2 or 3 Channel, Single Ended Class "A" Design Thread

[Spacehead]

 

[Ant1Xr1st]

R2 must be connected to V+ not to the Lout

 

[Spacehead]

OK. I don't have any spare IRF510 so I would need to remove LM317 and then solder BD139 in place.
 
Maybe a bare resistor gives good sound and this is all pointless?

 

[Leny]

Ant1Xr1st, thank you for the amp link and translation.
 
Avro_Arrow, perhaps with 24V (hence + / - 12V) a short of Source to true 0V would place 12V across the FET limited only by the internal resistance (which I think is about 0.5 R for IRF510). So 12 / 0.5 = 24 amps, which is too high. Or, maybe the instantaneous 0V across Gate-Source simply turns off the FET fast enough and nothing flows, and all is OK? (I really must learn more about FETs). If you do have the inclination to try with one of your spare FETs it would be interesting to hear the outcome!
 
Spacehead, I think that you are going in the right direction, and as stated by Ant1Xr1st "R2 must be connected to V+ not to the Lout". Size R2 to get, say, 10 mA through D1. Then if the voltage drop across D1 is about 1.4V, the 0.7V Vbe drop of Q1 leaves you with 0.7V across the 6 Ohms of R6. Then by I = V / R your current is 0.7 / 6 = 0.117 amps.

 

[Spacehead]

I had help from a Finnish friend and he fixed the circuit for me. Here is working Szekeres VE with Transistor CCS
 

 
 
 
here is the lt spice .asc file
right click and save as
http://dl.dropbox.com/u/7337908/szekeresve_fix.asc
 
If you are interested, please try it Point-2-Point should be easy?
 
All six transistors need to be heatsinked. 

 

[Avro_Arrow]

That one looks much better.
One design note, you don't need C1 and C2.
Just one cap across the rails. 
Now all it needs is a way to "fine tune" the
current to balance the DC offset.

 

[Spacehead]

yeah it is cool, I was pretty close to getting it working but had ugly looking layout too
 
I have the biasing circuit left from original Szekeres on left and right channels and with that trimmer I can adjust the offset to zero. 
 
It is your turn to do that. Get LTspice and download that 
http://dl.dropbox.com/u/7337908/szekeresve_fix.asc
 
Quote:

That one looks much better.
One design note, you don't need C1 and C2.
Just one cap across the rails. 
Now all it needs is a way to "fine tune" the
current to balance the DC offset.

 
 
I might try to change the LM317 ccs to BD139 or MJE3055 transistors. I have plenty of BD139 so I could match them too, if their legs with into my dmm. (if they don't I need to build a testing circuit!)
 

 

[Avro_Arrow]

Here is BD139 as a current source at 120mA.
You can put a 500R pot across R2 on the L/R
channels to tune the DC offset. I don't have any
of these transistors laying around or I would try it.
 

 

[burgunder]

It's great to see some investigation into this area with all the sensitive cans around these days Grado, Audio-Technica, IEMs and so on all that most people need will be  a source follower to drive their heaphones.

 

[kvant]

For those uninitiated, it could be of interest to read about the origins of the Szekeres VE circuit on the RJM's site or on headwize. In that headwize thread, one can find an important note (post 62):
 
Quote:

...the main design issue with the original VE where the relatively high impedance of the ground channel is shared by both L and R channels, causing a certain amount of stereo enhancement. (yes enhancement... the effect is akin to anti-crossfeed.)

 
The output impedance of the source follower is approximately 1/gm, where gm is the MOSFET's transconductance. The datasheet for IRF510 that I have does not specify this parameter anywhere near the (small) drain current used here. But I guess that 1 siemens is still an optimistic values and gives the output impedance of 1 ohm. This looks very excessive for a virtual ground circuit, especially for low impedance headphones, and thus it does not seem like a particularly good way to remove the output coupling capacitor from the Szekeres buffer, which I believe was the original motivation.

 

[Leny]

I understand your point kvant. However it will pass ground noise and psu noise equally to both sides of the transducer, and hence null them, which is a significant bonus.
 
If you have any alternative suggestions for this circuit, or anything similar in concept or objective, then please post them as I (and I am sure many others) would be very pleased to read about them.

 

[G.Trenchev]

Can you tell any significant difference between the AC and DC coupled version of this amp?I'm using it with the caps for almost an year(with some modifications to increase PSRR).I've integrated it with a preamp module,and it's a complete amplifier.I'm using WIMA Polypropylene input and Panasonic FM low-ESR output caps.I like it,though it's quite  revealing with powerful in-front mids,slightly colored upper mids.It depends almost to the opamps used only.
I wonder if it can drive high-impedance headphones ?Never tried.

 

[Avro_Arrow]

I'll bet your version sounds quite good.
I think the "look Ma...No caps" version sounds different but
like everything, I'm sure different people will favor one or the
other for their own reasons.
You can "tune" the amp for different impedance headphone by
changing the Voltage/Current values. For low impedance, use
more current and less voltage. For higher impedance use
lower current and more voltage.
If you are building the traditional style with a resistor as the
current source, Rsource = Headphone impedance.
A general value of Rsource = 100R drives most cans well
enough.
 
If you want to lower the output impedance you can add
an op amp (or jfet or tube) stage in front and put the MOSFET
in the feedback loop.
 
Quote:

Can you tell any significant difference between the AC and DC coupled version of this amp?I'm using it with the caps for almost an year(with some modifications to increase PSRR).I've integrated it with a preamp module,and it's a complete amplifier.I'm using WIMA Polypropylene input and Panasonic FM low-ESR output caps.I like it,though it's quite  revealing with powerful in-front mids,slightly colored upper mids.It depends almost to the opamps used only.
I wonder if it can drive high-impedance headphones ?Never tried.

 

[Avro_Arrow]

Here is the process for building the amp to get the lowest offset
and most stable result.
You are free to use any or all of the steps you wish.
These steps apply to the IRF510/LM317 version.
 
First. Match your LM317.
Depending on how many of these you have to play with,
match them as closely as you can. If you have several
of them try and match them to .1mA.
Mount the 317 to a heat sink and use the same resistor (8R25)
for each test. Use the power supply that you will use with the finished
amp if you can. Just solder the resistor between out and adjust, connect
it to Gnd and connect the In to + with your amp meter. You should get about
152.2mA for 8R25 resistor.
 
Second. Match your IRF510.
Use one of the current sources for all of the tests.
I just soldered the Gate and Drain together and connect to +.
Connect the Source to the current source.
Measure the voltage between Gate and Source.
Mine were in a range of 3.5 to 4.2 with the closest three being 3.69, 3.70
and 3.71. I used the middle value for the Gnd channel.
 
Third. Pre-adjust your trim resistor to about 50R
This resistor in parallel with the 10R should equal 8R25.
Your left and right channels are built with this adjustable
resistor.
 
Fourth. Build your amp.
Mount all the 510s on the same heat sink.
Mount all the 317s on the same heat sink
as the 510s or on a different heat sink. As
long as all the devices of the same type are
all mounted together.
 
Power it up and check the DC offset.
Use the trimmers to adjust it to zero.
Now, plug in your source and check the offset again.
Any offset from your source can be zeroed out too.
 
Plug in your headphones and listen to your new creation!

 

[Avro_Arrow]

Here are a couple pictures of the ugly mess I made.
I threw this together from leftover and salvaged parts.
When I get some spare time I will make a nice
case for it.
 
Thanks for not laughing too much 

.