leach amp vbe multiplier

[mikeks]

Just think about it for a while, and you, in all probability, will understand just why Leach is wrong.

Here's a clue:

The upper left 4-diode string should be connected to A & B....

 

[MikeBettinger]

Hi Mike K,

I have a question for you, since you do appear to be quite knowledgable as to the reasons for and the proper way to implement most aspects of amp design.

If I examine older amp designs I have noted that the bias for the output stage has been handled many ways, from a simple resistor or diode string to some very complex signal or temperature tracking approaches.

My question is, is there a real technical advantage to using a tracking type circuit relative to a few diodes and an adjustable resistor coupled to properly designed heatsink?

I have experimented with a signficant number of approaches and sonically prefer a simple resistor (which I've found is not practical to implement) but I'm currently using 3 1N4148's thermally coupled to a heatsink that is only luke warm after 5 minutes under a blow drier. I will admit to an hour or so warmup time playing music before it all locks in.

In your experience is there a measurable improvement aligned to a more active approach to supplying the bias? Can you siuggest any further reading on the topic?

Regards, Mike.

 

[mikeks]

If I examine older amp designs I have noted that the bias for the output stage has been handled many ways, from a simple resistor or diode string to some very complex signal or temperature tracking approaches.

With a diode string, your bias voltage varies significantly with variation in the current through the diodes.

This problem, of course, is vastly worse if a simple resistor is used to set bias.

This is why i recommended this approach.

 

[mikeks]

Note that figure 2c and 2d are NOT Vbe multipliers.

 

[AndrewT]

Hi Mike,

Leach's Vbe multiplier is not 2c or 2d either.

It is a circuit where

The output voltage is a function of the base emitter voltage.

The voltage across the circuit is a multiple of the Vbe of the transistor (just like a resistive multiplier), the diodes give a measure of temp compensation (just as a resistive multplier uses the transistor), the voltage across the circuit varies if the VAS current is not constant (just like a resisitive multiplier).

The voltage varies just like (not exactly like) a resistive Vbe multiplier, it just arranges the temp compensation differently.
It looks like a multiplier
It behaves like a multiplier
It fits the amplifier as a multiplier
It is a multiplier.

 

[mikeks]

Hi Andrew,

The diode string connected betwen A and B make this untrue, as the voltage across these diodes is not a linear function of the current through them.

This is why it is this string of diodes and not Q7 that is mounted on the heatsink for thermal compensation.

 

[AndrewT]

Hi Mike,
you are right,

the voltage across these diodes is not a linear function of the current through them

you are also right when you say

This is why it is this string of diodes and not Q7 that is mounted on the heatsink for thermal compensation

but

The diode string connected betwen A and B make this untrue

does not follow

It is still a voltage multiplier with a different arrangement of temperature compensation.

However, the current through the diode string is nearly constant over a big range of VAS current. The Vbe voltage holds the bottom leg resistor at nearly constant voltage. The base current is fairly small and variations in this change the diode current only a little.
the main variation is current coming in through the collector and this I think causes temp changes at the junction and in turn change Vbe. If this is the mechanism that causes the slope in the multiplier curve then it explains why your proposal, D.Self's solution with the added Collector R and Hagerman's all improve on the basic multiplier's voltage change with VAS current changes.

But back to where we started. The fact that any multiplier has a varying voltage with varying VAS current does not of itself stop the multiplier being (and behaving as) a multiplier.

 

[mikeks]

The Vbe voltage holds the bottom leg resistor at nearly constant voltage.

This is correct at a given temp.; however, with a true Vbe multiplier the bias voltage across Q7 is a function of Q7's Vbe.

This is not the case here since, as you've noted, the voltage across the diode string does not vary ''significantly with current''.

Viz. Diode current is not linearly related to diode voltage, which is why the Vbe multiplier equation is inapplicable: the diode string and not Q7 must, therefore, be mounted on the heatsink.

 

[AndrewT]

Hi,
why have you introduced this distraction?

the diode string and not Q7 must, therefore, be mounted on the heatsink

We all know that Leach intended the diode string be mounted on the main heatsink. He describes the mounting and the need for the two isolating resistors in detail.

We have no disagreement with this.

The disagreement is whether the Leach style of Vbe multiplier is entitled to be called a multiplier or not.

with a true Vbe multiplier the bias voltage across Q7 is a function of Q7's Vbe

applies to the Leach Vbe multiplier just as much as it would to a resistively (with no added diodes) loaded multiplier

 

[mikeks]

Hi,
why have you introduced this distraction? We all know that Leach intended the diode string be mounted on the main heatsink. He describes the mounting and the need for the two isolating resistors in detail.

He has no choice: mounting Q7 on the heatsink will not cause the bias to track the heatsink temp.

The disagreement is whether the Leach style of Vbe multiplier is entitled to be called a multiplier or not.

I don't think you've actually read my post.

It is called a Vbe multiplier because its bias voltage varies with Q7's Vbe.

How can this be the case if you include a diode string between Q7's base and collecor?

The diode string's voltage will vary hardly at all when Q7's Vbe varies. This should now be self-evident.

 

[jackinnj]

Pick any number and device -- do a temperature sweep -- left axis is the ratio of Vce.

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

 

[mikeks]

Hi jackinnj,

Run Temp sweep with respect to Q1's Vbe, and plot Vce.

Similarly, run Temp sweep for the diode string alone and again plot Vce.

Further, run temp sweep for Q1's Vbe with the diode string removed, and, ounce more, plot Vce.

You'll find that Vce varies most (with the diode string in situ) when the temp of the diodes is varied, rather than when Q1's Vbe changes.

 

[jackinnj]

Perhaps I should have labeled them better -- the Vbe plots with and without the diode string sit atop each other -- the Vce plots are shifted by the number of diode drops and have different slopes. Not enough time to do everything.

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

 

[AndrewT]

Hi,
I just spent 25minutes compiling a reply to Mike refuting his contention that the Leach version is not a Vbe multiplier.
But I lost it.

Mike,
I think this argument is not worth another 25minutes, I give up (not give in!).

 

[MikeBettinger]

Note that figure 2c and 2d are NOT Vbe multipliers.

Hi Mike,
Basically voltage references. I thought about trying to use one but then It was easier to fit a few of diodes.

I have a few obsrvations on the suggested circuit but would like to do my homework before speaking. We'll see how the week pans out. Thanks for the input.

Regards Mike.

 

[sreten]

Hi,

Hmmm....... A Vbe multiplier can have an effective "gain" of near 1.

So its very near a Vbe reference.

Personally I'm not keen on Leach's inacurracies regarding bias.

Seems in the case shown you have a string of diodes + an adjustable
Vbe multiplier to set the actual bias voltage, thus the output voltage
is not a direct multiple of the transistors Vbe.

D.Self covers the complete issue comprehensively and far more more
accurately than Leach, who simply misunderstands optimum aB bias.

/sreten.

 

[Ultima Thule]

It appears the Leach's bias circuit is a hybrid between a normal Vbe multiplier and Hagermans circuit 2c, 2d.

In Leach's bias circuit Rbe(R27+P1) is 3,2 kOhm and Rbc(R25+R26) 2 kOhm plus the diode string.

The background to Leach's "deviating" bias circuit is as follow quoted from his site :

Many amplifiers have the VBE multiplier transistor on the heat sink with the output transistors. This eliminates the need for the diodes. The wires which run from the circuit board to the transistor exhibit capacitance to ground which can affect the high-frequency response of the second stage. At worst, this could cause oscillation problems. With the diodes on the heat sink, resistors on the circuit board can be used in series with the wires to isolate this capacitance from the second stage. These resistors are R25 and R26 in Figure 10. If Q7 were mounted on the heat sink, isolation resistors could not be used because they would affect the voltage regulation between the collectors of Q8 and Q9.

Cheers Michael

 

[acenovelty]

Are there any practical suggestions for a solution for joe mobile's
stated question?

 

[joe mobile]

I think I really started something... although I don`t understand too much of the discussion.
I just blew the amp by accidently shortening R36 so for now my problem is solved... I`ll start a new leach amp. I guess thats learning the hard way.
I guess it will take quite some time till I finish the new one since I have to etch new boards and I don`t have that much time, but at least I got all the parts.
I`ll post when I got it finished
Thanks again to all the people who helped, and I`ll go on reading everything.

Joe

 

[mikeks]

From Leach's site :

Many amplifiers have the VBE multiplier transistor on the heat sink with the output transistors. This eliminates the need for the diodes.

The wires which run from the circuit board to the transistor exhibit capacitance to ground which can affect the high-frequency response of the second stage.

At worst, this could cause oscillation problems. With the diodes on the heat sink, resistors on the circuit board can be used in series with the wires to isolate this capacitance from the second stage.

These resistors are R25 and R26 in Figure 10. If Q7 were mounted on the heat sink, isolation resistors could not be used because they would affect the voltage regulation between the collectors of Q8 and Q9.

Leach's explanation is rather fanciful.