You have a point, I did forget about David Berning. He has taken OTL circuits (which is a niche market within a niche market) to another level by eliminating the need to have a gazillion output tubes in order to get any appreciable power output and lower the output impedance.
Are tube designs still evolving, or have all circuits effectively been invented?
- Thread starter ack
- Start date
You have a point, I did forget about David Berning. He has taken OTL circuits (which is a niche market within a niche market) to another level by eliminating the need to have a gazillion output tubes in order to get any appreciable power output and lower the output impedance.
I would like to see a schematic of that amplifier, because, the designer must have found some as yet undiscovered in 75 years method of lowering the plate resistance of a triode from several thousand ohms to less than one ohm. I suspect the claim is inaccurate.
I'm designing an OTL with 6P45 tubes (6KG6) which have some of the lowest plate resistance in the industry and I still must use a half dozen to get a good match into 8 ohm loads.
I'm designing an OTL with 6P45 tubes (6KG6) which have some of the lowest plate resistance in the industry and I still must use a half dozen to get a good match into 8 ohm loads.
It's based on a design by Tim Mellow, which appeared in Audio Express a couple of years ago. Bear in mind that the plate resistance of a 6C33C is much lower than you are quoting above; about 100 ohms or so. With negative feedback, a really low output impedance is pretty easy to achieve. So it's just using standard, well tried and tested, methods for lowering the output impedance. I've measured it myself, and it is indeed of order a fraction of an ohm.
Here's another one, by Hans Beijner, and which I've also built. http://www.tubetvr.com/otl.html
A modified Futterman with two 6C33C output tubes has about 11 ohms output impedance prior to introducing feedback to the input stage. With feedback, it can easily be reduced to a fraction of an ohm. Hans Beijner gets about 0.3 ohms, and that's about in line with what I've measured too.
Chris
Last edited:
Dave Berning circuits are not OTL circuits in the strict sense being used my Mark Weiss. As they now call them, they are "ZOTL" for Zero-Hysteresis Output Transformerless. As they say "operating at a fixed high-frequency without traditional audio output transformers, the ZOTL Impedance Converter eliminates the frequency-dependent performance limitations inherent in all transformer coupled tube amps".
There are endless debates in tube forums about Dave Berning amplifiers being classified OTLs or not, but it is all semantics ...
evolution
COME ON guys, the thread asked if there was evolution in tube designs: If ever there was one it is this one.
It does not go away, just simply because you hide behind the issue if it is an OTL.
I think the last response in all those VERY nasty discussion mostly based on inventors envy came down to the fact:
"If a design has no Output transformer but perfectly emulates one, it is not STRICTLY an OTL."
Can everyone live with that!
And no, I don't understand Davids white paper - but an engingeer friend of mine is ravin about it ever since studying it properly. I am simply a musician, using this design along with others. I listen.
COME ON guys, the thread asked if there was evolution in tube designs: If ever there was one it is this one.
It does not go away, just simply because you hide behind the issue if it is an OTL.
I think the last response in all those VERY nasty discussion mostly based on inventors envy came down to the fact:
"If a design has no Output transformer but perfectly emulates one, it is not STRICTLY an OTL."
Can everyone live with that!
And no, I don't understand Davids white paper - but an engingeer friend of mine is ravin about it ever since studying it properly. I am simply a musician, using this design along with others. I listen.
From a quick look at his patent, it looks like he is essentially using a switching-mode power supply with output tubes loading the output of the supply, and hence controling the current that flows through the switching transistors in the primary of the transformer (and hence control the current through the loudspeaker load). It looks like a very ingenious idea, but it is surely not helpful or useful to call it an OTL amplifier, since its principle of operation is so very different from what is normally understood by that term. Also, there is a transformer separating the output tubes from the loudspeaker load. Furthermore, the current through the loudspeaker is passing directly through the switching transistors, and is only indirectly controlled by the output tubes. Arguably, therefore, it lies within the broad class of "hybrid" amplifiers where the output stage is solid-state, and prior stages are vacuum tube. In that sense, it could perhaps be said to be strictly "output transformerless," but only because the output stage is solid state, controlled, through a transformer, from a vacuum tube driver.
Chris
Last edited:
I met David Berning at the CES show in 1980. David first started building tube amps when he was 16, I believe he is now about 55, give or take. I was a dealer of his back in the 80's and early 90's, sold a lot of his amps.
Since David was a Julius Futterman fan, David always wanted to get away from transformers since he understood their weeknesses. And since he always has worked and retired from electronic design for the Government of Bureau and Measures, he cross-bred his government work with his audio design work. Thus, the ZOTL design was born.
One thing you have noticed over the years, no one has ever attempted to copy his design, no other fully understands it. Many calls it a solid state design, but it is not. David is a perfectionist and will never build anything solid state. One definitely has to call David's work as genius.
Since David was a Julius Futterman fan, David always wanted to get away from transformers since he understood their weeknesses. And since he always has worked and retired from electronic design for the Government of Bureau and Measures, he cross-bred his government work with his audio design work. Thus, the ZOTL design was born.
One thing you have noticed over the years, no one has ever attempted to copy his design, no other fully understands it. Many calls it a solid state design, but it is not. David is a perfectionist and will never build anything solid state. One definitely has to call David's work as genius.
Well it is without doubt a very interesting and ingenious idea (I am looking at his patent, 5,612,646 "Output Transformerless Amplifier Impedance Matching Apparatus.") . It is, I think, essentially doing what I described in my previous post. Honestly, I think it is very heavily dependent on solid state devices for its operation. If you look at the circuits in the patent, it is clear that the current that passes through the loudspeaker is passing through, and is controlled by, the switching transistors in the output stage. So the "low impedance, high current" needed in the output stage to drive the loudspeaker is carried by the transistors. The "output tubes," by contrast, are operating at relatively high voltage and low current, and are effectively the driver stage for the solid-state output.
In effect, one might say that the usual output transformer of a conventional tube amplifier is replaced by a "solid state impedance transformer." The speaker is driven by a solid state switching output stage, which is in turn controlled, through a high-frequency transformer, by vacuum tube drivers. The use of solid-state devices in the output is essential to its operation.
I'd be happy to be corrected if I have wrongly interpreted the MO, but I think what I'm saying is right; the basic idea of how it works is ingenious, but rather straightforward. And again, let me emphasise that I am not in any way disparaging it; quite the opposite, it is a very clever idea, and apparently the sound it produces is wonderful. But please take a look at the schematics in the patent, and you will see that it has many MOSFETs playing a crucial role in its operation. It is without doubt a tube/solid-state hybrid.
Chris
Last edited:
The examples in the link provided above are just variations of Julius Futterman's designs. Basic cathode follower push-pull circuits.
I'm working with 6P45 tubes which have 80? plate resistance, and even THOSE can't compete with the sub 1/100th ohm source impedance of solid state devices. 80? is still 80?. You can lower it only by parallism of arrays of tubes.
Feedback will not turn a current source into a voltage source. It is an illusion, much like boosting the bass can produce the illusion of deep bass response at low volume levels with tiny speakers. However, when it comes to driving a real load, your output impedance is the plate resistance, divided by the number of parallel tubes in the output stage. No amount of feedback lowers actual output impedance. It's like asking a power supply to put out more than its rated current. You can fake it at low levels with voltage regulators, but once you're at the limit of the supply (in this case the plate load capacity of the tube) your amplifier can no longer drive the load. Oh, how I wish it were so (that adding feedback would enable a tube to drive a speaker without an output transformer), but unfortunately, physics gets in the way.
That said, there are ways to improve efficiency of OTL amplifiers, so that when you DO have enough tubes to drive a speaker, you can also maximize the power output. I'm expecting a conservative 125W per channel from my two sets of six 6P45 outputs, driven by 1000VA power supplies, each rail, +/-180VDC, and I'm modulating the screen grids with audio signal to enable me to get even more power without exceeding grid dissipation ratings. But all of this comes at a cost of many large tubes with huge cathodes with emissions levels that make 6550s look like miniature tubes in comparison. But there's no free lunch. It still takes massively parallel output stages to drive a conventional loudspeaker. Feedback won't increase the ultimate current capacity of the amplifier one iota.
Here's an early photo of my behemoth OTL amplifier, undergoing filament testing with my PWM filament drive direct from ac mains:
Well, yes and no. It is absolutely true that feeback won't boost the current-supplying capacity at all, and ulimately that is what limits the power capacity of the amplifiier. But that doesn't mean that output impedance is an illusion. As long as the current demanded into the load is less than the maximum that the tubes are capable of, then the output impedance is the feedback-determined one. (Of course there will be some detailed variation in output impedance as a function of load current, but the broad-brush statement would be that the impedance is the low one governed by the feedback unless the attempted current draw exceeds the capacity of the output devices.) One has to give an operational definition of output impedance, and essentially, any reasonable definition involves treating the amplifier's output as a perfect voltage source in series with a resistor whose value equals the "output impedance." As long as one does not attempt to draw more current than the output can supply, this model works reasonably well.
Exactly the same is true of a solid-state amplifier. Its output impedance of, say, 1/100 ohm is perfectly meaningful as long as one does not try to exceed the current capacity of the output transistors, but it will fail (along, probably with the transistors themselves!) if the current limit is exceeded.
Output impedance is just one of various parameters that one can use to characeterise the performance of an amplifier, and its meaning should be treated with appropriate caution; this is in essence what you are saying, I think, but I would not agree that it is an illusion or meaningless.
In the case of the an OTL using a pair of 6C33C tubes, the maximum current they can handle translates into about 25W rms into 8 ohms. If the demand on the amplifier is less than this limit, then the output impedance will roughly be the one calculated by taking the feedback into account. Thus, the low output impdence only ceases to be relevant or meaningful if one tries to get more than the 25W out of the amplifier. And one would never want to stray into that territory in any circumstances, since the amplifier would be clipping and distorting horribly. So, I would maintain that within the design ratings of the amplifier, the low value of impedance calculated (or indeed measured) including the effect of the feedback is the meaningful one.
Chris
Last edited:
The real world output impedance of an amplifier is the one at which maximum power is transferred when the load is matched. In transistor amps, that value is too low to be practical, due to power supply limitations and the dissipation of the transistors themselves. In practical amplifier designs, we find that 2/4/8? provides useful power transfer, though not optimal impedance match. But it's in the right direction, meaning that the amplifier will have close to the desired "near zero" source impedance. Transistor amps can enjoy the benefit of increasing power output with decreasing load impedance at *practical* speaker impedances. For a valve OTL amplifier to enjoy such advantages, speaker impedances would have to be scaled up by a factor of 100. But valve efficiency is comparatively low, and such 'under matching' would waste inordinate amounts of power.
With an OTL valve amplifier, the real output impedance greater than 100X that of a typical power transistor. While it is possible to get meaningful power into 8-16? loads, in fact, the optimal power transfer will occur somewhere around 50? load impedance with the tube types being discussed here. The tube's transconductance--not so much the power supply--is the limiting factor. The audible implications of this is emphasis of every speaker system resonance, as the output voltage will soar at resonance. Feedback can improve damping to a degree, but it's no substitute for low source impedance.
The only way to get substantially lower impedances from a tube would be to use a gas thyrotron. However, there are aspects of the ionization of gas producing threshold currents which are undesirable for audio applications. But they can drive a low impedance load for their relatively small size. Back in the 50s at Remington Rand, the R2 computer used 2D21 gas thyrotrons to drive relay coils. The 2D21 is a miniature 7-pin valve, to lend some perspective to how effective gas filled valves are at producing current flow at low voltages. And that's the real ticket to driving speakers, because they are inherently low voltage devices, whereas valves are high voltage devices. I still have a lot of left over 2D21s in my parts bins. One of these days, I may get to experimenting with them as audio amplifiers.
Probably one can go round and round on these issues endlessly. It seems to me that it is helpful to try to diagonalise the discussion of the two distinct, but related, questions of power capability and output impedance. Taking the power capability first, let's say that one wants to be able to achieve 25 W rms into an 8 ohm load. This means that the amplifier has to be able to provide a peak current of 2.5 amps, and in fact this is just about the maximum that a 6C33C can achieve. So an arrangement like an inverse Futterman with a pair of 6C33C output tubes will be able to produce the desired 25 W into 8 ohms. This part of the discussion is essentially independent of any questions about the output impedance.
Having settled on the desired output power, one can now discuss the output impedance. And for all of this subsequent discussion, it should be understood that we will never be trying to push the amplifier beyond its current handling capacity; i.e. we will never try to get it to produce a peak current in excess of 2.5 amps, in the example above. (Any attempt to do so would drive the amplifer into clipping, with huge distortion.) Having said that, the output impedance that is relevant is the one that is achieved by means of whatever feedback one chooses to use. In the absence of additional feedback, an inverse Futterman using a pair of 6C33C tubes will have an output impedance of about 11 ohms. With feedback, as discussed for example in the Hans Beijner or Tim Mellow articles, the output impedance can easily be lowered to a fraction of an ohm.
This feedback-generated output impedance is perfectly real and genuine (provided, as always, that we never ask the amplifier to output more than 2.5A peak current). In particular, it is this output impedance that will determine the damping factor, and it is this output impedance that will control how the amplifier performs in the case that there is a speaker resonance where the speaker impedance becomes very large. So for all operational purposes, the feedback-generated output impedance is the relevant one that governs the performance of the amplifier under frequency-dependent load-impedance conditions, etc. (Of course, if the speaker impedance is less than the nominal 8 ohms in some frequency range, then one really has to go back to step one and re-assess what current-handling capability one is needing for the system.)
Your remarks about gas thyratrons are interesting. I suppose for ordinary audio amplification they are of limited utility, since they are presumably essentially on/off devices? If they could switch at a high enough speed, I suppose one could build a tube-only version of the Berning ZOTL!
By the way, your OTL under construction looks great! It will be interesting to hear how it progresses. My latest, using series-connected heaters with 6082 tubes run direct from the mains, and all power supplies direct mains connected (so no transformers at all) has been running nicely for a few months now.
Chris
Last edited:
Although power in amplifiers is given by simple equations, in practical terms the import aspect is the perceived loudness of a system in typical listener conditions. And here things can become much more complicated, as aspects such as speakers load, listening conditions and listener preference can change the nice predictions due to simple theorems.
I have experienced conditions in which big amplifiers rated at 8/4 ohms as much more powerful were not able to drive the system adequately and tube amplifiers with one quart the rating did an excellent job. IMHO OTLs seem particularly adequate to drive electrostatic speakers - I have no reason to explain it, although my experiences with a copy of the Futterman OTL3, the Technics 20A (also an incredible sounding design), Graph GM20, Atmasphere MA50 and MA2 seem to confirm it.
I have experienced conditions in which big amplifiers rated at 8/4 ohms as much more powerful were not able to drive the system adequately and tube amplifiers with one quart the rating did an excellent job. IMHO OTLs seem particularly adequate to drive electrostatic speakers - I have no reason to explain it, although my experiences with a copy of the Futterman OTL3, the Technics 20A (also an incredible sounding design), Graph GM20, Atmasphere MA50 and MA2 seem to confirm it.
Yes, it all comes down to current delivery, at the most basic level, but current X voltage is the complete picture. 2.5A for the 6C33 is completely off the charts. I'm using the much more powerful 6P45 and the peak current is 1400mA. The 6C33 tops out at 875mA and that's with the control grid at -2V, running well outside the linear part of the curve and utterly useless for audio. At practical bias levels, the current limit is 450mA for this tube. You would need a half dozen of them on each half of the amplifier to achieve 2.5A output current, notwithstanding the efficiency factor and that some of that power is being burned up by plate losses. Plate resistance is 130?. At full conduction, it's like a transistor amp with a 130? resistor in series with each pass transistor. That won't deliver much power to an 8? load as most of it will be burned up in the plate resistance.
Servo feedback can only achieve the appearance of low source impedance at very low levels, up to the current delivery of the tube. I doubt much more than 450mA into the load. With push pull, double the voltage, but the current is still limited to 450mA.
The gas thyrotron would not be a practical amp unless a means of keeping the gas ionized full time, but at minimal standby current could be arranged.
Here's a OTL that I sketched out and built in the 1960s using 6AS7s (similar to the 6080 type), using no transformers. Note that it's limited to half wave rectification however:
This amp worked nicely into a 16? load (in my case a pair of Utah 12" speakers in series) and was pretty flat to 100KHz and had no visible distortion at 5Hz as seen on my Heathkit oscilloscope, which was going on 12-13 years old at that time.
Servo feedback can only achieve the appearance of low source impedance at very low levels, up to the current delivery of the tube. I doubt much more than 450mA into the load. With push pull, double the voltage, but the current is still limited to 450mA.
The gas thyrotron would not be a practical amp unless a means of keeping the gas ionized full time, but at minimal standby current could be arranged.
Here's a OTL that I sketched out and built in the 1960s using 6AS7s (similar to the 6080 type), using no transformers. Note that it's limited to half wave rectification however:
This amp worked nicely into a 16? load (in my case a pair of Utah 12" speakers in series) and was pretty flat to 100KHz and had no visible distortion at 5Hz as seen on my Heathkit oscilloscope, which was going on 12-13 years old at that time.
Question for Mark [Basspig] Weiss & Chris - cnpope & other knowledgeable people
Question for Mark [Basspig] Weiss & Chris / cnpope & other knowledgeable people please join in.
Amplifiers that use the 6c33c tend to have means to set the bias. Some tubes even if matched to begin with quickly lose the exacting match once they are in use.
Therefore, should one bother buying matched 6c33c tubes?
Or, do not bother with "matched sets" & just break-in singles & properly maintain the bias?
Thank you for your assistance.
Respectfully,
zz
Question for Mark [Basspig] Weiss & Chris / cnpope & other knowledgeable people please join in.
Amplifiers that use the 6c33c tend to have means to set the bias. Some tubes even if matched to begin with quickly lose the exacting match once they are in use.
Therefore, should one bother buying matched 6c33c tubes?
Or, do not bother with "matched sets" & just break-in singles & properly maintain the bias?
Thank you for your assistance.
Respectfully,
zz
Practically every 6c33 tube I bought in the open market failed in a matter of months. Those I got from Lamm and BAT for the ML1.1 and VK-150s respectively lasted a whole lot longer. The longest being 3 years or more. I think the problem is finding and getting "good" 6c33s. I've not seen a tube tester that does 'em. The premium asked by the manufacturers is small so I just go direct to them and dip into their stash for replacements.
Similar threads
