JFet vs bipolar

[sonnya]

Originally posted by rickpt


whit a 50k pot that gives you about 10k of output inpedance at mid point position, fet amplifiers should be at home, and bipolar amplifiers should perform much worse... that wasn’t the case!



did you verify this with a null test? remember that the null test isn’t a typical thd test! all non linearity’s are measured.

about that resistor I don’t believe that was the case... I believe that Mr. Bateman had a parasitic oscillation when he inserted the AD797 on the circuit, and that resistor dampened that oscillation, and off course, distortion was lower, so nothing more than a base stopper...

now look at the typical diy'er, normally the topology is the same and only opamps are swapped...that was what I did but instead of listening, I was measuring

it's fairly easy to do a null test , so do it yourself and find your results...

cheers

Ricardo

No i have not verified the test setup.

Not for being a pain. Still before you can compare the opamps, there optimum working conditions, as described by the designer have to be meet.

It could be bypassing caps, as already mentioned source impedance. (Jfet are subjected to higher distortion when the imbalance of the source impedance is not corrected). Notes on how to add a series resistor to the Cf like the one in the AD797 datasheet ... The addition of RS has actually great impact on the open-loop gain and phase margin as shown in the datasheet figure 14.

Then you have to have very high CMRR on your measurement tools, not to get the result distorted by a low CMRR.

When all this is okay, then you can start comparing.

Even though you still needs to take PRR's words into account.

Originally posted by PRR
> What is the file type?

I thank Ricardo for the numbers, but without knowing how much of the number is phase-shift, noise, low- or high-order distortion, or if the test rig was optimally adjusted for each amp (in ways we could do in the real world), I don't trust it without trying things on the scope.

Regards

The pain.

[rickpt]

I thank Ricardo for the numbers, but without knowing how much of the number is phase-shift, noise, low- or high-order distortion, or if the test rig was optimally adjusted for each amp (in ways we could do in the real world), I don't trust it without trying things on the scope

PRR, these measurements where made with a scope, I saw the residual wave! at lower frequencies it was dominated by noise on all opamps! at higher frequencies it was clear that the residual was dominated by harmonic disrtortion+noise, specially the fet devices and the op27!all opamps have good bandwidth and I didn’t use any input filtering, so phase wasn’t a problem!

that thing about optimizing topology sounds very nice on theory, but how can you be sure you are comparing the opamps only or you are comparing different topologies? where to set the line?

this test wasn’t made to measure the lower distortion I could achieve but compare different devices on a typical application! A Pre-Amplifier

Jfet are subjected to higher distortion when the imbalance of the source impedance is not corrected

how can you balance inpedance on both inputs of a opamp when you have a potentiometer on one input? it's just show another type of distortion that affects fet devices, common mode distortion!

instead saying my test is wrong, please do the test yourself and refute my findings! it would create a very nice discuss

Cheers

[Terry Demol]

Originally posted by Jocko Homo
Some of the stuff in this thread is ludicrous. But what do I know........30+ years of doing this has taught me nothing.

Jocko

You crack me up sometimes Jocko!

I'm not even buying into this thread.... it's an entertaining read
though

Terry

[Jocko Homo]

Most be losing my touch. Didn't think that I was getting that soft in my old age.

Oh well..........

Jocko

[johnferrier]

Originally posted by rickpt
how can you balance inpedance on both inputs of a opamp when you have a potentiometer on one input? it's just show another type of distortion that affects fet devices, common mode distortion!

One could approximate the typical volume control setting. Also, Borbely has described how cascoding the FET input fixes the problem with input capacitance.

See section titled "Cascode to the Rescue" on page 30 here:
http://www.tkhifi.com/div/Erno_Borbe..._articel_1.pdf

Borbely uses discrete circuitry; however, it seems that cascoding ought to help integrated devices as well.


JF

[johnferrier]

Douglas Self has a website with THD graphs of the NE5532 and the OPA627 (and other BJT and FET opamps).

See:
http://www.dself.dsl.pipex.com/ampins/webbop/opamp.htm



JF

[Terry Demol]

Originally posted by johnferrier


One could approximate the typical volume control setting. Also, Borbely has described how cascoding the FET input fixes the problem with input capacitance.

See section titled "Cascode to the Rescue" on page 30 here:
http://www.tkhifi.com/div/Erno_Borbe..._articel_1.pdf

Borbely uses discrete circuitry; however, it seems that cascoding ought to help integrated devices as well.


JF

OPA627 does have a form of cascoding on IP stage.

Cheers,

Terry

[Terry Demol]

Originally posted by rickpt


PRR, these measurements where made with a scope, I saw the residual wave! at lower frequencies it was dominated by noise on all opamps! at higher frequencies it was clear that the residual was dominated by harmonic disrtortion+noise, specially the fet devices and the op27!all opamps have good bandwidth and I didn’t use any input filtering, so phase wasn’t a problem!

For a given load, quiescent current and with both devices
degenerated for same gain, Jfets are more linear than BJT's
but not by much. I've measured it with AP.
However the jfets have much worse capacitance modulation with
voltage swing so I assume this is hard to deal with in monolythic
low supply voltage (+-15) implementations and at higher
frequencies there is less feedback to correct it.

Having said this, I feel a good newer opamp such as OPA627
is quite superior sonically than the 5534 in most low gain
applications.

I (and client) have compared 5534 vs 627 in mixing desk summing
point and the 5534 became much more sonically constricted
when more channels+level were introduced. The 627 was
cleaner and stayed cleaner longer with bigger
levels and more channels. No question.

We have compared 5534 vs 627 in DAC I-V, as stated before
the 5534 is a bit slow for this application, the 627 was way better.

instead saying my test is wrong, please do the test yourself and refute my findings! it would create a very nice discuss

Cheers [/B]

I'll have a play with it, looks interesting.

As a related point you appear to be a pro audio guy,
I know early series SSL's were riddled with 5534's and
these consoles are known for thier bad sound (but incredible
flexibility) They have since introduced the J series which was
much better and now the K series. Are these still full of 5534's?


Terry

[rickpt]

However the jfets have much worse capacitance modulation with
voltage swing so I assume this is hard to deal with in monolythic
low supply voltage (+-15) implementations and at higher
frequencies there is less feedback to correct it.

This can be partially dealt with a cascode...

I (and client) have compared 5534 vs 627 in mixing desk summing
point and the 5534 became much more sonically constricted
when more channels+level were introduced. The 627 was
cleaner and stayed cleaner longer with bigger
levels and more channels. No question.

We have compared 5534 vs 627 in DAC I-V, as stated before
the 5534 is a bit slow for this application, the 627 was way better.

for me those are just a subjective comments and doesn’t mean a thing, some like apples, others like oranges...
Remembers me one talk I had with a friend that is the Meyer Sound loudspeakers importer for Portugal, they had one model that had one type of compression driver that generated a lot of distortion because of the used topology of the horn, but people seamed to like very much that speaker wen they first listened to it, because of it's rich highs
As you can see the ear can be easily tricked...

As a related point you appear to be a pro audio guy,
I know early series SSL's were riddled with 5534's and
these consoles are known for thier bad sound (but incredible
flexibility) They have since introduced the J series which was
much better and now the K series. Are these still full of 5534's?

There are only 2 SSL consoles that I know of, in Portugal and are both 4000 series, and pretty old by now, so I really cant answer that question but there is the SSL 9000 preamp schematic floating around the web and uses the NE5534 and MAT02...
Here in Portugal we have a pretty small music market and it isn’t easy to rentable such an expensive console.

I know early series SSL's were riddled with 5534's and
these consoles are known for thier bad sound

All sound engineer I know say that those mixers sound great, and almost all studios around the world has a SSL4000 console, if its so bad why it is still used?

Cheers

Ricardo

[JFETglare]

Wow, what a thread...

I find the 5532's sound inoffensive and honest, for lack of a better word... It reminds me a bit of a 4556.

Anyway, I always find myself going back to the sound of a BJT over a JFET. If I listen to two circuits, each carefully designed and built to accomodate either transistor, I find the JFET one has something that slowly wears on my nerves. It is an artificial sound, mostly concentrated in the upper presence band.

The BJT sounds more neutral in comparison, and often times seems to sound more open and spacious.

Though not exactly on topic, I find MOSFETS configured as a gain device can often have a sweet sound, almost triode like. They are sensitive to circuit design however... I have listened to some MOSFET amps that sound "misty" and thin. I wish MOSFETs were not so static sensitive, and that they had a better reliability record--they are not forgiving devices, certainly nothing like a BJT.