An experimental battery powered, sub watt, tube guitar amplifier


1D8GT PP Amplifier with 1N5 preamp breadboard

I had never used dry battery, direct heat, tubes before so this was an opportunity to play with them. The 1D8GT was picked because it contains a 'power' pentode and triode. It also contains a detector diode but I'm not using it.

The tube caps are grid, I.E. grid caps, and not plate so they're not exposing high voltage and 1/4" fuse clips make convenient, inexpensive, grid cap clips. I wish I had some but 5x20mm fuse clips worked with a little coaxing.

Remaining parts are, currently, whatever was lying around, and not necessarily optimum, but it does work.

For those who've never used dry battery, direct heat, tubes one of the interesting characteristics is there's virtually no (very little) 'warm up' time, unlike indirect heat types, so the amp comes on with the flip of a switch 'like a light'.

The "Batman" name was inspired by the abbreviation for "battery."


Parallel filament operation precludes cathode bias as commonly used with indirect heat types so the bias 'trick' common to battery radios of the era is employed (see the battery eliminator, below). That also dictated the hybrid paraphase splitter and grid feedback.

The splitter is a cross between the simple paraphase, where a voltage divided signal from the first triode plate is feed to the second grid, and the floating, unity gain feedback, paraphrase, where R1 would equal R2 and differing outputs cause an error signal driving V2AA. The triode, however, does not have enough gain to make the floating paraphase error small enough, hence the offset resistor values. As such, visual similarity to a simple paraphase might lead one to believe the signal from R1 and R7, being reminiscent of the simple paraphase resistive divider, is 'too large' but that is reduced to balance by feedback through R2. Note that the paraphase resistors also conveniently provide bias to the output pentode grids.

One of the paraphase's failings is high frequency mismatch due to an extra triode in the second phase path but R2 feedback improves response and balance is good to over 20kHz. That's not particularly impressive for 'hi-fi' but it's more than adequate for a guitar amplifier. Alternately, a small capacitor of approximately 2pF could be placed across R1 to compensate for HF loss through V2AA.

On a related note, it appears that the C1 and C2 low frequency poles are 'mismatched' (and technically they are) but since the second triode is driven post C1 that pole dominates as long as the rest are sufficiently lower. Although, as currently implemented, the C3-R11 pole is much higher so the paraphrase doesn't see frequencies near the C1 pole.

Note: The schematic, drawn using existing types, is a bit misleading in showing separate cathodes when the 1D8GT triode and pentode are both on the same filament (and not an indirect heat cathode at all). So, for example, while one might think you could put a tail under the phase splitter it would be under the pentodes as well and wouldn't work. What's shown as the 'cathode' is the filament negative.

The output is a SPECO 5W line transformer giving 20k plate to plate and at these current levels bias mismatch doesn't seem to be a problem. And a good thing too since direct heat filaments preclude the current mirror balance circuit I often use for PP stages.

There is no tone stack, tone shaping, (and, originally, guitar amplifiers from 'way back when' didn't have any either) or even a volume control so, for the time being, the guitar's controls are the only ones.

Power output is around 300mW, which is perfectly fine because we aren't going for 'raw power' here. It's intended to be a 'quiet' practice amp where one can crank 'full power' without busting ear drums but output voltage is about right for feeding a power amp so one could use this for getting 'tube tone' and drive a P.A. or home stereo for volume.

Music plays well but I have yet to slap on a guitar to see what kind of tone it produces.


With dry battery filaments very good regulation is required to eliminate hum so both supplies are transistor regulated.

The double zeners (82V and 11V) were parts on hand, as was the 40VCT transformer. A simple voltage doubler brings it up for the 90V supply. Those familiar with the part numbers will notice high voltage transistor overkill but, again, parts on hand,

Bias comes from a resistor (network) between the B (plate) and A (filament) battery negative side. I.E. Amplifier B current (from all tubes) goes through the resistor (network) before returning to B-, which makes B- negative with respect to A- (circuit ground) and splitting the resistor into multiples provides varying voltage tap points.

Effective B+ is reduced by the amount of bias generated, which is similar to the loss associated with a traditional cathode bias arrangement, and C6 bypasses audio.

The circuit works, as is, with either AC or traditional batteries but 90V types are not made any more so one substitute idea would be to make a SMPS supply driven by, perhaps, a 12V battery.


(c) 2010 All Rights Reserved