Super Beta Vbe Multiplier

[ChocoHolic]

In my former experience the traditional Vbe multiplier often is showing a slight undercompensation of the thermal behaviour of the output transistors. This is coming worse as smaller emitter resistors are used.
Also sometimes the Vbe multiplier is suffering from low current gain of the Vbe multiplier transistor, when you choose a power transistor in order to screw it onto the heat sink, instead of a small signal type.

In practice I would even prefer to overcompensate the thermal a little bit, making the system more rugged against overheating.
Attached a PDF of my Complementary Super Beta Vbe Multiplier. Probably I reinvented the wheel, but I am not aware of having seen this so far.
Compared to the standard version, this super beta version is additional taking advantage of the increasing current gain of the 2nd BJT with increasing temperatures.

...readings of the real thing in the next post...

[Workhorse]

Hi Choco,

This is just like a Sizklai pair used in VBE multipleir application....I think it will overcompensate to a much greater extent.....


K a n w a r

[ChocoHolic]

Some more details tested this in DC condition deriving the 6.5mA from a 30V DC source and 2 x 2K2 series resistors:
I adjusted the voltage across the Vbe multiplier to 2.80V at 22C, assuming a darlington output configuration.
Both transistors where glued to a small copper plate and heated up to above 130 C.
The readings above 80C where not very precise, because of radily decreasing temperature after removing the light bulb, which I used for heating up. So I am just listing my readings up to 80C.

Standard Vbe Mult vs. Super Beta Vbe Mult

23 C: 2.80 V 2.80 V
30 C: 2.73 V 2.70 V
40 C: 2.65 V 2.61 V
50 C: 2.57 V 2.51 V
60 C: 2.49 V 2.43 V
70 C: 2.40 V 2.32 V
80 C: 2.30 V 2.22 V

In averge the temp. coefficient of the standard version is 8.6mV/K
while
the Super Beta has a temp. coefficient of 10mV/K.

...seems to match more or less to what I usually would like to have...
This circuit would even allow to adjust the temp coefficeint to any desired value between 8.6mV/K & 10mV/K
by adding a resistor from base to emitter of the second transistor.

[djk]

(1977)

http://www2.yamaha.co.jp/manual/pdf/...AFR/CR-820.pdf

[ChocoHolic]

Originally posted by Workhorse
Hi Choco,

This is just like a Sizklai pair used in VBE multipleir application....I think it will overcompensate to a much greater extent.....


K a n w a r

Hi Kanwar,

it cannot massively overcompensate, because it's a closed feedback.
If the voltage across the circuit is decreasing, this is feed back to the base of the NPN and the NPN is becoming less conductive.

But Sziklai is a good hint as it indicates a closed loop of 2nd order.
I will check if this circuit shows oscillations. Fortunately this would be easy to cure, because most applications have a large cap across the Vbe multiplier anyway.

[ChocoHolic]

Originally posted by djk
(1977)

http://www2.yamaha.co.jp/manual/pdf/...AFR/CR-820.pdf

..nothing new under the sun.... seems like it just got forgotten during the last 20 years.

[ChocoHolic]

uhps....
During my first measuremnts I had the PNP connected wrong, means C & E exchanged. This invers operation usually is resulting in much less gain. With correct connection the temp. coefficient increased a little bit to 11.7mV/K, which I appreciate very much. This offers to adjust any value between 8.6mV/K ... 11.7mV/K with an additional resistor between base and emitter of the PNP.
22 C: 2.82V
30 C: 2.71V
40 C: 2.59V
50 C: 2.46V
60 C: 2.36V
70 C: 2.24V
80 C: 2.14V

Regarding oscillation, the voltage across the Vbe multiplier looked perfectly flat.
Hm, so I tried if an additional signal could initiate some ringing.
Instead of 30V DC + 4.4k Ohms ,
I applied 30V DC +/-5 V rectangular - again with 4.4k Ohms in series. The frequency of the rectangular signal was 10kHz.

Time base of the attached picture is 20us/div.
The white trace is the 30V DC +/-5V rectangular signal. 10V/div.
The red trace is the resulting AC voltage across the Vbe multiplier. 100mV/div.

Very nice. No ringing, no overshoot.
I think I will keep this sort of Vbe multiplier in mind.

[AndrewT]

Hi,
D.Self showed this circuit as fig11.32 on P320 of amp design handbook.

He added a base emitter resistor (1k0) to the PNP. Which forces the NPN into constant current.

His graph fig11.33 showing Vbias is illuminating. Very consistent Vbias (1.7ohm slope) with varying Ic

I see Yamaha put the r adjust in the upper leg, why?

[ChocoHolic]

Originally posted by AndrewT


I see Yamaha put the r adjust in the upper leg, why?

...also wondering about that, I prefer to have it in the lower leg. Because in case of slider lifting, the position in the lower leg is fortunate to avoid overheating.

[darkfenriz]

Check out Pavel Dudek's amplifiers. All use Sziklai Vbe multiplier.
I achieve slightly negative thermal coeficient another way: I use complementary Vbe mulltiplier for cascade of three emitter followers and pre-drivers are not on heatsink while drivers and power are.

regards