Honestly I don't have my head completely wrapped around it yet - I get the basics, but there's one part that is still enough of a head-scratcher (for me, anyway) that I don't feel confident explaining it. Yet.
Maybe Nelson could chime in here...
2SK2013 and 2SJ313 require a Vgs of around 2.3V / -2.3V approximately for say 100mA bias.
This is normally provided by the resistor network R6/R15 & R23/R24.
The caps C17, C18 keep the bias more or less constant during a signal swing.
In many BJT output stages, R15, R23, C17, C18 are replaced by a so-called Vbe multiplier (or amplified diode).
So nothing new here.
R11, R19 are gate stopper resistors, R16, R22 source resistors for the follower, again nothing new.
What is new is the opto-coupler 4N35.
The LED of the 4N35 is supplied with a (class-A) current by the voltage across R16, R22.
This current can be adjusted by R18.
The LED then turns on the photo-transistor, to steal current from R15, R23.
So in case the bias current of the MOSFETs increases due to increase in temperature, the voltage across R16, R22 will rise.
This then increase the current through the opto-coupler, which drains current from R15, R23.
The voltage across R15, R23 will decrease as a result, thus reducing (stabilising) the current of the MOSFETs.
It also reduces part-to-part bias variations as well, to an extent.
This allows you to use the same resistor values for R15, R23 despite Vgs variations.
Of course the optocoupler responses to AC signals as well as to DC drifts.
In case the MOSFETs are well matched and provide the same AC current to the load, the voltage across R16-R22 remains largely constant.
In addition, C13 (as well as C17 & C12) is in place to reduce this effect on the Vgs bias.
Such bias stabilisation circuit has essentially the same function as the NTC in the original F5 circuit.
I personally do not use such stabilisation circuits. I prefer well-designed heat sinking instead.
But then I don't (have to) design circuits for volume production.
Patrick
This is normally provided by the resistor network R6/R15 & R23/R24.
The caps C17, C18 keep the bias more or less constant during a signal swing.
In many BJT output stages, R15, R23, C17, C18 are replaced by a so-called Vbe multiplier (or amplified diode).
So nothing new here.
R11, R19 are gate stopper resistors, R16, R22 source resistors for the follower, again nothing new.
What is new is the opto-coupler 4N35.
The LED of the 4N35 is supplied with a (class-A) current by the voltage across R16, R22.
This current can be adjusted by R18.
The LED then turns on the photo-transistor, to steal current from R15, R23.
So in case the bias current of the MOSFETs increases due to increase in temperature, the voltage across R16, R22 will rise.
This then increase the current through the opto-coupler, which drains current from R15, R23.
The voltage across R15, R23 will decrease as a result, thus reducing (stabilising) the current of the MOSFETs.
It also reduces part-to-part bias variations as well, to an extent.
This allows you to use the same resistor values for R15, R23 despite Vgs variations.
Of course the optocoupler responses to AC signals as well as to DC drifts.
In case the MOSFETs are well matched and provide the same AC current to the load, the voltage across R16-R22 remains largely constant.
In addition, C13 (as well as C17 & C12) is in place to reduce this effect on the Vgs bias.
Such bias stabilisation circuit has essentially the same function as the NTC in the original F5 circuit.
I personally do not use such stabilisation circuits. I prefer well-designed heat sinking instead.
But then I don't (have to) design circuits for volume production.
Patrick
Last edited:
Professional sound shop in Madrid, but I see on ebay & Michael Percy Audio U.S.A.. I use the Mogami 2534 Neglex.
2SK2013 and 2SJ313 require a Vgs of around 2.3V / -2.3V approximately for say 100mA bias.
This is normally provided by the resistor network R6/R15 & R23/R24.
The caps C17, C18 keep the bias more or less constant during a signal swing.
In many BJT output stages, R15, R23, C17, C18 are replaced by a so-called Vbe multiplier (or amplified diode).
So nothing new here.
R11, R19 are gate stopper resistors, R16, R22 source resistors for the follower, again nothing new.
What is new is the opto-coupler 4N35.
The LED of the 4N35 is supplied with a (class-A) current by the voltage across R16, R22.
This current can be adjusted by R18.
The LED then turns on the photo-transistor, to steal current from R15, R23.
So in case the bias current of the MOSFETs increases due to increase in temperature, the voltage across R16, R22 will rise.
This then increase the current through the opto-coupler, which drains current from R15, R23.
The voltage across R15, R23 will decrease as a result, thus reducing (stabilising) the current of the MOSFETs.
It also reduces part-to-part bias variations as well, to an extent.
This allows you to use the same resistor values for R15, R23 despite Vgs variations.
Of course the optocoupler responses to AC signals as well as to DC drifts.
In case the MOSFETs are well matched and provide the same AC current to the load, the voltage across R16-R22 remains largely constant.
In addition, C13 (as well as C17 & C12) is in place to reduce this effect on the Vgs bias.
Such bias stabilisation circuit has essentially the same function as the NTC in the original F5 circuit.
I personally do not use such stabilisation circuits. I prefer well-designed heat sinking instead.
But then I don't (have to) design circuits for volume production.
Patrick
Thanks so much. When you wrote "The caps C17, C18 keep the bias more or less constant..." I assume you meant C17 and C12.
Is there some advantage of using the optocoupler circuit in place of an NTC?
Steve
This is very very elegant and neat. Congrats Wayne and thanks for demonstrating this to such good effect.
The optocoupler is doing a couple ( ) of things - The LED side is acting as a voltage reference across the source resistors, setting bias current. The transistor is burning up the excess bias voltage on the gates - the string of 4 resistors from V+ to V- is making more bias voltage at the gate than needed, (about 1/2 the rail) and the transistor side of the opto trims it down to what's necessary. (about 3V for the Toshiba, a bit more for Fairchilds)
The fact that it helps with thermal compensation is just a happy side effect.
) of things - The LED side is acting as a voltage reference across the source resistors, setting bias current. The transistor is burning up the excess bias voltage on the gates - the string of 4 resistors from V+ to V- is making more bias voltage at the gate than needed, (about 1/2 the rail) and the transistor side of the opto trims it down to what's necessary. (about 3V for the Toshiba, a bit more for Fairchilds)The fact that it helps with thermal compensation is just a happy side effect.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.