Are there generalizations about the sound signature that can be made between the IRF510PBF, IRF520PBF, IRF530PBF, and IRF540PBF along with their IRF9xxxPBF complements?
I've searched and it looks like the gate capacitance effects the treble extension and the trans conductance effects the slew rate, but is it really that easy and clear cut?
I have all four, but before I go through the efforts of swapping them out, I wanted to have a theoretical idea of what to expect.
I've searched and it looks like the gate capacitance effects the treble extension and the trans conductance effects the slew rate, but is it really that easy and clear cut?
I have all four, but before I go through the efforts of swapping them out, I wanted to have a theoretical idea of what to expect.
Sounds like a job for a white noise generator and a spectrum analyser.
Unless your doing something silly with them they should all give a good sound.
Unless your doing something silly with them they should all give a good sound.
If you just use one of those as a follower, just for buffering,
without any feedback
then it might be diffences in sound.
Especially between 510 and 530, 540.
Such a follower 'amplifier' is for example Szekeres HeadPhoneAmp:
HeadWize - Project: A Class A MOSFET Headphone Driver by Greg J. Szekeres
When using MOSFET or any other power transistor
inside an amplifier with plenty of Feedback
then there is a correction taking place.
This will overrun most any quality of single components.
The 'sound' is here often dependant on the input stage
where the feedback comes in and is used to reduce distortion from the rest of Amp.
without any feedback
then it might be diffences in sound.
Especially between 510 and 530, 540.
Such a follower 'amplifier' is for example Szekeres HeadPhoneAmp:
HeadWize - Project: A Class A MOSFET Headphone Driver by Greg J. Szekeres
When using MOSFET or any other power transistor
inside an amplifier with plenty of Feedback
then there is a correction taking place.
This will overrun most any quality of single components.
The 'sound' is here often dependant on the input stage
where the feedback comes in and is used to reduce distortion from the rest of Amp.
Thanks guys. It's for a lower power P-P amp. A pair of complements per channel in SE for about 7W bias around 200mA on 30V rails. All DC coupled. With better heat sinking I can crank it up later.
Hi
Lineup has a point about the feedback. A feedback/feedforward drive circuit will 'adjust' Vgs as needed for adequate conductance. This Vgs vs Id value will not be the same for the different devices. Without any fb/ff, the distortion components will not be be the same. I think the driving circuitry will have greater influence on sound than the transistors by themselves.
Lineup has a point about the feedback. A feedback/feedforward drive circuit will 'adjust' Vgs as needed for adequate conductance. This Vgs vs Id value will not be the same for the different devices. Without any fb/ff, the distortion components will not be be the same. I think the driving circuitry will have greater influence on sound than the transistors by themselves.
I believe a 520 is just two 510 chips in the same package etc. - so everything more or less just halves or doubles as the case may be. If it is your own amp design then you can work out the likely effects of this.
Should it be current feedback only, or both with voltage feedback? And how many feedback (what gain) is needed (what maximum gain is needed)? Will this feedback auto adjust / correct DC offset too (e.g. due tome small miss match of complement)?
Thx,
Ervin L
Well, there is more than one way to skin that cat. CFA is one way but it works a bit differently than VCA. I've never tried local Vfb within a Cfb loop. Could be an interesting concept.
You can use local fb around the output stage, but that will not affect the DC offset or so much the rest of the amplifier. It will help to linearize the output transistors though. If you use vertical cellular type switching mosfets such as the IRFs or similar, there will be a significant amount of distortion components generated by this type of device due to non-linear Gm and quite a bit of it at very high frequencies. This is one reason they are not popular as linear output stages. Global fb will not be sufficient to reduce these components because of the many stages and time delay within the loop, and the BW of the amplifer gain stage limit how much 'correction' can be obtained. But then they were designed for switching after all, eh? One type of gate drive circuit I find works quite well is HEC (Hawksford error correction) adapted for use with these mosfets. Bob Cordell's EC mosfet pwr amp paper is a good reference for this. It is based on Malcolm Hawksford's paper, EC for Darlington output stage. One major advantage of this fb/ff EC topology aside from linearization is the amplifier output becomes much less load dependent as changes in the gate drive signal related to the variable load Z, as speakers are, is included in the error signal. Damping factor is significantly improved.
As for DC offset, I prefer a DC servo.😉
You can use local fb around the output stage, but that will not affect the DC offset or so much the rest of the amplifier. It will help to linearize the output transistors though. If you use vertical cellular type switching mosfets such as the IRFs or similar, there will be a significant amount of distortion components generated by this type of device due to non-linear Gm and quite a bit of it at very high frequencies. This is one reason they are not popular as linear output stages. Global fb will not be sufficient to reduce these components because of the many stages and time delay within the loop, and the BW of the amplifer gain stage limit how much 'correction' can be obtained. But then they were designed for switching after all, eh? One type of gate drive circuit I find works quite well is HEC (Hawksford error correction) adapted for use with these mosfets. Bob Cordell's EC mosfet pwr amp paper is a good reference for this. It is based on Malcolm Hawksford's paper, EC for Darlington output stage. One major advantage of this fb/ff EC topology aside from linearization is the amplifier output becomes much less load dependent as changes in the gate drive signal related to the variable load Z, as speakers are, is included in the error signal. Damping factor is significantly improved.
As for DC offset, I prefer a DC servo.😉
Last edited:
Wow, thanks CBS240. That's a lot of info for a newbie to digest.
Perhaps you could shed a little theoretical light on a given scenario:
If the local feedback was optimized for IRF540s, such that they are as linear as they can be, what is the effect on the sound signature when IRF510s are inserted in the circuit with no other changes? I understand that the THD will be higher, but does it skew the sound towards a bass emphasis or treble emphasis, or is there simply not enough information to draw any useful conclusions?
Of course, this leads to the next level of questions, what happens if the IRF540s are replaced by IRFZ24s or 2SK2013s...
Perhaps you could shed a little theoretical light on a given scenario:
If the local feedback was optimized for IRF540s, such that they are as linear as they can be, what is the effect on the sound signature when IRF510s are inserted in the circuit with no other changes? I understand that the THD will be higher, but does it skew the sound towards a bass emphasis or treble emphasis, or is there simply not enough information to draw any useful conclusions?
Of course, this leads to the next level of questions, what happens if the IRF540s are replaced by IRFZ24s or 2SK2013s...
Try it. To be honest I'd bet you can't hear any difference. I've swapped MOSFETs into BJT amps for audiophile friends and they couldn't tell any difference, and the difference between 500 series MOSFETs is even smaller.
The lower the Gate capacitance the lighter and brighter the sound, the more the capacitance (upto a limit set by the current available in the driver circuit) the more authoritative the bass and darker the background. Check some of the Pass Labs threads for the details. I have tried this in many circuits and can vouch for this.
In most amplifier circuits the gate capacitance makes little difference because it is too small at audio frequencies.
With a 390R gate resistor into a IRFP240 you have about a 170KHz low pass filter.
So with even smaller resistors the affect of the gate capacitance is negligible.
That is an incorrect assumption. Yes it is too small to worry about the RC rolloff introduced, but the load on the VAS is something to consider.
I agree with those that think there is no sound difference just by changing output transistors.
Not when we talk amplifiers with global negativ feedback.
Suppose one guy say he can hear that IRF540 is used.
Now one other guy can tell that this or that capacitor is used.
And a third guy can tell that those audiophile grade resistors are in amplifier.
This means that any component can be pinpointed by 'sound'.
Not only that, but we can pick it out from all the rest of amplifier.
But how can we know if it is transistor or capacitor that makes the sound???
And how do we listen for transistor, without letting all the other components confuse??
Not when we talk amplifiers with global negativ feedback.
Suppose one guy say he can hear that IRF540 is used.
Now one other guy can tell that this or that capacitor is used.
And a third guy can tell that those audiophile grade resistors are in amplifier.
This means that any component can be pinpointed by 'sound'.
Not only that, but we can pick it out from all the rest of amplifier.
But how can we know if it is transistor or capacitor that makes the sound???
And how do we listen for transistor, without letting all the other components confuse??
No we have them to lower bandwidth to prevent oscillation.
More gate capacitance puts a higher load on the VAS. It's a common misconception that FETs are "high impedance" and don't load the VAS. The gate resistors will "shield" the VAS to a degree but there will still be some load, especially as with bigger gate capacitance you need a smaller gate resistor to maintain the same -3dB point.
Last edited:
Assuming that transconductance is sufficient for the load, I doubt there would be any audible difference. If there is no fb, I suspect there may be. IMO for an audio output stage, devices with higher Gm would be more suitable to drive a speaker.
These mosfets make very effective oscillators due to their equivalent internal components. Even leaving the leads (+ PCB traces) too long can create a more efficient oscillator because at 20+ MHz the inductance of the lead pins is significant. Therefore the 'oscillator' has to be dampened. This can be done with a slight HF roll off limiting the effective BW to like less than 200KHz with just a gate resistor, or a gate RC filter with a steeper roll off allowing higher BW from the mosfet and still effectively dampen the oscillator. I believe the latter is the way to go. Don't be afraid of larger gate capacitance (BTW CISS = Cgs + Cgd) or take it directly as the limitation. This being said, it still requires a driver stage(s).
Last edited:
- Status
- Not open for further replies.