optimizing the VBE multiplier

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
sajti

I don't think it is a good idea to set bias like this..

Voltage drop on emiter resistors may be different for different current amplitude...


situation1:no input signal

let's assume idle current is aiming to be 50mA, then the voltage drop on two emiter 0.1 resistors is 10mV, right?

situation2: huge signal

say something like 2A from NPNout make 200mV drop on one emiter resistor and on the second it isn't negative
then the same from PNPout.... and again

optimisticly... something like 50mV mean voltage drop on emiter resistors.
what will do your servo? will 'correct' this unappropiate bias and force the amp to produce HEAVY crossover distortion


sorry if I'm wrong or have missed something
please correct me then

regards
 
It is possible that i made some mistake...not the first time, and i am sure, ...

Will not be the last time i make mistakes.

Well, i am human...and i am alive (for a while...hehe...old!)

The schematic, is the first Hienrich published in his first thread...unless you have modified, this is the one he told us first.

Sorry if i disturb a little, i guarantee that was one unhappy accident.

Carlos
 
The LT1166 is made for Hexfets, to use with BJTs you need to add a couple of base diodes to increase the voltage drop across the IC (or use a tripple as Jennice suggests). Let pin 2 float and drive both the top and bottom pins instead.

The part is way too much money, and there is no second source.
 
"What's the point of this circuit if this pin should be floating!? "

Uh, bias IC?

Do you really want to run a signal through a slow opamp running on ±2V?

Take the Krell drive circuit, remove the Vbe, and hook up to pins 1 and 4 and leave 2 float.
 
Jennice said:
No, I don't think I have missed the point.
On the contrary, someone else has missed the sense of realism.
...
Also, please explain to me, how acuate tracking / temperature compensation can be obtained with his circuit.

Actually, tracking will be fine since the diode string acts as a Vd 'multiplier' instead of the Vbe multiplier, but since he has resistors in series, they need to be optimized together with Re of the outputs to achieve stability. Looks like a needlessly finicky design to me...
BTW good points on all other accounts...

Jennice said:
Leach writes: 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. ..

A length of 10 cm wire, 1 mm diameter, run in parallel with, and 1 cm from reference ground, creates about 0.27pF of capacity.

I have to completely agree with you on this. Even if the wires were very close to ground and longer, the stray capacitance would be 10x less than the typical Cdom. Finally, driver stage input capacitances are far larger (and of course if you were using MOSFET outputs, the comparison would be laughable).

But, if capacitance is a problem, it is FAR more easily solved without a single additional electronic component - in a BJT output stage the voltage swing between the Vbe multiplier and output is essentially that of the voltage drop on the Re of the output transistors, far lower (one would hope!) than the voltage swing of the VAS respective to ground. Solution: use a shielded cable with the shield connected to the output of the amp, and thereby bootstrap the capacitances.
 
sajti said:
This circuit is planned to apply in existing amplifier. The amplifier contains the original Vbe multiplier, which helps to avoid the switch on overcurrent. Just set the Vbe multiplier for higher current (400-500mA), and this circuit will reduce it to the level, set by R10.

Just use the optocoupler transistor as a Vbe multiplier...
 
Strange VBE multiplier problem

I make an amp, using CFP output stage (Like RodE's P3A).
The bias generator is VBE multiplier, using MPSA06. The VAS current is 7.5mA. Accross C-E I put 10nF cap, Rbe=3kohm, Rbc=2k2ohm.

When I put the VBE multiplier transistor on the upper side of the PCB, the bias went rocketing high. It goes to 1V75, but strange is, there is no oscilation on the output trace. The output transistor sucks alot of current and becoming very hot.

When I put the VBE multiplier transistor on the lower side of the PCB, everything went normal, still with no oscilation on output.
The bias goes down to 1V2 (which I wanted to be)

What happened here?

When I put the VBE multiplier transistor on the upper side of the PCB, even when I plug-out the Rbe (3kohm), the bias still ramping upwards.
Putting 47uF between C-E doesn't help at all.

Is the VBE transistor pick up some RF or something? How to cure this?

In the first page, a member said to put small values R's for Rbe-Rbc to maintain those R's still becoming a voltage divider, not becoming current source for VBE multiplier's base. What's the difference if Rbe-Rbc is (3k-2k2) or (300ohm-220ohm) or (30k-22k)? The ratio is the same, but what is the difference between these combinations of R's?
 
Andrew T wrote some month's ago:


Mr evil, what effect (better or worse) does the shiklai or darlington Vbe have on the sound quality?

I've used this approach and I do not believe it makes any direct change in sound quality. What it did for me is make the quiescent current much easier in an output configuration. Even with a large heatsink, setting it even approximately correctly can be very difficult when the current reading wanders about a considerably. Using a Sziklai pair (and I supose Darlingto too) can make for greater quiescent stability. Whether stability is good or bad in itself is another question but I think is fairly moot when stability is so poor you can't really tell if you are anywhere near where you want to be.

In principle, A Sziklai pair can be "tuned" to increase/decrease tempco. It may also be subject to oscillation. I did not pursue either of these having concluded that precision in such maters is less important in actual practice than theory or simulation. Be sure the bias is reasionably close to optimal (or a little under biased) and stable -- let NFB do the rest since under best of conditions the bias will drift some anyway. This is a crude approach, I admit, but it works and chasing long-term perfect bias with just a Vbe multiplier is probably a prescription for futility and frustration.
 
Disabled Account
Joined 2005
Re: Strange VBE multiplier problem

lumanauw said:
In the first page, a member said to put small values R's for Rbe-Rbc to maintain those R's still becoming a voltage divider, not becoming current source for VBE multiplier's base. What's the difference if Rbe-Rbc is (3k-2k2) or (300ohm-220ohm) or (30k-22k)? The ratio is the same, but what is the difference between these combinations of R's?

The foundamental reason the Vbe works is that the current going through Rbc is about the current going through Rbe. Otherwise, the "voltage divider" formed by the two resistors wouldn't work.

It is critical to maintain that assumption when you chose the value of those two resistors. What I usually do is to make sure that a) the bulk of the current goes through the transistor, and b) the base current is as small as possible.

For example, a 300/200ohm combination in a Vbe configuration means the current going through the resistors would be about 3.5ma minimum. When your VAS runs at about 5-6ma, that is a little bit too much.

This is less of a concern if you are loading the VAS with a CCS but would be quite probamatic if your VAS is loaded with resistors.

I tend to use 10k/10k combination on a Vgs multiplier (most of my output stages are MOSFETs). If I have to use a bipolar, I make sure that I use a high beta type there.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.