> But when it comes to the headphones, that's likely all we need.
Output impedance for 4 pairs is ~4.5 ohm.
Max Class A output current is ~ 80mA when using V grade.
For 300~600 ohm phones, for sure.
Will probably also drive 30ohm phones, but it will lack authority.
Nelson used 250 pairs to drive speakers in his beast.
So you can scale from that.
Patrick
Output impedance for 4 pairs is ~4.5 ohm.
Max Class A output current is ~ 80mA when using V grade.
For 300~600 ohm phones, for sure.
Will probably also drive 30ohm phones, but it will lack authority.
Nelson used 250 pairs to drive speakers in his beast.
So you can scale from that.
Patrick
14 years ago, I did something similar with my preamp output stage (intended for 200R load min).
Today, with such limited supply of JFETs, I would save an octet for some more interesting applications.
(A Blowtorch like nalanced circuit comes to mind...)
And there are better alternatives for headphone buffers that only use readily available components.
But I have to admit it is still fun to try, if you have plenty of octet JFETs to spare.
Patrick
Today, with such limited supply of JFETs, I would save an octet for some more interesting applications.
(A Blowtorch like nalanced circuit comes to mind...)
And there are better alternatives for headphone buffers that only use readily available components.
But I have to admit it is still fun to try, if you have plenty of octet JFETs to spare.
Patrick
Attachments
That's a relatively poor measure of Vp because the FET is in the subthreshold region. It is better to measure Vgs at two currents one above subthreshold like 200uA and one below Idss and use the FET equations given for Id vs Vgs. Two measurements two equations and solve for Vp and Idss.
All the FET's in question deviate from the ideal equation at high and low Id.
IMO Borbly leaves out the most important graph gm vs Id which makes the point that if you set the operating current the gm (and noise) is the same for all Idss'.
Attachments
Last edited:
You don't need octets, just N/P pairs in parallel.
Edit, each JFET should have it's own source resistor.
Last edited:
I was not able to match octets. I am short of suitable idss p-channel jfets.
Take a look at this :
https://www.diyaudio.com/forums/pass-labs/128571-source-follower-configurations-10.html#post5147541
(Post #473)
Patrick
Well, while I have no overlap of idss for k170 and j74, I did find out I have large quantities of j113 and j176 complementary pairs. It was easy to match octets. So I build the circuit, exactly as posted. Good news is, it works and sounds good. Bad news is, it has ~200mV dc offset. While I was using dual regulated supply set to +/_ 3V, I was able to adjust different voltages on positive and negative side to bring it little lower. But still there is few tens of mV. I may need to put cap on output in order to use it with better headphones.
The board you held for that photo was 1/2 of a channel.
The Beast has over 1200 Jfets per channel.
After I put cap on the output, I listened to the circuit with my best headphones, and I must say it sounds great. I am thinking about scaling it up. I have few hundred of J113 and J176, so perhaps I should be able to find a dozen of each with similar Idss.
The performance may not be as good as Nelson's with k170/j74, but if it sounds good, it is good.
The performance may not be as good as Nelson's with k170/j74, but if it sounds good, it is good.