Hi Elso Kwak,
In what way?
That someone else had half invented the same light bulb on a different continent well before me?
I am not bothered about that myself. The point of the amplifier is still simple and direct and something that even non DIYers might feel like having a go at constructing.
That person obviously didn't make much headway after the patent else we would all be listening this topology amplification now, whereas to most people is is a novel and strange beast...
... but I would guess still of interest as there have been well over 20000 views to this thread.
It is after all something that many people who are into making amplifiers as a hobby or even professionally can quickly try out with bits probably already sitting on or under the bench.
People tend to forget how much the Internet has changed the way we think, interact and access information.
10 years ago I had a prized collection of at least 16 feet of electronics data books including pretty much the entire Phillips/Mullard, TI, Motorola, AMD, Intersil, MIPS, Intel, Zilog, Dallas, PMI, Analog Devices, Nat Semi, Burr Brown, Hitachi, Harris, Lattice, and many others including a specialize section on RISC processors and graphics parts.
Now I only have a quarter of that and it is easier to go to the manufactures website and download a PDF than it is to get up from the sofa and walk to the other room. Plus I can check availability, production status (is is going to be obsolete), price and get some samples which are as often as not (well from TI anyway) delivered DHL the next day from the US to my front door here in London.
Ten years ago - trying to get samples out of most manufacturers unless one was in a big company - forget it!
And this also relates to Patent searches and documents. It is not that long ago (a year or two) that the British Patent Office Internet server was closed down overnight and at the weekend.
Ten years ago doing a patent search meant going up to London and physically going though reference books. I know, I did it. And it cost money.
Anyway, as I said the point of the amplifier is that it is simple, non-feedback, simple, non-feedback (did I mention the lack of overall feedback?), and to my ear far better than anything else I could afford to buy or build.
I appreciate that it is not to everyones tastes and some will probably not like the "valve" quality although I do not hear it as warm. But this is beginning to be subjective again and hence very much open to personal interpretation.
In part because of this I have ordered a 192/24 sound card and upgrades and drivers for WinaudioMLS software so I can make some proper measurements rather than the basic % distortion that I can do with the HP.
So I am hoping that by the end of next week I will be able to scatter graphs left right and center 🙂
Best wishes,
Susan.
Elso Kwak said:
In what way?
That someone else had half invented the same light bulb on a different continent well before me?
I am not bothered about that myself. The point of the amplifier is still simple and direct and something that even non DIYers might feel like having a go at constructing.
That person obviously didn't make much headway after the patent else we would all be listening this topology amplification now, whereas to most people is is a novel and strange beast...
... but I would guess still of interest as there have been well over 20000 views to this thread.
It is after all something that many people who are into making amplifiers as a hobby or even professionally can quickly try out with bits probably already sitting on or under the bench.
People tend to forget how much the Internet has changed the way we think, interact and access information.
10 years ago I had a prized collection of at least 16 feet of electronics data books including pretty much the entire Phillips/Mullard, TI, Motorola, AMD, Intersil, MIPS, Intel, Zilog, Dallas, PMI, Analog Devices, Nat Semi, Burr Brown, Hitachi, Harris, Lattice, and many others including a specialize section on RISC processors and graphics parts.
Now I only have a quarter of that and it is easier to go to the manufactures website and download a PDF than it is to get up from the sofa and walk to the other room. Plus I can check availability, production status (is is going to be obsolete), price and get some samples which are as often as not (well from TI anyway) delivered DHL the next day from the US to my front door here in London.
Ten years ago - trying to get samples out of most manufacturers unless one was in a big company - forget it!
And this also relates to Patent searches and documents. It is not that long ago (a year or two) that the British Patent Office Internet server was closed down overnight and at the weekend.
Ten years ago doing a patent search meant going up to London and physically going though reference books. I know, I did it. And it cost money.
Anyway, as I said the point of the amplifier is that it is simple, non-feedback, simple, non-feedback (did I mention the lack of overall feedback?), and to my ear far better than anything else I could afford to buy or build.
I appreciate that it is not to everyones tastes and some will probably not like the "valve" quality although I do not hear it as warm. But this is beginning to be subjective again and hence very much open to personal interpretation.
In part because of this I have ordered a 192/24 sound card and upgrades and drivers for WinaudioMLS software so I can make some proper measurements rather than the basic % distortion that I can do with the HP.
So I am hoping that by the end of next week I will be able to scatter graphs left right and center 🙂
Best wishes,
Susan.
Re: Re: Re: Re: Re: Re: Oh, I forgot....
You are right they changed their policy.
Irfanview I only use for downloaded files. It is not incorporated in my browser.
Espacenet works with PDF files which you can print or download.
darkmoebius said:
The US Patent office offers the entire patent form in a ".tiff" image form for viewing. If you look under the "Images" link, it will lead you there.
I think the problem you are having is because IRfan is a viewer that is external to the browser, so you have to download each individual image in order to see them all. The menu on the blue bordered left side of the page provides links to view the "front page", "drawings", "specifications", and "claims". I assume that clicking these will launch the individual images in IRfan.
The big problem with that method is when viewing 10, 20, or 30 page documents. Very clinky.
On the other hand, if you use Alternatiff, it allows you to thumb through the entire document directly from the Patent webpage without any hassle by simply clicking the gold arrows on the left.
And as Netlist pointed out, it is very simple to zoom, print, save, etc. by simply "right-clicking" your mouse.
Download and install Alternatiff, I think you'll prefer it for this particular task. Be sure to set it to handle ".tiff filetypes during installation. Feel free to emial me if you need more help.
[And now, back to our regular programming]
You are right they changed their policy.
Irfanview I only use for downloaded files. It is not incorporated in my browser.
Espacenet works with PDF files which you can print or download.
Re: Re: Commentable thoughts
Hi SUSAN-PARKER
ThanX for Transformer details.
Mosfets are already a type of variable resistors whose conduction is controlled by gate to source voltage.
The Current Sharing is excellent performed by their POSITIVE TEMP. COEFF
In our history of amp fabrication we had never implemented any type of SOURCE or DRAIN resistors as long as Mosfet output stage is concerned .
The Case is different with BJT's.
With Regards,
WORKHORSE TECHNOLOGIES
Susan-Parker said:
Hi SUSAN-PARKER
ThanX for Transformer details.
Mosfets are already a type of variable resistors whose conduction is controlled by gate to source voltage.
The Current Sharing is excellent performed by their POSITIVE TEMP. COEFF
In our history of amp fabrication we had never implemented any type of SOURCE or DRAIN resistors as long as Mosfet output stage is concerned .
The Case is different with BJT's.
With Regards,
WORKHORSE TECHNOLOGIES
If you use lateral power MOSFETs, I agree that you can get to a positive Vgs tempco with a fairly modest bias current.
However, with the usual vertical MOSFET (an IRF540, for example), the positive tempco bias current is so high as to make it unattainable for practical applications, hence source resistors or temperature compensated bias schemes. For normally encountered bias currents, vertical power MOSFETs have a negative Vgs tempco. With small vertical MOSFETs, it is possible to bias them into the positive tempco region without cooking them. I don't have the data for the IR-type devices right at my fingertips, but I had a look at the vertical devices in the Supertex data book. Their 0.7ohm MOSFETs have a tempco crossover point of about 1A - not too bad as long as you run with a modest Vds, though the higher junction to case thermal resistance of the smaller FETs will work against you. If you go to a 0.3 ohm device, the crossover current jumps to ~5A - kinda scary for a TO-220. Maybe the device doesn't run away, but it's hotter than blazes and you have to bust your chops to keep it from melting. I wouldn't try it. You could try cascoding the FET with another device to clamp the VDS to a lower level, but your'e still going to be blasting the top of the cascode pair. A lower resistance device will have an even higher crossover point.
However, with the usual vertical MOSFET (an IRF540, for example), the positive tempco bias current is so high as to make it unattainable for practical applications, hence source resistors or temperature compensated bias schemes. For normally encountered bias currents, vertical power MOSFETs have a negative Vgs tempco. With small vertical MOSFETs, it is possible to bias them into the positive tempco region without cooking them. I don't have the data for the IR-type devices right at my fingertips, but I had a look at the vertical devices in the Supertex data book. Their 0.7ohm MOSFETs have a tempco crossover point of about 1A - not too bad as long as you run with a modest Vds, though the higher junction to case thermal resistance of the smaller FETs will work against you. If you go to a 0.3 ohm device, the crossover current jumps to ~5A - kinda scary for a TO-220. Maybe the device doesn't run away, but it's hotter than blazes and you have to bust your chops to keep it from melting. I wouldn't try it. You could try cascoding the FET with another device to clamp the VDS to a lower level, but your'e still going to be blasting the top of the cascode pair. A lower resistance device will have an even higher crossover point.
To Susan -
Though ecological considerations are nice, my real reason for trying to reduce the current in the bias FET is to get away witn a small heat sink on the current source driving it. You can soak up some of the power away from the current source with a series resistor (I'll probably do some of that in a very modest way for this design), but for large power dissipations, that's just another thing or string of things getting hot, and there's already enough of those with a class A design, anyway. ('
')
If I were a real ecofreak, I'd be running all class D, but that's another thread altogether. (Rant on) I design switching power supplies for a living, but I don't put them in or near the audio stuff I design - same thing for switching power amps, though I might make an exception for a sub amp. I liken using switchers to power an audiophile-type linear application to very carefully sweeping out every corner of a room untill it's spotless, then dumping a nice big bucket of offal in the middle of the floor I just cleaned. (Rant off.)
Though ecological considerations are nice, my real reason for trying to reduce the current in the bias FET is to get away witn a small heat sink on the current source driving it. You can soak up some of the power away from the current source with a series resistor (I'll probably do some of that in a very modest way for this design), but for large power dissipations, that's just another thing or string of things getting hot, and there's already enough of those with a class A design, anyway. ('
')If I were a real ecofreak, I'd be running all class D, but that's another thread altogether. (Rant on) I design switching power supplies for a living, but I don't put them in or near the audio stuff I design - same thing for switching power amps, though I might make an exception for a sub amp. I liken using switchers to power an audiophile-type linear application to very carefully sweeping out every corner of a room untill it's spotless, then dumping a nice big bucket of offal in the middle of the floor I just cleaned. (Rant off.)
Hi Wrenchone,
Thank you for your further detailed and insightful posts.
I have tried some of these lateral types but they are horribly expensive at 20 to 30 dollars each and the matching between same batch devices seems a lot poorer than with the cheepie IRF family parts.
A pair of APT20M20LLL for example were 100 mV apart when biased circa 250 mA.
I wouldn't mind paying the money for a matched pair, but this is too much when one might have to buy 25 pieces to get good matched sets.
From RS (A UK component distro) I get the following prices:
IRFP140N IR TO247 UKP 1.81
IRF540N IR TO220 UKP 2.93
IRFP150N IR TO247 UKP 4.36
IRF150 - TO3 UKP 12.27
(I have just discovered that I am wealthy, I have a tray of some 30 pieces of TO3 IRF150's! - These were the parts I did my original development work with.)
The answer here would be to use twisted multi filar (Litz style perhaps) wire for each primary and split the number of strands across the multiple devices. E.g. 4 off 0.5 mm instead of one 1.0 mm wire, or 3 off 0.71 instead of one 1.25mm wire.
This way each mosfet is tightly coupled but with an individual DC path to ground. Thus any bias mismatch will not cause current hogging as each device is "on it's own".
Simple really.
http://www.mwswire.com/twistite1.htm
BTW.
I have checked the prices of non-inductive low ohm power film resistors and they are expensive:
MP915 Series 15W in TO-126 are UKP 4.71 each
MP930 Series 30W in TO-220 are UKP 5.73 each
Available in 0.02, 0.05, 0.1, 0.2, 0.5, 1.0, etc. ohm
Even HS25 Series wire wound are UKP 1.45 each
Ah, my "ecological consideration" was meant to be humor 🙂
Large heat sinks are expensive and anything that reduces the amount of power unnecessarily being dumped into it is a good thing.
Also I am going to have a go at machining my own heatsinks out of aluminum bar as I can then make fancy fins with my name as part of the design [SUSAN] but I am limited to about 3.5" x 8.5" in XY as this is the range of movement of my milling machine.
The justification of course is to use variable profile fins to cause turbulent flow of the air across the heat exchanging surfaces.
Nice rant, and I am with you much of the way on this one.
However it would be interesting to drive the input of my amplifier with a PWM drive from a digital amplifier output. Or...
... as far as I can see one could also do this directly from a Direct Stream Digital (DSD) source.
Best wishes,
Susan.
Thank you for your further detailed and insightful posts.
wrenchone said:
I have tried some of these lateral types but they are horribly expensive at 20 to 30 dollars each and the matching between same batch devices seems a lot poorer than with the cheepie IRF family parts.
A pair of APT20M20LLL for example were 100 mV apart when biased circa 250 mA.
I wouldn't mind paying the money for a matched pair, but this is too much when one might have to buy 25 pieces to get good matched sets.
From RS (A UK component distro) I get the following prices:
IRFP140N IR TO247 UKP 1.81
IRF540N IR TO220 UKP 2.93
IRFP150N IR TO247 UKP 4.36
IRF150 - TO3 UKP 12.27
(I have just discovered that I am wealthy, I have a tray of some 30 pieces of TO3 IRF150's! - These were the parts I did my original development work with.)
The answer here would be to use twisted multi filar (Litz style perhaps) wire for each primary and split the number of strands across the multiple devices. E.g. 4 off 0.5 mm instead of one 1.0 mm wire, or 3 off 0.71 instead of one 1.25mm wire.
This way each mosfet is tightly coupled but with an individual DC path to ground. Thus any bias mismatch will not cause current hogging as each device is "on it's own".
Simple really.
http://www.mwswire.com/twistite1.htm
BTW.
I have checked the prices of non-inductive low ohm power film resistors and they are expensive:
MP915 Series 15W in TO-126 are UKP 4.71 each
MP930 Series 30W in TO-220 are UKP 5.73 each
Available in 0.02, 0.05, 0.1, 0.2, 0.5, 1.0, etc. ohm
Even HS25 Series wire wound are UKP 1.45 each
wrenchone said:To Susan -
Though ecological considerations are nice, my real reason for trying to reduce the current in the bias FET is to get away witn a small heat sink on the current source driving it. You can soak up some of the power away from the current source with a series resistor (I'll probably do some of that in a very modest way for this design), but for large power dissipations, that's just another thing or string of things getting hot, and there's already enough of those with a class A design, anyway. ('
Ah, my "ecological consideration" was meant to be humor 🙂
Large heat sinks are expensive and anything that reduces the amount of power unnecessarily being dumped into it is a good thing.
Also I am going to have a go at machining my own heatsinks out of aluminum bar as I can then make fancy fins with my name as part of the design [SUSAN] but I am limited to about 3.5" x 8.5" in XY as this is the range of movement of my milling machine.
The justification of course is to use variable profile fins to cause turbulent flow of the air across the heat exchanging surfaces.
Nice rant, and I am with you much of the way on this one.
However it would be interesting to drive the input of my amplifier with a PWM drive from a digital amplifier output. Or...
... as far as I can see one could also do this directly from a Direct Stream Digital (DSD) source.
Best wishes,
Susan.
Positive temperatue coeficient of " large " V - mosfets is only solving by temperature sensig device, conected to biasing circuit, not by multipling winding.
Simplicity begets simplicity.
Earlier tonight I was driving to the fish & chip shop and thinking about this amplifier along the way. It occurred to me that one of it's features, that of relatively high output swing for a given supply rail would be of great benefit in a car sound setup. You could swing 28 volts peak from (ideally) 14 volt battery supply. That's 50W into 8 ohms, 100W into 4 ohms *without having to use a switched mode DC/DC converter*.
Earlier tonight I was driving to the fish & chip shop and thinking about this amplifier along the way. It occurred to me that one of it's features, that of relatively high output swing for a given supply rail would be of great benefit in a car sound setup. You could swing 28 volts peak from (ideally) 14 volt battery supply. That's 50W into 8 ohms, 100W into 4 ohms *without having to use a switched mode DC/DC converter*.
Re: Re: Re: Commentable thoughts
Threshold voltage = NTC ~5.5mV per deg C.
Minimum Rdson is irrelevant if you are using the device in linear mode. Vth is everything.
On resistance = PTC.amp_man_1 said:
Threshold voltage = NTC ~5.5mV per deg C.
Minimum Rdson is irrelevant if you are using the device in linear mode. Vth is everything.
Susan-Parker said:
Hmm,
My telephone directory indicates that a person by this name lives literally several hundred feet from my home. Maybe I should knock on his door and ask him if he's still happily listening to his amp.
Pat in Madison, WI
Patent..
..I'd guess is lapsed now as it is 26 years old...
So Susan doesn't need to worry about it, unless she wants to patent her amp, which she can't in any case because she has now put the design into the public domain...
..I'd guess is lapsed now as it is 26 years old...
So Susan doesn't need to worry about it, unless she wants to patent her amp, which she can't in any case because she has now put the design into the public domain...
Lateral FETs
Susan,
> I have tried some of these lateral types but they are horribly expensive at 20 to 30 dollars each
You can get 2SK1529's or 2SK1058's for about 5 to 7 Euro's, unmatched, in Germany. 2SK1058 is more expensive, but probably better for your application. Check out the datasheets.
> and the matching between same batch devices seems a lot poorer than with the cheepie IRF family parts.
That is absolutely true. You would have little chance of good match even with a batch of 25 (in a tube). Last time we used any of these, we bought 200 and then there were hardly any rejects, with enough sets of 8's and 12's. : )
So my advice to anyone wanting to use lateral FETs -- do a group buy. One more tipp, measure transconductance as well as Vgs.
I have a very limited amount of either of above-mentioned type to spare. If you are not in a rush and tell me which one, what current and Vds, I can offer you 1 pair to try, at cost. I have an account at Barclays so no problem there. Send me a email @ ybpkwan@web.de if you have interest.
Keep the good work going. You have my full admiration for your generosity and enthusiasum.
Cheers,
Patrick
Susan,
> I have tried some of these lateral types but they are horribly expensive at 20 to 30 dollars each
You can get 2SK1529's or 2SK1058's for about 5 to 7 Euro's, unmatched, in Germany. 2SK1058 is more expensive, but probably better for your application. Check out the datasheets.
> and the matching between same batch devices seems a lot poorer than with the cheepie IRF family parts.
That is absolutely true. You would have little chance of good match even with a batch of 25 (in a tube). Last time we used any of these, we bought 200 and then there were hardly any rejects, with enough sets of 8's and 12's. : )
So my advice to anyone wanting to use lateral FETs -- do a group buy. One more tipp, measure transconductance as well as Vgs.
I have a very limited amount of either of above-mentioned type to spare. If you are not in a rush and tell me which one, what current and Vds, I can offer you 1 pair to try, at cost. I have an account at Barclays so no problem there. Send me a email @ ybpkwan@web.de if you have interest.
Keep the good work going. You have my full admiration for your generosity and enthusiasum.
Cheers,
Patrick
Vertical FETs
> From RS (A UK component distro) I get the following prices:
> IRFP140N IR TO247 UKP 1.81
> IRF540N IR TO220 UKP 2.93
> IRFP150N IR TO247 UKP 4.36
> IRF150 - TO3 UKP 12.27
Not really news for anyone on the continent :
IRFP140N Euro 1.45
IRF540N Euro 0.68
IRFP150N Euro 1.35
IRFP150 (TO247) Euro 1.65
All from www.Reichelt.de, price includes VAT. They do, as I understand, have a MOQ of 100 Euro outside Germany.
PS 1 GBP = 1.5 Euro approx.
Patrick
> From RS (A UK component distro) I get the following prices:
> IRFP140N IR TO247 UKP 1.81
> IRF540N IR TO220 UKP 2.93
> IRFP150N IR TO247 UKP 4.36
> IRF150 - TO3 UKP 12.27
Not really news for anyone on the continent :
IRFP140N Euro 1.45
IRF540N Euro 0.68
IRFP150N Euro 1.35
IRFP150 (TO247) Euro 1.65
All from www.Reichelt.de, price includes VAT. They do, as I understand, have a MOQ of 100 Euro outside Germany.
PS 1 GBP = 1.5 Euro approx.
Patrick
Hi Upupa Epops,
I am sorry I obviously didn't describe this very clearly.
The multiple windings are to ensure that one of a set of paralleled mosfets doesn't current hog, which can otherwise happen unless one uses low value e.g. 0.47 ohm power resistors.
Using a separate winding per paralleled mosfet removes the requirement of a separate power resistor - which may well cost more than the power device itself!
The close coupling between the windings will ensure that the mosfets all step in tune.
To track the bias against temperature changes will of course need a separate temperature sensor of some type - preferably a similar technology mosfet.
Best wishes,
Susan.
Upupa Epops said:
I am sorry I obviously didn't describe this very clearly.
The multiple windings are to ensure that one of a set of paralleled mosfets doesn't current hog, which can otherwise happen unless one uses low value e.g. 0.47 ohm power resistors.
Using a separate winding per paralleled mosfet removes the requirement of a separate power resistor - which may well cost more than the power device itself!
The close coupling between the windings will ensure that the mosfets all step in tune.
To track the bias against temperature changes will of course need a separate temperature sensor of some type - preferably a similar technology mosfet.
Best wishes,
Susan.
Re: Vertical FETs
Hi Patrick,
Thanks for the info and your helpful offer.
Seems a bit better 🙂
Um, I thought I was being generous in considering buying 25 of the IRFPxxx types to hunt among to mix n' match.
Thank you.
I believe house prices are a lot cheaper too. (BTW my comment about being rich because I have a half tray of IRF150s was meant to be ironic).
I use RS (and Farnell) as I can buy small quantities that would in many cases otherwise be subjected to MOQs... however...
DigiKey US are quoting USD 3.20 for IRFP150s. That translates into UKP 1.78 on the DigiKey UK site. Um, that is a LOT better.
... and they want UKP 1.00 for the IRF540N which is also a distinct improvement although they only have 78514 pieces available at time of writing.
... oh, the 10 up price is a mere UKP 0.651, and the 100 is UKP 0.50 - aagh!
... now what is the price on the IRFP150N's? Um, 10 up is UKP 1.262 and 100 up is UKP 1.039.
Food for thought.
DigiKey's 20W TO220 power resistors are still UKP 3.85 though.
I was going to write that if I was setting up production for the amplifiers I would go through "proper" distribution. However it looks like I should give DigiKey a try anyway.
Many thanks for the "nudge" 🙂
Best wishes,
Susan.
Hi Patrick,
Thanks for the info and your helpful offer.
EUVL said:
Seems a bit better 🙂
Um, I thought I was being generous in considering buying 25 of the IRFPxxx types to hunt among to mix n' match.
Thank you.
EUVL said:
I believe house prices are a lot cheaper too. (BTW my comment about being rich because I have a half tray of IRF150s was meant to be ironic).
I use RS (and Farnell) as I can buy small quantities that would in many cases otherwise be subjected to MOQs... however...
DigiKey US are quoting USD 3.20 for IRFP150s. That translates into UKP 1.78 on the DigiKey UK site. Um, that is a LOT better.
... and they want UKP 1.00 for the IRF540N which is also a distinct improvement although they only have 78514 pieces available at time of writing.
... oh, the 10 up price is a mere UKP 0.651, and the 100 is UKP 0.50 - aagh!
... now what is the price on the IRFP150N's? Um, 10 up is UKP 1.262 and 100 up is UKP 1.039.
Food for thought.
DigiKey's 20W TO220 power resistors are still UKP 3.85 though.
I was going to write that if I was setting up production for the amplifiers I would go through "proper" distribution. However it looks like I should give DigiKey a try anyway.
Many thanks for the "nudge" 🙂
Best wishes,
Susan.
Hi Pat,
If it is the same person it would be interesting to hear a bit about the background to the development.
I just hope he isn't too upset that everyone ignored him (I know this feeling rather well myself).
Best wishes,
Susan.
tubesguy said:
If it is the same person it would be interesting to hear a bit about the background to the development.
I just hope he isn't too upset that everyone ignored him (I know this feeling rather well myself).
Best wishes,
Susan.
Seriously, I'll drop off a copy of your circuit. May bring a smile to his face, who know? - Pat
Back to Susan -
My comments on MOSFET Vgs tempco were made mainly in response to the post immediately preceding that one. One can get in a lot of trouble by assuming that any old MOSFET has a positive Vgs tempco - I don't know where this myth actually started, except maybe as a hangover from the early to mid 80's when the Hitachi lateral devices were being extensively used. I think there is also some confusion generated by the claims of current sharing for MOSFETs used as switches, where the resistive nature of the device helps to promote sharing, as opposed to linear operation, where the negative Vgs tempco promotes thermal instability and/or current hogging, just like with bipolar devices. If one actually takes the time to read the IR device data sheets, they plainly state that the Vgs tempco is negative. At any rate, I wanted to point out the implications for thermal stability in general.
This is not so much a problem with the current version of the push-pull amp, as the bias current is comparatively low, and some form of thermal tracking is being used anyway. I would expect the resistance of the output transformer windings would add some degeneration, since they are in the mosfet sources, but not too much for really low loss windings. The real problem is with class A and AB amps with really hot bias currents on the order of an amp or two or with high rail voltages, where significant power is dumped into the output heat sinks. Adequate thermal stability in these applications is the difference between an amp that hums along serenely with a 50C heat sink temperature and one that runs away and destroys itself. As an exercise, I went to irf.com and looked up the point at which the Vgs tempco crosses between negative and positive for an IRF540. You can get it from the graph of transfer function vs. temperature. This number is typical for all the common devices with an IR size 4 die (used extensively in designs mentioned on this list). In this case, the crossover point is at approximately 12A, making it really mandatory to use some sort of bias thermal compensation. In practice, a MOSFET of this sort would fry up like a sausage long before it reached the Vgs tempco crossover current.
My comments on MOSFET Vgs tempco were made mainly in response to the post immediately preceding that one. One can get in a lot of trouble by assuming that any old MOSFET has a positive Vgs tempco - I don't know where this myth actually started, except maybe as a hangover from the early to mid 80's when the Hitachi lateral devices were being extensively used. I think there is also some confusion generated by the claims of current sharing for MOSFETs used as switches, where the resistive nature of the device helps to promote sharing, as opposed to linear operation, where the negative Vgs tempco promotes thermal instability and/or current hogging, just like with bipolar devices. If one actually takes the time to read the IR device data sheets, they plainly state that the Vgs tempco is negative. At any rate, I wanted to point out the implications for thermal stability in general.
This is not so much a problem with the current version of the push-pull amp, as the bias current is comparatively low, and some form of thermal tracking is being used anyway. I would expect the resistance of the output transformer windings would add some degeneration, since they are in the mosfet sources, but not too much for really low loss windings. The real problem is with class A and AB amps with really hot bias currents on the order of an amp or two or with high rail voltages, where significant power is dumped into the output heat sinks. Adequate thermal stability in these applications is the difference between an amp that hums along serenely with a 50C heat sink temperature and one that runs away and destroys itself. As an exercise, I went to irf.com and looked up the point at which the Vgs tempco crosses between negative and positive for an IRF540. You can get it from the graph of transfer function vs. temperature. This number is typical for all the common devices with an IR size 4 die (used extensively in designs mentioned on this list). In this case, the crossover point is at approximately 12A, making it really mandatory to use some sort of bias thermal compensation. In practice, a MOSFET of this sort would fry up like a sausage long before it reached the Vgs tempco crossover current.
Lateral FETs
Susan,
I can recommend Digikey if you buy enough, as they charge some GBP20 extra for post and adminstration to keep small orders away.
Reichelt is very popular here because they are almost as good as digikey for price, and in Germany it is 10 Euros MOQ, 4 Euros P&P.
> Um, I thought I was being generous in considering buying 25 of the IRFPxxx types to hunt among to mix n' match.
25 pieces if IRFP from the same batch would give you quite a few pairs, but not the lateral FETs, from my experience.
> DigiKey's 20W TO220 power resistors are still UKP 3.85 though.
If you are referring to Caddock 925's or similar, you won't get it much cheaper, I can assure you.
> Many thanks for the "nudge"
The offer for 2SK's is still on. Why not give it a try ? : )
You don't get them from Digikey.
I guess I have some private interest (NOT financial) to see how lateral FETs compares in your application. : )
Patrick
Susan,
I can recommend Digikey if you buy enough, as they charge some GBP20 extra for post and adminstration to keep small orders away.
Reichelt is very popular here because they are almost as good as digikey for price, and in Germany it is 10 Euros MOQ, 4 Euros P&P.
> Um, I thought I was being generous in considering buying 25 of the IRFPxxx types to hunt among to mix n' match.
25 pieces if IRFP from the same batch would give you quite a few pairs, but not the lateral FETs, from my experience.
> DigiKey's 20W TO220 power resistors are still UKP 3.85 though.
If you are referring to Caddock 925's or similar, you won't get it much cheaper, I can assure you.
> Many thanks for the "nudge"
The offer for 2SK's is still on. Why not give it a try ? : )
You don't get them from Digikey.
I guess I have some private interest (NOT financial) to see how lateral FETs compares in your application. : )
Patrick