Hello -
No, it doesn't show up on the data sheet. If it did, I wouldn't have anything to complain to IR about, because it would meet their specs.
Maybe they've fixed it by now. I don't really know because I don't use vertical MOSFETs any more.
Best regards,
Charles Hansen
No, it doesn't show up on the data sheet. If it did, I wouldn't have anything to complain to IR about, because it would meet their specs.
Maybe they've fixed it by now. I don't really know because I don't use vertical MOSFETs any more.
Best regards,
Charles Hansen
Charles,
What has been your opion of the Fairchild MOSFETs. So which vendors MOSFET's dos you use in your amps.
Oh the last time I spoke to an application engineer, he told me the the IRF244 and 240 were not recomended for audio amps, HAHAHA. That was the last time I called them.
What has been your opion of the Fairchild MOSFETs. So which vendors MOSFET's dos you use in your amps.
Oh the last time I spoke to an application engineer, he told me the the IRF244 and 240 were not recomended for audio amps, HAHAHA. That was the last time I called them.
Hello -
We used to use Harris almost exclusively. They had licensed the "IRF" designs from International Rectifier. The Harris power semi division was spun off as Intersil, which then sold their discrete product line to Fairchild. I don't know if they're the same as they used to be, but I would assume so. I had forgotten, but the N-channel output devices was from IR. The schematic is now posted at:
http://www.diyaudio.com/forums/showthread.php?postid=334963#post334963
Best regards,
Charles Hansen
We used to use Harris almost exclusively. They had licensed the "IRF" designs from International Rectifier. The Harris power semi division was spun off as Intersil, which then sold their discrete product line to Fairchild. I don't know if they're the same as they used to be, but I would assume so. I had forgotten, but the N-channel output devices was from IR. The schematic is now posted at:
http://www.diyaudio.com/forums/showthread.php?postid=334963#post334963
Best regards,
Charles Hansen
Charles Hansen's experience is documented in my article on
testing Mosfets ( www.passdiy.com ) and he is in fact correct
that only IR P channel parts do this.
However my interpretation is that the transconductance of the
device declines at midband audio, shelving down a couple of
dB. Is a 2 dB gain shelf a problem? Not particularly. We use
P channel devices from both Harris and IR and nobody has
ever noticed or measured a difference in follower applications.
As pointed out in the article, you can conceivably take advantage
of this in Common Source applications if you want more feedback
at lower frequencies (for greater damping factor) and less
at the higher frequencies (for the usual reasons).
Another point worth mentioning - I have seen this effect
intermittently on push-pull N channel circuits, but have not
had the time to chase it down properly.
testing Mosfets ( www.passdiy.com ) and he is in fact correct
that only IR P channel parts do this.
However my interpretation is that the transconductance of the
device declines at midband audio, shelving down a couple of
dB. Is a 2 dB gain shelf a problem? Not particularly. We use
P channel devices from both Harris and IR and nobody has
ever noticed or measured a difference in follower applications.
As pointed out in the article, you can conceivably take advantage
of this in Common Source applications if you want more feedback
at lower frequencies (for greater damping factor) and less
at the higher frequencies (for the usual reasons).
Another point worth mentioning - I have seen this effect
intermittently on push-pull N channel circuits, but have not
had the time to chase it down properly.
Charles,
thanks for your reply!
One thing I forgot to mention is that the Qgd is much greater for IRF9540 compared to 540, is the Qg something you pay attention to?
FET's have a non linear capacitive behaviour.
One other thing I dscovered for couple of years ago when working with a SMPS was that the booster FET, an IRF 840, showed up very badly in radiated EMC measuring.
I had put a lot of work solving the EMC issues with that PSU and later on we ordered a couple of factory made from our subcontractor, back in the EMC room the measuring was realy bad and was up by some 20 dB around 50-300 MHz.
Took me some time to track it down until I discovered that at our laboratory where I used a second source from ST(if I remember it right..) while our subcontractor used IR.
A replaced that IR with ST and everything was back on track.
Well, this is not so much about audio but I have been since that suspicious with IR's FET's, I belive they just switch "harder", I'm just guessing but I suspect this is not good for audio.
BTW, I recently red the Stereophile test between your V3 against M.L. 334(?), just curious if you ever tested your amp's also with NFB?
My question is, if so, did you experience a more "steel fisted" and perhaps a more harsh sound or so?
Regards
thanks for your reply!
One thing I forgot to mention is that the Qgd is much greater for IRF9540 compared to 540, is the Qg something you pay attention to?
FET's have a non linear capacitive behaviour.
One other thing I dscovered for couple of years ago when working with a SMPS was that the booster FET, an IRF 840, showed up very badly in radiated EMC measuring.
I had put a lot of work solving the EMC issues with that PSU and later on we ordered a couple of factory made from our subcontractor, back in the EMC room the measuring was realy bad and was up by some 20 dB around 50-300 MHz.
Took me some time to track it down until I discovered that at our laboratory where I used a second source from ST(if I remember it right..) while our subcontractor used IR.
A replaced that IR with ST and everything was back on track.
Well, this is not so much about audio but I have been since that suspicious with IR's FET's, I belive they just switch "harder", I'm just guessing but I suspect this is not good for audio.
BTW, I recently red the Stereophile test between your V3 against M.L. 334(?), just curious if you ever tested your amp's also with NFB?
My question is, if so, did you experience a more "steel fisted" and perhaps a more harsh sound or so?
Regards
Ultima Thule said:
In a vertical MOSFET, both the capacitance itself and the non-linearities thereof are both many, many times worse than for a lateral MOSFET. That is why I no longer use vertical MOSFETs for audio.
I have never tried explicitly adding feedback to one of our zero feedback designs. In my experience, this is a dead end.
You can take a mediocre circuit and add feedback to produce a good measured result. So one could say that the feedback itself becomes the dominant factor in the circuit's measured performance. In my experience this is also true of the circuit's sonic performance. Using feedback imparts a strong sonic coloration that overwhelms almost any other factor. (Most people don't hear this as a coloration, simply because they have never heard a zero-feedback music system. They think that the "feedback sound" is normal.)
On the other hand, a zero feedback circuit becomes very chameleon-like. If you change anything in the system, that change is easily heard.
For example, the review you referred to was over 4 years old. Since that time we have made two significant upgrades to that amplifier (Ayre V-1x). We have been able to address the areas of weakness that Michael Fremer noted in the original review, but without losing the inherent musicality and natural sound of the zero feedback circuit. For a review of the current version, please refer to:
http://www.soundstage.com/revequip/ayre_v1x_followup.htm
Where the reviewer states, "It’s the most musical solid-state amplifier I’ve heard and a significant improvement over the V-1."
Best regards,
Charles Hansen
I have to disagree that lateral Mosfets are more linear than
vertical, as I see it they simply have lower transconductance,
and some applications prefer that. Put a lateral Mosfet in
a Zen amp and watch the distortion increase dramatically.
vertical, as I see it they simply have lower transconductance,
and some applications prefer that. Put a lateral Mosfet in
a Zen amp and watch the distortion increase dramatically.
Hello Nelson,
I mostly agree with your conclusions, although that's not quite what I said. I said "the non-linearities thereof", the key word being "thereof" where I was referring to the interelectrode capacitances. I stand by that statement.
You are absolutely correct about a vertical MOSFET having much greater transconductance than a lateral MOSFET, and that this is of great importance in the output stage of a power amplifier. So you are correct that if you replace a single vertical MOSFET in that application with a single lateral MOSFET that the distortion will rise significantly (at least under most circumstances).
However, if you replace the single vertical MOSFET with multiple lateral MOSFETs you can achieve the same transconductance while still retaining a significant advantage in the interelectrode capacitances (and the non-linearities thereof).
The drawback now is cost. It typically requires 5 to 10 lateral MOSFETs to replace a single vertical MOSFET, at least in terms of transconductance. Since the cost per device is similar, there is an overall price penalty that is significant when using lateral devices.
Best regards,
Charles Hansen
I mostly agree with your conclusions, although that's not quite what I said. I said "the non-linearities thereof", the key word being "thereof" where I was referring to the interelectrode capacitances. I stand by that statement.
You are absolutely correct about a vertical MOSFET having much greater transconductance than a lateral MOSFET, and that this is of great importance in the output stage of a power amplifier. So you are correct that if you replace a single vertical MOSFET in that application with a single lateral MOSFET that the distortion will rise significantly (at least under most circumstances).
However, if you replace the single vertical MOSFET with multiple lateral MOSFETs you can achieve the same transconductance while still retaining a significant advantage in the interelectrode capacitances (and the non-linearities thereof).
The drawback now is cost. It typically requires 5 to 10 lateral MOSFETs to replace a single vertical MOSFET, at least in terms of transconductance. Since the cost per device is similar, there is an overall price penalty that is significant when using lateral devices.
Best regards,
Charles Hansen
I take your point, but consider that a number of characteristics
of laterals simply do not point to superiority.
Let's take an example of a 2SK2220, a modern lateral part, and
it's vertical counterpart, and IRF230.
Both are rated about the same voltage, current, and wattage,
but the lateral's transconductance is about1 and the vertical's
is about 5.
Ciss capacitance of both parts is the same at 600 pF, but the
Cdss of the lateral is about 3 times higher. The Crss of the
lateral is 1/10 the vertical (score a point there).
The turn-on turn-off times are quite different, with the vertical
being 30 and 50 ns and the lateral being 240 and 90 ns.
So depending on the application, paralleling several lateral
Mosfets to get the transconductance figure of a comparable
vertical device may give you trouble with the capacitance
numbers that pile up.
I don't have a big agenda with laterals vs verticals, I just get
tired of the popular assumption that laterals are always
superior.
😎
of laterals simply do not point to superiority.
Let's take an example of a 2SK2220, a modern lateral part, and
it's vertical counterpart, and IRF230.
Both are rated about the same voltage, current, and wattage,
but the lateral's transconductance is about1 and the vertical's
is about 5.
Ciss capacitance of both parts is the same at 600 pF, but the
Cdss of the lateral is about 3 times higher. The Crss of the
lateral is 1/10 the vertical (score a point there).
The turn-on turn-off times are quite different, with the vertical
being 30 and 50 ns and the lateral being 240 and 90 ns.
So depending on the application, paralleling several lateral
Mosfets to get the transconductance figure of a comparable
vertical device may give you trouble with the capacitance
numbers that pile up.
I don't have a big agenda with laterals vs verticals, I just get
tired of the popular assumption that laterals are always
superior.
😎
Hello Nelson -
Well, we are drawing different conclusions from the same data.🙂
If we parallel five 2SK2220s to achieve the same transconductance as the IRF230 in your example, I assert that the lateral devices will give better performance for audio.
This is because a source follower is almost always used for an audio power amplifier output stage. In this circuit the only interelectrode capacitance that has any real effect is Cgd. The composite lateral device will have about half the capacitance as the vertical device. Furthermore, this capacitance is far more linear in the lateral device as Vds changes.
As an added bonus, the lateral devices are much easier to bias as their zero tempco point is at a very convenient idle current. The vertical device will always require a temperature-compensated bias scheme. The only disadvantage to the composite lateral device is price.
Try it, you may like it.
Best regards,
Charles Hansen
Well, we are drawing different conclusions from the same data.🙂
If we parallel five 2SK2220s to achieve the same transconductance as the IRF230 in your example, I assert that the lateral devices will give better performance for audio.
This is because a source follower is almost always used for an audio power amplifier output stage. In this circuit the only interelectrode capacitance that has any real effect is Cgd. The composite lateral device will have about half the capacitance as the vertical device. Furthermore, this capacitance is far more linear in the lateral device as Vds changes.
As an added bonus, the lateral devices are much easier to bias as their zero tempco point is at a very convenient idle current. The vertical device will always require a temperature-compensated bias scheme. The only disadvantage to the composite lateral device is price.
Try it, you may like it.
Best regards,
Charles Hansen
I forgot to address the switching time issue. If you refer to the data sheet for the 2SK2220, you will see that this is measured with a common source circuit. Therefore these numbers will not apply to the common drain circuit typically used for an audio power amplifier output stage.
I refer you to my previous comment: The transistor doesn't
know what mode it's in, and so the switching time is apples
to apples. Not that I actually care about this spec - I'm
responding more to people who wave it around.
Thermal Vgs stability? A non-issue for a decent designer, and
I've never had to give it more than cursory attention, but then
I use big output stages biased very high and big heat sinks.
( I pities the fool who makes Mosfet AB amps - I pities em! 😉 )
Drain-Source voltage dependent non-linearities show up in
both devices, more at low frequencies with laterals, more at
high frequencies with verticals. Life is tough. Crss nonlinearity?
Cascoding makes that go away, along with some other
distortions. Works miracles for laterals, too.
OK, I say the best argument for Laterals is that they look
more like triodes than the verticals. Assuming you want
triodes.
know what mode it's in, and so the switching time is apples
to apples. Not that I actually care about this spec - I'm
responding more to people who wave it around.
Thermal Vgs stability? A non-issue for a decent designer, and
I've never had to give it more than cursory attention, but then
I use big output stages biased very high and big heat sinks.
( I pities the fool who makes Mosfet AB amps - I pities em! 😉 )
Drain-Source voltage dependent non-linearities show up in
both devices, more at low frequencies with laterals, more at
high frequencies with verticals. Life is tough. Crss nonlinearity?
Cascoding makes that go away, along with some other
distortions. Works miracles for laterals, too.
OK, I say the best argument for Laterals is that they look
more like triodes than the verticals. Assuming you want
triodes.
Commentable Thoughts
THANX NELSON PASS FOR CLARIFYING THE STATEMENT ABOUT VERTICAL MOSFETS ESPECIALLY N-CHANNEL TYPE FROM IRF TO ALL OTHER PERSONS WITH MISCONCEPTIONS.
With Regards
Ampman
THANX NELSON PASS FOR CLARIFYING THE STATEMENT ABOUT VERTICAL MOSFETS ESPECIALLY N-CHANNEL TYPE FROM IRF TO ALL OTHER PERSONS WITH MISCONCEPTIONS.
With Regards
Ampman
Not really, Cgs is what limits how fast you can drive them. Generally MOSFETs are driven through a gate resistor to quench oscillations, and the RC time constant of the gate resistor and Cgs limits the speed of the device. I have played around with very small gate resistors and ferrite beads for gate drive to speed up the devices, and it does speed them up. Cgd cannot be ignored, simply because the all the device capacitances, strays and lead inductance can make a seemingly stable design oscillate nicely.Charles Hansen said:
Granted the tempco is positive for lower currents, does bias current drift enough to really matter, if you have source resistors to compensate for it as well as the typical Vbe multiplier bias, which tracks ambient? ICharles Hansen said:
Don't get me wrong, I like the laterals, but they are very limited in current capability (8A max), voltage ratings (200v max) and power dissipation per device (125W max), in addition to being higher priced and having limited sources. Vertical mosfets easily exceed these parameters and have much higher transconductance, multiple sources and about half the cost per device. It becomes a complex tradeoff, not a simple choice.Charles Hansen said:
May I put my foot in mouth with a naive question?
Two capacitors in series have reduced combined capacitance.
If capacitance is such a problem in driving Mosfets - what would happen if one used a gate capacitor with, or instead of, a gate resistor? Would that not decrease total capacitance (Input - gate - source and input - gate - drain being caps in series then)?
(I know little about all this, just learning)
MBK
Two capacitors in series have reduced combined capacitance.
If capacitance is such a problem in driving Mosfets - what would happen if one used a gate capacitor with, or instead of, a gate resistor? Would that not decrease total capacitance (Input - gate - source and input - gate - drain being caps in series then)?
(I know little about all this, just learning)
MBK
ABO said:
Mostly joking of course, but it is true that Mosfets excel at
high bias Class A, and generally have more distortion than
BJT's at low bias AB operation. 😎
slowhands said:
Nope, not in a source follower configuration. In this case the source "follows" the gate (clever naming scheme, eh?). There is (essentially) no change in voltage between gate and source, and therefore Cgs is never charged or discharged in a source follower.
Hi MBK,
Well I like people who try to think of solutions to problems. Like my ideas, most do not pan out. But keep up the effort since that way is how new discoveries come about.
If two caps are placed in series, then a capacitive voltage divider effect occurs, in other words, drive signal is lost and is distributed in proportional to the values of the caps. It would take more voltage to charge and discharge the gate. Such a technique may be able to match a higher voltage drive to drive the gate though.
Well I like people who try to think of solutions to problems. Like my ideas, most do not pan out. But keep up the effort since that way is how new discoveries come about.
If two caps are placed in series, then a capacitive voltage divider effect occurs, in other words, drive signal is lost and is distributed in proportional to the values of the caps. It would take more voltage to charge and discharge the gate. Such a technique may be able to match a higher voltage drive to drive the gate though.
- Status
- Not open for further replies.