thoughts on this limiter?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
G

Guest

Guest
http://twin-x.com/groupdiy/displayimage.php?pos=-332

I'm curious what might have to say about the limiter described in the paper above. It seems very interesting, I haven't breadboarded this yet but it seems like it might be worth a shot....
 
He is using current source to limit frequency response of controlling loop stating that oscillations as if are caused by high gain in detector. It is not true, high gain itself does not cause them.
 
The use of the term "frequency response" is a little ambiguous here... I prefer response speed.

I don't believe that the frequency/amplitude response (which is what is normally understood by that term) is an issue.

The insertion of a low-order low-cut filter before the input of A2 wouldn't be a half-bad idea though... Nor would the implementation of an input gain-before-threshold control.

Keith
 
As a sidebar, Dr. Leach is also the same person who wrote extensively on TIM or slewrate limited distortion in tube and transistor amplifiers.
 
[quote author="SSLtech"]The use of the term "frequency response" is a little ambiguous here... I prefer response speed.

[/quote]

...or Slewrate. :cool:
 
Thats what i was wondering matthias... guess I'll knock it up as is on breadboard as soon as I get time.
 
IMO a little too many parts in the side chain and not enough in the variable gain element. There are some fairly simple tricks to improve the linearity of FETs used as variable Rs. Check out the AGC in Cordells low distortion sinewave oscillator or other FET app notes. There is a trick to add the AC component of the channel voltage back into the gate down 6dB to reduce distortion.

You can make a crude FET limiter even simpler than that for like a guitar pedal (albeit without the fancy math). If you are serious about sound quality, VCAs are reasonable these days.

JR

PS: Yes, that is the same EE Prof from Ga Tech. who published papers on sundry aspects of amplifier design.
 
[quote author="mr coffee"]Could this be re-done to use a Siliconix VCRN2 n-chan JFET
?

Aren't they supposed to be better than p-chan?[/quote]

From the PDF page 4 - "The circuit is designed for a p-channel FET. A n-channel device cannot be used because the polarity of the control voltage is not correct."

I'm only quoting, I don't know that you couldn't use an n-channel if the circuit was modified (perhaps by using A4 in non-inverting mode, I'm sure there are other issues too)

cheers,
Ruairi

Thanks for the link JCM, really enjoyed studying this tonight.
 
[quote author="JohnRoberts"] There are some fairly simple tricks to improve the linearity of FETs used as variable Rs. Check out the AGC in Cordells low distortion sinewave oscillator or other FET app notes. There is a trick to add the AC component of the channel voltage back into the gate down 6dB to reduce distortion.

[/quote]

He is doing that John, it's just not as obvious as some realizations.
 
[quote author="ruairioflaherty"][quote author="mr coffee"]Could this be re-done to use a Siliconix VCRN2 n-chan JFET
?

Aren't they supposed to be better than p-chan?[/quote]

From the PDF page 4 - "The circuit is designed for a p-channel FET. A n-channel device cannot be used because the polarity of the control voltage is not correct."

I'm only quoting, I don't know that you couldn't use an n-channel if the circuit was modified (perhaps by using A4 in non-inverting mode, I'm sure there are other issues too)

cheers,
Ruairi

Thanks for the link JCM, really enjoyed studying this tonight.[/quote]

It would be sufficient I think to reverse diode and transistor polarities, and the rail for the Q's and R8 would become the negative 15 rather than +15. But there isn't much to be gained for the voltage-controlled resistor applications between P-ch and N-ch IMO. The situation is much different with amplifiers, where the higher mobility of electrons versus holes confers advantages on silicon N-FETs in terms of gain-bandwidth and noise, all other things equal.

Now, if you extended the design to stereo and used a dual FET then you might well want to go with the sex change for parts availability. But as John pointed out you might well reconsider FETs as the active (i.e., ~gain-variable) elements anyway.

BTW JCMaudio: is the complete document only those 5 pages? I should have thought Leach would have provided some performance data.
 
I vaguely recall some discussion about long channel vs. short channel JFETs as being preferred for use in these applications. But I don't understand enough semi- physics to make sense of that, if I'm even remembering it correctly.

I also looked at using duals, in an attempt to linearize and get predictable gain laws. There are much better ways to do that these days (even were back then) . :?

JR
 
Thanks for replies all.

Hi JohnRoberts,

That was my interest in the VCRN2 - it is a long channel device designed specifically for use as a voltage controlled resistor (VCR).

I've got some VCRN2s in the junk box and I was looking for a low current approach for, as you surmised, a guitar effect.

Unfortunately, Leach's fancy math doesn't do me much good since the VCRN2 data sheet doesn't provide the specs used in his equations.

Hi bcarso,
I was thinking the same thing in terms of reversing polarities.

Hi Ruairi,
I think it might be possible to re-do the A4 part, although it might be a little tricky to keep the feedback of the output signal (the linearizing voltage fed to the FET along with the control voltage).

I'm going to fool with this some. I've tried it in spice and it doesn't seem to like modeling JFETs as VCRs very well.
 
The trick to reduce distortion is to simply add back in the ac component down 6dB. There are tradeoffs between the Leach (inverting) topology vs. Cordell (non-inverting) topology, in that with inverting, increased conductance of the FET increases noise gain, while non-inverting runs at an elevated but constant noise gain (pick your poison).

For guitar effects (like bass guitar compressors) the control side chain can be as simple as a bipolar transistor Vb-e threshold, or a pair of them to full wave detect. Attack and release can be set by ear for desired sound.

Have fun and note, linearizing the gain element for a fast release compressor may be an exercise in futility as envelope modulations will dominate. But in a guitar effect a little grunge is expected.

JR
 
[quote author="Wavebourn"]Smart dances around a 50/50 voltage divider from drain to gate? :cool:[/quote]

I prefer the implementations when you can pick off a buffered version of the channel voltage. In Leach's it is inverted but predictable from resistor ratios around opamp, likewise in non-inverting topology you can pick it off from the opamp output, then divide down by the feedback network ratio.

Directly coupling it in, requires cap coupling to prevent DC contamination of audio path and impedance's must be scaled pretty high to avoid errors due to source impedance. Also control voltage changes will leak into the audio path based on the impedance ratios used.

Perhaps worth considering in a stipped down circuit watching every penny, but not my inclination... although I'm not inclined to use FETS all that much. Note: I have used FET shunts in cheap noise gates and had to add linearization compensation because distortion was audible during very slow release rates (with test tones).

JR
 

Latest posts

Back
Top