Bob Cordell Interview: Error Correction

Jump to Latest
jan.didden

jan.didden

AX tech editor
Joined 2002
linearaudio.net
Paid Member
traderbam said:
Jan, I feel the penny is going to drop very shortly...so let's persist.

Disagree.
I am saying that figures 1, 2 and 3 (below) are equivalent. Fig 1 is H.ec as shown in his model. Fig 3 is a normal NFB system.

If they are all the same, if fig 1 can be nulled, so can fig 3.

No.
Click to expand...


OK, getting closer. If, in your figs 2 and 3 in this quoted post, you give the global nfb block the value 'beta'.
Then, the remaining 'a' s should be called '1/N' to be equivalent with my fig that you quoted. Agreed?

Jan Didden
 
B

Bob Cordell

Member
Joined 2006
cordellaudio.com
traderbam said:
Bob wrote: Yes, that would be useful. I want to try to mimmick your results as closely as possible. By the way, which of the resistors did yout tune if any? I notice in your paper that you suggest a resistor in the PFB loop, between the junction of the 330 ohm resistors and the 27 ohm resistors. I'd like to make the LF loop gain comparable...I realise this may be tricky unless you've measured it in the specific circuit you built. Based on your THD improvement measurement I'd guess the loop gain of the EC was about 30dB at 20kHz.
Click to expand...


Hi Brian,

Here's some more detailed info on the actual MOSFET amplifier prototype I built and measured. I needed to actually go back and look at the old board.

TRANSISTORS:

Q1,2: NPD5566
Q3: 2n4401
Q4,5: MPSA06
Q6,7,8,9,10,11: 2n3906
D2,3: 1n4148
Q12,13: MPSU57
Q14,15,16: 2n4401
Q17: MPSU07
Q18: MPSA56
Q19: MPSA06
Q20: MPSU07
Q21: MPSU57
Q22: MPSU06
Q23: 2n3906
Q24: 2n3904
Q25: 2n3906
Q26: MPSU07
Q27: MPSU57
Q28: IRF132
Q29: IRF9130

Note that EC transistor Q22 is on heat sink for thermal compensation.

Note that the EC transistors are not as fast as they could be, especially Q22.

R38,39: 26.1
R40,41,42,43,44,45: 261
trim pot between R38-39 and R44-45 typically set at 28 ohms
R47 implemented as a pair of 147 ohm resistors in series
(test point at junction of them.)

Cheers,
Bob
 
traderbam

traderbam

Member
Joined 2002
OK, getting closer. If, in your figs 2 and 3 in this quoted post, you give the global nfb block the value 'beta'.
Then, the remaining 'a' s should be called '1/N' to be equivalent with my fig that you quoted. Agreed?
Click to expand...
Disagree.

In my diagrams (all three of them) the input to N, let's call it Vn, is given by:

Vn = (Vi - a.Vo)/(1 - a)

In your diagram,

Vn = Vi + (1/A - B).Vo

I don't see how to make these the same.
 
andy_c

andy_c

Banned
Joined 2003
andycpublic.50webs.com/
lumanauw said:
What happened if the EC tries to compensate for the nature rolling down response of the output stage? I think it is very possible, because the EC transistors usually smaller size (have much higher fT) than the output transistors.
Click to expand...

You're absolutely right. If you refer to the schematic in this post you'll see the single capacitor to ground from the junction of the two resistors in the driver stage. So instead of subtracting the output stage's output voltage from its input, the output is subtracted from a low-pass filtered version of its input. If that capacitor is not in place, you can look at the frequency response from the circuit's input to the gate of one of the output devices and see it rising as frequency increases, just as you said. What I do is just increase the capacitor until it is just large enough so there is no rise in this response. That seems to result in a nice stable system according to the sim.

The idea is to have this low-pass filter approximate the frequency response of the output stage as closely as possible so that what you are saying doesn't happen
 
andy_c

andy_c

Banned
Joined 2003
andycpublic.50webs.com/
lumanauw said:
In Bob Cordell's schematic, there is C10 (39pF) from R38-R39 junction to ground. Is this C works the same as your C3?
Click to expand...

Yes, that's it.

Since you have made amp with Hawksford EC...
Click to expand...

I haven't made one :). I was working on the design of one but couldn't get the sim results to come out right, so I gave up. I am working on the design of one now again.
 
traderbam

traderbam

Member
Joined 2002
Andy_c wrote:
When I do a sim of my circuit with a 2 uF load, this overshoot also appears and is cured the same way, with an input LPF. With the LPF, its square wave response into a 2 uF load looks like a single-pole circuit - no overshoot, no ringing.
Click to expand...
I infer that your o/p stage overshoot does not change much when you add the 2uF to the 8 ohm resistor. Which is very good. The overshoot you get (you didn't say how big it is) will be something outside of the EC feedback loop.
 
mikeks

mikeks

Disabled Account
Joined 2004
en.wikipedia.org/wiki/Censorship
andy_c said:


Mike, I don't think you need those current mirrors. For each diff amp, you can take one collector and hook it up to the parallel RC. Then you can take the other collector and hook it up to the emitter of the emitter follower at the input. Notice this results in a constant current of the emitter follower, even as the error correction current varies.

Then you might imagine floating the diff amp current sources so they're bootstrapped to the output voltage somehow. This may be more practical with FET output devices, as the bipolars develop so little voltage drop that VCE crowding of the error correction transistors or their current sources could result.
Click to expand...


Hi Andy,

As you've noted regarding your arrangement: how does one accommodate the headroom required by the diff. pairs without over-biasing the output stage, especially if an all-BJT output stage is used?

I suspect, for a start, returning the tails of the diff. stages to the supply rails, and, perhaps, adding a resistor to connect the emitters of the complementary followers, between which the diff. pairs are sandwiched, to better define the followers' standing current.

What think ye? :scratch2:
 
traderbam

traderbam

Member
Joined 2002
Bob,
I can't find LTSpice models for all of the original parts. I can't find datasheets for the MPSU types.
Here are my substitutes at the moment.

Q20: MPSU07 -> MPSA06
Q21: MPSU57 -> 2N3906
Q22: MPSU06 -> MPSA06
Q23: 2n3906
Q24: 2n3904
Q25: 2n3906
Q26: MPSU07 -> MJE15030
Q27: MPSU57 -> MJE15031
Q28: IRF132 -> IRF9410
Q29: IRF9130 -> IRF9Z24S

I've done some measurements to find close equivalents to the output FETs. These two are in the standard LTSpice library. I think these are close enough.

I am more concerned about Q26 and Q27 because I have no idea what their characteristics are. If you happen to know what close equivalents may be that would be helpful. But don't go to too much trouble because this is icing on the cake.

Brian
 
mikeks

mikeks

Disabled Account
Joined 2004
en.wikipedia.org/wiki/Censorship
darkfenriz said:
Andy and Mike
LTP in Hawksford's correction is no better than original single bjt. Look carefully: predriver's emitter is a part of correction circuit's collector load. Then, if correction is a sigle bjt it kind of 'diode-linearizes' itself.
Adam
Click to expand...


Not really; look more carefully:
 

Attachments

  • hawksford_nfbec.png
    hawksford_nfbec.png
    75.3 KB · Views: 346
andy_c

andy_c

Banned
Joined 2003
andycpublic.50webs.com/
mikeks said:
Hi Andy,

As you've noted regarding your arrangement: how does one accommodate the headroom required by the diff. pairs without over-biasing the output stage, especially if an all-BJT output stage is used?
Click to expand...

I was thinking in terms of MOSFETs, so I'm not sure what would happen with bipolars. In my referenced schematic, the DC Vcb of each diff pair will be about 8V. If you set up a sim and try it, you'll see that's way more than needed. This also leaves 8V for the current sources and that's plenty too. If the output stage transconductance is large enough, the voltage across the diff amps' collector resistors will consist (approximately) of only DC plus distortion, so the dynamic range requirement is less than one might think. In the sim I'm looking at now, the diff amps can only swing +/- 4V, and this is more than enough to accommodate an output that's swinging +/-80V, even with a 4 Ohm load.

I suspect, for a start, returning the tails of the diff. stages to the supply rails, and, perhaps, adding a resistor to connect the emitters of the complementary followers between which the diff. pirs are sandwiched.
Click to expand...

Yikes! I'm not sure how I'd adapt this to bipolars to tell the truth. Bipolar output stage with MOSFET drivers maybe? That's what John does in his big amps apparently. I'm not sure why you'd want a resistor between EF emitters though. If you bias the diff amps hot so that the collector resistors are small, you may find them pulling too much current through the EFs as is. Neglecting base current, that current is constant even as the error correction current varies. But I may be thinking of something different from what you're describing.
 
B

Bob Cordell

Member
Joined 2006
cordellaudio.com
traderbam said:
Bob,
I can't find LTSpice models for all of the original parts. I can't find datasheets for the MPSU types.
Here are my substitutes at the moment.

Q20: MPSU07 -> MPSA06
Q21: MPSU57 -> 2N3906
Q22: MPSU06 -> MPSA06
Q23: 2n3906
Q24: 2n3904
Q25: 2n3906
Q26: MPSU07 -> MJE15030
Q27: MPSU57 -> MJE15031
Q28: IRF132 -> IRF9410
Q29: IRF9130 -> IRF9Z24S

I've done some measurements to find close equivalents to the output FETs. These two are in the standard LTSpice library. I think these are close enough.

I am more concerned about Q26 and Q27 because I have no idea what their characteristics are. If you happen to know what close equivalents may be that would be helpful. But don't go to too much trouble because this is icing on the cake.

Brian
Click to expand...

Hi Brian,

Yes. Bummer. The MPSU parts are pretty old. They were a modest power device, smaller than a TO220, with a metal tab for heatsink. I think they were rated at maybe 10 Watts and maybe had an ft in the neighborhood of 30 MHz. Capacitances were not too bad. I think they were used in video amplifiers.I see andy pointed you to some spec sheets, hopefully that will help.

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