Hi Space,
The reason I say select on test is that I dislike pre-sets, you never know what they get up to when your not watching
However I would initially use a good quality multiturn pot for R8 and when alls O.K. make the value up from a couple of parallel resistors, R8 and R8A on the PCB. A 470 ohm preset should be fine, initially set to Zero ohms for minimum current.
The only Zener is the 12 volt one for the IC supply, nothing else to add. If you were testing with a much lower supply, say +/- 20 or so then the 10 K feeding it would need dropping-- thats all.
The reason I say select on test is that I dislike pre-sets, you never know what they get up to when your not watching
However I would initially use a good quality multiturn pot for R8 and when alls O.K. make the value up from a couple of parallel resistors, R8 and R8A on the PCB. A 470 ohm preset should be fine, initially set to Zero ohms for minimum current.
The only Zener is the 12 volt one for the IC supply, nothing else to add. If you were testing with a much lower supply, say +/- 20 or so then the 10 K feeding it would need dropping-- thats all.
Hello Lumba Ogir,
"Why does this topology create a distortion spectrum so close to the ears liking" That surely is the ultimate question ! The ear ( well certainly mine ) does not seem to like technical perfection.
I suspect going back to a single ended input stage has a lot to do with it, yes the distortion is initially higher, but it is mainly even harmonic and rises at a much lower rate with increasing frequency than the more usual LTP ( which produce more odd harmonics ).
It never ceases to amaze me that the ear can be so discriminating, but it really does seem to have this ability.
"Why does this topology create a distortion spectrum so close to the ears liking" That surely is the ultimate question ! The ear ( well certainly mine
) does not seem to like technical perfection. I suspect going back to a single ended input stage has a lot to do with it, yes the distortion is initially higher, but it is mainly even harmonic and rises at a much lower rate with increasing frequency than the more usual LTP ( which produce more odd harmonics ).
It never ceases to amaze me that the ear can be so discriminating, but it really does seem to have this ability.
Hello Otto88,
I can't really say it's designed for sustained output into 3 ohms certainly not with a single output pair. Theres a world of difference between a real speaker with impedance dips to 3 ohms reproducing music and a 3 ohm resistive test loading.
As I say at the start, it's designed for a domestic setting where it will cope just fine with any "normal" speaker loading.
You can always parallel outputs or use "double die" FET's and tweak the driver stage to cope with the additional extra capacitive loading but I can't recommend doing ANY of this without listening tests.
The whole reason for the amps existance is it's sonics, if you start changing things it's almost certainly going to effect this.
I can't really say it's designed for sustained output into 3 ohms
certainly not with a single output pair. Theres a world of difference between a real speaker with impedance dips to 3 ohms reproducing music and a 3 ohm resistive test loading.As I say at the start, it's designed for a domestic setting where it will cope just fine with any "normal" speaker loading.
You can always parallel outputs or use "double die" FET's and tweak the driver stage to cope with the additional extra capacitive loading but I can't recommend doing ANY of this without listening tests.
The whole reason for the amps existance is it's sonics, if you start changing things it's almost certainly going to effect this.
Mooly said:
Samuel Jayaraj said:
I can add to this about MOSFET gate stoppers.
Nelson Pass uses, almost constantly, 220 Ohms for IRF240 / IRF9240.
(I know, that Lateral is a bit diffrent than IRF HEXFET.)
--------------------
But anyway,
some time ago I did some spice mosfet amp design, using IRF240 / IRF9240.
I tried different values for Gate Stoppers.
Then I had a look at Bandwith Curve for the transistor ( AC Analysis ).
I started out low 22 Ohm, 47 Ohm, 100, 150, 220, 330, 470.
Result was almost Frightening
Master nelson certainly has not chosen 220 Ohm, just by chance, randomly.
Because at or around 220 Ohm (150-300) my simulation showed the most nice frequency curve sloop, without overshoot and bad PHASE Angle.
Of course I have seen Nelson Pass use a few other values, than 220 Ohm, in some circuits.
For example 100 Ohm for IRF240/ IRF9240.
So, to a certain degree, it depends on the circuit use, what value is optimal.
-------------------
Another thing.If the P-MOS is way different regarding capacitance, we should theoretically,
use different resistor value for this. Compared to the N-MOSFET
regars
lineup
Hello Lineup,
It's nice to know I'm not too wide of the mark with 270 Ohms
I must confess that I don't use a simulator, I tend to work I suppose empirically, particularly on matters like the gate stoppers.
The nice thing with the laterals ( and HEXFET'S ) is that the gate stopper is a user chooseable value, allowing you to define your own upper limit for the particular application.
JLH once mentioned something as well, that because of the extremely high Ft for a lateral that a Lecher Line type oscillator could be formed around a single device depending on the surrounding values of inductance/capacitance etc. Now they are weird things--- I first came across Lecher lines in UHF tuners, and I always wondered why sticking a pin through a piece of coax TV downlead thus shorting the inner and outer could "improve" the signal seen by the receiver on certain channels.
With anything like this amp I put a lot of faith in squarewave testing, it really is very revealing and well, 270 Ohm seemed just right all things considered.
It would be an interesting exercise to modify the design for HEXFET's including a vgs multiplier and see whether the sonics are unchanged.
It's nice to know I'm not too wide of the mark with 270 Ohms
I must confess that I don't use a simulator, I tend to work I suppose empirically, particularly on matters like the gate stoppers.
The nice thing with the laterals ( and HEXFET'S ) is that the gate stopper is a user chooseable value, allowing you to define your own upper limit for the particular application.
JLH once mentioned something as well, that because of the extremely high Ft for a lateral that a Lecher Line type oscillator could be formed around a single device depending on the surrounding values of inductance/capacitance etc. Now they are weird things--- I first came across Lecher lines in UHF tuners, and I always wondered why sticking a pin through a piece of coax TV downlead thus shorting the inner and outer could "improve" the signal seen by the receiver on certain channels.
With anything like this amp I put a lot of faith in squarewave testing, it really is very revealing and well, 270 Ohm seemed just right all things considered.
It would be an interesting exercise to modify the design for HEXFET's including a vgs multiplier and see whether the sonics are unchanged.
Mooly said:
I omitted the vgs multiplier from my first amp and it sounded distorted at low levels, although didnt sound too bad louder.
On the scope I found the negative half of the wave form to be badly distorted which is odd coz it didnt sound that bad.
I later added a vgs multiplier and blew up all the output MOSFETs ! I had put the pot to the wrong end to start with and put far too many volts to the output FET's
I put the pot right and after fitting a new set of MOSFET's I turned the bias up slowly and the negative phase distortion melted away. The amp now sounds great at low levels but also sounds a lot cleaner at louder levels too.
I would definitely recommend the vgs multiplier but be careful when setting it up !!!!
Do you have the multiplier in thermal contact with the outputs ?
It should be possible to get a quiescent current that is reasonably stable against temperature changes in the outputs. Although I have never built a HEXFET amp, I imagine it can't be to different to a BJT output amp in terms of temperature compensation. You would need to match the multiplier characteristics to the tempco of a pair of HEXFET's which will be around -0.3%/deg C.
Regards Karl
It should be possible to get a quiescent current that is reasonably stable against temperature changes in the outputs. Although I have never built a HEXFET amp, I imagine it can't be to different to a BJT output amp in terms of temperature compensation. You would need to match the multiplier characteristics to the tempco of a pair of HEXFET's which will be around -0.3%/deg C.
Regards Karl
Mooly said:
I dont have the vgs transistor on the heatsink.
It is on the driver board.
It is not a mountable transistor anyway, its just a MPSA42.
I dont set it up to an exact current, I just tweak the pot until crossover distortion goes. I would have thought this was OK and on the safe side rather than whacking lots of current unnecesarily through the output transistors.
I am using IRFP240/IRFP9240
Hello Andrew,
Think you would have to derate the PSU voltage as well and think about adding Zeners to limit the gate voltage seen under loading. But the CFP output does stand up well to low impedance loads--- no gain droop, but 3 ohms resistive is tough, not suitable for a single pair of FET's for realistic output levels.
And lets not confuse Space, with any more talk of Zeners
Your not thinking of having bash at it Andrew ? It won't disappoint musically
Think you would have to derate the PSU voltage as well and think about adding Zeners to limit the gate voltage seen under loading. But the CFP output does stand up well to low impedance loads--- no gain droop, but 3 ohms resistive is tough, not suitable for a single pair of FET's for realistic output levels.
And lets not confuse Space, with any more talk of Zeners
Your not thinking of having bash at it Andrew ? It won't disappoint musically