MOSFET input stage?

[Mr Evil]

I'm frustrated by the poor availability of SMT dual JFETs, so I wondered if MOSFETs might make an acceptable alternative for an op-amp input stage (LTP).

There tend to be significant differences in gain and input capacitance, but what sort of differences in noise and distortion could be expected? Any particular difficulties to be aware of?

 

[Joe Berry]

I'm using a ZVN3310 (TO92) LTP input stage in one of my power amps and like the sound. I haven't measured noise, but subjectively, it's not an issue with the LPT working open-loop and using source degeneration to give 20-26 dB stage gain.

It appears that a SOT23 version of this device is available with a reduced current rating, which may better suit your purposes. My experience has been that you have to start with maybe 10 pieces to get a couple of really well-matched pairs.

 

[ilimzn]

I've used 2N7000/7001, BS107 and BS170 in this role with good success. It pays to have them cascoded to reduce nonlinear Cgd capacitance problems unless you can guarantee small output swings and high and Vgd.
They also work best with relatively high tail current, but you sould watch the dissipation. The gm improves with tail current, and you ave ample current to feed any kind of VAS you care to put there. Gm is also higher than in JFETs but still considerably lower than BJTs. Pairing the MOSFETs for Vgs is mandatory if you want low output offset voltage and minimum distortion. Fortunately, all the types mentioned above are cheap and you can buy 10-20 and get a few good pairs.

 

[PMA]

I'm frustrated by the poor availability of SMT dual JFETs, so I wondered if MOSFETs might make an acceptable alternative for an op-amp input stage (LTP).

There tend to be significant differences in gain and input capacitance, but what sort of differences in noise and distortion could be expected? Any particular difficulties to be aware of?

MOSFETs as input stage are great. You have to drive them from pre with very low output impedance, to eliminate non-linear input capacitance. Just do it and you will be surprised by a very good result. They are more noisy, but it does not matter for the gains used in a power amp.

 

[Mr Evil]

Re: Re: MOSFET input stage?

...It appears that a SOT23 version of this device is available with a reduced current rating, which may better suit your purposes. My experience has been that you have to start with maybe 10 pieces to get a couple of really well-matched pairs.

I was looking at the monolithic dual devices specifically to avoid having to hand-match them, otherwise I would stick to JFETs. NDC7002 is the one that looks most promising at the moment, having low-ish input capacitance (20pF) and being very cheap. It looks to be virtually identical to 2N7002.

MOSFETs as input stage are great. You have to drive them from pre with very low output impedance, to eliminate non-linear input capacitance. Just do it and you will be surprised by a very good result. They are more noisy, but it does not matter for the gains used in a power amp.

They will be going into an op-amp which could be used in any situation, dirven by any source and with any gain, so the input capacitance and noise might be a problem.

 

[Joe Berry]

Re: Re: Re: MOSFET input stage?

I was looking at the monolithic dual devices specifically to avoid having to hand-match them, otherwise I would stick to JFETs.

Most understandable. I've avoided monolithic duals in the past because I need Vds > 50-60V and would rather work with more generic through-hole package types. But for you, these are non-issues.

 

[StevenOH]

<<They will be going into an op-amp which could be used in any situation, dirven by any source and with any gain, so the input capacitance and noise might be a problem.>>

In this case you should consider some Sanyo JFET pairs:
2SK332 - 40 V, 0.2 W/ch, S~17 mS, Ci/Cr ~13/3 pF
2SK333 - 80 V, same, S~15 mS, Ci/Cr ~11/2 pF
both come in 4x3x2 mm packages, with leads on both sides.
FC11 - 40 V, 300 mW total, S~9 mS, Ci/Cr ~9/2 pF (common Source).
Also, check the Fairchild catalog, my impression they had SM dual JFET too.
Linear Systems is still another source for SM dual JFET.

 

[Mr Evil]

2SK332 - 40 V, 0.2 W/ch, S~17 mS, Ci/Cr ~13/3 pF
2SK333 - 80 V, same, S~15 mS, Ci/Cr ~11/2 pF

I hadn't seen them before; they look good. I can find very little information on them though - perhaps you know where I might find a datasheet? They do still have very poor availability, with only one place I can find selling 2SK332 in the UK and none selling 2SK333. Maybe worth getting some for testing, but I would rather stick to something more common for the final design.

FC11 - 40 V, 300 mW total, S~9 mS, Ci/Cr ~9/2 pF (common Source).

That one, though good, has the two sources commoned, which means no way to use degeneration resistors.

FC11 - 40 V, 300 mW total, S~9 mS, Ci/Cr ~9/2 pF (common Source).
Also, check the Fairchild catalog, my impression they had SM dual JFET too.
Linear Systems is still another source for SM dual JFET.

Fairchild don't seem to have any dual JFETs. At least not listed on their website. What they do have is some very interesting dual MOSFETs (FDC6301N/FDC6302P) that appear to be lateral types, with low input capacitance (9-10pF), but unfortunately only 25V Vdss (I want at least 35V).

I've looked at LS before, but again their devices are not easily available to buy.

 

[StevenOH]

When I search for components, I use Google - a lot of useful info.
I also have a list of distributors, that won't always show up in Google. There are plenty of places to shop for 2SK's, I guess FC11 can be a problem to find.
How about dalbani.co.uk, or ceitron.com to start.
http://72.4.160.82/cpcares/pdf/2SK332.PDF or 333.PDF
http://katalog.elektroda.net/download.php?id=100916

 

[Mr Evil]

I did Google for info, but found very little, and what I did find wasn't very useful due to my lack of ability to understand languages other than English.

I noticed Dalbani sell them, and cheap too. They are a good source for all sorts of more obscure parts. But still they aren't very common, and although I can get them I am not sure if it will always be that way, so I would instead base my design on components that are more common.

 

[tlf9999]

I have often used MOSFETs as either input or VAS devices. My favorite are Fairchild small signal MOSFETs: they have very low gate capacitance. However they are hard to find to I use IRF510 from time to time.

I also run them hot: 5 - 10ma for input devices and upto 20ma for VAS (depends on what I want to drive).

I couldn't tell them apart from BJT devices.

 

[ilimzn]

Yes, they do need to be run hot to get into the favorable part of the gm curve, though an IRF510 needs to go even higher (50mA or more, not always feasible).
Small signal MOSFETs sound more BJT-like than JFET-like, that has been my experience too...

 

[Mr Evil]

Looks like 2SK332 is even harder to obtain than it at first appeared: Dalabani say they can't get them anymore, as have a couple of other places I've tried. www.mark5.co.uk show 20 in stock, but they cost an unreasonable £4.71 there.

 

[djk]

In the past, Dalbani has sold me many counterfeit output transistors.

I will never buy anything from them again.