hybrid amps

[uneumann]

I'm looking into hybrid ss/tube amps and have found some interesting ideas. The hybrid output stages of music man amps is very interesting and seems to have been successful. Are there any opinions or experience with those amps?
The MM hybrid output stage seems to be widely used in their line of amps - for example:


There must be other examples of hybrid stages out there.
Sequential stand-alone SS and tube stages are common, I know.
I'm specifically looking for hybrid-stages that use both SS and tube elements together in one overall amplifier stage as the MM design does.

I'm also somewhat aware of the Vox VT series, but schematics are hard to come by so my knowledge is limited. Is the Vox VT approach a sequence of SS/tube stages or is it a hybrid stage with both SS and tube elements working together?

[g1]

Peavey Classic VTX uses a similar cathode drive (via transistors) power amp circuit.
Silvertone Jockey had a thread recently about it, he was very impressed with the sound:
http://music-electronics-forum.com/t37831/#post367475

[loudthud]

Peavey had a number of amps in the 70's with a solid state preamp and phase inverter driving 6L6GCs. The Mace was top of the line.

[Wombaticus]

Peavey had a number of amps in the 70's with a solid state preamp and phase inverter driving 6L6GCs. The Mace was top of the line.

I just had a Mace in my shop recently... not the original discrete transistor model, but the later Mace VT with opamps. That amp sounded just OK, but wow, with 6x6L6GCs it sure is LOUD. The guy brought it in because it was too quiet... five out of the six screen grid resistors were shot... it shipped originally with only 100R resistors in that position. The original Mace's were evidently a preferred amp for some Southern Rock type bands to take on the road back in the day, even if they recorded with Marshalls.

[teemuk]

Is the Vox VT approach a sequence of SS/tube stages or is it a hybrid stage with both SS and tube elements working together?

What's the difference? There's a solid-state phase inverter driving two 12AX7 tubes configured in push-pull. Their plate load is a reflected impedance of the speaker load, only scaled proportionally higher in impedance. After the transformer, or "virtual transformer", handling the impedance reflection the signal is amplified by a solid-state power amp of some form. That amp is designed to operate rather "linearly" besides current feedback reflected to the power amp stage.

Additionally, in the Valvetronix series of amps there are controls to bias the tube push-pull stage to class A or class AB, to disconnect one of the feedback loops, and some other stuff. These functions are controlled by the microprocessor so the signal processing of the Valve Reactor stage is rearranged to mimic performance of the modeled amps.

It's a pretty clever circuit. New ones have a genius comparator/switching scheme that allows to run a single pentode stage as two individual sections of a push-pull stage.

[uneumann]

What's the difference?

Thanks everyone for the pointers and comments.
I think you are asking about how to define a "hybrid stage with both SS and tube elements working together". I don't have a precise definition of a "hybrid stage". It's certainly a fuzzy line. I know it's not an opamp feeding a complete tube PA, or a CC tube stage driving a complete SS power amp. I'd say a hybrid-stage has SS/tube circuit elements that behave together to accomplish a function often considered a "stage". So the MM circuit qualifies since the elements work together to form a power amp. I know it's somewhat arbitrary, but these circuits operate the mixed technologies in a more interdependent fashion.

[g1]

Another question to ponder is whether any difference in the (later) MM power amps are due to the mix of SS and tube, or is it because they are cathode drive rather than grid drive? I think the latter would have the greater impact.

[teemuk]

Indeed. The biasing arrangement is quite different too and as is rather free of bias shifting and crossover/blocking distortion from sustained overdrive. IMO, the output tubes in this design contribute very little aside higher output impedance.

Interestingly, Peavey's VTX series have another variation of this circuit. But Peavey uses their patented driver circuit, which switches between two alternate bias sources and effectively "softens" the crossovering region, similarly to all tube amps. Additionally, Peavey's driver also makes at least a modest attempt to limit driver transistor current. Overall, IMO, theirs was a better, and from its characteristics more "tubey", design than the Music Man version.

[J M Fahey]

IMO, the output tubes in this design contribute very little aside higher output impedance.

True.

In fact I have been pondering for a long time, and having not built it just out of lazyness, that an MM style power amp, cathode driving suitable high voltage MosFets, with output transformers and all the trimmings, should sound exactly the same as the tubed ones.

Maybe in February which is a "dead" month for us, sales wise, since it's the middle of Summer holidays, I'm bored enough that I build and test one .

But I expect no surprises.

Interestingly, Peavey's VTX series have another variation of this circuit. But Peavey uses their patented driver circuit, which switches between two alternate bias sources and effectively "softens" the crossovering region, similarly to all tube amps. Additionally, Peavey's driver also makes at least a modest attempt to limit driver transistor current. Overall, IMO, theirs was a better, and from its characteristics more "tubey", design than the Music Man version.

Yes, they clearly have studied what overdriven tubes do and mimic that.

Way beyond than simply sticking a tube amidst SS circuitry and hope it imparts some flavour.

Which often is made in a terrible, self defeating way.

Old ADA Microtube power amps used a linear 12AX7 as input stage in an otherwise full SS power amp, within the feedback loop.

I remember scoping the tube plate waveform while increasing drive into clipping, and just at the limit point it started shaping like a condom tip or a tit nipple (sorry for the examples but these are the most accurate) which at first I did not understand ... until I realized that as the output transistors or Mosfets started limiting the tip of the waveform, clipping it down compared to what it "should" be ... NFB dutifully created a waveform "boosted" in that area trying to compensate.

"Tubyness" contribution by that tube? : none, zero, zilch, take your pick.

Old errors , when forgotten, tend to be repeated over and over, this is a "hybrid" amp which today is all the rage at DIY Audio Forum, makes the same mistake or worse, adds all linearizing tricks to an already linear used dual triode differential amplifier.

Tubyness? ... zero:



not to mention other terrible design choices made there, such as MosFets driving Bipolar outputs (if anything, the opposite is better), non thermally compensated outputs ... which force the next terrible decision: output transistors run fully class B (notice the 0.3 micro Ampere output idle current ) which probably is just leakage current, not forgetting the Zobel network on the far side of the output inductor, etc.

Oh well ... but it glows in the dark

I bet it has a small chassis window showing the tube.

At least the ADA did

[teemuk]

The ADA design...

Conceptually Ampeg, Crate, Randall, Mesa Boogie, etc. hybrids seem pretty similar: tube voltage amp followed by a small signal buffer (single ended or push-pull) drives a solid-state (often MOSFET) current amplifier, "buffer". I think that the scheme was principally mimicked from the "MosCode" hifi amps. ADA's biggest difference is probably wrapping up the tube stage to global feedback loop, whereas most designs run the output buffer in open loop, which slightly softens the clipping characteristics and increases output impedance.

Yeah, the design can "fail" in some ways. ADA designed the circuit so that the voltage amp stage actually has a headroom extending way above the capability of the current amplifier (or even the emitter follower output stage buffering the voltage amp), all potential to get tube clipping is pretty much destroyed right there. The clipping characteristics of the output buffer or, worse yet, the emitter follower, are not really impressive and yes they are pretty solid-state-ish in the clichey meaning of that. Then the effect is worsened by global feedback.

I think they may have aimed to this intentionally. The amps are more akin to generic PA amps than instruments as themselves so I can see the point for design goal of linearity and high output power. It's not too different from putting tubes to hifi amps to contribute miniscule amounts of distortion but greater amounts of "wow factor". Ampeg/Crate did introduce a very similar design to ADA (sans global negative feedback) put many models featured a user adjustable B+ voltage regulator so that the headroom of the tube voltage amp could be reduced. For example, Ampeg SVT3-PRO (If I remember right) basically blends from ca. 300W "hard" solid-state clipping amp to ca. 60W "softer" tube clipping amp with the dial. Do note the cost of this as severe reduction of "linear" output power rating. Not always so good.

Then there's another, more inherent drawback: A single-ended capacitively coupled triode voltage amp, the "driver" of these designs, from its characteristics, is far from a push-pull transformer coupled pentode amp. So if the aim of these "hybrid" power amps is to mimic characteristics of traditional tube power amps then this circuit topology pretty much fails from the start. Yes, with proper design you can extract true and genuine "tube clipping" from it, but this does not mean that it behaves or sounds even remotedly similarly to, say, classic tube power amp designs like JTM45/Bassman, Vox AC30, Champ....

[uneumann]

Hybrid PA

Appreciate this discussion. I'm in the midst of a project with a hybrid PA. I'm taking the following approach...



This is still in process so the schematic values are approximate. I've built a version using triodes in the PI and it works pretty well. The hope is that using pentodes will bring their compression/distortion qualities to the tone.

Comments or suggestions are welcome.

Note: V+ = 24v @4a regulated B+ = 210v. Output target is 40w RMS into 4 ohms with switch reduction to 20w.

[J M Fahey]

The hope is that using pentodes will bring their compression/distortion qualities to the tone.

Cool, but you'll need to drive them hard to get that; in the linear area they are not that different to transistors ... specially if using pentodes.

[uneumann]

Cool, but you'll need to drive them hard to get that; in the linear area they are not that different to transistors ... specially if using pentodes.

Yes - I'm planning to drive the PI into clipping. The full clipped PI output swing is scaled to fall just within the range of the transistor followers. I aim to avoid any clipping in the SS section. The PI produces more than enough voltage swing, so everything after the PI is just a follower (or voltage attenuator) of some sort, it's all about current-gain and preserving the PI waveforms.

It's turning out too simple (IMO). No OT, no power tubes, no hi-watt B+ supply... so many advantages. What am I missing?
The prototype works and sounds great (also IMO), but I'm wondering what I'm overlooking and whether this concept has been tried before or rejected for some reason.

[loudthud]

In reference to the best classic all tube guitar amps

Two key elements of compression (aka feel, touch sensitivity and dynamics) are:
1) Distortion and soft clipping
2) Power output reduction and gain reduction from power supply sag

The enemies of compression are:
1) Gain stabilizing feedback (local and global)
2) Regulated or stiff power supplies

Hybrid amps usually suffer from some poor combination of these elements. Sometimes additional circuits are added to mimic the characteristics of tube amps. Diode clipping is one example.

In reference to the schematic in post #11, there is very little drive on the negative side to turn on the Q2 and Q6. The best you can do in this situation (impedance wise) is the 4.7K resistor divided by the combined beta of the 3055 and the NTE129. Optimistically the output impedance is several ohms and the current capability is limited. Even with a bootstrap, the grids of V7 can't go negative so there is little hope of pulling the load within 5V of the rail (ground). Short answer is you need to rethink the level shifting and current demands of the output stage.
　
To get compression out of the EF86s, you need to let the screen voltage sag a little.

[uneumann]

In reference to the best classic all tube guitar amps

Two key elements of compression (aka feel, touch sensitivity and dynamics) are:
1) Distortion and soft clipping
2) Power output reduction and gain reduction from power supply sag

The enemies of compression are:
1) Gain stabilizing feedback (local and global)
2) Regulated or stiff power supplies

In reference to the schematic in post #11, there is very little drive on the negative side to turn on the Q2 and Q6. The best you can do in this situation (impedance wise) is the 4.7K resistor divided by the combined beta of the 3055 and the NTE129. Short answer is you need to rethink the level shifting and current demands of the output stage.
　
To get compression out of the EF86s, you need to let the screen voltage sag a little.

Thanks LT - good points. I avoided NFB, but I'm relying on the EF86's for distortion/compression since I don't expect PS sag (not much at least). My understanding is that compression effects increase with higher values of anode-load and screen resistances. Is that a reasonable expectation?

The drive-to-ground problem is a good catch. I"ll mull on that one. I'll look for a higher gain driver and consider other options.
Also, as for PI swing not getting low enough. The tube curves show that Vak gets down to ~10v at saturation. Is that reasonable to expect in an actual circuit? I could also reduce the zener for grid bias to 6 or 9v.

thanks