Legal issues
Demian, my experience is directly in line with your post. I have spent a lot of both my own and my company's money defending patents, and in more than one case where I demonstrated an innovative circuit without patenting it first, the client went on to patent and capitalize on the concept DESPITE non-disclosure agreements.
Without bags of money I could never win any of these suits, and in more than one case was unable to get lawyers to take the case on a contingency basis due to the deep pockets of the thieves. The small guy has a really tough time winning in this legal climate, where $$ for legal representation=judgement in your favor, or interminable delays and re-filings until you give up due to lack of $$...
I am currently involved with two patent infringement cases, one we initiated and one brought against us. Between the legal costs for both I may find myself out of a job soon... At least then I can spend more time experimenting with audio...
Great thread, and I have learned a lot from all of you guys...
Howard Hoyt
Demian, my experience is directly in line with your post. I have spent a lot of both my own and my company's money defending patents, and in more than one case where I demonstrated an innovative circuit without patenting it first, the client went on to patent and capitalize on the concept DESPITE non-disclosure agreements.
Without bags of money I could never win any of these suits, and in more than one case was unable to get lawyers to take the case on a contingency basis due to the deep pockets of the thieves. The small guy has a really tough time winning in this legal climate, where $$ for legal representation=judgement in your favor, or interminable delays and re-filings until you give up due to lack of $$...
I am currently involved with two patent infringement cases, one we initiated and one brought against us. Between the legal costs for both I may find myself out of a job soon... At least then I can spend more time experimenting with audio...
Great thread, and I have learned a lot from all of you guys...
Howard Hoyt
HHoyt, thanks for the input. I was very naive in my younger years, and I paid the price for it. My associate Jack Bybee is equally careful, these days. Once he lent a sample of one of his devices to a major German company, encased in a hard resin. He finally got it back, after they attempted to break it open.
I once let a high fi dealer in Berkeley try one of my Vendetta preamps, it was given SOTA, a strong competitor at the time, and returned partially disassembled. And so it goes.
Of course, this is one of the reasons that I don't publish schematics. Even so, virtually everything I have ever designed is most probably reverse engineered by somebody, sooner or later. Hopefully, later, after I have a run with it.
I once let a high fi dealer in Berkeley try one of my Vendetta preamps, it was given SOTA, a strong competitor at the time, and returned partially disassembled. And so it goes.
Of course, this is one of the reasons that I don't publish schematics. Even so, virtually everything I have ever designed is most probably reverse engineered by somebody, sooner or later. Hopefully, later, after I have a run with it.
Well, now that the patent situation is 'resolved', I might point out something that seems to not be well understood about nonlinear capacitance in j-fets.
Nonlinear capacitance in j-fets is one characteristic that makes j-fets inferior to vacuum tubes. While it can be fairly small, and the circuit designed to minimize its contribution to the distortion in a design, it is often overlooked, especially by amateurs.
There is a range of capacitance in j-fets that is usually related to the voltage noise of the same j-fet. In reality, many 'low noise' j-fets are actually several j-fets in parallel in the same package. This lowers the noise, and raises the Gm, BUT it also multiplies the nonlinear input capacitance. Paralleling these devices must be done with care, or you will get more distortion, and little useful improvement in the design.
Nonlinear capacitance in j-fets is one characteristic that makes j-fets inferior to vacuum tubes. While it can be fairly small, and the circuit designed to minimize its contribution to the distortion in a design, it is often overlooked, especially by amateurs.
There is a range of capacitance in j-fets that is usually related to the voltage noise of the same j-fet. In reality, many 'low noise' j-fets are actually several j-fets in parallel in the same package. This lowers the noise, and raises the Gm, BUT it also multiplies the nonlinear input capacitance. Paralleling these devices must be done with care, or you will get more distortion, and little useful improvement in the design.
I agree, Joachim, that cascoding can solve much of the problem with non-linear stages, especially IF the load on the fet is a current source. However, with a low value resistive load, the gain is minimal so the Early Effect is minimal, and the added advantage of higher voltage across the input fets, can offset the need for a cascode.
I think it depends on what you want. More then 3V from Source to Drain may increase the noise but linearity will improve up to 7V and then it does not get much better. That is the experience i have with the typical low noise Toshiba Fets.
One topology that should work is a cascode, then current source, than folded cascode in the second stage. That has low impedance nodes all over. A phonostgae could be made that way with transimpedance RIAA out of the folded cascode and then a buffer.
I whould do that parallel symmetric for low noise and 2nd harmonic supression. It is a pitty that P channel J-Fets are not made any more for Audio in quantity.
One topology that should work is a cascode, then current source, than folded cascode in the second stage. That has low impedance nodes all over. A phonostgae could be made that way with transimpedance RIAA out of the folded cascode and then a buffer.
I whould do that parallel symmetric for low noise and 2nd harmonic supression. It is a pitty that P channel J-Fets are not made any more for Audio in quantity.
There is virtually no market. The few nuclear engineering and optical applications are served by N-channel devices.
Why complementary j-fets? Why complementary differential j-fets, or any complementary differential input stages at all?
The same could be asked why 12 cylinder autos are made. For ultimate quality that is both intellectual and sonically satisfying in solid state circuits.
I happened across the complementary differential bipolar input stage when I was trying to make the best 10W amp that I possibly could, in order to power my K-horn. I had experimented with tubes, and commercial transistor amps, but neither completely satisfied me, sonically. What did, was this Radiocraftsman triode amp that was at a hi fi store, and not for sale. I tried to make at least as good a 10W amp as the triode amp. I really tried hard. Class AB1 biasing at 1/2A, beta matched output transistors, selected drivers, and an all balanced topology. And guess what? It took me years to do it right, and ONLY when I listened more carefully to Dr. Otala and his recommendations, did I get it to sound about as good.
Accident? No, just good luck.
IF you look at Jan Lohstroh's amp design, you will find the same thing. A good typically engineered design is not necessarily a GREAT design. Without Otala's input, the project at Philips Research, would have been a waste of time, subjectively, and soon forgotten.
Now where does that put Dr. Otala? Guru? Not really, he was just strong enough in his personal opinions as to what he and others before him had found, worked, to insist on them, in an audio design, to see what happens.
It is the same with me, I moved from single sided complementary bipolar, like some video amps, today, to bipolar complementary differential, then to j-fet complementary differential, and even experimented with single sided complementary j-fet input, like the Vendetta phono input stage. Personally, I like the Vendetta approach best, but it is NOT practical without excessive fet matching and selection. That is why Parasound amps and most preamps that I design use complementary differential j-fet input stages.
By using a fairly sophisticated input stage, like this, it sounds better, all else being equal, probably because the input stage is often underestimated in its need for extreme linearity, as most test measurements will not bring it out, but extended bandwidth music will.
The same could be asked why 12 cylinder autos are made. For ultimate quality that is both intellectual and sonically satisfying in solid state circuits.
I happened across the complementary differential bipolar input stage when I was trying to make the best 10W amp that I possibly could, in order to power my K-horn. I had experimented with tubes, and commercial transistor amps, but neither completely satisfied me, sonically. What did, was this Radiocraftsman triode amp that was at a hi fi store, and not for sale. I tried to make at least as good a 10W amp as the triode amp. I really tried hard. Class AB1 biasing at 1/2A, beta matched output transistors, selected drivers, and an all balanced topology. And guess what? It took me years to do it right, and ONLY when I listened more carefully to Dr. Otala and his recommendations, did I get it to sound about as good.
Accident? No, just good luck.
IF you look at Jan Lohstroh's amp design, you will find the same thing. A good typically engineered design is not necessarily a GREAT design. Without Otala's input, the project at Philips Research, would have been a waste of time, subjectively, and soon forgotten.
Now where does that put Dr. Otala? Guru? Not really, he was just strong enough in his personal opinions as to what he and others before him had found, worked, to insist on them, in an audio design, to see what happens.
It is the same with me, I moved from single sided complementary bipolar, like some video amps, today, to bipolar complementary differential, then to j-fet complementary differential, and even experimented with single sided complementary j-fet input, like the Vendetta phono input stage. Personally, I like the Vendetta approach best, but it is NOT practical without excessive fet matching and selection. That is why Parasound amps and most preamps that I design use complementary differential j-fet input stages.
By using a fairly sophisticated input stage, like this, it sounds better, all else being equal, probably because the input stage is often underestimated in its need for extreme linearity, as most test measurements will not bring it out, but extended bandwidth music will.
Now what are the limitations of complementary differential j-fet input stages?
For one thing, the P channel fets have much more input capacitance than the N channel equivalent, used in parallel. Therefore, the complementary differential input stage is really not ideal for inputs of over 100K ohm drive impedance. This problem also limits the usefulness of paralleling complementary j-fets as well.
We once made a line stage with double sized complementary j-fets, because that was all that was available. This was the 2SK146 and the 2SJ73 part pair that are equivalent of at least two 2SK170's for each device, and of course, its complement, the 2SJ74, the same. When the lower capacitance Toshiba matched parts, the 2SK240 and the 2SJ75, that are really two 2SK170's and two 2SK74's respectively, were swapped into the same circuit, the worst case distortion, even with a 10K volume control pot, was reduced, EVEN THOUGH the Gm was reduced as well. This was due to the reduction of nonlinear capacitance loading the input pot.
When you have an EXTREME range of impedances, up to 1 meg ohm, for example, it is better to not use the complementary differential j-fet input stage, and just use an N channel matched pair, instead.
For one thing, the P channel fets have much more input capacitance than the N channel equivalent, used in parallel. Therefore, the complementary differential input stage is really not ideal for inputs of over 100K ohm drive impedance. This problem also limits the usefulness of paralleling complementary j-fets as well.
We once made a line stage with double sized complementary j-fets, because that was all that was available. This was the 2SK146 and the 2SJ73 part pair that are equivalent of at least two 2SK170's for each device, and of course, its complement, the 2SJ74, the same. When the lower capacitance Toshiba matched parts, the 2SK240 and the 2SJ75, that are really two 2SK170's and two 2SK74's respectively, were swapped into the same circuit, the worst case distortion, even with a 10K volume control pot, was reduced, EVEN THOUGH the Gm was reduced as well. This was due to the reduction of nonlinear capacitance loading the input pot.
When you have an EXTREME range of impedances, up to 1 meg ohm, for example, it is better to not use the complementary differential j-fet input stage, and just use an N channel matched pair, instead.
I wonder what if one wanted to have a complementary JFET output stage to drive 50 ohm headphones. How would it work to match say 25 pairs of k170/k74s and use a second stage (say 4 parallel pairs) that have an output impedance of <=100 ohms? (Complementary differential input stage as you mention.)
I'm thinking about something that will work with 50 ohm Audeze LCD2 planar magnet headphones.
Gallery :: Audeze
Thank you for your input, John.
- Status
- Not open for further replies.