This is a project which is not only relevant to my interests, but also has similar rate of progress to mine... I got the input JFETs off ebay two years ago, so now I'm down to sourcing cascodes and outputs...
For those who are so wise in the way of building F5 amps for Stax headphones, a few questions:
I see Mouser has the FqP3N80 in stock but not the FQP3P50. Is my best bet just to wait for the P types to be re-stocked?
The not very different FQP4P40 is available at Mouser - capacitanes similar, VGS(th) is spec'ed to the same limits, and the gFS looks like it's based on the same story. A drop-in replacement, if I were to lower the supply to +/- 200V?
Are there better types available, anyway? It has been a few years - I have no idea if better types have been introduced.
Should I get several and match them up for VGS(th)?
For cascodes, I would start with MOSFETs. Any preferences / hints that people can share?
Any help appreciated
Jens
For those who are so wise in the way of building F5 amps for Stax headphones, a few questions:
I see Mouser has the FqP3N80 in stock but not the FQP3P50. Is my best bet just to wait for the P types to be re-stocked?
The not very different FQP4P40 is available at Mouser - capacitanes similar, VGS(th) is spec'ed to the same limits, and the gFS looks like it's based on the same story. A drop-in replacement, if I were to lower the supply to +/- 200V?
Are there better types available, anyway? It has been a few years - I have no idea if better types have been introduced.
Should I get several and match them up for VGS(th)?
For cascodes, I would start with MOSFETs. Any preferences / hints that people can share?
Any help appreciated
Jens
Hi Jens, haven't tried the 4P40 but it seems a good choice, at lower voltage as you write of course. In my build, with the 3P50 (and 5N50), the margin is questionable, as 2x250V = 500V and so it's all good until you go to high volumes, at which point the peak voltage on the transistors could actually reach 500V for a short time. In industrial electronics, this would not be advisable, but we are in "Burning Amp" territory here 😀
Yes, MOSFETs can be used for the cascodes, haven'T tried but it should be possible. You could use the same transistors as used for the outputs. I will try and dig out that simulation, to see what the impact would be, in a couple days or so.
Unfortunately, these high voltage p-channel transistors are a dying breed, and a similar thing can be said for the n-channels. While lots of n-channel HV transistors exist, they are more and more being optimized for sharp switching, but we are looking for good analog characteristics here. So, if you can , you should stock a few of those, as better transistors are unlikely to come. The SiC MOSFETs (and before those, the JFETS while they existed) where even better n-channel devices, but no p-channel device to be seen....
Hope that helps!
Yes, MOSFETs can be used for the cascodes, haven'T tried but it should be possible. You could use the same transistors as used for the outputs. I will try and dig out that simulation, to see what the impact would be, in a couple days or so.
Unfortunately, these high voltage p-channel transistors are a dying breed, and a similar thing can be said for the n-channels. While lots of n-channel HV transistors exist, they are more and more being optimized for sharp switching, but we are looking for good analog characteristics here. So, if you can , you should stock a few of those, as better transistors are unlikely to come. The SiC MOSFETs (and before those, the JFETS while they existed) where even better n-channel devices, but no p-channel device to be seen....
Hope that helps!
I would not compromise on rail voltage.
FQP3P50 is back in stock 2nd half of January 2022 at RS :
https://uk.rs-online.com/web/p/mosfets/1455369https://uk.rs-online.com/web/p/mosfets/6715118
If I were you I would buy 50 pcs and match.
You will not regret when they all become obsolete in a few years.
Patrick
FQP3P50 is back in stock 2nd half of January 2022 at RS :
https://uk.rs-online.com/web/p/mosfets/1455369https://uk.rs-online.com/web/p/mosfets/6715118
If I were you I would buy 50 pcs and match.
You will not regret when they all become obsolete in a few years.
Patrick
Hi Jens, haven't tried the 4P40 but it seems a good choice, at lower voltage as you write of course. In my build, with the 3P50 (and 5N50), the margin is questionable, as 2x250V = 500V and so it's all good until you go to high volumes, at which point the peak voltage on the transistors could actually reach 500V for a short time. In industrial electronics, this would not be advisable, but we are in "Burning Amp" territory here 😀
While the currents should certainly never impress these types, I'd rather play it safe with the voltage.
I think the smallest possible device for cascodes (we don't want any of the precious signal current from the input to go off into the parasitics), so I would expect output devices to be grossly oversized.Yes, MOSFETs can be used for the cascodes, haven'T tried but it should be possible. You could use the same transistors as used for the outputs. I will try and dig out that simulation, to see what the impact would be, in a couple days or so.
Looking at Mouser's catalog, yes, the P types are on the way out. While analogue characteristics are grand, I'll settle for devices with parasitic capacitances which leave me with bags of open-loop gain at 20 kHz.Unfortunately, these high voltage p-channel transistors are a dying breed, and a similar thing can be said for the n-channels. While lots of n-channel HV transistors exist, they are more and more being optimized for sharp switching, but we are looking for good analog characteristics here.
At these prices, I'll grab a bag of the 400V types now and see if I'm ready to use the 500V types when they show up in spring.So, if you can , you should stock a few of those, as better transistors are unlikely to come. The SiC MOSFETs (and before those, the JFETS while they existed) where even better n-channel devices, but no p-channel device to be seen....
Everything helps 😎Hope that helps!
Best Regards
Jens
Hi Patrick,I would not compromise on rail voltage.
FQP3P50 is back in stock 2nd half of January 2022 at RS :
https://uk.rs-online.com/web/p/mosfets/1455369https://uk.rs-online.com/web/p/mosfets/6715118
If I were you I would buy 50 pcs and match.
You will not regret when they all become obsolete in a few years.
Patrick
Thanks for the tip! January is certainly soon enough for me 🤣 . And, indeed, getting a large number and matching is a plan - especially at these prices.
Best Regards
Jens
MOSFETs designed for switching will fail at much lower power rating in analogue applications than specified, because of local hot spots.
It is not just about characteristics.
You can now see some devices from IXYS that emphasis their suitability for analogue / DC operations.
Cheers,
Patrick
It is not just about characteristics.
You can now see some devices from IXYS that emphasis their suitability for analogue / DC operations.
Cheers,
Patrick
Hi Patrick,
That makes sense.
Another thing, the cascode dvices. I see in your V1a schematic you had the same transistors as in the outputs. Would devices with the same voltage rating but much smaller power rating not be better (lower parasitics, is what I'm thinking)?
Best Regards
Jens
That makes sense.
Another thing, the cascode dvices. I see in your V1a schematic you had the same transistors as in the outputs. Would devices with the same voltage rating but much smaller power rating not be better (lower parasitics, is what I'm thinking)?
Best Regards
Jens
I was using the same BJTs as hesener, in TO126 case.
You can use the output MOSFETs as well.
No real difference, but I have not tried.
TO92s (such as ZVP0545A) will get too hot.
Patrick
You can use the output MOSFETs as well.
No real difference, but I have not tried.
TO92s (such as ZVP0545A) will get too hot.
Patrick
Oh, yes, at these voltages the SOA of the cascodes definitely needs to be kept in mind.
Best Regards
Jens
Best Regards
Jens
Yes, too long in fact.... Thank you very much, and the same for you , health and happiness!Long time no hear. 🙂
Happy New Year to you. Stay healthy.
Patrick
(and a great choice of new, good transistors for audio use 😀)
(couldn't resist to throw that in 😀)
Here you go, the simulation virtually shows no difference. There is a small difference in very high frequency > 1MHz, no doubt due to the different parasitic capacitances. Virtually no impact on distortion performance, the three versions simulate identical:
Green = original circuit
Red (overlapping with blue) = Cascodes with the same MOSFETs as the output transistors
Blue = 3P50 changed with 4P40 (curve hidden under the red curve) (don't worry about the phase, I mistakenly picked the "other" output for the picture)
Judging from circuit performance, I would see no reason why not to do that.
Back then, I chose the bipolars because they are a little smaller than the TO220 packages for the MOSFETs, while still providing good heat transfer. A TO92 package won't cut it, these parts would get too hot. Transistors in TO252 or similar would work, but they have more or less the same size as TO126 (the bipolars) and no hole, so mounting on a heat sink is slightly more complex, at leat for DIY. And, I had a bunch of them in the drawer 😀
Green = original circuit
Red (overlapping with blue) = Cascodes with the same MOSFETs as the output transistors
Blue = 3P50 changed with 4P40 (curve hidden under the red curve) (don't worry about the phase, I mistakenly picked the "other" output for the picture)
Judging from circuit performance, I would see no reason why not to do that.
Back then, I chose the bipolars because they are a little smaller than the TO220 packages for the MOSFETs, while still providing good heat transfer. A TO92 package won't cut it, these parts would get too hot. Transistors in TO252 or similar would work, but they have more or less the same size as TO126 (the bipolars) and no hole, so mounting on a heat sink is slightly more complex, at leat for DIY. And, I had a bunch of them in the drawer 😀
Attachments
Hi Patrick, I know I have no rights to ask, after taking 8 years to assemble mine, but did your friend finish his build, have you been able to listen?The pair of boards were given to a friend who is in the process of housing them properly.
With regulators, etc. Slow progress.
Patrick
He has everything, including the case ready drilled.
Just no time, as he is a "Pendler" now.
And he cannot do home office.
Horrible, isn't it ?
Patrick
Just no time, as he is a "Pendler" now.
And he cannot do home office.
Horrible, isn't it ?
Patrick
That's too bad... I was in "hard" home office the last three months, and achieved some progress on a few phono preamps (as if I don't have enough of them already...) until my wife found out 😀
Also, I admit that my own build is no longer working - the whole power supply setup in that amplifier was shaky, and now broke down. Now, I have ordered a new transformer and will rebuild the power supplies for a fourth time (!). Classical story of "still had those parts in the drawer", "just a quick try", ..... Should have one it right from the beginning. At least I have a strong motivation - bought three records and promised myself that I would only listen to them with this amplifier working, on my headphones. So those records arrived, I washed them, and they sit on the shelf since two years now, untouched 😀
Horrible, isn't it?
alfred
Also, I admit that my own build is no longer working - the whole power supply setup in that amplifier was shaky, and now broke down. Now, I have ordered a new transformer and will rebuild the power supplies for a fourth time (!). Classical story of "still had those parts in the drawer", "just a quick try", ..... Should have one it right from the beginning. At least I have a strong motivation - bought three records and promised myself that I would only listen to them with this amplifier working, on my headphones. So those records arrived, I washed them, and they sit on the shelf since two years now, untouched 😀
Horrible, isn't it?
alfred
Hi Hesener,
Interesting, thanks. I did look at a few devices (other MOSFETs with lower power rating but similar voltages) and they appeared to have higher parasitic capacitances than the output devices. So, no obvious reason to use other than the output devices (or, indeed, the bipolars from your original suggestion).
Patrick, you mentioned regulators etc, is that something you have available to share? I guess some app note circuit for MOSFET output and a spare output device from a bulk purchase would do nicely, but if there is a known working solution there's no reason for me to go out on another limb.
Best Regards
Jens
Interesting, thanks. I did look at a few devices (other MOSFETs with lower power rating but similar voltages) and they appeared to have higher parasitic capacitances than the output devices. So, no obvious reason to use other than the output devices (or, indeed, the bipolars from your original suggestion).
Patrick, you mentioned regulators etc, is that something you have available to share? I guess some app note circuit for MOSFET output and a spare output device from a bulk purchase would do nicely, but if there is a known working solution there's no reason for me to go out on another limb.
Best Regards
Jens
Check post #12 from this thread to see the regulators I used in my build, working like a charm 😉
Mine is essentially the same as post #12, except that I used LM329 in place of TL431.
Then just change R3 to set the output voltage.
I used the same BJTs as the cascode, and same MOSFETs as the output stage.
And R8 is actually a LND150 TO92 wired as CCS.
Patrick
Then just change R3 to set the output voltage.
I used the same BJTs as the cascode, and same MOSFETs as the output stage.
And R8 is actually a LND150 TO92 wired as CCS.
Patrick
- Home
- Amplifiers
- Pass Labs
- X5HV - the F5 principle applied to a headphone amp