jfet input stage versus op-amps: opinions

[Wounded Paw]

I'm building a fairly extensive bass pre amp with it's own drive section, effects loop, clean blend, baxandall tone controls and it runs off +/- 15V. 
It's using op-amps for splitting and mixing signals, filters, and gain.
My question is about the input stage.  Currently it's an op-amp buffer which may use an NE5532 bi-polar based op-amp or possibly an OPA2134 fet-based op-amp.  This stage also splits at the output of the op-amp to different sections of the pre-amp. 
The question is will I gain anything by adding a discrete jfet transistor before this op-amp as the input buffer? 
The op-amp will still be there to provide the drive for the signal splitting.
The OPA2134 is fairly expensive so adding the discrete jfet transistor is the cheaper way to get a jfet input.
I like the idea of having a high impedance jfet input and possibly skipping the input capacitor all together.
So I'm looking for opinions based on experience.  Sound quality, etc.

[DavenPaget]

TL062 , TL072 , LF353 
Don't forget those .

[Wounded Paw]

I've already decided against TL062, TL072 in favour of higher quality, lower noise op-amps.  But they are all pin compatible so I can swap them at will.  If I use a fet op-amp it will be the OPA2134, specialized for audio.

[Seljer]

the NE5532 has a relatively low input impedance as far as guitar type signals are concerned

[DavenPaget]

I've already decided against TL062, TL072 in favour of higher quality, lower noise op-amps.  But they are all pin compatible so I can swap them at will.  If I use a fet op-amp it will be the OPA2134, specialized for audio.

Wow , you sure want even lower noise stuff . Come on , it's probably a design flaw if you can hear the minute amount of noise a 72 makes .

[Earthscum]

Wow , you sure want even lower noise stuff . Come on , it's probably a design flaw if you can hear the minute amount of noise a 72 makes .

+1 on that. Take a look at some Marshalls... particularly the IBS amps: http://drtube.com/marshall.htm#IBS

Worked fine for them.

[Wounded Paw]

well op-amps asides, I'm really asking about the use of a discrete jfet as the input stage as opposed to straight to the first op-amp.

[Pablo1234]

The only thing you will get from adding a transistor is Asymmetrical gain and probably added noise. Alot of people like transistor gain as it slightly colors your sound.

[Wounded Paw]

that's more of the kind of info I was looking for.

[DavenPaget]

well op-amps asides, I'm really asking about the use of a discrete jfet as the input stage as opposed to straight to the first op-amp.

1) Probably higher ratings .
2) Not any better slew rates .
3) Transistor colouring .
4) Silicon resistors are alot more accurate .
5) Everything has been matched internally .

[amptramp]

It is easy to bias an op amp.  JFET's have bias variations that go all over the place, which is why some people still use bipolar transistors - at least you know what a base-emitter voltage is going to be.  JFET op amps have matched input transistors so that you get a small variation, like 10 millivolts, between them.  Just look at the JFET matching problems some people have with phasers.  You also get good noise data with the typical op amps we tend to use - like 18 nanovolts per root Hertz for a TL072.  Theoretically, the input FET's in the TL072 would have 0.707 times this noise if they were available as discrete devices, but the TL07X series is good enough that most applications don't require anything better.

[merlinb]

well op-amps asides, I'm really asking about the use of a discrete jfet as the input stage as opposed to straight to the first op-amp.

A TL07x has an equivalent input noise voltage of about 4uV.
OPAx134 is about 1.2uV
NE5532 will be about the same, but it has much more current noise, making it a poor choice.
A single JFET is likely to be around 1uV or even a little less, which is good, but you of course sacrifice all the advantages of using an opamp. Maybe that doesn't matter to you.