Marsh headphone amp from Linear Audio

Disabled Account
Joined 2012
Let's see, 5 MHz bandwidth at a gain of 4X (12 dB), is a GBW product of 20 MHz. Almost as good as the FET input opamps Jan Didden used in his 1995 Super Regulator, the AD817 (GBW= 50 MHz) and the AD825 (GBW= 41 MHz). We need to remember that the ICs are voltage feedback amplifiers but the headphone amp is a current feedback amplifier.

These chips cost about five dollars each if you pay extra for thru-hole packaging: Mouser link

_


And, the point is? LOL :)

There are lots of IC VFA opamp circuits out there. Many even attempting use for HPA here in this DIYAudio HPA forum.

Try those two for your HPA and compare and let us know.


THx-RNMarsh
 
Last edited:
Yes I am laughing too, at the idea that a 20 MHz current feedback amplifier in 2012 is some sort of amazing achievement. LOL.

Well, you're right, it's not exactly a new concept that you can outperform a discrete amplifier with an IC composite amplifier. But, is that really a good reason to poke fun at this design? I'm not the most civilized poster here, but I generally try not to go into threads about specific designs and rag on it unless it's actually broken or dangerous.
 
Member
Joined 2011
Paid Member
Chris the design's okay, I'm not ragging on the design. Just giggling at the proud daddy who feels that his CFA achieving 5 MHz bandwidth at a gain of 4, is worth highlighting in bold text.

All my topologies are inherently linear, fast, wide bandwidth. Today, I got around to measuring the BW.

With the original transistors I tried, the BW = 5 MHz at -3 dBv
 
Disabled Account
Joined 2012
MJ... the amp is more than OK. had I thought BW was the #1 critieria for good sound, I would know what it's BW is from the beginning. not 5 years after publication.

Inherently, linear circuits withOut using hundreds of DB's of gNFB and done with only 8 transistors and having vanishingly low THD, if you wish to spend the time, driving low Z loads. Is a very nice accomplishment. If you add-in the EQ'ed sound from my article, you will get a bonus in sound quality and accuracy.

One other nice thing about such topologies is how stable they are. Without any tricky phase shifting/comp technique to keep the amp stable.




THx-RNMarsh
 
Last edited:
MJ... the amp is more than OK. had I thought BW was the #1 critieria for good sound, I would know what it's BW is from the beginning. not 5 years after publication.

Inherently, linear circuits withOut using hundreds of DB's of gNFB and done with only 8 transistors and having vanishingly low THD, if you wish to spend the time, driving low Z loads. Is a very nice accomplishment. If you add-in the EQ'ed sound from my article, you will get a bonus in sound quality and accuracy.

One other nice thing about such topologies is how stable they are. Without any tricky phase shifting/comp technique to keep the amp stable.

:cheers:


THx-RNMarsh
 
Disabled Account
Joined 2012
When I invented this particular fully complimentary push-push topology, now called current feedback, the CFB circuits evolved into single gain stage, using current-mirrors. My original CFB amp. topology, used here for the HPA, was later produced as a fully complimentary two-gain stage IC op amp as a AD8011 (1995) and others in that family.

I havent tested it or measured the THD/IM but it deserves a look-see one day.



THx-RNMarsh
 
Last edited:
Disabled Account
Joined 2012
Yes I am laughing too, at the idea that a 20 MHz current feedback amplifier in 2012 is some sort of amazing achievement. LOL.

If you were around in the late 70's when i first developed and published this topology, I used all bipolar and I did not have a signal generator with wide enough bandwidth to tell. BUT, i did have a scope and a fast Tr square wave. I measure 20nS Tr. Then I measured the gen Tr and it was about 20nS also. So, it was faster than 20nS and very wide bandwidth for sure. Again with no instability issues. The THD+N was below the HP339A would test. This was CF mode of operation. It was a line stage for my preamp. As the load Z didnt need to be as low as this HPA, I did it with only 6 transistors 50 years ago.

You are a little late to the game. "Those who laugh last, think slowest"

:)


THx-RNMarsh
 
Last edited:
Member
Joined 2011
Paid Member
Have some people at Lawrence Livermore explain the "risetime bandwidth equivalency" relationship to you. They can start with the Wikipedia article.

You will discover that 20 ns risetime corresponds to a bandwidth of 17.5 MHz. Congratulations, when going from FETs to bipolars your bandwidth only dropped 12%, from 20 MHz to 17.5 MHz.

At that time in history, Elantec and Comlinear were shipping all bipolar Current Feedback Amps whose bandwidth was 80 MHz. Soon National and AD would enter the fray with even greater bandwidth at even lower power.
 
Disabled Account
Joined 2012
Yep. Like I said, it was the same Tr as the generator. So i dont know how fast it actually was. 20ns is not its Tr. It was faster. With the devices I used 50 years ago, it should be near 100Mhz.

If you read back -- you will see that i preceeded them in CFB amp topology and in fact the designers called me on the phone to discuss it when i was at LLNL... before they completed their design because they had been working on similar circuit and was about to go to patent app when my article came out. This is well before any IC shipments. They got my phone number from Ed Dell (TAA) and called me at my office at LLNL. They were really upset.

They modified the circuit slightly and got patent. They added a diode to make 2nd stage a current mirror. This is all old news... i have described this before in detail and in publication. Now the speeds using CMode topology are in GHz range with MOSFET etc.




THx-RNMarsh
 
Last edited:
I see that Mouser has all the semi's in thru hole.
Does anyone have a pcb for this design?
I was thinking of making one, I will buy the article from LA.

I've been working on one. It is self contained with a wallwart power supply. I designed it to slip into a 100 x 160mm Hammond extruded aluminum enclosure. I haven't actually built one yet but expect to soon.
 

Attachments

  • MARSH_00.JPG
    MARSH_00.JPG
    411.8 KB · Views: 412