• The site will be going down for scheduled maintenance at 2PM GMT / 6AM PST Tuesday, Nov. 12 for approximately 30 minutes to perform some minor software updates.

Weird Design: need help!

Status
Not open for further replies.
Hello,

I found this design at a convention in Japan. I won't bother explaining how I got them.

It is a design for a low power amplifier for headphones. It uses a pair of mosfets and a pair of something called VFET. I don't know what it means, but the original designer suggested that I should use something else. I have been putting off for nearly 2 years now.

After messing around with Pspice, I came up with something workable with 4 mosfets per channel. I have been asking people at Headwize, but they seem to be more interested in something else. I am not sure it really works. I am not quite sure I am using right parts either, I just grabbed IRF510 and IRF9510 because I used them before.

<img src="http://www.people.cornell.edu/pages/ti20/HPAInvPPReal2.JPG" img>

Thanks,

Tomo
 
Hello,

This amplifier uses 4 independent dual power supplies. 2 to set the gate voltages and 2 "floating" to supply current to the mosfets. (for stereo)

+-15V PSU is completely independent of +-20V PSU. In fact, 20V PSU is floating. Gnd is connected to +-15V PSU. Common of 20V PSU, can swing, I guess. As you can see, it gives me some good output with some gain. (Input was 1V AC)

Tomo
 
Hey,

It works with some very rare mosfet called VFETs (2SK82 and 2SJ19). I do not know what it is. As for current post with all mosfets, the design is essentially extension of original design (w/ Vfets). The picture shows my results of bias point calculation and freq analysis. The latter isn't reliable but former should be somewhat correct.

But no, I have never build this thing. I am very curious about what you guys would say. Besides, I am not even sure why it give me gain 🙂

I guess I almost am posting to entertain me ... But I do hope you are getting entertained as well.

Tomo
 
This circuit is interesting for three of reasons:

First, it is single stage, and it looks a bit like one of the
upcoming Zen Variations projects.

Second, it makes use of the "inverted power supply"
we remember from Strickland's TNT amps.

Third, it has an extremely clever use of cascoding.
 
But but ...

Hello,

I wonder it actually works. Because, as I have mentioned, I am not using the original parts which are MOSFETS (2SK134 and 2SJ49) and VFET (2SK82 and 2SJ19).

Also, I really wonder how the gain is controlled. It has to do with the R14 resistor. But nothing conclusive came up.

Please note that this is not my creation. This is made by a unique individual who reside in Yokosuka City, Kanagawa Prefecture, proximity of Tokyo. (I can't read his Kanji name ... a man forgets, ok!?) He built this for a headphone amp convention in Tokyo by a group called Aedio. Unfortunately, the track end at Aedio because I am not good enough to earn their trusts and I am not very good with Japanese.

I won't be building this amp any time soon due to economical and situation problems. However, I would like to know more about the new variation when it comes out.

Thanks,

Tomo
 
I think the Vfets you are discussing use a V shaped stucture, mainly used for RF.

From memory they had a problen with uneven charge buildup in the V channel??

The similar current devices might be the <a href="http://www.onsemi.com/parametrics/selector/0,4093,10201,00.html?LevelName0=Product&LevelName1=Discrete&LevelName2=Power+MOSFET+%28TMOS%29&AssociatedParametrics=Power+MOSFET+%28TMOS%29+10201">TMOS from On Semi</a> or TrenchFET from Vishay/Siliconix.


Regards
James
 
VFETS have a solid-state channel structure which is vertically oriented and the lateral MOSFETS have a channel which is....well, lateral. Motorola (ONSEMI) has a series called VMOS which are this type of device. As for the circuit it looks as if it is running open-loop....but that's not giving too much to the cause.

BeanZ
 
Weird design

The evolution from V fets to the newer lateral or trench types or what have you is the result of the quest for lower Rdson. The downside is that the topology leading to the lower Rdson also increases the capacitances. So, for audio, the Vfet types may well be *better*.

Jan Didden
 
I've been working up a circuit that is on this type of vien. Tho, I havn't cascoded the output. I started with a circuit and the idea of building a PP zen type of amp. This didn't work out so well and I gave up when I blew the 540/9540's. I'm not sure these are the easiest critters for this type of topology. There is no thermal compensation that I can see, and the IR's definately seem to need this. I've now started to actually use some Japanese transistors, and low and behold some things are starting to work out. I believe that what I have are VFETs, but alas the differentiation can be hard to sort out at times. I have the Toshiba 2SJ200/ 2SK1529 Comp. Pair. I know that they are not Lateral types, and I presume that they are V types. The spec sheets say that they are for 'high power amplifier'. Anyway point is that I got something in a 2 stage PP to work with these. I substituted in IRF devices once the circuit worked and was not pleased with the results. I suspect that the bias was not correct for the IRF's. And I also suspect this because it seemed to be crossover distortion that I was getting (at least that's what I felt that it was). However, with the toshibas the sound was very similar to that of a Zen circuit but with some audible distortion in the extreeme high end (sounds like > 8K). Anyway, do a Google serch for the original 2SJ/2SK pair and you should find quite a bit of info on them as well as sources if dig hard enough. Hope this is of some help.

-D.
 
For the cascode function you definitely want the
parts with the 4 volt Vgs. As to using the IR type
parts for the gain devices, the 510's already are
fairly low in capacitance, and they are highly
degenerated with the Source resistance, so I
doubt you'll get great improvement with other
Mosfet structures. But then again, it's always
worth trying.
 
Hello,

I think thermal variations can be ignored if I heatsink the fets well. That I indeed will. Am I correct? ... I want to use TO-220AB on-board heatsinks (1.5X1/4X1.5 inch, Identical ones are used for SDS headphone amp.)

Do I have to thermal-couple some components?

So I guess it would work the way it is. Sound quality might be entirely different matter. I will have to see it for myself.

I will try to build it when I have come up with 4 trannies, 2 high current psu and 2 precision ones without cutting into my food budget. (Yeah, I could use Gate's pocket money any day.)

Tomo
 
Status
Not open for further replies.