update on clipping detection.

[thch 10+ year member]

i've been working on a clipping detector. since people are willing to buy $150+ oscopes after all!

i've got a circuit designed that uses a DMM and maybe $5 of parts from radioshack that seemed to do overall as well as the oscope for detecting the onset of clipping.

in the lab, with the labs DMM, it was able to detect clipping right before the visible onset*. if the test frequency was lowered, the detection was not as good and the results were detected at the visible onset of clipping. When coupled with an OScope the clipping* was detected well before the visible onset.

In anycase, it was shocking at how well the circuit worked. it may end up being the downfall though. it will indiscriminately detect any distortion or noise. if the amp has a lot of noise or distortion, it may be hard to use.

the final circuit is something i theorized about, a bandstop filter. basically play a tone, then build a circuit to block the tone. measure anything in the output that isn't the tone.

battery powered implementations should also be possible for people who don't want to buy a DMM. the cost of parts would probably be in the $20 range at radioshack.

*I used a "hard limiter" type circuit with LED's. the LEDs aren't are good as switching diodes, and as a result my clipping was a bit soft -- the wave would round off instead of fully clip. i had to crank it up past the rounding to get flat tops. but flat tops were easily determined.

Click to read more...

 

[Raven 10+ year member]

Awesome job. $20 in RadioShack parts for a real working clipping detector. I can see everyone who bought a $150 DMM within the last 2 weeks smacking themselves. Will you offer finished versions?

 

[squeak9798 5,000+ posts]

Cool beans man. I'm definitely interested.

Just curious....must this be used with an amplified signal (i.e. amplifier output), or could it also be used on things like preamp outputs?

 

[FoxPro5 5,000+ posts]

Schweeeet...wonder how this would compare to an amps on-board clipping dectector. Have you cross referenced it against anything else?

 

[thch 10+ year member]

Cool beans man. I'm definitely interested.

Just curious....must this be used with an amplified signal (i.e. amplifier output), or could it also be used on things like preamp outputs?

for the simple dector, its only up to the DMM used, maybe within some noise limits (the circuit i made is very noisy, measuing about 3mV AC at lowest.) for reference, high clipping measured about 600mV, but clipping was detected at 15mV. this is for a 20V peak signal.

basically, a sound wave is played which increases linearly. if the DMM's reading starts to increase quickly, then there is distortion.

but you should be able to use the amp to give some gain to the preamp signal so you can get something measurable on a DMM. but active filters can probably give higher resolution. the issue with a passive filter is the you can have no complex poles. thus my bandstop filter will block 60hz (by a factor exceeding 1000), but also blocks 180hz by a factor of 2. if I could boost 180hz by a factor of 100, the resolution of the circuit could improve. as it stands, it picks up very low levels of compression. the OScope coupled with the sensor picked up on the distortion well before it was visually apparent on a real OScope (not a handheld with limited screen resolution).

an on board clipping detector is more generic and can work on more signals. this only works for certain test tones. my test used 50 and 60hz tones. 50hz does not work as well as 60hz because the distortion component at 150hz is more attenuated then the distortion at 180hz. it is the 3rd harmonic that I am most interested in.

 

[thch 10+ year member]

As it turns out, i don't have a handheld OScope or oscope near a bay for testing. i was hopeing for some real world testing of the basic circuit. the circuit can be made from a "Twin T" circuit, which is all over google.

and you can cascade two of them in series to allow testing over more frequencies.

I will make a special test tone tonight and find someplace to store it. basically you need a test tone that LINEARLY increases in amplitude. either that or the ability to see logrithmic trends.

as the tone increases in volume, the DMM output increases linearly. as clipping occurs, the output should increase much faster.

in anycase, i'm wanting to get help testing the circuit out, like real testing in the feild. because it is so simple, i'll probably give out the schematic with values. I may end up selling them for maybe $5 or $7.50 if i can find suitable components. the componenets need high voltage ratings for high voltage output amps. i'd like to use 600V rated parts which might be required for the big JBL amps.

if you're interesting in this, i can give the schematic out with parts and you can build this and test it out.

 

[slim j 10+ year member]

How difficult will it be to build?

 

[PatFitz9 10+ year member]

i wouldn't mind trying to build it, i am an electronics dork //content.invisioncic.com/y282845/emoticons/smile.gif.1ebc41e1811405b213edfc4622c41e27.gif can you email it to me, along with a little idea on how to use it? LAXattk9@yahoo.com thanks.

 

[flakko 5,000+ posts]

^^^^ is going to steal your design for personal monetary gain //content.invisioncic.com/y282845/emoticons/wink.gif.608e3ea05f1a9f98611af0861652f8fb.gif

anyways im mos def interested. keep me updated //content.invisioncic.com/y282845/emoticons/smile.gif.1ebc41e1811405b213edfc4622c41e27.gif

 

[thch 10+ year member]

its very easy to build. as i said, look up "Twin T" on google.

http://www.radio-electronics.com/info/circuits/rc_notch_filter/twin_t_notch_filter.php

use 4 resistrors and 4 caps. to get R/2, place two resistors in parallel. to get 2C, place two caps in parallel. try something like 100Kohm resistors and 26nF caps. note that you'll have to change the test tones accordingly. note that you need some high voltage rated film or metalized film capacitor. it should be rated for line use!

on the downside, note that i've only tested this in my lab, and not one real amps. this circuit doens't seem to present a signifigant HF capacitance to the amp, but you may concider placing a 100ohm resistor in series with the input just in case. if the amp makes odd noises when you attach the circuit, you should remove the circuit.

you will still need linearly increaseing (amplitude) sine waves at various frequencies.

 

[AcidicDreams 10+ year member]

its very easy to build. as i said, look up "Twin T" on google.
http://www.radio-electronics.com/info/circuits/rc_notch_filter/twin_t_notch_filter.php

use 4 resistrors and 4 caps. to get R/2, place two resistors in parallel. to get 2C, place two caps in parallel. try something like 100Kohm resistors and 26nF caps. note that you'll have to change the test tones accordingly. note that you need some high voltage rated film or metalized film capacitor. it should be rated for line use!

on the downside, note that i've only tested this in my lab, and not one real amps. this circuit doens't seem to present a signifigant HF capacitance to the amp, but you may concider placing a 100ohm resistor in series with the input just in case. if the amp makes odd noises when you attach the circuit, you should remove the circuit.

you will still need linearly increaseing (amplitude) sine waves at various frequencies.

You should build 10 of them and sell them for $35-40 shipped... then I might end up with one //content.invisioncic.com/y282845/emoticons/smile.gif.1ebc41e1811405b213edfc4622c41e27.gif

 

[JimJ 5,000+ posts]

for the simple dector, its only up to the DMM used, maybe within some noise limits (the circuit i made is very noisy, measuing about 3mV AC at lowest.) for reference, high clipping measured about 600mV, but clipping was detected at 15mV. this is for a 20V peak signal.
basically, a sound wave is played which increases linearly. if the DMM's reading starts to increase quickly, then there is distortion.

but you should be able to use the amp to give some gain to the preamp signal so you can get something measurable on a DMM. but active filters can probably give higher resolution. the issue with a passive filter is the you can have no complex poles. thus my bandstop filter will block 60hz (by a factor exceeding 1000), but also blocks 180hz by a factor of 2. if I could boost 180hz by a factor of 100, the resolution of the circuit could improve. as it stands, it picks up very low levels of compression. the OScope coupled with the sensor picked up on the distortion well before it was visually apparent on a real OScope (not a handheld with limited screen resolution).

an on board clipping detector is more generic and can work on more signals. this only works for certain test tones. my test used 50 and 60hz tones. 50hz does not work as well as 60hz because the distortion component at 150hz is more attenuated then the distortion at 180hz. it is the 3rd harmonic that I am most interested in.

Maybe this is a complete n00bx question, but would this sort of detector work with amps that have suppressed odd harmonics?

 

[thylantyr 10+ year member]

Inspired by this post {the same post on CAF that I saw, but the forum is down},

I spent 5 minutes rethinking the peak detector and have a different approach.

How it works-

You connect the detector to the amplifier speaker terminals, 2 wires and if the

LED blinks, you hit your maximum voltage peak of the amplifier.

Peak Detector Calibration-

*The unit is calibrated per individual amplifier.

*You connect the detector to the amplifier speaker terminals, 2 wires.

*You over-drive the amp severely using a test tone where you assume

the amp is clipping.

*You adjust the calibration potentiometer so the peak detector LED triggers,

this sets your reference voltage for later use.

Circuit idea -

* Battery operated

* An IC comparator

* Potentiometer for reference voltage setting

* A zener diode at the reference pin to isolate battery voltage fluctuactions

* Optional - low battery voltage indicator {yellow LED}

 

[theCybe 5,000+ posts]

/ subscribed. twin-t schematic, no modifications? can something like this be inline with a real world load ie; voicecoil?

 

[thch 10+ year member]

some replies:

@AcidDreams, i actually had planned to build active versions of the detector and have pricing in the $30 area. but at this time i'm still working on a full wave, low noise voltage controlled amplifier. I know of one way to do a VCA, but i'm not sure it'll work. the problem is that the circuit must work over a wide range of inputs. from 2V to 200V (Preouts of HU to some of the high powered amps).

@JimJ, the amp cannot supress the odd harmonics, if it could then it would not clip. however the amp could clip asymetrically. this would be even harmonics, and would be picked up.

@thylantyr, the circuit uses the amps output terminals. the circuit you describe doesn't seem to actually detect clipping. It sound more like an automated DMM method.

@theCybe, the circuit is wired in shunt with the speaker. it will pick up distortion, so if the woofer can cause signifigant distortion from the amp, it will show. likewise the circuit should not be placed at the woofer end, as non-linear current flowing through the speaker wire will generate a signal and you will measure this distortion.