Dumb Biasing Mod, applicable to F6 and other Papa Amps. Possibly My Dumbest Idea Yet

2 picoDumbs

Member
2013-09-06 9:35 am
Pico's Dumb (ie Pico is Dumb) Biasing Trick for F6 and other Amps

As you may or may not appreciate many devices have a positive, negative, or virtually zero temperature coefficient.
Most mosfets have a positive temperature coefficient, ie as they heat up they draw more current when biased with a constant voltage source applied to the gate of the mosfet (provided the bloody voltage is adequate to turn it on).

To reduce the effect of this increasing current draw as the device gets hotter we can do a few things.

1) Reduce Delta T of the devices, eg larger heatsinks, better thermal pads ie improving Case to Heatsink performance, etc

2) Using source degeneration which acts as a form of negative feedback, ie as the device attempts to draw more current, an increasing voltage drop occurs across the source resistor which effectively reduces Vgs which then stops or reduces the amount of thermal current drift.

3) Using NTC thermistors in the bias circuitry, which effectively reduces the Voltage at the gate as the thermistors heat up, eg as seen in F5, and elsewhere. I used a similar method to this, in F4 beast builders ie building push-pull amps (mostly with hockey pucks) with zero degeneration.

4) Active bias control circuitry, optocouplers, hall sensors, discrete designs.

You can obviously also combine some or even all of these methods, to get the result you need.
There are probably quite a few more techniques than this, but I am telling the story and I don't want it to be batshit boring.

Even though I have successfully used method 3, and 4 in the past, I don't like "intelligent solutions", I like dumb simple solutions that don't require maths or too much intelligence.

I have often thought, there has to be a dumber way to do this.

So I am sitting on the toilet (it always happens like this), and I am considering my options with regards to voltage references, TL431, LM329, Zeners, Leds.
I start thinking about Zeners and leds, since they are dumber than the other devices, and remembering that Zeners below 5V have a negative temperature coefficient and zeners above 5V have a positive temperature coefficient, the light bulb turns on "aahhh you bloody dumb bastard" and gave myself an uppercut for not considering this earlier.

Many here have either personally built the F6, or are at least aware that in many cases, the 5.1V zener shown in the original circuit diagram is not quite adequate to produce the required voltage at the gate to bias the F6 to the required value.
Most of us have used 5.6V, 6.2V, 6.8V zeners etc, some have used LM329.

Well, all these devices have a positive temperature coefficient, you could probably say that LM329 is effectively zero but it has a very small positive temp coefficient.

So, what is stopping us from using 2 smaller valued Zeners to achieve the required zener voltage eg 2 x 2.7V zeners, and achieve a slight negative temperature coefficient.
LEDs also have a negative temperature coefficient with regards to Vf, but we will first consider comparing Zener diodes.

Anyway, so I decided to compare the observed measured differences in biasing up a Vishay IRFP150 mosfet (with zero degeneration) using 3 different zener configurations:

1) 2 x 2.7V zeners in series (effectively 5.4V)
2) 5.6V zener
3) 6.2V Zener
4) 3 x Green LED LTL 4231N - tested at a later date


The zener diodes were fed around 5mA in each case (more detailed information below)

A multiturn trimpot was used in each case (just like F6 circuit) to achieve exactly 3.925V at the gate.
I used zero degeneration on the mosfet to better illustrate the effect.

Results
------------------------------------------------------------------------

Case (1) 2 x 2.7V Zener Configuration
Initial Vgs: 3.925 V
Ambient Temperature: 21.8 deg C

Id at turn on: 1.32 A
Id at thermal equilibrium: 1.68 A

Vgs at thermal equilibrium: 3.899 V

Delta Id after thermal equilibrium: 1.68-1.32 = 0.36 A
Delta Vgs due to Zener: 3.899-3.925= -0.026V (negative 26mV)

------------------------------------------------------------------------

Case (2) 5.6V Zener

Initial Vgs: 3.925 V
Ambient Temperature: 21.8 deg C

Id at turn on: 1.32 A
Id at thermal equilibrium: 1.95 A

Vgs at thermal equilibrium: 3.936 V

Delta Id after thermal equilibrium: 1.95-1.32 = 0.63 A
Delta Vgs due to Zener: 3.936-3.925= 0.011 V (positive 11mV)

------------------------------------------------------------------------

Case (3) 6.2V Zener

Initial Vgs: 3.925 V
Ambient Temperature: 21.8 deg C

Id at turn on: 1.32 A
Id at thermal equilibrium: 2.25 A

Vgs at thermal equilibrium: 3.954 V

Delta Id after thermal equilibrium: 2.25-1.32 = 0.93 A
Delta Vgs due to Zener: 3.954-3.925= 0.029 V (positive 29mV)

------------------------------------------------------------------------

Brief Discussion of Results

As predicted during my toilet brainstorming, the 2 x 2.7V zener in series produces an excellent result, with a delta Id between turn on and thermal equilibrium of 0.36 Amps. This was achieved using the negative temperature coefficient of a 2.7V zener to our advantage.

The 5.6V zener is twice as bad, and the 6.2V, almost 3 times as bad as our 2.7V zener setup.

The test was performed on a large heatsink (relative to the heat dissipation of the device) flat on the floor, with excellent ventilation.
If the same test was repeated on a relatively smaller heatsink with regards to dissipation like one you might be using, and performed inside the chassis of an amp, we would expect to see an even greater positive affect with regards to using 2 x 2.7V zeners.

However, if I were to repeat this test exactly as performed above using degenerating resistors the results would be closer together.
This was merely done the way it was to clearly illustrate the effect, and I always try to avoid any kind of degeneration, so this kind of thing is of interest to me.

You could also try 3 x 1.8V Zeners for an ever greater effect.
Or you could use 2 x 3V or whatever you might have. Basically anything below a 4V Zener configured for the voltage you need, is going to have a nice affect.
EDIT: A 3 x LTL 4231N LED combination was also tested and gave favourable results like 2 x 2.7V zeners, however it has better regulation than the Zener combination.

It's up to you if you want to try this out. I am not into preaching, just sharing.



Test Conditions:

Zeners tested:
5.4V (2.7V x 2 in series) BZX55C2V7-TR
5.6V BZX79-B5V6.113
6.2V BZX79-B6V2.113
LED: LTL 4231N

Mosfet:
Vishay IRFP150

Trimpot:
10k 0.5W Bourns 3299Y series

Resistor feeding current to Zener
3.3k CMF55 (zeners biased at approximately 5mA - close enough)

Powersupply Feeding Circuit
Linear regulated lab powersupply set to 23V

Heatsink
MF35-151.5 just laying flat on ground.

Ambient Temperature 21.8 deg C

Biasing Circuit
F6 basic biasing circuit.
 
Last edited:
I'm still dumbest!

:clown:

Papa's F7 is having 11 resistors , 4 actives and no caps ..... while I've just made Babelfish F7 , having 29 resistors, 6 actives , 5 ICs and 4 caps

go figure!

I'm going to call my next business venture - MLevinson! and my logo will be - No Less Than Two Parts For Each Position!!

:devily:

edit: of course that I'm going to adopt manner of using two resistors in parallel everywhere , with tol. ring oriented on opposite sides !
 
Last edited:

2 picoDumbs

Member
2013-09-06 9:35 am
Yeah leds will work too.
I could test that out now, since I am already set up to do it.
Give me a couple of hours and I will be back with a result

The only reason I didn't do leds initially cause people might whine about puting 4 in series.

I will do it now.
 
Last edited:
Congrats for an elegant idea.

Zener voltage references are a hobby for volt-nut calibration enthusiasts. I own two of these that are 40+ years old. Very stable. This stability requires averaging many measurement over some period of time. You need a really good DMM to see the drift and temperature dependence of these animals. Mine are much less than 1ppm / deg C.

If you plot the voltage over a very long time, you will occasional spikes which are thrown away when calculating the "actual" voltage of the reference. I have seen these spikes on my references.

The spikes are sometimes called popcorn noise.

So, your zener voltage reference across your MOSFET gate may induce random pops and clicks which may or may not be audible. A large cap across the zener may absorb these spikes.
 

2 picoDumbs

Member
2013-09-06 9:35 am
Congrats for an elegant idea.

So, your zener voltage reference across your MOSFET gate may induce random pops and clicks which may or may not be audible. A large cap across the zener may absorb these spikes.

Thanks

On F6 we have an RC filter after the zener comprising of a 10k resistor and 1000uF capacitor.
In my experience with the F6, I have yet to experience any popcorn noise.
 

2 picoDumbs

Member
2013-09-06 9:35 am
Ha! So an element with negative tempco in the Vgs bias circuit.
Will have to think about that.
What’s the tempco of a string of little green LEDs?

Vgs set at 3.925 V at ambient 21.8 C
Id at turn on 1.33 A

3 x green leds (LTL4231)

Delta V: 3.925 - 3.895 = negative 30mV
Delta Id: 1.61 - 1.33 = 0.28 A

The only thing that needs to be checked is a curve I vs V to check regulation. I can measure that now.
 
Last edited: