2N3820 replacement

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

dogears

Well-known member
Joined
Nov 15, 2017
Messages
548
Location
Texas
The 2N3820 seems to have a pretty unique Id / Vgs curve. Does anyone know if there is a new production fet that would be close-to-equivalent for a FET compressor like the Compex?
 
> 2N3820 seems to have a pretty unique Id / Vgs curve.

Where do you see that??

The datasheet says it is a P type from process 89 (generic), has no curve, and such w-i-d-e specs that "anything goes".
 
Yes!! Thanks rackmonkey!

Forgive my slow learning, im a mech engineer pretending to be an EE. How would you go about determining a suitable fet replacement in a compressor?  The sidechain amp is going to be configured for the specific fet, right?
 
dogears said:
Yes!! Thanks rackmonkey!

Forgive my slow learning, im a mech engineer pretending to be an EE. How would you go about determining a suitable fet replacement in a compressor?  The sidechain amp is going to be configured for the specific fet, right?

JFET limiters are crude animals... there are distortion reducing tricks (adding 50% of AC drain voltage, cap coupled into gate drive). Vgs differences can be selected for (hard) or tweaked with component values or trim (easier).

See you are learning already.

JR
 
All that tracks, but isn’t the slope of that curve on the graph rackmonkey posted going to mean some fets are going to behave differently than others for a given gate voltage? If the curve isn’t parallel, your compression volts per dB reduction can’t be adjusted by bias alone right?
 
Which is why JFET limiters are far-far-far from the best you can do today.

You get a baggie all the same. Sort for same Vto so they change at the same point. Then usually you match Idss because this makes Ron similar; you can instead change the series resistor, but this tends to foul other parameters.

The 2N3820 looks like the same shape of a curve but very "bad" next to newer devices.
 
Yet another back in the day anecdote...

I wasted some bench time back in the 70s/80s trying to linearize the resistance of a JFET. I started with a dual JFET (something like a 2n3957/8 IIRC). I grounded the Source of the dual, and wrapped an op amp around one to make the JFET track a control current input using NF.  With the two gates also tied together the second JFET was now a grounded variable resistance tracking the first within the matching performance of a dual JFET.

It worked but was still inferior to even the crude VCAs available back then (for S/N and THD).

JR
 
I have a VCA compressor design. I like it, it works well I think. But having a log responding RMS sensor almost seems like cheating.

Trying to cut my teeth a bit, and learn something besides.
 
I just remembered the last (only) JFET limiter I ever designed.

Back in the 90's while managing mixer engineering inside Peavey I oversaw the transition between spring reverbs and low cost digital multi efx, for use inside my high volume powered mixer skus. To deliver better than horrible S/N they used copious amounts of HF pre-/de-emphasis in the digital path. Sadly in some settings even modest amounts of HF signal content would overload the digital registers (my speculation). They provided an overload LED driven by the digital processor, but it was not linearly related to input level, but HF content "and" algorithm selected, so all over the place in real world use.

In trade show recording demo's where I had to show the Peavey digital multi-efx in a good light, I was able to make them suck less by inserting a de-esser in front of the digital box's audio input...

Since we could not afford to put a full de-esser inside a cheap digital multi-efx card, I did the next best thing... I added a simple JFET shunt limiter in the front end of the efx processor audio, and used the overflow LED to turn on the JFET with appropriate fast attack, slow release.   

So for around $0.50 we made the digital efx all but impossible to overload (making a silk purse out of a sow's ear).

JR

PS; I didn't literally do the design myself (I was the engineering manager after all). I told the junior digital engineer in charge of the efx subassembly project what to do, and then blessed the result when it didn't suck. Not very heavy lifting. 
 
A different idea for a cheapo FET limiter, rather than the shunt type John used.  Was messing with this on a breadboard a few weeks ago. As suspected, probably not the best design for this sort of thing.

qCzrOj4l.jpg




 
I've been hunting some of 2N3820 and I got an offer for what seems to be a fair price. I would get al they got, Those are from Central Semi - I want to know if anyone else is interested - the price is $0.165/pcs
 
@rackmonkey's JFET comp design. Looks very similar to what Scholz R&D had in their Rockman. To this day I can't figure out why I bought one....................the more I listened to it , the worse it sounded. I don't think that was because of the compression circuit, more like the clipping.
 

Latest posts

Back
Top