I am currently experimenting with and modifying the Aleph 0s circuit and am about to start prototyping. I am using IRFP240 and IRFP9240 output devices. I have purchased a number of Fairchild SFH9240 devices because they available and relatively cheap. They also look good judging from the Fairchild data sheet. I have two questions to ask the forum:
1) Has anyone tried the SFH9240 yet? How do they compare to IR parts or more importantly the old Harris/Intersil parts?
2) Where did the SFH9240 come from? The Harris MOSFET operation became Intersil in 1999. Fairchild then bought Intersil in 2001. Fairchild also bought the Samsung MOSFET operation a couple of years ago. I've recently noticed that Fairchild seems to have been thinning out their MOSFET eliminating some IRFP part numbers. In fact the only part that comes up on a search for "9240" on the Fairchild website is the SFH9240. Comparing the data sheets for the Harris and Intersil IRFP9240 and the SFH9240 shows that there are more than trivial differences between all the parts. I guess my question boils down to this: Is the SFH9240 really the same die as the old Harris/Intersil IRFP9240 or not?
I will be doing my own experiments to see how they behave and sound, but I thought that it wouldn't hurt to see if anyone out there already had results or knowledge they would share. Thank you.
Graeme
1) Has anyone tried the SFH9240 yet? How do they compare to IR parts or more importantly the old Harris/Intersil parts?
2) Where did the SFH9240 come from? The Harris MOSFET operation became Intersil in 1999. Fairchild then bought Intersil in 2001. Fairchild also bought the Samsung MOSFET operation a couple of years ago. I've recently noticed that Fairchild seems to have been thinning out their MOSFET eliminating some IRFP part numbers. In fact the only part that comes up on a search for "9240" on the Fairchild website is the SFH9240. Comparing the data sheets for the Harris and Intersil IRFP9240 and the SFH9240 shows that there are more than trivial differences between all the parts. I guess my question boils down to this: Is the SFH9240 really the same die as the old Harris/Intersil IRFP9240 or not?
I will be doing my own experiments to see how they behave and sound, but I thought that it wouldn't hurt to see if anyone out there already had results or knowledge they would share. Thank you.
Graeme
sfh9240
Hi Graeme,
SFH9240 is supposed to be a replacement for the old IRFP9240. Its data sheet specs look somewhat better (except for power dissipation) than those of the old IRFP0240 - especially Rds(on) and capacitance as well as linearity as you have noticed.
I have been thinking about getting some of these devices and match them with IRFP240 in the Stochino amp I intend to built early next year. Looks that nobody here has much knowledge about these devices except you - or experts have somehow decided to keep quiet.
My question to you is - did you try to match SFH9240s with IRFP240s? If yes, was it easy to match Vgs and how do these pairs work? Are you happy with SFH9240?
If you haven't tested them so far then it's likely that I will try to match these with IRFP240s sometime next year unless I get a word from you or others to forget the whole idea and stick with the old IRFP9240s.
Cheers,
Hi Graeme,
SFH9240 is supposed to be a replacement for the old IRFP9240. Its data sheet specs look somewhat better (except for power dissipation) than those of the old IRFP0240 - especially Rds(on) and capacitance as well as linearity as you have noticed.
I have been thinking about getting some of these devices and match them with IRFP240 in the Stochino amp I intend to built early next year. Looks that nobody here has much knowledge about these devices except you - or experts have somehow decided to keep quiet.
My question to you is - did you try to match SFH9240s with IRFP240s? If yes, was it easy to match Vgs and how do these pairs work? Are you happy with SFH9240?
If you haven't tested them so far then it's likely that I will try to match these with IRFP240s sometime next year unless I get a word from you or others to forget the whole idea and stick with the old IRFP9240s.
Cheers,
Janusz,
Nice to hear you mention the Stochino amp. I think that not many have built this great sounding amp and they have really missed something. By far it is close to the best you can get to. All the best as you give your try.
On matching of Mosfets: You only need to match the N channel Mosfets as a bank and likewise the P channel Mosfets and not across the types. This is to ensure proper current sharing between all output devices.
I have built quite a few channels of the Stochino amp using IRF640 and IRF9640 as suggested by the designer himself. I am not able to source IRFP9240s or 240s at a relatively cheap price in India or this amp with some rugged output devices and +-58 volts for the Voltage Amplifer stage and about +-42 to 45 volts for the Current Amplifier stage will produce 120 watts into 8 ohms and 240 watts into 4 ohms, and at the same time prove a lot more reliable than the smaller TO220 package devices.
Nice to hear you mention the Stochino amp. I think that not many have built this great sounding amp and they have really missed something. By far it is close to the best you can get to. All the best as you give your try.
On matching of Mosfets: You only need to match the N channel Mosfets as a bank and likewise the P channel Mosfets and not across the types. This is to ensure proper current sharing between all output devices.
I have built quite a few channels of the Stochino amp using IRF640 and IRF9640 as suggested by the designer himself. I am not able to source IRFP9240s or 240s at a relatively cheap price in India or this amp with some rugged output devices and +-58 volts for the Voltage Amplifer stage and about +-42 to 45 volts for the Current Amplifier stage will produce 120 watts into 8 ohms and 240 watts into 4 ohms, and at the same time prove a lot more reliable than the smaller TO220 package devices.
Hi janusz,
Thank you for responding to this thread. It's great to see it come back to life.
Since my post I have matched 150 IR IRFP240 devices and 150 Fairchild SFH9240's. All the parts for each device type came from unbroken tubes and had the same date code. I found that the SFH9240's had a much narrower spread of Vgs values than the IRFP240's. In fact so many devices had exactly the Vgs to within .01 volts that I stopped and rechecked the test rig with a couple of IRFP240's. I don't have the statistics here at hand but I will post more info tonight.
I am currently building a current output SOZ (95% done), a JFET/MOSFET redesign of the A40 (75% done), and two 150W Aleph X monoblocks (10% done). Nothing has been fired up yet and no listening tests have been done.
The one thing I have discovered is that the TO-3P package of the SFH9240 will not accept a 6-32 bolt like the TO-247 will. I had to use M4 hardware to mount them. Both packages do, however, use the same heat sink insulators.
Graeme
Thank you for responding to this thread. It's great to see it come back to life.
Since my post I have matched 150 IR IRFP240 devices and 150 Fairchild SFH9240's. All the parts for each device type came from unbroken tubes and had the same date code. I found that the SFH9240's had a much narrower spread of Vgs values than the IRFP240's. In fact so many devices had exactly the Vgs to within .01 volts that I stopped and rechecked the test rig with a couple of IRFP240's. I don't have the statistics here at hand but I will post more info tonight.
I am currently building a current output SOZ (95% done), a JFET/MOSFET redesign of the A40 (75% done), and two 150W Aleph X monoblocks (10% done). Nothing has been fired up yet and no listening tests have been done.
The one thing I have discovered is that the TO-3P package of the SFH9240 will not accept a 6-32 bolt like the TO-247 will. I had to use M4 hardware to mount them. Both packages do, however, use the same heat sink insulators.
Graeme
sfh9240 and Stochino
Hi Graeme and Sam,
So it looks that SFH9240 is a nice device. I'll get some of these from Mouser in January. I remember reading in AudioXpress (?) that International Rectifier IRFP9240 had some linearity problems but Harriss version did not. Anyway, I'll probably get IRFP240s from IR as Fairchild does not make IRFP240s any more. They make IRF240Bs which have somewhat different Vgs settings and lower capacitance.
Although N and P devices need to be matched among themselves some time ago I read that cross matching N with P devices has some merits but I forgot details. It was a technical paper written by someone form IR or Fairchild or some other manufacturer. Although I did engineering course including 2 semesters of electronics it was long time ago and then I switched into economics (and that's where I am today) so in most cases I prefer to rely on much more experienced people. The problem is that sometimes they contradict each other. Anyway, it looks that in the case of Stochino amp (which you praise so much, Sam) it is enough to match Ns and Ps independently.
When it comes to Stochino amp its speed and low TIM attracted me to it straight away. Alex Soton shared some of his experience with Stochino amp with me and others on the net so hopefully I will have less problems. I intend to use +/-56V unregulated supply for the output (the same voltage as in Stochino's prototype), more powerful and rugged output devices and high gain low noise transistors in the input differential stage. I also thought of replacing 2N driver stage transistors with 2SB649AC / 2SD669AC devices as the latter seem to me a more "natural" choice (they are much more easier to match - especially gainwise which has a lot of merits) but I may be wrong. What do you think Sam?
Anyway I'll be glad to hear how the Alephs sound once put together. It would be also interesting to compare them to the Stochino amps. Unfortunately, I do not know anyone in Perth who has alephs. But maybe one day ...
Cheers,
Hi Graeme and Sam,
So it looks that SFH9240 is a nice device. I'll get some of these from Mouser in January. I remember reading in AudioXpress (?) that International Rectifier IRFP9240 had some linearity problems but Harriss version did not. Anyway, I'll probably get IRFP240s from IR as Fairchild does not make IRFP240s any more. They make IRF240Bs which have somewhat different Vgs settings and lower capacitance.
Although N and P devices need to be matched among themselves some time ago I read that cross matching N with P devices has some merits but I forgot details. It was a technical paper written by someone form IR or Fairchild or some other manufacturer. Although I did engineering course including 2 semesters of electronics it was long time ago and then I switched into economics (and that's where I am today) so in most cases I prefer to rely on much more experienced people. The problem is that sometimes they contradict each other. Anyway, it looks that in the case of Stochino amp (which you praise so much, Sam) it is enough to match Ns and Ps independently.
When it comes to Stochino amp its speed and low TIM attracted me to it straight away. Alex Soton shared some of his experience with Stochino amp with me and others on the net so hopefully I will have less problems. I intend to use +/-56V unregulated supply for the output (the same voltage as in Stochino's prototype), more powerful and rugged output devices and high gain low noise transistors in the input differential stage. I also thought of replacing 2N driver stage transistors with 2SB649AC / 2SD669AC devices as the latter seem to me a more "natural" choice (they are much more easier to match - especially gainwise which has a lot of merits) but I may be wrong. What do you think Sam?
Anyway I'll be glad to hear how the Alephs sound once put together. It would be also interesting to compare them to the Stochino amps. Unfortunately, I do not know anyone in Perth who has alephs. But maybe one day ...
Cheers,
FWIW, I built my A75 using Fairchild IRFP240B and SFH9240 outputs and am very happy with the result. I don't have access to a distortion analyzer or a comparison to IR devices, though. I bought them because they were cheaper than IR, blindly accepting that they substitute for IR devices.
In the assembly process I blew some devices. Once I got my mistakes ironed out they have been completely reliable, the amp is on 15+ hours a day most of the time (operating at reduced bias these days)
In the assembly process I blew some devices. Once I got my mistakes ironed out they have been completely reliable, the amp is on 15+ hours a day most of the time (operating at reduced bias these days)
Hi janusz,
It turns out I matched 150 IR IRFP240's and only 100 of the Fairchild SFH9240's. The IRFP240's had a Vgs spread of 3.78V to 4.20V with at least one or two parts every .01V and a few holes. There were numerous groups with 4 to 6 parts and several with more. There were two groups with 12 parts. It was a wide bell curve with a noticeable bump around 3.85V to 3.90V. The SFH9240's fell in the Vgs range Vgs of 4.24 to 4.33 with groups of 6 to 15 parts.
From what I have heard, Harris and later Intersil had a licensing deal with IR to use their HEXFET designs. As would be expected, this made the Harris/Intersil parts very similar to and compatable with the IR parts. It's interresting that parts made from the same masks in two different Fab's would have differences that people have noted. I also heard that when Fairchild bought Intersil that this licensing deal ended or went into dispute. Hence my original question at the top of the thread.
Nelson Pass comments on what you label the "linearity problems" difference between IR and Harris parts in his paper on MOSFET's on the PassDIY site. In the paper and elsewhere he comments that he believes that this artifact is of little consequence. It seems that this is particularly so for source follower applications. This is the sum of what I've gleaned from this and other sites. There are a number of interesting comments on this issue on this site if you're interested in searching for them.
My concern about the SFH9240 comes from the fact that the price is so much lower than the IR IRFP9240. The implication is that the die has to be different. That in turn makes one want to question it's applicability as a complement to the IR IRFP240. A number of the newer generation MOSFET's have been further optimized for switching applications which further seems to "unoptimize" them for linear use. Again, many people have commented on this topic around here. I don't know what generation the SFH9240 falls into WRT the Harris/Intersil IRFP9240.
Good luck with your Stochino amplifier project. I have no experience with the input transistor types you mention. Please keep us all posted on how things go. I should mention that there seems to be quite a number of diyAudio members in Perth. I'm thinking that you should be able to find someone there who has an Aleph or Aleph X and who wouldn't mind offering you a listen.
To BobEllis: Thank you for you input. I find your experience very reassuring.
Regards,
Graeme
It turns out I matched 150 IR IRFP240's and only 100 of the Fairchild SFH9240's. The IRFP240's had a Vgs spread of 3.78V to 4.20V with at least one or two parts every .01V and a few holes. There were numerous groups with 4 to 6 parts and several with more. There were two groups with 12 parts. It was a wide bell curve with a noticeable bump around 3.85V to 3.90V. The SFH9240's fell in the Vgs range Vgs of 4.24 to 4.33 with groups of 6 to 15 parts.
From what I have heard, Harris and later Intersil had a licensing deal with IR to use their HEXFET designs. As would be expected, this made the Harris/Intersil parts very similar to and compatable with the IR parts. It's interresting that parts made from the same masks in two different Fab's would have differences that people have noted. I also heard that when Fairchild bought Intersil that this licensing deal ended or went into dispute. Hence my original question at the top of the thread.
Nelson Pass comments on what you label the "linearity problems" difference between IR and Harris parts in his paper on MOSFET's on the PassDIY site. In the paper and elsewhere he comments that he believes that this artifact is of little consequence. It seems that this is particularly so for source follower applications. This is the sum of what I've gleaned from this and other sites. There are a number of interesting comments on this issue on this site if you're interested in searching for them.
My concern about the SFH9240 comes from the fact that the price is so much lower than the IR IRFP9240. The implication is that the die has to be different. That in turn makes one want to question it's applicability as a complement to the IR IRFP240. A number of the newer generation MOSFET's have been further optimized for switching applications which further seems to "unoptimize" them for linear use. Again, many people have commented on this topic around here. I don't know what generation the SFH9240 falls into WRT the Harris/Intersil IRFP9240.
Good luck with your Stochino amplifier project. I have no experience with the input transistor types you mention. Please keep us all posted on how things go. I should mention that there seems to be quite a number of diyAudio members in Perth. I'm thinking that you should be able to find someone there who has an Aleph or Aleph X and who wouldn't mind offering you a listen.
To BobEllis: Thank you for you input. I find your experience very reassuring.
Regards,
Graeme
The IRFP9240 (and some other, if not all, IR P channel
devices) don't have what I would call a nonlinearity
as such, rather they exhibit a shelved frequency
response deviation in their transconductance. If
you measure the distortion (say, as a follower) you
don't see this, but you do see about a 3 dB gain
variation when you measure the transconductance.
I have no idea where it comes from, and have not seen
this in other manufacturer's P channel parts. It has not
been the source of performance issues for me, but I know
that Charles Hansen won't use the IR P channel devices.
We use both IR and Harris parts interchangeably, but the P
channel outputs are not part of the Aleph or XA output stages,
which are all N channel. We use P channel parts in the
Class AB X series follower output stages and they work fine.
😎
devices) don't have what I would call a nonlinearity
as such, rather they exhibit a shelved frequency
response deviation in their transconductance. If
you measure the distortion (say, as a follower) you
don't see this, but you do see about a 3 dB gain
variation when you measure the transconductance.
I have no idea where it comes from, and have not seen
this in other manufacturer's P channel parts. It has not
been the source of performance issues for me, but I know
that Charles Hansen won't use the IR P channel devices.
We use both IR and Harris parts interchangeably, but the P
channel outputs are not part of the Aleph or XA output stages,
which are all N channel. We use P channel parts in the
Class AB X series follower output stages and they work fine.
😎
- Status
- Not open for further replies.