Hi,
I wondering if anyone knew of a source for Harris/Fairchild made IRF9610?
NP has opined that the IR made versions are inferior to the Harris/Fairchild second source versions, so I was hoping to replace the 9610's in my Aleph 30's while they are stripped for repairs.
Any leads greatly appreciated!!
cheers
Paul
I wondering if anyone knew of a source for Harris/Fairchild made IRF9610?
NP has opined that the IR made versions are inferior to the Harris/Fairchild second source versions, so I was hoping to replace the 9610's in my Aleph 30's while they are stripped for repairs.
Any leads greatly appreciated!!
cheers
Paul
The FQP3P20 is Fairchild's cross reference for IR's 9610. I think you can also go to Fairchild's website and order directly. If anyone here has tried them, i'd be interested in their subjective evaluation. Nelson has stated somewhere on here that Fairchild's P channel devices do not exhibit "shelving", while IR's P channel devices do exhibit this phenomnea. As to what "shelving" is as it pertains to audio, I don't have any idea what it is and why its bad.
Mouser.com
FQP3P20
edit: for the IRF610, i'd try Fairchild's FQP4N20L. The "L" is the lead free package, which I think is the only type available.
Mouser.com
FQP3P20
edit: for the IRF610, i'd try Fairchild's FQP4N20L. The "L" is the lead free package, which I think is the only type available.
As to which MOSFETs do which, you'll find that Fairchild has a cross reference function on their website: www.fairchildsemi.com.
The problem with the IRF P-ch MOSFETs is that the gain changes with frequency. So how come you don't see it in the frequency response graphs? Well, that comes back to the relation between gain and feedback...
Start with a device. Any device. Let's invent a new one and call it a NGD (New Gain Device). For simplicity's sake, we'll say it's a new type of FET so as to be able to use the same terminology: Gate, Drain, Source. Okay, so we hook up the N-ch NGD in common Source mode with a resistor for a load. Apply a signal and--Presto!--out comes an amplified signal. The "standard" test signal is 1kHz. We provide a 1kHz input and observe a 1kHz output (a little larger, because it's amplified) on the oscilloscope. Cool. Now, just because we like the nifty wiggly lines on the oscilloscope, we twist the knob on the frequency generator and note that the amplitude remains the same with frequency. This is pretty much what you'd expect and so all is right with the world.
Now stick the equivalent P-ch NGD into the circuit (remember to swap + for - rails) and run the test again. There's the 1kHz signal. Looks good. But being a sucker for anything with curves (bikinis were invented for guys like me to watch), we run the frequency up again. Whoa! Waitammit! Whuzzat? The output dropped as the frequency rose. That's weird. Imagine having the treble turned down a bit, if you want a quick fix on the problem.
Why doesn't an amplifier's gain change with frequency? Because nearly all amps are designed with a surplus of gain. The majority of amps have a gain of around 26dB, which is 20 times the input signal. There's no law that says that has to be so, it's just customary. (I'm going to ignore THX certification, which I believe actually specifies a relatively high gain.) If you were to design an amp with an eye towards having 26dB of gain from input to output, and you decide to use, say, 30dB of feedback, then the circuit will need 56dB of open loop gain. That 30dB of feedback comes in very handy, as it equates to a ratio of nearly 30 that you can apply to correcting distortion and...you guessed it...frequency response problems. Apply enough feedback and you can yank a droopy NGD back into line in no time.
Moral of the story: The IRF parts will still give decent performance if you use them with feedback. Actually, the IRF P-ch problems vary somewhat depending on operating conditions, so you have a little control over the gain variation situation. If you intend to use them in a low-to-no feedback circuit, you might want to use parts from another manufacturer. In a circuit with enough feedback, you won't even notice that there's a problem.
Grey
The problem with the IRF P-ch MOSFETs is that the gain changes with frequency. So how come you don't see it in the frequency response graphs? Well, that comes back to the relation between gain and feedback...
Start with a device. Any device. Let's invent a new one and call it a NGD (New Gain Device). For simplicity's sake, we'll say it's a new type of FET so as to be able to use the same terminology: Gate, Drain, Source. Okay, so we hook up the N-ch NGD in common Source mode with a resistor for a load. Apply a signal and--Presto!--out comes an amplified signal. The "standard" test signal is 1kHz. We provide a 1kHz input and observe a 1kHz output (a little larger, because it's amplified) on the oscilloscope. Cool. Now, just because we like the nifty wiggly lines on the oscilloscope, we twist the knob on the frequency generator and note that the amplitude remains the same with frequency. This is pretty much what you'd expect and so all is right with the world.
Now stick the equivalent P-ch NGD into the circuit (remember to swap + for - rails) and run the test again. There's the 1kHz signal. Looks good. But being a sucker for anything with curves (bikinis were invented for guys like me to watch), we run the frequency up again. Whoa! Waitammit! Whuzzat? The output dropped as the frequency rose. That's weird. Imagine having the treble turned down a bit, if you want a quick fix on the problem.
Why doesn't an amplifier's gain change with frequency? Because nearly all amps are designed with a surplus of gain. The majority of amps have a gain of around 26dB, which is 20 times the input signal. There's no law that says that has to be so, it's just customary. (I'm going to ignore THX certification, which I believe actually specifies a relatively high gain.) If you were to design an amp with an eye towards having 26dB of gain from input to output, and you decide to use, say, 30dB of feedback, then the circuit will need 56dB of open loop gain. That 30dB of feedback comes in very handy, as it equates to a ratio of nearly 30 that you can apply to correcting distortion and...you guessed it...frequency response problems. Apply enough feedback and you can yank a droopy NGD back into line in no time.
Moral of the story: The IRF parts will still give decent performance if you use them with feedback. Actually, the IRF P-ch problems vary somewhat depending on operating conditions, so you have a little control over the gain variation situation. If you intend to use them in a low-to-no feedback circuit, you might want to use parts from another manufacturer. In a circuit with enough feedback, you won't even notice that there's a problem.
Grey
Nelson has posted on the subject a few times...
http://www.diyaudio.com/forums/showthread.php?s=&postid=1273543&highlight=9610#post1273543
http://www.diyaudio.com/forums/showthread.php?postid=1248500#post1248500
http://www.diyaudio.com/forums/showthread.php?s=&postid=1219682&highlight=#post1219682
My knowledge of amp design is almost nil (i'm reading the A75 pdf), but referring to the last link above NP comments that the P channel IRF's are "somewhat flawed for use in Common-Source applications, but it is acceptable in Common-Drain use". In the Aleph 30 aren't the differential pair Q1 and Q2 common-source, and therefore prime candidates for a non IR P-channel device? Or am I misunderstanding the design?
Paul
http://www.diyaudio.com/forums/showthread.php?s=&postid=1273543&highlight=9610#post1273543
http://www.diyaudio.com/forums/showthread.php?postid=1248500#post1248500
http://www.diyaudio.com/forums/showthread.php?s=&postid=1219682&highlight=#post1219682
My knowledge of amp design is almost nil (i'm reading the A75 pdf), but referring to the last link above NP comments that the P channel IRF's are "somewhat flawed for use in Common-Source applications, but it is acceptable in Common-Drain use". In the Aleph 30 aren't the differential pair Q1 and Q2 common-source, and therefore prime candidates for a non IR P-channel device? Or am I misunderstanding the design?
Paul
Mouser stocks and sells the SFP9610 and at prices much cheaper than the IR parts go for at most places. I have used these parts with great success.
In my small experience (1 tube) they showed a very small spread in Vgs values and for some reason they seemed to be very stable on my matching jig. The IR's on the other hand and some original Harris 9510's had Vgs voltages that jumped and drifted a lot while I was trying to measure them. The IR's were the worst in this regard. YMMV.
Graeme
In my small experience (1 tube) they showed a very small spread in Vgs values and for some reason they seemed to be very stable on my matching jig. The IR's on the other hand and some original Harris 9510's had Vgs voltages that jumped and drifted a lot while I was trying to measure them. The IR's were the worst in this regard. YMMV.
Graeme
From what I've seen posted to diyaudio it's only the IRF P channel devices that are problematic. The IRF N-channel devices are apparently fine. As the SFP9610 is a second source alternative to the IRF9610 it should have same characteristics - less the transconductance shelf NP mentions - so I wouldn't think there would be any problem using them with IRF610's.
the "not for new designs", and the "lifetime buy" status suggest the part will be obsoleted in the near future. Grey and others have recommended not using parts with near EOL status as it makes future maintenance and parts replacement an expensive proposition.
the "not for new designs", and the "lifetime buy" status suggest the part will be obsoleted in the near future. Grey and others have recommended not using parts with near EOL status as it makes future maintenance and parts replacement an expensive proposition.
spzzzzkt said:
Nelson Pass said:
As of this date, I am still not sure about the production status of the Siliconix J310/J271 parts. When I was first getting the J310s, they were leaded. Then they were unleaded. This with no appreciable lag time when I ordered the parts. The J271, however, was a horse of an entirely different color. I tried repeatedly to get J271s, only to be told any number of reasons as to why they were unavailable. Finally, I was able to get some J271s (unleaded), but then they vanished entirely. For all I know, I got the last available in the world.
Yes, it seems that Vishay bought Siliconix and that undoubtedly played a part in this little drama. But why go to the trouble to run off a small batch of unleaded parts then stop? Surely Siliconix would incur more cost in the setup for such a run than they could hope to make, especially on the eve of takeover by Vishay. As for Vishay, if there was enough profit for it to be worth Siliconix to make the switch to unleaded, why wasn't it worth it to Vishay to keep going? Did they have another part in mind for that production line that would make them more money than the J271?
Mysteries abound. I gave up on the parts.
The Lovoltech/Qspeed power JFETs? Hmmm. Seems as though they're gone from what I can tell. Glad to have a few on hand, but I would have liked to see enough demand that Lovoltech made a P-ch complement. Presumably that chance is gone.
Bummer.
Grey
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.