Vbe multiplier sensitivity

[Kilentra]

I'm using Anthony Holton's symmetrical amplifier which uses a BD139 in a Vbe multiplier circuit mounted on the heatsink with the IRF Mosfet output stage, to control the bias. I've noticed that this Vbe multiplier is really sensitive to temperature, so much so that blowing a little breeze over the heatsink causes a significant increase in bias; playing music reduces the bias significantly even though there is little change of heat (I have pretty large heatsinks that are barely warm). At least I won't have any thermal runaway problems, but I bet this hurts the sound quality.

Is there any way to modify the circuit to reduce the sensitivity of the bias circuit to temperature, so the bias remains steadier? There is a .47uf capacitor across the collector and emitter of the BD139, but I imagine increasing that value would only slow the rate of change of the bias, and not affect temperature sensitivity.

Thanks for your help. As you might have guessed, I'm a newbie.

[BrianL]

Read Doug Self's amplifier books regarding the topic.
He's no fan of power FETs but the info on bias stability
should be helpful to you.

[sajti]

This is absolutely normal with this circuit.
The speed of the bias regulation depends by the heating up, and cooling down speed of the BD139 on the heatsink. If You have big heatsink with large cross sectional area, the change will be slow. If You want to speed up the regulation, You can put the BD139 directly to the case of the output device.
You can increase the capacitor over the BD139, but I think it will do nothing with the regulation speed.
Sajti

[peranders]

If you are using a BJT as temp sensing element you will get more mv/deg than you need but you are also VERY sure not to get thermal runaway. You should use a mosfet instead. Mosfets have -8-10 mV/deg. Use any type in a case which you can attach. You can even use a small BS170 in TO-92 case and clue this thing to the heatsink. Not the you have to change resistor values a bit. Calculate for 2.5-3 V Vgs (instead of 0.6-0.7 for the bjt)

[Badeck]

Hi All,

Thread's rather old, but as I came across this just now while designing a biasing for the output stage of a power amp I'm building, I'd like to share my finding on Vbe multiplier sensitivity.

Sensitivity of Vbe multiplier is defined as the change of the quiescent current of the output transistor stage with respect to change in temperature. When Vbe mult transistor is mounted on the heatsink where the output transistor is mounted (as it should be), the bias stability of the OPT will depend also on the Vbe mult transistor bias statbility.

Therefore, to change the sensitivity of the system, we can design the Vbe multiplier to have an emitter leg resistor, just like the normal emitter resistor stabilization technique. Increasing the emitter resistor value will decrease the sensitivity.


I didn't go into the math of the design, so I can't have formulas here.

In my design I use Orcad and on circuit simulation, Iq tends to go down with increasing temperature. The goal is to have no increase or decrease in Iq with temp changes.

I increased the Vbe mult transistor's emitter leg resistor until Iq flatlined.


Thanks All,

Dexter

[eyoung]

Hello,

Are the darlingtons driving the 2n3904's or visa versa???

Regards, Elwood

[lineup]

the 2N3904 and 2N3906 are normally inactive
only when too much OUTPUT current across darlington Emitter resistors
these will trigger the 3904/06 to reduce the base current to Darlingtons.

= current limit protection transistors

[Badeck]

Hi,

The 2n3904's, along with the emitter resistors of TIP142/147, are used as active current limiters.

Aside from another important function when placing transistors in parallel, the emitter resistors translate the emitter currents of the TIP142's/147's into voltage. When the current increases such that the voltage drop across the resistors approaches 0.6~0.7V, the 2n3904's turn on and divert the base drive currents of the TIP's, rendering the output current limited.

[fotios]

Hi

From experience, the speed of thermal reaction of bjts reduced according to their Vce increasing. Maybe for this reason D.Self he is using a big Vce such MJE340 = 300V as Vbe multiplier (from memory, he doesn't explain why MJE340).
You are using a BD139; the BD137 is more faster and BD135 even much faster in thermal reaction.

Fotios

[destroyer X]

When you heat a transistor it will give you a reduction of voltage drop that will reduce your stand by bias...but it is a single junction operating... so you have some percentual of decreasing.

When you install many diodes in series and put them touching the heatsink you gonna have, as result, added the reduction of resistances that is generated into each one of the diodes... when heated of course... this will allow you bigger control..bigger reduction of bias when hot... seems to me much better.... more reduction of resistance... a bigger percentual of action.

Other way is to use more than one transistor as heat sensor.

This is effectiveness, of course beeing effective we can call the thermal control a more sensitive too.

Into that sittuation effectiveness and sensitiveness goes confusing one each other, as they run together..when one increases the other increases too... a direct proportion.... direclty proportional

Carlos