1W SE Class A – Powerful Headphone or Weak Power Amp

[VladimirK]

I believe that the only way of getting close to tube’s sound with transistor designs is using of very low Ciss transistors. Although there is no theoretical or measurable support to this statement, listenings do prove it, and it makes a point to me.
With this idea, the KP907A lateral MOSFET (old Russian HMOS technology transistor, used up to 1,5GHz frequencies, that have specific linear Id vs Vgs) seduces me for already some time. I enclose some info from its datasheet. With max power dissipation 11W of this transistor, one can expect only around 1W output power with the Follower-like output stage, depending on the load value. It is too much for Pre and Headphone amps, and ridiculously small for a Power amp. But, finally I have convinced myself, that:

- even 1W could be enough for sensitive 97…105 dB speakers,
- there are current demanding 25…32 Ohms headphones,
- even preamps sometimes need to drive around 100 Ohms input impedance of power amps or other consumers,
- without paralleling of active small Ciss output transistors one might get the best sound,
- at this heat dissipation range, one can use a pair of RF power jFETs KP903A as a simplest current source, being a load for KP907A.

On this basis, I have put together the following prototype (the as-built schematics is enclosed), being equipped with all kinds of output connections (Eichmann RCA input connectors, Eichmann speaker posts, Neutrik silver headphone jack socket, RCA output connectors, DACT volume control). Soldering is done point-to-point, without PCB. I was afraid about possible ringing with UHF transistor, but everything is working without any additional adjustment. Enclosed are scope shots of 1 MHz sine and square wave signals reproduction, at 4V p-p level with 11 Ohms resistive load.

Even at 1V RMS output voltage it is possible to estimate sound quality with Castle Acoustics Pembroke speakers, and even with big PMC EB1i. The sound has lived up the expectations, it is extremely live-like. In spite that some average tinfoil caps and metallized polypropylene caps were used, the sound is excellent.
We also agreed with a friend of mine to arrange listening via his Denon AH-D7000 top level headphones. Presumably, the present schematics can get some justification from the low (25 Ohms) impedance of these headphones.

Enclosed are also shots of the prototype in its actual state, some case work is pending still.

[adason]

Very nice VladimiK!
1 watt can be enough in certain situations. I have been living with 2 watts OTL amp for many years. However, I only use it for mids and tweets. Not for bass, that's where there is never enough power. I use active crossover.
At normal listening levels we rarely need more than 100 to 500 milliwatts with reasonably sensitive speakers.

[AKSA]

Nice work, Vladimir!

I hadn't seen any of these HMOS devices before. Very linear but very low gm, probably not too important as a source follower.

Interesting that you use 2SJ103 on the front end, and even more that source feedback on the input side carries a high 23.5mA, which dwarfs the current through the device.

Thank you for publishing your excellent design,

Hugh

[tvrgeek]

Well, if you are into the super efficient single driver thing, yea, a Watt is actually a bunch. Big Louther horn?

[Bigun]

Originally Posted by VladimirK

I believe that the only way of getting close to tube’s sound with transistor designs is using of very low Ciss transistors.

I agree too that much of the magic is that 1W. But your special FETs are unobtainable outside of USSR (unless you gonna send me some )

Why settle for 'getting close to tube' when you can get your 1W from a tube ?

Single Ended Class-A Power Amplifier using 6C45Pi - Electronics Now! schematics, Electronic circuit, PCB or project for Electronic.

[VladimirK]

Originally Posted by Bigun

I agree too that much of the magic is that 1W. But your special FETs are unobtainable outside of USSR (unless you gonna send me some )

Why settle for 'getting close to tube' when you can get your 1W from a tube ?

Single Ended Class-A Power Amplifier using 6C45Pi - Electronics Now! schematics, Electronic circuit, PCB or project for Electronic.

Hello, Bigun
Sending of the FETs is possible, but would be nice to proceed via group buy, since getting them and sending them take much time, and shipment costs are to be remembered.
As for using tubes themselves, this would cause some additional complications and drawbacks, and I find that proper choice of UHF transistors allows to really overtake tube designs soundwise and conveniencewise. In particular, I doubt that the choke power supply in tube schematics overtakes the follower-like approach at the output stage of my schematics. All tube designs that I listened up to now, they are already behind soundwise.

[VladimirK]

Originally Posted by AKSA

Nice work, Vladimir!

I hadn't seen any of these HMOS devices before. Very linear but very low gm, probably not too important as a source follower.

Interesting that you use 2SJ103 on the front end, and even more that source feedback on the input side carries a high 23.5mA, which dwarfs the current through the device.

Thank you for publishing your excellent design,

Hugh

Hello, Hugh
The low Gm does matter, therefore I use current NFB, even with the follower-like output stage. With NFB, output impedance value goes below 1 Ohm. There are around 10 various types of these russian HMOS lateral FETs (KP901, 902, 904 (75W 100V 300pF 0,3S), 905, 907, 909, 911, 913 (100W 100V 200pF 1S) ... ), some of them are up to 70...100W heat dissipation. All of them have relatively low Gm, and in order to get to optimal working point, one must spend 5...7V of positive gate bias. This is not convenient for low-wattage schematics. All of them could be substituted by modern NXP parts, but Id vs Vgs curves are different for NXP.
As for power jFETs types, there is only one RF part - KP903.

[Bigun]

Originally Posted by VladimirK

Hello, Bigun
Sending of the FETs is possible, but would be nice to proceed via group buy, since getting them and sending them take much time, and shipment costs are to be remembered.
As for using tubes themselves, this would cause some additional complications and drawbacks, and I find that proper choice of UHF transistors allows to really overtake tube designs soundwise and conveniencewise. In particular, I doubt that the choke power supply in tube schematics overtakes the follower-like approach at the output stage of my schematics. All tube designs that I listened up to now, they are already behind soundwise.

I have some good chance to get directly from former USSR so if it's some hassle then I am OK to wait. You see, I have 3 chasis in my basement, they may be last SS projects I do for sometime. One of these chasis is set up for a Class A power follower (it will be tube driven so gate threshold voltage is not going to be a limitation) and it is waiting for me to choose that 'special' single FET as the output device (two total for stereo). It will be cascoded so that it doesn't need to dissipate a lot of power. Plus, it doesn't have to be a KP903, but you have got me all excited about it and it would be pitty to use something else

[VladimirK]

Originally Posted by Bigun

One of these chasis is set up for a Class A power follower (it will be tube driven so gate threshold voltage is not going to be a limitation) and it is waiting for me to choose that 'special' single FET as the output device (two total for stereo). It will be cascoded so that it doesn't need to dissipate a lot of power. Plus, it doesn't have to be a KP903, but you have got me all excited about it and it would be pitty to use something else

What idle current do you plan to use? With a single KP903A it could be 0,3...0,4A only.
With KP907A 1...2A. With KP913A 2...4A.
KP904A also could be used at 2...3A, but it is not UHF, it is VHF part with around 350pF Ciss. The KP904A I could get locally for around 7...8USD per pcs. All the parts have all pins isolated from the mounting stud.

[Bigun]

I'll likely have only +/-10V swing at most and worse case into 4R load that's 2.5A peak. If I don't see myself running this at more than 2.5A idle so that is worse case.

You mentioned that there are NXP equivalent parts ?

By the way, Happy Christmas this weekend