44 & 45 are understandable and I see where Don talks about the complexities of the design. Thanks.
Tim
Today's most innovative amplification technology: tubes or solid state?
- Thread starter ack
- Start date
44 & 45 are understandable and I see where Don talks about the complexities of the design. Thanks.
Tim
Direct-coupled tube preamps and amps have been around for ages but rarely commercialized, probably due to reliability issues. At least that was the problem with my direct-coupled tube preamp. The other problem was that I cleverly used a fully differential design, and the resulting sound (while measurably better than most other tube gear) was deemed "too SS-like".
Class D amplifiers use a non-linear (extremely nonlinear) output stage, require fairly high (for audio) switching rates, and a reactive filter at the output, all of which makes closing the feedback loop around the circuit very challenging. Phase shift is nonlinear and getting a stable design is tough, particularly if the load is varying (loudspeaker, anyone?) The combination of switching output stage and output filter makes them pretty load-sensitive. The output devices are subject to switching transients, exacerbated by the output filter, so need special protection. The output devices' timing (of their switching) is critical to avoid blowing them up. Finally, the ideal response depends upon a perfect power supply. An analogy is to think of switching between "0" and "1" but in this case the "0" and "1" levels are set by the power supply rails since that's what you are switching the output to and from. If the supply voltage "wiggles", it is translated to the output. Now consider you are switching gobs of current very fast, leading to large current spikes, and envision how hard it is to keep those power rails from wiggling. There's a lot more devil in a lot more details but those are the big ones I recall off-hand. It has been a while since I did a class D design.
Done well, they are great, but by and large I think they have a bit to go.
Class D amplifiers use a non-linear (extremely nonlinear) output stage, require fairly high (for audio) switching rates, and a reactive filter at the output, all of which makes closing the feedback loop around the circuit very challenging. Phase shift is nonlinear and getting a stable design is tough, particularly if the load is varying (loudspeaker, anyone?) The combination of switching output stage and output filter makes them pretty load-sensitive. The output devices are subject to switching transients, exacerbated by the output filter, so need special protection. The output devices' timing (of their switching) is critical to avoid blowing them up. Finally, the ideal response depends upon a perfect power supply. An analogy is to think of switching between "0" and "1" but in this case the "0" and "1" levels are set by the power supply rails since that's what you are switching the output to and from. If the supply voltage "wiggles", it is translated to the output. Now consider you are switching gobs of current very fast, leading to large current spikes, and envision how hard it is to keep those power rails from wiggling. There's a lot more devil in a lot more details but those are the big ones I recall off-hand. It has been a while since I did a class D design.
Done well, they are great, but by and large I think they have a bit to go.
Last edited:
Don-I think you did a great job of summarizing the challenges inherent in a Class D amplifier. That is why I previously said that if it wasn't for their super-high efficiency and low cost, no one in the high-end would think about using them. Not only do you have to make a pig fly, it has to look like a super model at the same time.
Thanks for the details Don, well put. But one the reasons I harp on the DDFA design is that it addresses most (all?) the issues raised around Class D, as the architecture integrates the load and maintains performance essentially equivalent to a very high quality DAC.
Interesting, I had not heard of DDFA designs exactly like Zetex for audio, though a lot of manufacturers appear headed that way. I dug around a little, not finding their patent but a white paper or two and a very interesting discussion over on a DIY forum that described it pretty well. It appears very similar to other schemes but details are hard to find (no surprise there!) What I read indicates a fairly complex scheme (not implying that is bad) that manages to compensate for power supply and load variation. Other schemes I have seen (bearing in mind this is not my primary field of expertise) seem similar but I am not sure any one incorporates all I am reading about the Zetex design. With limited time for searching I did not see a lot of data about how it operates with real-world loads, but it appears to use noise shaping (delta-sigma loop(s)) and very high (for audio power amps) sampling rates to reduce distortion and suppress power supply noise. I have seen similar designs (using DS feedback loops and other schemes for both load and supply compensation, as well as digital and analog circuits for compensating amplifier nonlinearities) in other applications and results have been a mixed (some good, some bad, and some that would work over a range of conditions then fail). Be interesting to see the products that arise from this idea, it is certainly worth watching!
Thanks - Don
Thanks - Don
Bruno discusses in some detail the Zetex circuit at 36:28. He uses Zetex as a contrast to the hypex approach which only uses PWM which is an analog signal.
We came out with the first balanced line preamplifier in 1989. It also featured a direct-coupled output. We still make it, its called the MP-1.
FWIW IMO, innovation has nothing to do with tube or transistor. It has to do with innovation. It really is that simple.
You and me both! I do know that they have a backer with deep pockets- at least that was what they told me at THE Show (they were right across the hall). Their tubes will be more expensive, but from what I can make out actually cheaper in the long run as they should last considerably longer than most modern tubes which are not built all that well.
Thanks, have to check that out tonight. I have heard PWM and PFM designs that have only two amplitude levels (or three, for RTZ) called digital modulation, but semantics aside the whole DDFA approach is intriguing.
Ralph-Their backer better have very deep pockets and lots of patience. Was anyone at any of the shows using those tubes or were the tubes just on static display?
As for saying the tubes should last considerably longer than most modern tubes, I don't know how you could be sure of that statement. Unless a tube dies for reasons unrelated to cathode depletion (shorts, gas, or micro leaks in the vacuum envelope) which should be uncommon these days, the only reason the tube would last longer than other modern tubes would be the cathode was capable of more emission over a longer period of time. In a perfect world, all vacuum tubes would die a natural death caused by cathode depletion. In other words, the cathode can no longer sustain an electron beam with sufficient current for the amplifier circuit to work. The secret to great cathodes are the recipe for the doping compounds. Are these new guys onto something? If so, where did that knowledge come from?
Until these guys actually have production tubes in production amplifiers with over 2.000 hours of life, I will remain skeptical about whether their company is going to be a real business and what the quality of their tubes will really be. I think the jury should be out on this one. I also wonder why they would want to build a KT-88 when the new KT-120 and KT-150 tubes show so much potential to be great tubes. But then, I also wondered why they chose to bring a 6L6 to market first (even though I'm not sure they really have any ready for sale) when that tube is not common for stereo tube amps being used today. It's much more common to be used in a guitar amp and as I already mentioned, good luck trying to convince the average musician to fork over $200 each for a 6L6.
^^ I think your concerns are well-founded.
In the old days though it took tens of thousands of dollars to fabricate the dies used to make tooling for the tube structures. The guy explained to me that now they are using computer controlled lazer techniques to fabricate the same structures- or ones slightly different. In this way they have been able to experiment with minor changes in the tube contruction in order to optimize the tube without costs that break the bank.
He also explained that as recently as 1960, while RCA knew what the optimal construction and composition of the cathode structure was, they didn't know why. They just knew from experience that it worked. Due to advances in materials science, that has changed also. He showed me examples of how Russian tubes were overfired during the getting process, which leads to premature cathode failure due to contamination. He also confirmed some of my theories about why some modern Russian signal tubes seem to fail so quickly (due to contaminants in the cathode that out-gas during the first few hundred hours of the tube's life). He seemed to feel that his grid structures were more accurate than the NOS tubes, since modern computer controlled machining techniques allow for greater precision in their manufacture.
In short, he seemed to know his stuff- and this is why his tubes should last longer. But you are right- we do need to see some examples in the marketplace!
In the old days though it took tens of thousands of dollars to fabricate the dies used to make tooling for the tube structures. The guy explained to me that now they are using computer controlled lazer techniques to fabricate the same structures- or ones slightly different. In this way they have been able to experiment with minor changes in the tube contruction in order to optimize the tube without costs that break the bank.
He also explained that as recently as 1960, while RCA knew what the optimal construction and composition of the cathode structure was, they didn't know why. They just knew from experience that it worked. Due to advances in materials science, that has changed also. He showed me examples of how Russian tubes were overfired during the getting process, which leads to premature cathode failure due to contamination. He also confirmed some of my theories about why some modern Russian signal tubes seem to fail so quickly (due to contaminants in the cathode that out-gas during the first few hundred hours of the tube's life). He seemed to feel that his grid structures were more accurate than the NOS tubes, since modern computer controlled machining techniques allow for greater precision in their manufacture.
In short, he seemed to know his stuff- and this is why his tubes should last longer. But you are right- we do need to see some examples in the marketplace!
I would say this is great innovation wrt the construction of tube devices, but agree, yet to be proven. By contrast, anyone know of significant innovations with solid state devices recently that manufacturers take advantage of? What's the essence of Pass's work? Anyone have links to share on diyaudio.com regarding this? (I know he's very active on it, but don't have the time to really follow that forum).
^^ That is something I would like to know as well. The IGBTs seem to be the latest thing. The troubling thing is, semiconductor manufacturers appear to be rapidly moving away from linear devices: I can see a day when class D may someday be the only game in town for solid state amplifiers.
Finding silent semiconductors for my Mellotron has been a real pain- such that it prompted me to design an opamp-based equalizer for it.
Finding silent semiconductors for my Mellotron has been a real pain- such that it prompted me to design an opamp-based equalizer for it.
Ralph,
I know this subject is not innovation, as I read about it long ago and can not fin the source anymore - I think it was research carried by the British post office during the 40's. But perhaps you can remember it - have you read about checking tube wear comparing measurements obtained with normal and reduced filament voltage?
That sounds great, but I don't buy it. Look at any RCA tube receiving tube manual and numb your mind with all of the different types of tubes that were designed for highly specific applications. The old tube engineers could make tubes do anything damn thing they wanted them to. RCA had some of the best tube engineers in the business. RCA didn't arrive at an optimal cathode design by accident. And this new guy who is talking the talk about how he can make tubes better than anyone ever did needs to walk the walk.
The science of cathode poisoning is old and well understood. Just because he claims the Russians are screwing it up without offering evidence of what company and what tubes doesn't mean that the great companies from the glory days were making these mistakes. After all, I thought he was claiming to make better tubes than the American originals and not better than the Russians. When are his awesome tubes going to hit the market?
^^ I would really like to know that too.
As far as the Russian contamination thing I have known about that for a long time. The Russian plants don't seem to be in the habit of giving their workers gloves, so parts in the tubes have fingerprints on them which out-gas. As a result, depending on your luck, you can get a new Russian 6SN7 that may last 2 years or only 2 weeks. They have been so problematic that we have had to warn our customers to stay away from them.
As far as the Russian contamination thing I have known about that for a long time. The Russian plants don't seem to be in the habit of giving their workers gloves, so parts in the tubes have fingerprints on them which out-gas. As a result, depending on your luck, you can get a new Russian 6SN7 that may last 2 years or only 2 weeks. They have been so problematic that we have had to warn our customers to stay away from them.
Similar threads
