Marsh headphone amp from Linear Audio

Member
Joined 2008
Paid Member
Thanks. I wish I could find one as good as the Juli@ with a USB interface...

I'd be interested, too, but only in the same price range...

Jan Didden't do it. ;)

Yeah. Sure. He never did. :p


Here're some box shots FYI. I'm no fan of those Hammond cases actually, but they're cheap and easily modified.
 

Attachments

  • perfmarchbox.jpg
    perfmarchbox.jpg
    626 KB · Views: 680
  • perfmarchbox2.jpg
    perfmarchbox2.jpg
    451 KB · Views: 660
Disabled Account
Joined 2012
In 99.9% of the time there are two mistakes --

1. Did not use same part values... including matching of input jFETs. The thd will be high if not matched.

2. The transistor is misplaced in pcb holes... the e-b-c and d-g-s are not placed correctly in the pcb. often the they are reversed ... it will work but not well.... thd will be high. Check the transistor leads and their placement in pcb.

Check the input level being used for your tests.... may be too high?


THx-RNMarsh
 
Last edited:
Member
Joined 2008
Paid Member
Richard,

thanks for chiming in!

2. The transistor is misplaced in pcb holes... the e-b-c and d-g-s are not placed correctly in the pcb. often the they are reversed ... it will work but not well.... thd will be high. Check the transistor leads and their placement in pcb.

I'm pretty sure that this is not the case here. I've checked it several times now, consulting several datasheets and measuring the actual devices. Sure, there might still be an error somewhere, since I'm the one who made it and the only one who checked for it, but I doubt it.

Check the input level being used for your tests.... may be too high?

Input level was 1VRMS, checked with oscilloscope, attenuated via Alps to yield 1VRMS at the output.

1. Did not use same part values... including matching of input jFETs. The thd will be high if not matched.

This is likely the culprit. I just tested my remaining FETs (a handful [by count, not by volume :(])with the circuit you suggested.

The NFETs are 2SK246BL and in the range of 2.4V to 2.6V, which is really far away from the 0.85V you suggested!

The PFETs are 2SJ103BL and in the range of 2.0V to 2.8V, with one device (out of 5....) measuring only 0.62V!

I did not measure the devices currently used in the amp, but it seems to me that getting a big bunch off the 'Bay might be the best thing to do now.
 
I am away from home so I do not have my Linear Audio copy with me.
But I used the schematic image at the LA website to make up a Spice file for the unity gain version, just out of curiosity.
The frequency response indicates too much gain in HF (> 1MHz) resulting in oscillation or very marginal stability.
Load is 60 ohm resistive.

Is it only a Spice problem and not in reality ?
Or does it only apply to the version with unity gain ?
Anyone did any Spice simulation before ?

I am not posting the Spice files unless Mr. Marsh and Mr. Didden gave their approval.


Patrick
 
Member
Joined 2008
Paid Member
Putting 100pF each between Base and Collector of Q1 and Q2 (I used BC550/560) results in the green trace.
Sims well, but I didn't actually test it.

EDIT: 47pF should be sufficient, yielding a higher bandwidth. Anything lower still shows peaking.
 

Attachments

  • unity_march_02.png
    unity_march_02.png
    19.6 KB · Views: 658
Last edited:
Disabled Account
Joined 2012
Dont forget to include the stray C from pcb layout..... especially if going to use a ground plane/shield.

BTW -- Which jFET P/N was modeled and whose model was used.... they arent very accurate as a rule.

What does the SIM's show for THD?

THx-RNMarsh
 
Last edited: