FET topology - why so superior?

Hello tha'!

Thought it was good to perhaps come over here for a change...=)

As I've mentioned earlier (in gearslutz) I am writing a paper related to limiters, and really would need some views on one of my main questions. Please...

One often reads about that the JFET topology was seen as much superior than the older vari-mu toplology at its time. UREI called their 1176 a superor limiter and in another article (written by two norwegian guys in 1965) they said that there was a need for a tranisistorized topology within broadcasting, now when so many other devices were tranistorized.

But my question is, why was the JFET topology seen so much superior, when it featured most of the cons that the vari-mu topology featured - thump, distortion, non-linearities etc.?

And the FET limiter wasn't necessarly much faster than a vari-mu limiter either, and still had the same problems with higher distoriton for higher attack time, anyway. But from reading old articles it seems as if the arrival of a FET limiter seemed so important, especially for the broadcasting industry.

Any ideas?

Danke

:idea:

It's the same old story, it's always better if it's new. Then the dust settles and the objective analasis begins. I guess you could say that even if tubes and FETs suffer from the same problems that the FET wins on lower noise price and reliability.
Most of the tube thing is nostalgic but I love them and they do SOUND different, same with transformers.
Steve

Do you really expect somebody made something cheaper (JFETs vs. tubes, cheaper PSUs, etc.) and wouldn't find reasons why it is superior? :shock: :shock:

I never actually saw any manufacturer saying that it's just different...

Does a FET limiter like the 1178 thump?

I'd expect it to bee much less than variable Mu (but then again I just start exploring variable Mu)

If a FET thumps, where would the CV feedthru come in?

I thought with the FET working in its "linear" region, there isn't any DC offset - it's a VC resistor AC coupled to the signal path.

The CV (as long as the GS junction does not conduct) is only coupled
to the signal path by a very tiny capacitance. Should this be enough to cause any considerable CV feedthru for fast transients?

In a vari mu setup, you always shift large amounts of DC in the signal
path, and only by symmetry of the circuit this cancels in the ideal case. Imperfections of symmety can only be trimmed out to a certain degree - what remains is the CV feedthru.

A FET limiter has no _such_ problem, at least.

JH.

> the JFET topology was seen as much superior than the older vari-mu toplology at its time. UREI called their 1176 a superor limiter

"Buy our new 2005 model, with ActiveZ technology! It isn't any better than last year's ActiveY technology, but we need the money so just buy it!"

They never say that. There is always some "improvement", no matter how meaningless. In retrospect, many "improvements" have not been good for sound.

However, tube-based limiters are BIG and HOT and in 24/7 duty they need new tubes often. Compare that with a Shure LevelLoc: the guts fit on 2 fingers and run on batteries, and they never need repair or calibration. If you are upgrading or downsizing a studio, or converting to stereo, the new solid-state limiters WERE better than the old beasts.

A lot of the faults of limiters are about MONEY. After all, the Fairchild 670 isn't a bad machine, just unaffordable. When you change from $1000 of tubes and transformers to $10 of FET, you can put $90 into a better side-chain design and still shave $900 off the price. So it is possible for an FET limiter to be a better box at a lower price.

The FET limiter does not have thundering DC current shifts to cancel. The Gate is higher-Z than most Grids, so the sidechain load is lighter (none of those 2W or 10W sidechain amps).

> the FET limiter wasn't necessarly much faster than a vari-mu limiter

To a first approximation, either Tube or FET response speed is under a microSecond, "instant" for any audio purpose. However tubes may have more stray capacitance, grid and transformer, so it may be harder to drive them really-fast (not an issue in audio limiting, even Fairchild's fast attack).

> still had the same problems with higher distoriton for higher attack time

That can be reduced (not eliminated) by more sophisticated sidechain designs. See remarks on "MONEY" above.

The Aphex 1001 VCA is the "best in the world", because they say so and we must believe them, being leaders in their field, and naturally we all believe what our leaders tell us without question, do we not? Do we know what is inside the 1001? Maybe the odd FET perhaps!
Stephen

> Do we know what is inside the 1001? Maybe the odd FET perhaps!

It was widely "discussed" when new. We know it is a BJT diff-pair variable current mirror. It is not the same as DBX (now THATcorp) modules, and the differences were loudly argued when the field developed, but for basic grasp of principle you can look at a DBX.

The Aphex/DBX modules (with associated opamps) were made to be variable gain/loss amplifiers for wide-range level control deep inside consoles. You do not need all that range and flexibility for a limiter, which only turns one way (down). Using a gain/loss VCAmp as a limiter is a little awkward: hard to find the Unity-gain point and keep it from drifting. With an R-FET VCAtt, the unity-gain point is set for life, even if the FET falls out. Sure, excellent limiters are built on VCAmps: some make-up gain is often handy and voltage-control avoids using a dual-pot for stereo. But you can't beat the FET for simplicity with good performance.

Perhaps then combining Aphex sidechain technology and good FETs, one ought to come up with something half decent.
Stephen

Thanx

[quote author="PRR"]> Do we know what is inside the 1001? Maybe the odd FET perhaps!

It was widely "discussed" when new. We know it is a BJT diff-pair variable current mirror. [/quote]
... Yes, transilinear principle.
But this performs the same with FET pairs like the bipolars.
But for the bipolars is only one possibility.
For the FETs second (and simplier) possibility is triode-like (or
resistive) part of characteristic.
And vairable mu can be third possibility with fets, not as good
asi with selectode - tubes because no producers products
selectode - FET.

But transilinear FET voltage controlled gain stage is not known in audio.
(and is widely used everywhere...)
And it can be superior and easy-integrable and ......
I mean, that transilinear with FETs are too new to use it in
analog audio processing. Main factories are on DSP.

It is the same problem as using current - mode (and not voltage
amplifiers) in audio circuits. It is too late (or too soon ?)

Best regards, PRR;
xvlk

.....................

PRR mentioned that more sophisticated side chains could solve the problem of distortion at faster attack times.

I recently read an interesting article that mentioned a design that BBC came up with in 1968. To achieve non-overshoot limiting, without distorting the signal, they combined a feed-forward and a feedback configuration, using two variable gain stages (using JFET's). The first stage was the fast acting, connected in a feedback loop. This stage was providing the second and main variable gain stage, slow acting in a feed-forward configuration, with the CV. The second stage was also delayed.

But I'm just curious would this configuration have been possible before a solid-state topology (like the FET) came around, using valves? Because BBC stated that this configuration solved the problem of the, far away from perfect, gain/bias relationship of the FET.