Greetings-
I've recently let the magic smoke out of my A75, and I've decided to build 2 more.
I've acquired a nice pile of the Fairchild SFH 9240's and SFH 9250's, since I've read that the P-channel mosfets from IRF have an irregularity in their response curve that increases distortion.
So the questions:
Does the irregularity matter when the mosfet is used as a follower, like in the A75 output? (I think I know the answer to this, but I just remembered this point today...)
Has anyone checked the Fairchild parts with the IRF knockoff numbers to confirm that they don't have the squiggly response characteristics?
If these Fairchild parts are OK, anyone needing 9610's might want to check out the Fairchild SFS9610's at Mouser -- they are in stock now, but there is a note saying Mouser will not be stocking those parts any longer.
Does anyone have any recommendations for alternate parts to use as input mosfets? I've looked at Zetex and Supertex in the TO-92 packages, and they have specs that look good, but I was wondering if anyone has used them? The supertex TP2535's look good with a typical Ciss of 65pF. They have a higher 'RDS on' figure (25 ohms) than some of the other TO92 mosfets -- will this have any impact on gain or power handling?
And last but not least, can I substitue jfets for the input devices without having to modify the circuit? I assume I would have to change some resistor values, which is not a problem.
Thanks,
JJ
I've recently let the magic smoke out of my A75, and I've decided to build 2 more.
I've acquired a nice pile of the Fairchild SFH 9240's and SFH 9250's, since I've read that the P-channel mosfets from IRF have an irregularity in their response curve that increases distortion.
So the questions:
Does the irregularity matter when the mosfet is used as a follower, like in the A75 output? (I think I know the answer to this, but I just remembered this point today...)
Has anyone checked the Fairchild parts with the IRF knockoff numbers to confirm that they don't have the squiggly response characteristics?
If these Fairchild parts are OK, anyone needing 9610's might want to check out the Fairchild SFS9610's at Mouser -- they are in stock now, but there is a note saying Mouser will not be stocking those parts any longer.
Does anyone have any recommendations for alternate parts to use as input mosfets? I've looked at Zetex and Supertex in the TO-92 packages, and they have specs that look good, but I was wondering if anyone has used them? The supertex TP2535's look good with a typical Ciss of 65pF. They have a higher 'RDS on' figure (25 ohms) than some of the other TO92 mosfets -- will this have any impact on gain or power handling?
And last but not least, can I substitue jfets for the input devices without having to modify the circuit? I assume I would have to change some resistor values, which is not a problem.
Thanks,
JJ
You can check your Mosfets easily enough, the IRF parts have a
shelf in the transconductance vs frequency curves. If you need some
hints you can look at the Mosfet testing article at
www.passdiy.com
Of course it makes a difference, but it is minimized in follower applications.
shelf in the transconductance vs frequency curves. If you need some
hints you can look at the Mosfet testing article at
www.passdiy.com
Of course it makes a difference, but it is minimized in follower applications.
Hi JJ I have found same laterals that may be worth looking at the source is connected to the sink so no mica washers and such don't know enough to judge the rest so I will apreciate your Judgment
Will it be worth getting one or 2 each and sent them to you?
I have proper PDF data sheets but cant load them due to the limits on the forum can I Email to you?
Al
Will it be worth getting one or 2 each and sent them to you?
I have proper PDF data sheets but cant load them due to the limits on the forum can I Email to you?
Al
I've used IRF610 and 9610 for the input pair, although you can still get the irf110 or irdfd210 at Mouser in lead free versions. I couldn't tell the difference in sound. I used them because I thought I needed more power and bumped the rails up to 65V and wanted something that could handle the dissipation a bit better. Ciss is lower on the irf parts, too.
Jfets would be an exercise in cascoding unless you lower the rails. Current setting resistors would change as well, even though you could change the input pair topology a bit and lose the CCS.
You could use the ZVN3310 and ZVP3310 for your input pair as well. The ZVP3310s were reported to sound very good in an Aleph.
AL - Laterals would need some changes to the bias circuit. You'd still need isolators unless you isolate the sinks from the chassis, since they will be at your output potential.
Jfets would be an exercise in cascoding unless you lower the rails. Current setting resistors would change as well, even though you could change the input pair topology a bit and lose the CCS.
You could use the ZVN3310 and ZVP3310 for your input pair as well. The ZVP3310s were reported to sound very good in an Aleph.
AL - Laterals would need some changes to the bias circuit. You'd still need isolators unless you isolate the sinks from the chassis, since they will be at your output potential.
Last edited:
Al,
You'd need to rework the Vgs multiplier to reduce the bias voltage. You only need a volt or so with those devices to bias them at over an amp.
They'd probably work fine if you get the biasing right. They'd cost a lot more than Verticals, though.
Direct mounting of output devices
Pro: Better heat transfer, less heat sink needed for same junction temperature
Con: Heat sinks are electrically connected to the device - Theoretically the anodizing on the sink should isolate the sink, but it's not good practice to assume it works. Your heat sink must be electrically isolated from the chassis and may have full output signal voltage on it - not an issue for internal sinks, could be for external.
You'd need to rework the Vgs multiplier to reduce the bias voltage. You only need a volt or so with those devices to bias them at over an amp.
They'd probably work fine if you get the biasing right. They'd cost a lot more than Verticals, though.
Direct mounting of output devices
Pro: Better heat transfer, less heat sink needed for same junction temperature
Con: Heat sinks are electrically connected to the device - Theoretically the anodizing on the sink should isolate the sink, but it's not good practice to assume it works. Your heat sink must be electrically isolated from the chassis and may have full output signal voltage on it - not an issue for internal sinks, could be for external.
Bksabath-
I am happy that Mr Ellis answered your post -- I don't know enough about the lateral mosfets to give a reliable answer. Experience here has taught me not to suggest things that I haven't tried, at least not without lengthy disclaimers.
I pulled up the datasheets for those Mosfets -- they do look like pretty interesting devices, but frankly, the only way to really tell would be to build up a circuit with them, give it a listen, and run it through a distortion analyzer (which I don't have). The datasheet mentions something about them being grouped together for ease in paralleling, but I suspect you would still have to match them via the procedures in the A75 article.
If you are building an F5 I would try a pair in that amp. No matching required for one pair of output devices.
I do know for a fact that the IRF9630 and IRF630 (or IRF634) make excellent output devices for the A75 -- that is what I used. They are available from Mouser and are not overly expensive. I recall that I chose those because their properties best matched those listed for the TO3 packaged IRF230 and IRF9231 output devices used in the article.
JJ
Lateral mosfets aren't generally subject to thermal runaway like verticals, and you may be able to get away without matching or Source resistors. Many commercial amps such as the Hafler DH-500 didn't use Source resistors. It would probably be a goof idea to match for sonics, though.
JJ, How many 630/9630 pairs did you use? With their higher thermal resistance junction to case and lower power dissipation ratings, you may have simply pushed them a bit too hard. I used the SFH9240 in my A75s and am very happy with them.
Just to clarify my earlier comment - without isolators the sink is at output potential only with the devices Al linked to and similar. With IRFP devices, the Drains would make contact, requiring separate isolated heat sinks for N and P channel devices, with the sinks at rail potential.
JJ, How many 630/9630 pairs did you use? With their higher thermal resistance junction to case and lower power dissipation ratings, you may have simply pushed them a bit too hard. I used the SFH9240 in my A75s and am very happy with them.
Just to clarify my earlier comment - without isolators the sink is at output potential only with the devices Al linked to and similar. With IRFP devices, the Drains would make contact, requiring separate isolated heat sinks for N and P channel devices, with the sinks at rail potential.
Bob-
I used 12 pairs per channel. I never did run with full bias across the output mosfets because the heatsinks were too small and the mosfets were placed much too close together on the heatsinks.
The amp smoked because I had left a wire hanging off of one of the source resistors that I would attach a multimeter probe to so I could check the bias. One day the wire touched the heatsink and grounded out the bank of mosfets.
JJ
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.