[luvdunhill]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif P.S. They're all FETs, whether MOS or J, they're still being used for the same purpose (as a JFET) - the first diagrams I posted should work fine.

Perhaps this is more appropriate in another thread, but...

When the GS junction is shorted on a JFET, the device saturates. When this happens on a MOSFET, it is off. Again, they are not equivalent and you aren't testing anything if you test a MOSFET this way.

 

[bhjazz]

Whoa...crap. I forgot I still have boards for this. Yikes! Gotta get going!

 

[tomb]

Quote:

Originally Posted by luvdunhill /img/forum/go_quote.gif Perhaps this is more appropriate in another thread, but... When the GS junction is shorted on a JFET, the device saturates. When this happens on a MOSFET, it is off. Again, they are not equivalent and you aren't testing anything if you test a MOSFET this way.

Ah, OK - understood. I was led to believe they would work as a JFET. Then perhaps they aren't appropriate in this application and we should just stick with the SK/SJ's.

 

[tomb]

Let me make this clear since I confused things quite a bit in the last few posts: Luvdunhill is correct about the Zetex MOSFETs and the testing circuit he explained. There is a nice article by Nelson Pass that explains this further:
http://www.passdiy.com/pdf/mos.pdf

In it, you will find similar schematics to the one Luvdunhill provided and an in-depth explanation for testing MOSFETs. (Ironically, Pass states at one point that perhaps "matched parts is the best approach doesn't always hold up.") The Zetex's are also specifically mentioned.

I apologize for causing confusion over this.

 

[BoilermakerFan]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif Let me make this clear since I confused things quite a bit in the last few posts: Luvdunhill is correct about the Zetex MOSFETs and the testing circuit he explained. There is a nice article by Nelson Pass that explains this further: http://www.passdiy.com/pdf/mos.pdf In it, you will find similar schematics to the one Luvdunhill provided and an in-depth explanation for testing MOSFETs. (Ironically, Pass states at one point that perhaps "matched parts is the best approach doesn't always hold up.") The Zetex's are also specifically mentioned. I apologize for causing confusion over this.

No worries Tom, I wasn't confused. I was planning to keep my prototype the BC546C/C556C for small signal and try the Zetex on my MOSFET-MAX on a production board. Seriously considering TO-92 sockets for that build so I can easily swap small signal transistors though...

I think the fact that the MAX can be built so many ways may confuse a few peeps, but that flexibility is what drew me to the MAX in the first place. The support of the MAX community is just the icing on a really big cupcake!

 

[luvdunhill]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif Ah, OK - understood. I was led to believe they would work as a JFET

Perhaps what you mean to say is depletion mode MOSFETs behave similar to a JFET in that an electric field is used to deplete the channel of current carriers. However, these devices are enhancement MOSFETs, and work a bit differently. As you said, there really isn't any reason to match devices in this circuit, as the output devices aren't paralleled. Furthermore, if you're looking to pick a complimentary pair from your matching efforts, good luck. Complimentary MOSFETs (with the very, very rare exception) are sorta up there with unicorns and pink ponies.

 

[tomb]

Quote:

Originally Posted by luvdunhill /img/forum/go_quote.gif Perhaps what you mean to say is depletion mode MOSFETs behave similar to a JFET in that an electric field is used to deplete the channel of current carriers. However, these devices are enhancement MOSFETs, and work a bit differently. As you said, there really isn't any reason to match devices in this circuit, as the output devices aren't paralleled. Furthermore, if you're looking to pick a complimentary pair from your matching efforts, good luck. Complimentary MOSFETs (with the very, very rare exception) are sorta up there with unicorns and pink ponies.

Now you're confusing me further. There are plenty of both paralleled and complimentary devices in the MAX's diamond MOSFET buffers. The real question about matching is whether the juice is worth the squeeze. For output devices, this is almost always not the case. However, at the sensitive location of the paralleled/complementary signal input pair - matching usually makes a big difference. This is because while the individual channels can be biased independently, the complementary halves of each buffer channel cannot. Offset at the signal input pair is magnified the most. Unfortunately, the procedure discussed on the last page or two makes me think we'd still be better off with the low-noise Toshiba's, despite their supposed scarcity. At least the effort of finding matches is much less severe and totally repeatable.

 

[luvdunhill]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif Now you're confusing me further.

Wait, you want to substitute the Zetex MOSFETs for the input JFETs? Er, why would you expect that to work? I think everyone in this thread is confused

I'll just let you guys work it out, or PM me if I can help.

 

[tomb]

Quote:

Originally Posted by luvdunhill /img/forum/go_quote.gif Wait, you want to substitute the Zetex MOSFETs for the input JFETs? Er, why would you expect that to work?

I was told that they might be good replacements for the Toshiba 2SJ74/2SK170 JFETs that were specified by AMB for the JFET-mod MOSFET-MAX, which pretty much forms the basis of the MOSFET-MAX. They substitute for QB2L/R and QB3L/R in the diamond buffer circuit, which were formerly BJT's in the BJT-version of the diamond buffer. The result is that they provide both low-noise, small signal capability along with super-high input impedance. The gate capacitance of the output MOSFETs creates a low-pass filter with the input impedance of "normal" BJT's to cause a high-frequency roll-off. AMB's JFET-mod solves this, but there is an issue with the minimum bias currents/resulting voltage at QB2L/R and QB3L/R that may be less than optimium. I was told the Zetex's may be a better fit there.

We are proposing to change the complementary output MOSFETs for similar reasons - lower gate capacitance - to the IRF510/9510's, which are the same as used on the SSMH (IRF510).

 

[BoilermakerFan]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif I was told that they might be good replacements for the Toshiba 2SJ74/2SK170 JFETs that were specified by AMB for the JFET-mod MOSFET-MAX, which pretty much forms the basis of the MOSFET-MAX. They substitute for QB2L/R and QB3L/R in the diamond buffer circuit, which were formerly BJT's in the BJT-version of the diamond buffer. The result is that they provide both low-noise, small signal capability along with super-high input impedance. The gate capacitance of the output MOSFETs creates a low-pass filter with the input impedance of "normal" BJT's to cause a high-frequency roll-off. AMB's JFET-mod solves this, but there is an issue with the minimum bias currents/resulting voltage at QB2L/R and QB3L/R that may be less than optimium. I was told the Zetex's may be a better fit there. We are proposing to change the complementary output MOSFETs for similar reasons - lower gate capacitance - to the IRF510/9510's, which are the same as used on the SSMH (IRF510).

Whoah, whoah, whoah...

Why don't we just switch to the LSK170s then? I'll check and see if they have the SJ74s...

 

[luvdunhill]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif I was told that they might be good replacements for the Toshiba 2SJ74/2SK170 JFETs that were specified by AMB for the JFET-mod MOSFET-MAX, which pretty much forms the basis of the MOSFET-MAX.

oh ok, I wouldn't guess that would work so well, or require some sort of (undesirable) compensation. I guess you'll find out. Funny, all the hits on google for "MOSFET diamond buffer" are MAX related

 

[tomb]

Quote:

Originally Posted by luvdunhill /img/forum/go_quote.gif oh ok, I wouldn't guess that would work so well, or require some sort of (undesirable) compensation. I guess you'll find out. Funny, all the hits on google for "MOSFET diamond buffer" are MAX related

MOSFET-MAX website
EDIT: I'm not aware that anyone else has actually built one (a MOSFET diamond buffer), except us - much less used it as a primary design for a headphone amp.

The Toshiba JFETs work great, btw and have served quite well as the basis for the MOSFET-MAX. However, there are issues - not the least of which that they are no longer made and quite scarce. That said, Beezar.com sells them and so does AMB Labs.

 

[BoilermakerFan]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif I was told that they might be good replacements for the Toshiba 2SJ74/2SK170 JFETs that were specified by AMB for the JFET-mod MOSFET-MAX, which pretty much forms the basis of the MOSFET-MAX. They substitute for QB2L/R and QB3L/R in the diamond buffer circuit, which were formerly BJT's in the BJT-version of the diamond buffer. The result is that they provide both low-noise, small signal capability along with super-high input impedance. The gate capacitance of the output MOSFETs creates a low-pass filter with the input impedance of "normal" BJT's to cause a high-frequency roll-off. AMB's JFET-mod solves this, but there is an issue with the minimum bias currents/resulting voltage at QB2L/R and QB3L/R that may be less than optimium. I was told the Zetex's may be a better fit there. We are proposing to change the complementary output MOSFETs for similar reasons - lower gate capacitance - to the IRF510/9510's, which are the same as used on the SSMH (IRF510).

No 2SJ74 equivalent from Linear Systems...

Are the BC546C/BC556C or the low noise equivalents "normal" BJTs?

 

[tomb]

Quote:

Originally Posted by BoilermakerFan /img/forum/go_quote.gif No 2SJ74 equivalent from Linear Systems... Are the BC546C/BC556C or the low noise equivalents "normal" BJTs?

In the case of the MAX/MOSFET-MAX/MiniMAX, "normal" BJT's are 2N5087 and 2N5088.

 

[BoilermakerFan]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif In the case of the MAX/MOSFET-MAX/MiniMAX, "normal" BJT's are 2N5087 and 2N5088.

So I'm wondering if the BC's have a higher input impedance to minimize or prevent the high freq. roll-off or do BJTs inherently have low input impedance? I did not see the input impedance in the data sheets for the Fairchild parts I have, they only listed capacitance.