LDR Limiter for TA2020, TA2024, spirited playback without clipping.

[Eva]

To get the loudest "transparent" limiting, the limiting threshold and gain reduction slope have to be dynamically adjusted depending on the amount of overdrive, how long the clipping events last and how frequent they are. The limiter has to learn about input signal. This imposes the need for a digital computer, or a thousand analog parts. Analog computers are a big no-no for mid/low-bandwidth mid/low-accuracy applications. That's the ideal application for a micro-controller. Each resistor, capacitor, transistor, diode is replaced by a few instructions, it's much cleaner in every aspect.

[nanonymous]

Originally Posted by Eva

To get the loudest "transparent" limiting, the limiting threshold and gain reduction slope have to be dynamically adjusted depending on the amount of overdrive, how long the clipping events last and how frequent they are. The limiter has to learn about input signal. This imposes the need for a digital computer, or a thousand analog parts. Analog computers are a big no-no for mid/low-bandwidth mid/low-accuracy applications. That's the ideal application for a micro-controller. Each resistor, capacitor, transistor, diode is replaced by a few instructions, it's much cleaner in every aspect.

Sounds interesting!
I'd love to see a design that can compete with the analog one in this thread

[danielwritesbac]

Oh, what have I done? T-amp loudness wars? lolz! This is so much like lawnmower races.

Digital could do the psycho-acoustic bass harmonic thing set to be engaged right on time when the amp is all run out of other resources to replay bass fundamental (aka fool the ear by playing the bass harmonics). That would sound as if louder cleaner bass. I don't quite know how to do that in analog.

[keantoken]

Here. I took me a while to track this one down...

HeadWize - Project: The Psychoacoustic Bass Enhancer by Jan Meier

[danielwritesbac]

Thanks for finding that!
Digital Ultrabass sidesteps the patent due to "a significant difference," but analogue does not and could not be sold for profit. The large size circuit, which doesn't come equipped with sensor and controller, is only one out of three analog circuits (limiter, harmonic bass, soft clipper) needed to supplant a small digital solution, with a huge analog solution. The size of the circuit seems to prove Eva's point. . . except that any size analog solution is still tunable with ordinary tools.

I'm actually still using the clipnipper limiter in post 1 because it works and sounds good. With that easy simple little limiter, the TA2020 does a great job on rockin the old RV on solar power. . . without clipping. The reason I didn't use Sure's much louder TA2024 is because the alternator voltage is 14.8v.

P.S.
Oh, terrible and good thought--the car radio also needs a clipnipper!

[danielwritesbac]

From sx881663 posted here (link).

If it can operate an LED, it can also operate a 4n25 or LDR, right?

[stang1125]

Originally Posted by danielwritesbac

Here's my ClipNipper limiter adjusted for TA2020, TA2024 with ~13vdc operating voltage. It is adjustable. I've got it running right now on the clipping prone TA2020. . . that's playing very loud without clipping. ClipNipper is a simple passive limiter that doesn't consume your battery power and it doesn't reduce useful output power at all, but what you get is a boost.

This schematic is currently up to date.

(log in to see schematic)

Notes ----------------------------------------------------------------------------------------------
Input:
As shown in the schematic above, the location of this fixed divider and its function is just like an ordinary 20k volume control turned almost all the way up. . . except for momentary LDR activations that prevent clipping.

Series resistance at input is within the range of 330 ohms to 1K whereby the larger value allows more volume fluctuation--set this to personal taste. I like 820R with the limiter version (schematic above) or 330R to 470R instead of 1k with the soft clip version (shown at post 12). Some optional gain management documentation is at post 17.

Substitutions:
3.1v Led: The 20ma+ 3.1v LED can be a white, Warm White, Green, Blue, Pink, or Purple LED.
2v Led: The 20ma+ 2v led in red, yellow, amber, orange can be used as four leds series plus two ordinary diodes series like (4*2v=8v)+(2*0.7v=1.4v)=9.4v total forward voltage drop, but I didn't test that--see settings/tuning below.
Schottky: You can use almost any Schottky or fast diode you happen to have instead of the Schottky diodes in my schematics.
Rectifier: The bridge rectifier doesn't have to be made of 1n4007. A miniature prefabricated bridge rectifier would work nicely too--the ~~ AC marks are for the speaker jack, but the +- marks run the ClipNipper limiter's LED's.

Settings/tuning:
It takes a bit of fine tuning, so have extra schottky (or fast silicon diodes) handy for that. If you need to trim a larger step, an amber or red LED is about 2v. More or less diode voltage drops in series to the LDR (led section) is how you set it. You'll have to try a variety of tracks before arriving at the setting that works best on average for your amplifier.

DIY CDS Optocoupler:
If there's not an LDR optocoupler handy, you can make one so easily. Just file the top of a white LED flat, then glue it to a CDS cell with clear glue, tape over the wires and paint that thing black. After it dries and after tape removal, you've got a DIY LDR optocoupler.

Battery management parts:
As a caveat of of its Zero standby drain design, ClipNipper will not reliably retard clipping for battery powered units with run down batteries unless the amp is run from a boost (or buck boost) type regulator. Eva's post #13 is a great idea for battery powered amp, but I couldn't figure out how to make that type of sensor work for bridge amp.

Larger scale amplifier applications:
Actually 3 LED's isn't enough for a larger scale amplifier, so for using ClipNipper to scrape the screech off of an LM3886 or even give the clean LM1875's a boost, you could start by simply adding more LED's and possibly a small change the drainer resistor value as needed for appropriate LED current. If one needs greater abruptness, one could use combination of zener and led instead of all led. Not all amplifiers/applications are equally dynamic. To track good dynamics without squashing, use more led, but to track lesser dynamics plus firmer control, employ some abrupt zener. Balance this out to track with your application and simply aim for pretty (pretty is better than clipping). The ClipNipper is interoperable with other limiters and soft clippers as well.

[Elvee]

I tested the Clipnipper on a Tringlophone with the components kindly provided by Daniel.

Results
In mode "A" only (no comparison), the thing seems to work transparently, with no apparent distortion, pumping, or any other noticeable artefact.

However, in A/B mode (with/without), the sound is clearly duller with, in particular the attacks, that it seems to kill, or at least to blunt seriously (test was done with rock music).

After some trial and errors, I ended up with the version shown below: the capacitor across the coupler has been increased to 100µ, and a discharge resistor has been added to somewhat help the recovery and delay the onset of AGC action.
To compensate for the threshold, one of the LEDs has been shorted

On a purely technical level, I noticed that the I/V function of these couplers is non-linear; even the residual dark resistance is "S" shaped, which must introduce distortion. However, my tin ears failed to detect any distortion (not a surprise).

[danielwritesbac]

Thanks for the update! Much appreciated!

I'm sorry that I forgot to mention that the original works with level tone only if you allow some of the amplifier's natural clipping pattern. However, the TV amp that I'm working on right now (for my mother) needs the AGC version you just posted. Thanks again!!

P.S.
The shorted D4. . . it might be interesting to replace that short with a small rectifier schottky(s) or a fast silicon diode(s) or 1n4007(s). Possibly that could add some more pretty/transparent flex usable for a touch more headroom. Any diodes with highly current reliant knee voltage can be used for trim--they don't actually have to be LED's.

P.P.S.
It could be interesting to put a 1k variable resistor in series with the light dependent resistor, so that the resistance can be fine tuned and "relaxed" as needed to preserve dynamics. LDR's vary a lot. Maybe the LDR that I sent was overly effective.

[Elvee]

Originally Posted by danielwritesbac

P.S.
The shorted D4. . . it might be interesting to replace that short with a small rectifier schottky(s) or a fast silicon diode(s) or 1n4007(s). Possibly that could add some more pretty/transparent flex usable for a touch more headroom. Any diodes with highly current reliant knee voltage can be used for trim--they don't actually have to be LED's.

P.P.S.
It could be interesting to put a 1k variable resistor in series with the light dependent resistor, so that the resistance can be fine tuned and "relaxed" as needed to preserve dynamics. LDR's vary a lot. Maybe the LDR that I sent was overly effective.

That's completely up to the user's taste, the power of the amplifier, how nasty it sounds when clipping, etc.
Practically all parameters can be made tweakable, attack, decay, levels, etc.
It is impossible to provide a universal setting that would suit everyone, every type of music, equipment, etc.

The best option is to start with the basic parameters, and correct them to taste until the result is satisfactory