From EDN: Current source makes novel Class A buffer deleted

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

clintrubber

Well-known member
Joined
Jun 3, 2004
Messages
5,982
Location
The Netherlands
I saw that one... (the EDN-e-subscription that you can download the whole issue or parts as pdf is quite nice)

IIRIC they didn't mention any advantages over existing approaches, if these had been possible it would have been more convincing.
 
Interesting.

The expected Vb-e change with load current in Q1 is mitigated somewhat by an opposing reduction in the voltage drop across R1 attributed to the reduced Vb-e of Q2 as Q1 draws more base current resulting in less current going to Q2. This improvement is in the right direction, I don't know how the I bias and value R1/2 impact the sensitivity of this correction voltage. If the beta is constant this might give a rough first order correction.

The log of 1+X is not the same as -(log of 1-X), but perhaps close enough for modest changes to be beneficial.

For high impedance line level applications the temptation is to make the class A current high wrt load current. It might be interesting to see if that out performs dialing it in for max correction at some lower current density.

This looks like something you might do inside an IC if the correction tracks adequately over a usable range. Kind of complicated for discrete use. IMO.

JR
 
Actually the input Z is much higher than it looks at first glance. This is in fact the Leach/Rossiter MC preamp without a floating supply and strategic capacitors. The concern I had with Rossiter's aXp article was in fact his analysis of the input Z of a stage fragment, which did not take into account the floating supply. When you do it a la Leach the input Z is low, as supposedly desired for his application. Here it can be ~ few megohms, depending on the choice of R1 and R2 and bias. EDIT: and the output loading.

As far as what the advantages may or may not be, I'm not really sure. My suspicion is a pair of opposed complementary feedback pairs, although a bit tricky to bias, would outperform it. Note that this circuit's output Z is not particularly low, really only about R1 || R2.

EDIT: Distortion does look quite low for small signals and light loading. If you cap-couple across each of R1 and R2 you can get the output Z down, although the distortion rises.

This arrangement is basically a reduced complementary Schlotzhauer stage pair. Although I wouldn't recommend using that as a pickup line in a singles bar.
 
> What's the input impedance....

Given in the article explicitly, tho there is a typo.

"The source driving the buffer stage sets {sees} hFE(Q1) * (R1 + RLOAD) ohms, where hFE is forward-current gain."

If pulled into class B, Q2 or Q3 reduces to a diode, asks the source to drive the load directly, and may take some time to come out of saturation.
 
[quote author="PRR"]If pulled into class B, Q2 or Q3 reduces to a diode, asks the source to drive the load directly, and may take some time to come out of saturation.[/quote]

Definitely a drawback, if things come out of class A.

The distortion performance is interesting. Very low, but energy at a lot of harmonics, which I thought initially was just spice numerical problems.

Another circuit along the same lines of derivation from current generators takes two complementary Wilson mirrors (replace R1 and R2 with diode-connected transistors) in push-pull. But now, although the distortion is low with a low source Z, the input Z is fairly low at about 4 times Rload. But the distortion is very low at harmonics higher than third.
 
The aXp Rossiter with assistance from Cherry version is closer to the first thread-linked schematic. I scanned the article for another forum member but I don't think it's appropriate to post it. I'll send on PM request.

Another circuit reminiscent of these is the Hawksford output device error correction topology. I don't have that at hand at the moment.
 
Back
Top