Controlling input signal current loops in balanced circuits.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've been pondering a headphone amp for a while now, similar to a BOSOZ CCS'd. With a good CCS (I use a cascoded CCS) the output current loop is extremely well controlled. The input current loop is bugging me, however.

The normal resistor from drain to grid and reistor from grid to ground voltage divider to bias the mosfets creates an alternate current path from just below the load resistors to ground, circumventing the CCS. This puts the power supply into the input current loop.

Noise on the power supply should be totally rejected if you have a good CCS.

What bothers me though is that even if the input impedances are perfectly balanced, if the + and - inputs aren't perfectly level matched, you get input signal current running up through the drain-grid resistor, the load resistor, and to ground via the power supply.

I can't figure out if this is a fundamental issue or if there's a way around it somehow via an input transformer. At the moment I'm figuring on using a CCS-fed shunt reg on the power supply to close the loop and keep the power supply caps and such out. After going to the trouble of keeping the power supply out of the output loop, it annoys me that it still sneaks in.
 
Any CCS'd amp without split supplies really, look at the Zen v7 for instance. (Though a split supply doesn't really 'fix' it)

An externally hosted image should be here but it was not working when we last tested it.


The little boxes are IRF610's.

I had a second picture clearly showing the current loop I'm talking about, but imageshack.us suddenly stopped resolving... Anyway, if we drive only one of the inputs, say V6 (the left one), the input current loop goes through C3, R11... and then there are two parallel paths, one is R7, but there's a lower impedance path through R8, R3, and the power supply.


Dug up another picture of the X-BOZOS-CCS, and the same current loop is there too, through the upper power supply. In the 4.7k, then there's the parallel combination of.. straight to ground via 100k, and the lower impedance route of 47k to the output, through the 10uF coupling cap, then the 680 ohm load resistor, and to ground through the 10,000uF power supply cap.
 
I'll try to though I'm having trouble w/ imageshack at the moment

Here you can see the two parallel paths (one through a power supply) on the X-BOSOZ-CCS:
 

Attachments

  • xccs-bls.gif
    xccs-bls.gif
    13.3 KB · Views: 209
The one and only
Joined 2001
Paid Member
Thank you, much better. I think the issue you're grappling with
is Kirchoff's law which states that the sum of all currents must
be zero, and that current (electron flow) is not created or
destroyed. You have to consider the sum of all currents in this
system, and you have chosen just a sub-net.

The Drain-Gate resistance does indeed shunt off some of the
electrons that would go elsewhere, but the effect is linear,
small, and cancelled out on the other side. The primary thing
is that it is linear and small - it's effect is miniscule and doesn't
add distortion because it's linear. We could easily discuss
the amplifier in general as a system of alternate shunts which
create both amplification and control, after all, that's pretty much
what's going on anyway.

I think that after you live with it for a while, you'll realize that
you just have to look at it another way, and it will fall into place.

:cool:
 
Yeah, I've been thinking about it for most of the night, messing around with switchercad and such... with tubes (what I normally deal with) we don't need feedback resistors and the grid's awfully high impedance, so it's easy to ignore there.

The way the CCS-loaded diff circuits sort of float has always bugged me a bit, with biasing up the grids and where the input current flows. Just SEEMS that if you were looking at a tube amp and said 'The grid resistor is gonna be a 47K, then a 10uF cap, then a 680 ohm resistor, then a 1000uF electrolytic to ground and then parallel all that with a separate 100K' people would look at you funny.

As you said, I don't really see the 1000uF electrolytic's possible nonlinearities messing up the input signal or anything, just that going from input (cap or interstage xformer), single 47k resistor to ground, to grid of tube... to a much more complicated system of current loops, is bothering me. I like being able to keep the current loops very small and controlled, using CCS's to separate them from the power supply and such.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.