Traditionally relays have been used for speaker protection in amplifiers to quickly disconnect the speaker in case there is an offset on the output, due to componet failure.

Relays are simple and probably to no contribute to distortion, but their contacts age and can cause problems.

Because of this some designers use mosfets instead:

speaker protector

This circuit (attached) has the first overtone -90dB below which is quite OK and inaudible, but I was more of wondering what causes distortion in these cases and how to select the optimal mosfet for the lowest distortion.

One assumes that V2 drives the mosfets to that they are fully on. One possible distortion source is Rdson variability with Id. For most mosfets this is a flat line for a very great variation in Id so it should be insignificant, but are there other possible sources?

Would a reasonable strategy be to look for mosfets with the lowest possible Rdson and least variablity with Id?

Edit: Changed to a true floating drive, V2 is generated by a VOM1271T:

floating drive

Distortion is now undetectable. The one I saw before was caused by Vgs not being constant:

enter image description here

LTSpice model:

Version 4
SHEET 1 880 680
WIRE 112 128 -16 128
WIRE 256 128 208 128
WIRE 480 128 352 128
WIRE -16 144 -16 128
WIRE 480 176 480 128
WIRE 192 208 192 176
WIRE 240 208 192 208
WIRE 272 208 272 176
WIRE 272 208 240 208
WIRE 240 224 240 208
WIRE -16 320 -16 224
WIRE 240 320 240 304
WIRE 240 320 -16 320
WIRE 480 320 480 256
WIRE 480 320 240 320
WIRE 240 336 240 320
FLAG 240 336 0
FLAG -16 128 IN
FLAG 480 128 OUT
SYMBOL nmos 112 176 R270
SYMBOL nmos 352 176 M270
SYMBOL voltage -16 128 R0
WINDOW 123 24 124 Left 2
WINDOW 39 24 152 Left 2
SYMATTR Value SINE(0 1 1000 0 0 0)
SYMBOL voltage 240 208 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR Value 10
SYMBOL res 464 160 R

But back to my initial question.. Am I correct in assuming that I should look for a mosfet with as little change in Rdson with Id? So not this one?


But rather this one:


For a typical power amplifier, the curve would approximate a flat line.

  • 1
    \$\begingroup\$ Rds(on) also varies with Vgs. Consider how you might exploit this to keep Rds(on) approximately constant. \$\endgroup\$ – Brian Drummond Sep 8 '19 at 15:28
  • \$\begingroup\$ Yes. I know. Vgs will be constant in this scenario and well above Vgsth \$\endgroup\$ – AndersG Sep 8 '19 at 17:01
  • \$\begingroup\$ I'm afraid Vgs constant is not accomplished by the diagram you posted. Vg is constant instead which gives Vgs=Vg-Vout while the MOSFETs are on. Driving the floating gates is in fact the problem with that circuit. \$\endgroup\$ – carloc Sep 8 '19 at 17:32
  • \$\begingroup\$ Yes. You are correct. The real circuit will use a VOM1271T which is a floating mosfet driver. \$\endgroup\$ – AndersG Sep 9 '19 at 5:22
  • \$\begingroup\$ And making the drive floating made all the difference! See above. This is in fact what I have in the lab prototype. With that I have measurable additions, but I suspect that they are mostly hum and noise because of the way the prototype looks (solderless breadboard) \$\endgroup\$ – AndersG Sep 9 '19 at 5:38

Am I correct in assuming that I should look for a mosfet with as little change in Rdson with Id? So not this one?... But rather this one:

The 320A (?) FET has much lower Rdson to much higher current than the 5A FET, so it would seem like the obvious choice due to its lower distortion and power loss. However there may be other factors involved such as voltage ratings, Gate charge, physical size and mounting requirements, cost etc.

A specification of 'as little...' is generally not good design philosophy for a commercial product. Your 320A FET looks great, but there may be another one that's even better (and costs even more). When do you decide what is good enough?

Once the resistance gets low enough to be 'inaudible', going lower is probably a waste of money - unless you can convince customers that the over-specified component is worth paying for.

| improve this answer | |
  • \$\begingroup\$ Yes. I was specifically asking about "what" would affect distortion. That other factors play in goes without saying. The voltage and current ratings of the mosfet need to match the power output of the amp, taking into account a safety margin. Vgs needs to be low enough to be fully driven by the 1271. Gate charge is probably irrelevant as the mosfet will be switched on after a delay and then remain on indefinitely, unless the amp develops a fault. So, yes, "good enough" in this case would be a circuit that has a distortion say 10x less than the amp it protects. Going any lower is indeed a waste \$\endgroup\$ – AndersG Sep 11 '19 at 7:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.