• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Simplest OTL?

I was just thinking: it would be pretty cool to try an OTL amp, but a really small one, easy to build, maybe only of practical use as a tweeter amp. I searched a bit, but didn't come up with what I expected. Does anybody know of such a design: a simple OTL, low parts count? I was thinking one of the newer Russian tubes might be a natural for this application. Any leads/links appreciated.
Simple OTL? This is the best you can do. Note that it doesn't belong in this forum.

(Hmm, that's actually a 2N6254. Darn my typos.)



  • se_transistor_amp.gif
    2.8 KB · Views: 3,037


Paid Member
2003-06-12 7:04 pm
Maine USA
> an OTL amp, but a really small one

And in the Tubes forum, eh?

Tubes are limited current. Most speakers are low-impedance. In such a situation, power is proportional to square of current, i.e. square of number/size of tubes. Big amps have much better economics than small amps.

And straining tubes into a low-Z load is not going to be best for sound.

But taking you at your wish: the best current per buck is probably the russian monster, 6c33. You need two per channel. Working around 80V, you can easily pull peak current of 0.7A, about 0.5A RMS. Drive voltage will be around 20V peak per grid, so a simple driver would suffice. One small bottle, two large bottles, a large bipolar DC supply, a very large heater supply: as "easy to build" as any other push-pull tube power amp. Power in 8Ω is about 2 watts.

> maybe only of practical use as a tweeter amp

Tweeters don't take large average power, true, but their peak power demands may not be much less than woofers. In pro-audio we overpower woofers about 2X, but tweeters about 10X above their long-term thermal rating. Even little dinky active speakers with 2KHz crossovers will run 60W woof, 30W tweet. So to get away with two watts, you need either very low listening levels or very efficient tweeters. (I suspect 2 watts into a JBL 2440 would be ample in any home system, but the 2440 is a very heavy-duty device, and not particularly sweet.)

If you like Piezo tweeters (ack!), the math looks better. The 3.5" ones run around 100Ω at 4KHz. 6C33 is actually more tube than they need: 2-in-1 6080 might be nice.

If you like line-array or other mass-tweeter rigs, the math can be good. Put eight 8Ω tweeters in series, 64Ω, around 16 Watts. Combined with array beaming, this could put ample levels at the listener's ear. Find twenty 16Ω tweeters, 320Ω, you could make ample level with lesser tubes like 6BQ5.


Paid Member
2003-06-12 7:04 pm
Maine USA
Minimalist OTL speaker amp.

Fooling with 6c33 curves, I realized there was little audio advantage to running into Class B areas. With just one pair of 6c33, good linearity demands fairly high idle current, and the mighty 6c33 can stand the strain. Working Class A, there is an alternative to complex Futtermans: the White Cathode Follower.

The WCF is "flawed" in that it can not get out of Class A. But it is dead simple to drive. It is not true Push-Pull, because the drive for the bottom comes from the top, but linearity is excellent up to 80% of what you can get with a true push-pull stage stuck in Class A. It is in fact two voltage amplifiers bound-up in a unity gain feedback loop. Although the 6c33 is a poor voltage amp, if you get the two halves balanced (not an easy thing) then 2nd harmonic vanishes and 3rd harmonic tends to be low up to a dB or so below gross clipping.

This seems to be good for about 2 very clean watts into 8 ohms, 3 watts at a couple % THD. Power is higher in 16Ω.

It is not easy to get biased-up for best power in very-very low load impedance. I suspect my 6c33 model is poor, and that real 6c33s vary a lot. So it has to be hand-biased.

Start without the 6c33 tubes in socket. Trim the 12AX7 cathode resistor so that the plate voltage is 40%-42% of the supply voltage.

R3 can be a series string of say fourteen 10Ω 2-Watt resistors: we want to start around 120Ω to be sure the tubes don't burn. Put a voltmeter on the bottom 10Ω resistor. We want to see 2.9 to 3.2V here (290mA-320mA cathode current). With 140Ω it will probably be low. Let it cook a few hours to stabilize the tubes. Then start shorting-out the top 10Ω resistors for total resistance of 130Ω, 120Ω, 110Ω, etc, until you get to about 3V on the bottom resistor (300mA cathode current).

R1 should be 60% of R3, whatever R3 ends up being.

Working this way, you should get a couple watts at 1%, a mix of 2nd and 3rd, higher harmonics well under 0.1%. Damping factor will be about 2. Frequency response will be below 20Hz to above 100KHz. If you use it for tweeter-only work, you could make all the amplifier caps 10V smaller.

BTW: you may notice that the claimed power is higher than the Class A bias implies. Tube nonlinearity raises the peaks more than the troughs, so we get over 700mA peak with 300mA bias current.

One fault of the WCF is poor power supply rejection. You will need a VERY clean power supply. I have sketched a 3-stage R-C filter to get you started: it should not be a buzz-box, but it won't be dead-clean on high efficiency speakers. Replacing one or two of the 50Ω resistors with something like Hammond 159V (1.5H, 500mA, 27Ω ) should clean it up good. Yes, the last capacitor MUST be big, like 470uFd, because the top of R1 has to be held solid to develop the correct drive for the lower tube.

The "320V" raw supply is meant to be a standard 220V-240V winding plus rectifier and capacitor. Normal winding and utility variations don't matter.

If you bias a heater winding up to about 70V DC, you can heat both 6c33s from one winding without exceeding heater insulation breakdown limits. If you wire the 12AX7 for 6V, twist and route the heater leads well, it can be fed from the same winding.

Heat output per channel is about 72 Watts heaters, 110 Watts in the plate circuit. For stereo, you want a 200VA heater transformer, at least 400VA plate transformer.


Paid Member
2003-06-12 7:04 pm
Maine USA
Typo corrections:

> R3 can be ....we want to start around 120Ω to be sure the tubes don't burn.

Should be "start around 140Ω". (I originally considered a 20% initial under-bias, then changed to 40%, and didn't catch all the changes.)

> for tweeter-only work, you could make all the amplifier caps 10V smaller.

Should of course be "10X smaller" or "10 times smaller". Fumble-fingers.

I have assumed a 6c33 can run self-bias with a 100K grid resistor. If anybody knows different, speak up.

C4 can be 0.05uFd if desired. C4 has no effect on small-signal frequency response: when it gives up, the amp runs as a simple cathode follower with reduced maximum power and higher 2nd harmonic. 0.05uFd will pass music fine.

> R1 should be 60% of R3, whatever R3 ends up being.

I will add that "workable" values for R1 can be 50% to 75% of R3. At low levels, changing R1 nulls the 2nd harmonic. At high levels, changing R1 changes clip-point symmetry (though TU1 plate bias is more important).

There is no single value to optimize for both low-level and clipping balance. If you have test gear: load in 8Ω, attach a scope. Bring a tone up about 2dB past the start of clipping. Adjust TU1 cathode resistor for symmetric clipping. Adjust R3 for maximum peak voltage, top and bottom. These two adjustments interact, so repeat several times. When the clipping levels are maximum and near-equal, the clipping shape will probably be different top and bottom. Bring it to about 1dB past clipping and adjust R1 for best visible symmetry.

Now reduce level to 6dB below clipping and watch your distortion meter. Small changes in both R4 and R1 will reduce 2nd harmonic from small to nulled. In the simulator, 2nd harmonic can be nulled as low as you want. In real life, this won't be easy and sure will not stay nulled on a loudspeaker load. For best numbers, get the 2nd well below the 3rd. I suspect best sound will be with 2nd harmonic at 6dB below clipping to be similar to or a little higher than 3rd harmonic. Then at 12dB below clipping (typical music level) the 3rd will be very-small and masked by 2nd; as you approach clipping the 3rd will rise above 2nd and add an "edge" on the crescendos.

> parallel cathode followers into a ...choke

An effective plan. I would point out that it has some of the disadvantages of a transformer, particularly cost and weight. And even-order distortion will dominate (this may not be bad!). Power efficiency is similar to the WCF: the CF-choke needs twice the standing current at half the supply voltage. If you buy them by the case, 6c19 may be cheaper per Amp than 6c33. But the 6c19 is a fairly small tube, not much power in something like 8Ω, so you need a heap of them and some low-paid labor to stitch all the connections. For a minimal DIY OTL, I think the mighty 6c33 gets results with less effort and not much difference in cost.
I´m sure that 6C33 is better suited for OTL amps AS LONG AS THEY WORK PROPERLY!!
A friend of mine has a 6C33 SE amp (with OPT) and he told me that they take hours to stabilize, if they stabilize at all.

With a cap on teh output we at least have some protection for the speaker, but I would never use them in a OCL/OTL amp


Paid Member
2002-01-15 12:57 am
Tyrone Ga. U.S.A.
PRR, that was a cute little circuite thanks!

On the same topic ARG Labs which I think is now out of
business had a 5watter using only 1 6c33c-b and a 6h6p.
I have asked several places but no one seems to have the
shematic. I am assuming they use an inductor for the cathode
load. Would this raise the effency a great deal?



Paid Member
2002-01-15 12:57 am
Tyrone Ga. U.S.A.
She lives!

I just built my first otl. I just need a few watts and my speakers
were built assuming a source impedience of 6 ohms so this let
me try something real simple. I am using a choke loaded 6dj8
feeding a 6c33c-b using a choke load with about 6 db of feedback.

I estimate about 2 watts using 350ma bias. I have just done a
verry verry Crude mono amp but I think I could like this sound!
Crude? Well I used batteries for the tube filiments and the -grid

If I wanted to go with a little higher power but stay with single
ended I wonder if it would sound better to use a pair of
6c33c-b in parallel or say 12-16 6c19pi in parallel?

Thanks for all the help in the past.