Can anyone comment as to why most Lateral FET amps seem to use just a simple trimmer and parallel capacitor to bias the output devices? Wheras most class AB amps using BJTs seem to use a VBE / Bias Servo to do this? Other than a risk of trimmer-wiper failure destroying your amp, there are some other issues I have observed when testing both of these methods with FETs - most notably:
- With a simple trimmer and capacitor in parallel, the bias voltage between the two gates creeps up as the amplifier is fed a bigger signal.
- example circuit here: Project 101 - High Power, High Fidelity Lateral MOSFET power amplifier
- With a VBE / servo the bias voltage remains pretty constant with signal size
- example circuit here: 60-80W Power Amplifier
The problem I’m wrestling with is heat dissipation. If the bias voltage of my output FETs is increasing as the input signal increases, then so is the amount of heat being dissipated. I would like to keep the bias voltage of the FETs constant if possible so I am inclined to go with the VBE (or VGS) - but I don’t know if there any implications with doing this with lateral FETs. I can see the Bias servo is necessary with BJT output devices in class AB as they are current controlled so need a constant bias current, as well as thermal protection. Conversly Lateral FETs are voltage controlled and thermally negative, so don’t need this.
Are there any implications of holding the Lateral FET gates at a constant Bias voltage using a VBE or VGS Servo? I would like to do this if it does not cause any other issues.
Many thanks for reading, NICK
- With a simple trimmer and capacitor in parallel, the bias voltage between the two gates creeps up as the amplifier is fed a bigger signal.
- example circuit here: Project 101 - High Power, High Fidelity Lateral MOSFET power amplifier
- With a VBE / servo the bias voltage remains pretty constant with signal size
- example circuit here: 60-80W Power Amplifier
The problem I’m wrestling with is heat dissipation. If the bias voltage of my output FETs is increasing as the input signal increases, then so is the amount of heat being dissipated. I would like to keep the bias voltage of the FETs constant if possible so I am inclined to go with the VBE (or VGS) - but I don’t know if there any implications with doing this with lateral FETs. I can see the Bias servo is necessary with BJT output devices in class AB as they are current controlled so need a constant bias current, as well as thermal protection. Conversly Lateral FETs are voltage controlled and thermally negative, so don’t need this.
Are there any implications of holding the Lateral FET gates at a constant Bias voltage using a VBE or VGS Servo? I would like to do this if it does not cause any other issues.
Many thanks for reading, NICK
Typical lateral FETs are thermally stable at and above about 200mA, and are biased around 100mA. This means that as they heat up the bias will increase a bit, but cannot runaway. There is no need for temperature compensation with a something like a Vbe multiplier attached to the heatsink, constant voltage is good enough, and a simpler circuit.
The amplifer you link to with load problems doesn't have a constant current circuit providing the bias current, this is no doubt why its bias varies with load.
Heat dissipation due to biasing ought to be much less than that due to the load at anything but low signal levels.
A Vbe multiplier does not provide good constant current, note, due to the temperature dependence of the voltage. A good constant current source can reduce temperature dependence by using a precision voltage reference, or at least an LED (rather than diodes).
However using a Vbe multiplier will lead to reduce biasing when hot, which would fix your problem, perhaps at the expense of distortion which increases with under-biasing.
The amplifer you link to with load problems doesn't have a constant current circuit providing the bias current, this is no doubt why its bias varies with load.
Heat dissipation due to biasing ought to be much less than that due to the load at anything but low signal levels.
A Vbe multiplier does not provide good constant current, note, due to the temperature dependence of the voltage. A good constant current source can reduce temperature dependence by using a precision voltage reference, or at least an LED (rather than diodes).
However using a Vbe multiplier will lead to reduce biasing when hot, which would fix your problem, perhaps at the expense of distortion which increases with under-biasing.
Last edited:
Thanks Mark
I'm doing some more testing right now that seems to support that view i.e. that the increase in bias voltage may be to do with heat itself, rather than the signal going into the devices.
Thanks, NICK
I'm doing some more testing right now that seems to support that view i.e. that the increase in bias voltage may be to do with heat itself, rather than the signal going into the devices.
Thanks, NICK
Note that the level of current where the temperature coefficient reverses is usually way higher in any VFET, 10's of amps or so, which is one of the reasons why laterals are valued, simple stable biasing.
They have a reputation for robustness and lasting because they are used in circuits where the bias is less problematic, with fewer failure modes, and because of the lack of secondary breakdown which means highly reactive loads are less likely to destroy them.
The minus points are high bias current, and large voltage drop under load (both source-drain and source-gate), and the need for full anti-static precautions when handling.
They have a reputation for robustness and lasting because they are used in circuits where the bias is less problematic, with fewer failure modes, and because of the lack of secondary breakdown which means highly reactive loads are less likely to destroy them.
The minus points are high bias current, and large voltage drop under load (both source-drain and source-gate), and the need for full anti-static precautions when handling.
True, but there is also the hidden cost of unreliability to factor in, and the fact one lateral might handle a load that two BJTs are needed for due to secondary breakdown limitation.
It seems both power bjt's and latfet's are, or are, becoming single sourced parts these days.
I do not know Mark, are you preferring the Exicon (250w Dual Die) over one or more of 200/230W Onsemi MJL... types for your OPS?
With the latfet you are saving on a pair of MJE1503x drivers too, so that factors into the total $.
I do not know Mark, are you preferring the Exicon (250w Dual Die) over one or more of 200/230W Onsemi MJL... types for your OPS?
With the latfet you are saving on a pair of MJE1503x drivers too, so that factors into the total $.
I hear this said often on the forum but Renesas types 2SK1058/J162 are still offered by Ampslab in the U.S.
Power Mosfets for Audio Amplifiers @ AmpsLab
Doubtless, they're the same as Exicon's single die package types with a similar high price but at least you have a second source of probably a more familiar product from a vendor who's closer to home.
Last edited:
You might want to take a look at this application note -- it was Nat Semi and the author's name has been removed (but he was on DIYAUDIO years ago).
http://www.ti.com/lit/an/snaa045a/snaa045a.pdf
http://www.ti.com/lit/an/snaa045a/snaa045a.pdf
I use leds with bypass cap for bias so that when the rails droop under load,
the bias stays constant.
V+
r
e
s
---------gate
l c
e a
d p
---------gate
r
e
s
V-
the bias stays constant.
V+
r
e
s
---------gate
l c
e a
d p
---------gate
r
e
s
V-
Last edited:
Interesting. What sort of bias voltage are you achieving with that arrangement? I have built something similar in my previous interations, except I used diodes - two in series with a bypass capof 100nf, in fact; so bias voltage of arond 1.3v. Sounded very good.
There does seem the (relatively low) risk of a blown diode taking your bias network open circuit and toasting one's amp! I am leaning towards a resistor in parallel with a trimmer pot, so if the wiper on the trimmer fails, the bias will increase but not to the point of smoke! I have this running right now and it sounds fine. I also have a regulated power supply as I get smaller filter caps that way, so rail-droop is fairly small.
I did a similar LED/cap thing in a pre-amp recently and I saw the LED flicker a lot when big transients came through, suggesting rail droop might not be completely mitigated by the diode. Seems that the sure way to mitigate this is with a Vbe or Vgs multiplier. Your milegae may vary, as they say.
Thanks again, NICK
Last edited:
Thank you for this interesting link. I did never suspect TI supporting these obselete parts in 2013😱