JFet vs bipolar

[Asbjorn]

Hi!

I'm from Denmark and is working on a highend preamp sparetime project.

I've been doing some listening tests of jfet vs bipolar transistors, and found some generel sound differences.
Jfets tends to sound softly, while bipolar sounds more dynamic.
Do you have any experiences like this, and which do you think sounds the best?

Regards
Asbjorn

[DrG]

Both can be made to sound excellent in a well-designed circuit. The JFET would be more forgiving in a low/no feedback application I think... Chat to John Curl... he's very well versed on things of this nature.

[richwalters]

You should be able to get the noise level equal to standard bipolars or better. Jfets provide an easy opportunity as a voltage controlled resistor (VCR) for channel selection without clicks.
When the bias points are properly selected, the low noise of high gfs junction fets is obtained.
The fet is voltage operated while the transistor is current operated, so FEts have a lower noise current. They should be preferred in audio appls where now noise is a requirement but not always. The industry standard IC 5534A bipolar is still used practically in most studio consoles mic/ line interfacing circuits. One point to watch out with fets, they are basically RF devices and will "take off" with bad layouts.
I adore them, as a Jfet is in many ways similiar to a triode.
Fets have very high dynamic range greater than 100dB with very low IM distortion products.
Go for it.
): rich

[Asbjorn]

Actually I believe jfets sounds alot more like triodes than bipolar too! But this smooth, soft and nice sound it delivers - is it correct or some kind of "coloration" that's being added to the original signal? I'm afraid it is!

Btw, I think ne5532 is better than those popular fet opamps from AD and BB. I think all the AD and BB opamps I've heard sounds bright and undynamic comparede to ne5532, which is really the only opamp I know with good dynamics from 500hz and downwards (sonically). Also it is more detailed, but it is detailed in another way than the AD and BB's. It is detailed in a way like the sounds coming from the real world are; you really don't hear them - they just are there! With the AD and BB opamps there is some kind of "focus" on the details, and they therefore sound more detailed at first, but really the details are seperated from their natural context - at least that's what I think!

But what do you think about jfets and bipolars? Especially sonic experiences have my attention.

[richwalters]

from the pit......When I was in audio some years ago, the little thing we used to do in design work was to create a composite op-amp i,e strap Jfets to a 5534. This will create dammned good slew rate of 20V/uS upwards with unity gain stability. A reason that the 5534 (low noise version suffix A) is so often used is that it is designed for 600ohm loads. There are IC's around with this configuration, but you will have to sift through the catalogues.

Pays to visit www.vishay.com/company/brands/siliconix/ for more physics understanding. (analogue small signal box)
The bonus of the 5534 as a building block is the high current bipolar o/p with an o/p resistance of 0.3 ohms and can operate from +/- 15 V supply.
The application note you need to get hold of is AN81-3 which goes back to the Siliconix design catalogue dated July 1983. I couldnt' find this although I have the book.

The 5532 is two internally compensated 5534 chips, fitted in the same die. The 5532 is then technically close to max dissipation from a +/- 15V supply if loading 2x 600 ohm outputs. pin 5 & 8 has a compens cap of about 47pF, a figure which I find a bit high.
As you probably know the THD from these blocks is virtually nothing, we are down into noise typ 0.001%.
The TL071/2 family suffers the disadvantage of not being able to slam drive into 600 ohm outputs, and somwhat noisier. I don't use these for audio work.

I am not up to comparing manufacturers similiar devices, like for like, the sonic differences are interesting especially when everyone shuns class B operation, and these IC's use it!


Unfortunately all this technical progress makes the conventional performance and bandwidth of the signal triode tube look rather silly. I do like using tubes but I am more driven to pentodes as they are more efficient and one can do alot more with them. It is a disadvantge in audio designs that I am always fighting to get the noise level to respectable low level. The question of 2nd 3rd 5th harmonics is interesting as this is far more easier to observe in tubes than IC's. Naturally we are up all getting mixed and cancelled in in the output transformer-. on when did it all matter at board level ??? More and more studio performers utilise 2nd harmonic in the mixing process especially in the bass end.

More anon
):rich

[TimA]

I have listened to the NE5534 in three different circuits, one where it was configured as a buffer, another where it performs I/V conversion (with no buffer following it), and another where it was used as a balanced to unbalanced converter. In the second two circuits mentioned above I have also listened to many BB fet input, AD bipolar input and AD fet input devices, I have also built a discrete op-amp where the input can be either fet or bipolar. My own preference so far has been for the bipolar devices every time; NE5534, AD797 and the discrete circuit. I find them clearly superior in... their ability to set a natural perspective that has a realistic portrayal of depth, their 'focus' and tonal/textural accuracy, their bass quality. To put thinds into perspective a little more, I have not heard a fet op-amp that didn't sound artificial, cold, sterile and fatiguing to some degree, the OPA627 and AD8610 are partial exceptions here, though I still prefer the sound of the NE5534 for its sheer naturalness!. In my experience the NE5534 is unmatched in its ability to portray the acoustic signature of the recording venue, the AD797 is also a favourite of mine due to its refined sound. I find that a CCS on the output for class A operation improves 'smoothness' and 'purity', particularly in the treble.

Tim.

[PMA]

JFETs have wider linear range of input voltages compared to BJTs. Input linear voltage range of Gm characteristics is about +/-800 mV for JFETs compared to some +/-200mV for BJTs (I speak about input differential pair stage). BJT pair has more steeply sloped Gm curve and more narrow input dynamic range.

[richwalters]

I put it to a vote. The 5534 is the best op amp to build audio circuits around. As to circuit details, for optimum low noise performance it pays to keep the o/p feedback resistor back to inverted input less than 10K in this mode of operation. 5K is actually optimum.
The LM1875 power amp is also a good specimen for a general purpose headphone amp.
rich

[macka]

Somewhere I have an old article from Audio Ameteur my a studio engineer about this topic. I will try and locate and scan for you.

The gist of it was how the 5532/34 is set up in terms if feeback and input filtering regards open/closed loop gain HF Margin. Also input filtering, this is why the 5532 earned such a bad reputation in the CD player with the audio press, so called designers were basically clipping the chip in the HF area.

John Stammer also wrote an excellent article on the application od audio opamps for pro mixer console in AE.

He Chose to add class A biasing and also servos to eliminate outptu capacitors. In comparisons he reckoned the Jenson discrete opamp and and the Borbely were the best sonically.

Not to mention regulators the 3 pins 7815, no way baby, use the Jung hi performance regulators in these applications.

In my own experience I thought these opamps were bearable untill I built the BOSOZ

Ian

[Richard C]

For the most natural musical presentation I've found simple SE bipolar designs to work extremely well and are certainly superior to op-amps in this respect.

I wouldn't use an op-amp in an audio amplifier it just isn't necessary: why use 50+ transistors when 5 will do? The sonic performance of a simple discrete circuit can be astounding compared to an equivalent op-amp design.

Who cares whether an amp produces 0.1% THD or 0.001% there are more important factors.