Marsh headphone amp from Linear Audio

Jump to Latest
#1,042
pchw said:
If the reference is to page 37, figure 8, then the above advise is invalid. There is no 'R6', R5 is 1K to ground, don't think R4 should be this value. There is another labeling error in the schematic, there are 2 R3's, one below J1 and one below J2.

If I read it correctly, P1 is there to balance the unmatching of J1 and J2. The unknown J3 and J4 can be something like 50R each. Remember they in are in series to each side of P1 anyway. FWIW, Jack's PCB labels both as 51R. In theory, if you have perfectly matched J1 and J2, then you can use 2 perfectly matched 500R for R3 and R4 and get rid of (jumpered) P1 🙂
Click to expand...

Thanks, the errors were pretty confusing 😀
 
#1,044
jackinnj said:
if you have to regulate (as Dick has mentioned, more juice is better, so you are trading off sound quality vs the regulator's overhead), I would much prefer you use LT1963A and LT3015.
Click to expand...

Hi Jack, thanks for Your comments. Regarding juice&regulation, I plan to use trafo with 2X18 secondaries connected in serial for one rail, so I will have plenty of juice (2X36AC) to spend 🙂. Probably overkill but why not. I prefer to have caps multiplier on board as a last power stage to vacuum all HF junk.
 
#1,047
sikahr said:
Thanks Jan, food for thought.

For how much rail current varies with output current in this circuit?
Click to expand...

Depends on the Zout of the cap multiplier, which depends on the device and standing current and such.
According to Ohms law, for every ohm Zout, 1 mA load current generates 1mV ripple.
Most cap multipliers have several ohms Zout up to 10 or more.

Jan
 
Last edited:
#1,051
sikahr said:
Sorry Jan, maybe I put wrong question. I mean:
For how much rail current varies with input signal in this circuit?

I had an impression this circuit works mostly in class A (maybe I am wrong)?
Click to expand...

Depends on the load you draw. The output voltage across the load requires a current, which has to come from the supply. Know the Vout and Zload and you know Isupply (talking about the signal only of course, not DC).

Jan
 
#1,052
I happened to have a Spice model of a cap multiplier on hand.
At 15V 150mA DC bias, and +/-100mA 1kHz AC current draw, the ripple is 70mV pk-pk.
So that works out to be 0.35 ohm ?
Maybe 0.5 ohm if DC bias reduced to 100mA.....
But not quite 10 ohms.

Output transistor is BC550/D44H11 Darlington.
You'll kill the BC547 with 100mA.

😉


Patrick
 
#1,053
EUVL said:
I happened to have a Spice model of a cap multiplier on hand.
At 15V 150mA DC bias, and +/-100mA 1kHz AC current draw, the ripple is 70mV pk-pk.
So that works out to be 0.35 ohm ?
Maybe 0.5 ohm if DC bias reduced to 100mA.....
But not quite 10 ohms.

Output transistor is BC550/D44H11 Darlington.
You'll kill the BC547 with 100mA.

😉


Patrick
Click to expand...

OK so maybe it's 0.5 ohms or even less. Still two orders of magnitude more than a good reg with remote sensing. My point is that I can't see why I should build a super-duper reg and then negate it with a cap multiplier after it.

Jan

PS Never ever blew up a BC547 in the sim yet; don't expect to either 😉
 
#1,056
But if you LP filter that twice and put it to a low noise emitter follower, you will only have the noise of the transistor.

I think John uses something similar in his commercial products.

Plus that at least for the F5 headamp, someone did have stability problems with a high NFB regulator (Salas shunt).
Not so with a cap multiplier.


Patrick



Patrick
 
#1,057
EUVL said:
But if you LP filter that twice and put it to a low noise emitter follower, you will only have the noise of the transistor.
Patrick
Click to expand...

Yes but I don not believe that it would be lower than the noise of a good reg. You can use the same low noise parts in either. You would increase signal related ripple in the hope that the noise would be less - well I know what I would chose!

EUVL said:
Plus that at least for the F5 headamp, someone did have stability problems with a high NFB regulator (Salas shunt).
Not so with a cap multiplier.

Patrick
Click to expand...

Quite possible - the signal related ripple may, depending on phase, counteract instability if it was a result of coupling through the supply lines.

So JC uses it, right? Well that would make me suspicious 😉

jan
 
#1,059
EUVL said:
Yes, he uses a 317/337 followed by RC filtered follower.
Me too, for certain applications where PSRR is good.

I guess we shall always agree to disagree on this ..........
🙂


Patrick
Click to expand...

No I don't necessarily disagree for such a case - if PSRR is already very good, you get into the realm of diminishing returns, agree. And then a 317/337 combo with a cap multiplier will be perfect.
It's just that I don't think it is a good move to first build a superreg and then follow it with a cap multiplier.

Jan
 
You must log in or register to reply here.
Top Bottom