My new amp design- video demo

[Chuck H]

Absolutely capitol. It sounds like a good single ended practice amp but there's some magic when the volume goes up and the shimmer and watery mids are actually interacting with the guitar. I've noticed on some of your other clips that you seem to prefere a little more low end grunge on heavy clipping than average, so tone is subjective, but it sure seems to be doing what any good "tube" amp should. Nice retro looking packaging too. What's especially impressive is that your not just using the output stage for reference, your actually clipping it. And it sounds good. You can still hear some play between the speaker and the power amp.

[ThermionicScott]

Interesting approach. It's sort of like taking the small-amp-into-the-PA setup and putting it all into one package, isn't it?

- Scott

[TDS]

Hybrid PA output stage

Your amp sounds amazing!

I'm very interesting in your output stage, particularly your arrangement for feeding the output transformer to the transistors. I'm currently in the stages of building a 5 watt 2 channel practice combo (SLO inspired preamp, 6V6 output) for my son away at Uni. It will include an 100mW attenuator for bedroom use, but as the speaker (Celestion Tube 10) is rated at 30W, it might be interesting to consider a switchable S/S power boost to make it gigworthy as well.

[Steve Conner]

Thanks! That encouraged me to get around to drawing up the schematic.
http://scopeboy.com/scopeblog/wp-con...etteHybrid.pdf

[tubeswell]

Thanks for posting that Steve. Interesting DIY custom tranny and associated paraphernalia for the attenuator supply. (Did you work out the number of turns by using inductive or deductive reasoning? LoL - sorry couldn't resist ;-)

[TDS]

Thanks! That encouraged me to get around to drawing up the schematic.
http://scopeboy.com/scopeblog/wp-con...etteHybrid.pdf

Thanks for sharing that Steve. Your design is quite a lot different from the Vox Super Beatle style output stage. You have obviously improved the original design and seem to have put in features to give it more stability and protection than existed back when Vox designed their power amp.

I may have a go at trying your design, but I my limited knowledge of S/S amps is confined to a 50 watt Linn topology amp that I built a few years ago and improved with some extras as my knowledge progressed. Your power amp is interestingly different, especially not needing any neg feedback. I am sure there are others in addition to myself who would like to understand some more about it.

Any chance of a short walk-through of your design and how you got it to where you are now.

[kg]

i think one of the keys to steve's design here, and the successful hybridization, is the fact that there is absolutely no global NFB.

steve, one thing that came to mind while looking at the circuit is that you could easily couple the bottom of c16 to the el84's plate and net triode mode operation, while retaining the variable DC level on vr4.

hell, add a couple more caps and another pot and you've got yet another continuously variable tone knob (triode to pentode).

[imaradiostar]

Great approach- it's like "power steering" for a low wattage amp- it pushes harder but the speaker still interacts with the power amp.

Am I seeing this right? It looks like the output transistors are biased to conduct more or less equally and the tube output TF wiggles each base through a resistive drop, turning the top one and the bottom off and vice versa. Is the transistor output more or less class A then, with each output conducting (at least a little bit) 100% of the time?

jamie

[pdf64]

I don't think it will be in class A, as the output transistors are only biased to 55mA (.026/0.47).
But this has exposed my lack of understanding of transistor power amp design, in that the output devices seem to be operating in common collector mode, therefore voltage gain of 1 - so why is the transistor mode so much louder (as the voltage across the speaker will be the same)? Pete.

[Steve Conner]

Well, I had the same problem at first. My original idea gave the output stage a gain of 1, and needed a custom OT to provide the voltage gain. It would have needed an 80 ohm tap to drive the output stage.

But I realised I could just connect the end of the OT secondary that would normally be grounded, to the speaker output. That's what the wattage switch does in the 30W position. The output stage bootstraps itself and essentially turns into a common-emitter amp, which again is how the old Super Beatles etc. worked. Jamie's "Wiggling of the bases" explanation is about right.

It's not as theoretically pure in terms of tube/speaker interaction, but I ignored that in the interest of being able to use a $20 off-the-shelf OT.

The diodes and 4.7 ohm resistors provide some local NFB to linearise it: they make it into a push-pull version of the diode/transistor current mirror seen in many SS amp circuits. Therefore the current gain is stable and doesn't vary too badly with temperature, transistor beta, etc. This also limits the short-circuit current.

The effect is not unlike a pentode output stage. It has a well-defined current gain, but the voltage gain is simply "whatever the load dictates". For a 10x power boost, the current gain should be 3.2, but I think mine is nearer 10. That makes it a bit more loud and unruly than it should be, and is probably a mistake.

The two 360 ohm resistors to ground lower the output impedance and damp it a little. I think they lower the output impedance to around 18 ohms. Without that it made so much high end, it was like an icepick to the eye socket. They also set the DC conditions well enough that it can run with no output coupling cap, and an acceptable DC offset.

It is Class-AB. Actually AB2 in tube speak. The bias supply has to deliver quite a lot of "grid" current.

Yes, I figured out the transformer turn count by induction, in both senses of the word. I add a 10 turn test winding, and measure the voltage on it, to work out how many turns I need.

[TDS]

Thanks for making the details of your amp available for us all to ponder over. Could I ask a few questions regarding your circuit based on your experience with it.

I would like to try it out using bits and pieces to hand, Is it forgiving on the choice of components?

For instance:

Power Transistors - Would it be okay to use 2 pairs of lower power more common types such as 2N3055/MJ2955's or a pair MJ11015/MJ11016's.

Input Caps - What is affected by the choice of their value, would it be okay to use 470uF for instance?

DC offset - How much DC offset could I expect and how much is reasonable to live with?

Would varying the centre grounding point of the two 360 Ohm resistors help, or matching/trimming the resistors be a beneficial?

Would a small change to 330 Ohm resistors change things dramatically?

Bias Regulator - Could a 7805 be used and the and the values of the R3, R6 and VR3 totem be adjusted to compensate for the higher voltage?

[greekie]

Great work Steve. Very inspiring! You are indeed an asset to the community and this forum.

Best regards Jake

[Steve Conner]

Thanks for the kind words, Jake!

TDS: The short answer is yes. The circuit is pretty forgiving of component choice. Most of the choices were just determined by what was lying around the lab at work, anyway. The only critical things are that the top and bottom halves have to be the same, and the bits that I've called out as being in thermal contact.

Transistors: My previous prototype used a MJ15024 and MJ15025, and the results were much the same. (Just louder.) I don't see any reason why the more common ones wouldn't work. I only used those MJL3281/MJL1302 because I had them.

Input caps: They roll off the bass a little, helping to save the speaker from flapping to death. 470uF should be fine, it's within the tolerance range of a 330uF electrolytic anyway.

DC offset: My prototype measured less than 0.1V at idle. While working on the DC offset under drive problem, I added a 150 ohm resistor between the PT centre tap and the rest of the circuit. This reduced it by another factor of 20, but the effect on the cone flapping has yet to be tested.

Yes, dangling a 100 ohm pot in the middle of the 360 ohm resistors would allow you to adjust the DC offset, but it should be low enough without adjustment.

330 or 470 would probably do fine. They don't just set the DC conditions, they also damp the speaker. The bigger these resistors, the higher the output impedance gets. My first prototype had 1k, and a separate 100 ohm damping resistor. For this one I started out with 180, but they smoked a lot at full volume. So I added another 180 in series with each, and the smoke stopped.

Regulator: I only used the LM1117 because I had some of them, and its low voltage meant less time spent adding windings to a toroid by hand. A 7805 would work fine too. You don't need to adjust the resistors in the bias circuit, because it's a regulator in its own right, with a 1.4V output. The upstream regulator is just to remove every last trace of ripple.