall off shelves, etc...
Placing the speaker in the other room (where only more mundane objects could be damaged -ponderance of the day: "... if a crystal platter falls in the china cabinet, and no one is there to hear it... will it still take out the wine glasses???" )

The sound held together quite well as the volume was increased. Distortion was not even noticeable until well into clipping. By this time I was holding my ears and thinking about what I was going to say to the cops...

However, the thing that really suprises me is that this amp truely sounds good at lower volume levels. I would have expected some of the clarity and dynamics to fade with decreasing volume. It seems that below about 6W is where some of the "life" starts to go out of it. But with 100W+ on tap... that's a pretty wide range. Makes this a quite versatile amp. Imagine what it could do in STEREO!!!


Email your comments/suggestions/critiques/questions to Bob. D. at robert.m.danielak@lmco.com

Back to Bob. D.'s Music/Audio Page
Back to Bob. D.'s Home Page

This page, and all contents, are Copyright (C) 1997 by Bob. D.