Go Back   Home > Forums > >

Headphone Systems Everything to do with Headphones

Rush Cascode Headphone Amp + JLH Output Stage
Rush Cascode Headphone Amp + JLH Output Stage
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools
Old 24th January 2010, 06:02 AM   #1
keantoken is offline keantoken  United States
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Default Rush Cascode Headphone Amp + JLH Output Stage

Hello everyone.

I've spoken wbold things, things of dreams, things that never happened...

Well this time I did something that actually worked.

I've always been fascinated with the Rush Cascode. It has interesting characteristics that distinguish it from the LTP, but basically it's the same idea, an inverting and noninverting input. The JLH output stage was chosen because it had the lowest impact on the harmonic signature. This is a class A design.

It was named after Christopher Rush, the one who used it in a NAD preamp some time ago. But the name is misleading, because (trusting the experts here), it's not really a cascode. I've ventured to give it a new name which is more to the point. So I've decided to call it the NTP, standing for "No-tail Pair".

Many experts would have the NTP banned from amplifiers altogether, but this is audio; we're given some slack as long as it sounds good. Sometimes it even sounds better.

The three interesting things about the NTP are:

1: It has a exponential transfer function, as opposed to the LTP which has a parabolic (looks like it anyways) one. This means:

a: It does not cancel even harmonics, which instead balance the odd harmonics. (in the present circuit, odd harmonics are actually canceled to an extent by Q7 and Q6)

b: It can deliver more power on positive strokes. This isn't really that great, because by this time the negative strokes are clipping... It may be better than the alternative though, the LTP, which clips both ways and is thus current limited both ways (and distorts both ways too).

2: It doesn't need a current source like the LTP does; this makes it faster at HF and there is less loss as well. The LTP is not as suited for HF work (I've simulated several HF buffers)

3: You don't need to worry about balancing the transistors! YAY! (well, this could be said to be a bad thing, since in a conventional LTP balance will decrease its distortion)

4: Contrary to popular belief, the NTP is does NOT necessarily produce more distortion than the LTP. Since the NTP is faster, BW is increased, and the OLG BW can be increased, which decreases HF distortion. Or you can go the other way, and increase open-loop gain, which will decrease LF distortion at the expense of relative HF distortion.

The largest hurdle with implementing the NTP has been DC offset; on its own, the rush cascode has 1.2V between its two inputs. This is a serious problem and is hard to correct without being clever. If you manage to do that, temperature drift might or might not be a problem.

The presented solution I don't think can be beat in terms of temperature drift cancelation, offset correction, ease of use and component number. An interesting note is that the buffers Q7 and Q6 even increase speed. Offset can easily be corrected via a single potentiometer.

My reasons for designing this amp were:

1: To design a good amp using the NTP, for others to try out and see whether it works well; to see what the consensus was on its value in audio amplification.

2: To see if using the NTP is practical in real life.

So far I have been blown away. Not that it means much, because my best source has only been my Soundblaster X-fi with Koss Portapro heasphones. But what that DOES mean, it that it's not bad, and better than most consumer equipment in the low-$ range. Unfortunately, my bench is still young and I don't have sophisticated signal sources for measuring distortion, BW... This is where I would like others to try.

The first difference I noticed was between the snares and high frequencies, and brass instruments; they sound more like they are supposed to. I can discern between a clap and a snare. It also sounds much more dynamic. I can discern this through the loud hum; I have a ground loop between my power supply and soundcard source (NOT a design flaw, only a testbench problem). But it sounds good enough that it doesn't really bother me, although eventually it will have to be fixed.

The current design is the one that is on the bench. I had stability problems using MJE340/350 outputs, but I decided I could do better with component choices and went with a few 669A's I had laying around from a computer monitor. So far, there have been no signs of instability.

For low DC offset it is important that Ic for Q7 and Q6 is equal to the Ic of Q5 and Q8. Offset at the moment is 40mV, because I haven't added the adjustment trimmer. To add offset control, place a trimmer resistor between the bases of Q7 and Q8 (you may have to decrease R2 and R3 to give it enough current).

I think many will see C2 and scratch their heads. This is certainly NOT the normal method of stability compensation. So how does it work?

Well, first of all, C2 is dependent on C1. As frequency increases, C1 draws more current, which causes the Vbe modulation of to increase, which causes C2 to suck up more current. This combination allows us to create a quick decline of open-loop bandwidth, rather than a slow one that spans the audio spectrum. So we get less distortion in the 5kHz area. (an attachment shows C2's affect on OLG).

You might also ask why C1 is connected directly to the output instead of Q3's collector, as is standard. Well, it is better this way if you can get away with it, because otherwise strange harmonics from Q1's Vbe will be injected into the feedback loop - this a much worse issue in class AB and B amplifiers, because it adds a lot of transient trash to the feedback loop, causing harshness, brightness, HF distortion... (perhaps the idea of Cdom in itself is not what is harmful to Blameless amplifiers as some say - rather, its implementation could be better?)

So, now that I've got it working, I want others to build it and see for themselves of my own observations matches up. To an extent they come from simulation; so I might not be totally right.

The current source I1 is a standard 2-transistor CCS with a 20 turn trimmer.

I have pictures and the schematic attached.

While visions of sugarplums dance in his head,
- keantoken
Attached Images
File Type: png Rush_Headamp.PNG (26.5 KB, 2803 views)
File Type: png Rush_Headamp_OLG_C2.PNG (30.2 KB, 2612 views)
File Type: jpg DCP00667.JPG (384.9 KB, 2497 views)
File Type: jpg DCP00668.JPG (336.6 KB, 2227 views)
__________________
The Kmultiplier rail filter! -=- The Simple Kuartlotron Superbuffer!
Need something built, repaired or modded? PM me and ask what I can do!

Last edited by keantoken; 24th January 2010 at 06:11 AM.
  Reply With Quote
Old 24th January 2010, 06:37 AM   #2
keantoken is offline keantoken  United States
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Here is an FFT plot of the output spectrum. Taken at 1KHz, 1V pk-pk.

At higher voltages, 3rd harmonics are canceled. If you want stop this and minimize distortion (which then will be very low, possibly inaudible), you can replace R2 and R3 with Jfet CCS's. As it is, it is kind of my experiment on whether dominant 2nd harmonics sound good as opposed to dominant 3rd harmonics.

Harmonics decrease very quickly after the 2nd and 3rd; this indicates that multiple transfer curves with different parent functions aren't interfering with each other. Instead, the output is a single exponential function, except for the 3rd harmonics which are canceled.

Simulations are made with about 200mA class A bias. Guessing by the heat, I think I've been listening at about 100mA.

The music I used to compare is here. I used the tracks "Nocturne", "Empty Frame feat", and "Heart of Vlad".

http://baddudesmusic.com/symphony.html

Also, I have to clarify. I didn't use the Portapros, I actually used my old in-ear headphones that I modified, which are more detailed. I don't have two amps so I couldn't do stereo, so I connected right and left together, which presents the headphone amp with about a 9 ohm load.

- keantoken
Attached Images
File Type: png Rush_Headamp_FFT1kHz_1Vpkpk.PNG (35.5 KB, 2079 views)
__________________
The Kmultiplier rail filter! -=- The Simple Kuartlotron Superbuffer!
Need something built, repaired or modded? PM me and ask what I can do!

Last edited by keantoken; 24th January 2010 at 06:42 AM.
  Reply With Quote
Old 24th January 2010, 06:54 AM   #3
keantoken is offline keantoken  United States
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Here is an FFT plot of the output spectrum.

At higher voltages, 3rd harmonics are canceled. If you want stop this and minimize distortion (which then will be very low, possibly inaudible), you can replace R2 and R3 with Jfet CCS's. As it is, it is kind of my experiment on whether dominant 2nd harmonics sound good as opposed to dominant 3rd harmonics.

Harmonics decrease very quickly after the 2nd and 3rd; this indicates that multiple transfer curves with different parent functions aren't interfering with each other. Instead, the output is a single exponential function, except for the 3rd harmonics which are canceled.

Simulations are made with about 200mA class A bias. Guessing by the heat, I think I've been listening at about 100mA.

- keantoken
Attached Images
File Type: png Rush_Headamp_FFT1kHz_1Vpkpk.PNG (35.5 KB, 587 views)
__________________
The Kmultiplier rail filter! -=- The Simple Kuartlotron Superbuffer!
Need something built, repaired or modded? PM me and ask what I can do!
  Reply With Quote
Old 24th January 2010, 09:01 AM   #4
keantoken is offline keantoken  United States
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Sorry, I made an incredible mistake on the OLG plot in my first post... This should make more sense...

- keantoken
Attached Images
File Type: png Rush_Headamp_OLG_C2.PNG (30.1 KB, 484 views)
__________________
The Kmultiplier rail filter! -=- The Simple Kuartlotron Superbuffer!
Need something built, repaired or modded? PM me and ask what I can do!
  Reply With Quote
Old 24th January 2010, 11:34 AM   #5
revintage is offline revintage
diyAudio Member
 
Join Date: May 2007
Hey keantoken,

In order to make your circuit more understandable, please add the *.acs . Also where is the NTP in your schematic?

Last edited by revintage; 24th January 2010 at 11:44 AM.
  Reply With Quote
Old 24th January 2010, 09:41 PM   #6
keantoken is offline keantoken  United States
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
The NTP is Q5 and Q8.

I've attached the simulation file here.

I am going to switch to battery power to get around grounding problems.

- keantoken
Attached Files
File Type: txt Rush_JLH.txt (5.4 KB, 211 views)
__________________
The Kmultiplier rail filter! -=- The Simple Kuartlotron Superbuffer!
Need something built, repaired or modded? PM me and ask what I can do!
  Reply With Quote
Old 24th January 2010, 10:07 PM   #7
AKSA is offline AKSA  Australia
diyAudio Member
 
Join Date: Sep 2001
Location: Melbourne, Australia
Anthony,

It's a very clever circuit, like the use of EFs to level shift!

Keep up the outstanding work,

Hugh
__________________
Aspen Amplifiers P/L (Australia)
www.aksaonline.com

Last edited by AKSA; 24th January 2010 at 10:10 PM.
  Reply With Quote
Old 24th January 2010, 10:13 PM   #8
Wavebourn is offline Wavebourn  United States
diyAudio Member
 
Wavebourn's Avatar
 
Join Date: Sep 2006
Location: Pleasant Hill, CA
I like it.
__________________
!!!Warning!!!
Single Ended Class-A-Holism is addictive!!!
  Reply With Quote
Old 24th January 2010, 10:53 PM   #9
keantoken is offline keantoken  United States
diyAudio Member
 
keantoken's Avatar
 
Join Date: Aug 2006
Location: Texas
Thanks guys!

The circuit looks and IS very, very simple, but I have been thinking it over for months and months, exploring many options and I don't think it can be beat in terms of performance vs. part count. (at least for the same topology of design)

- keantoken
__________________
The Kmultiplier rail filter! -=- The Simple Kuartlotron Superbuffer!
Need something built, repaired or modded? PM me and ask what I can do!
  Reply With Quote
Old 24th January 2010, 11:19 PM   #10
Wavebourn is offline Wavebourn  United States
diyAudio Member
 
Wavebourn's Avatar
 
Join Date: Sep 2006
Location: Pleasant Hill, CA
Quote:
Originally Posted by keantoken View Post
Thanks guys!

I don't think it can be beat in terms of performance vs. part count.
Contest?
__________________
!!!Warning!!!
Single Ended Class-A-Holism is addictive!!!
  Reply With Quote

Reply


Rush Cascode Headphone Amp + JLH Output StageHide this!Advertise here!
Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
JLH Headphone Amp ashok Headphone Systems 859 2nd November 2020 06:47 PM
The Rush Cascode: Possible Wiki page keantoken Solid State 8 10th January 2017 07:38 AM
JLH Headphone amp aethelbert Headphone Systems 4 14th July 2009 12:14 AM
cascode output stage what device? Stefanoo Pass Labs 24 27th February 2009 04:35 PM
JLH output stage sajti Solid State 20 21st January 2008 08:25 PM


New To Site? Need Help?

All times are GMT. The time now is 05:50 PM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2021 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.00%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2021 DragonByte Technologies Ltd.
Copyright ©1999-2021 diyAudio
Wiki