Excellent communication. Totally understandable. Thanks!!
Slight problem: The voltage tolerance is so tight that there's very little curve and a lot of flat. I like the adjustable idea, but we need "more roundness, less flat" You might want to build a clipping indicator to see that real music signal is so very dynamic, that your device will be giving peaks an audible flat top more often than expected.
Can you do this multi-step approach?
A mild (tiny current) soft clipper takes effect first, followed by. . .
A medium current soft clipper takes effect next, followed by. . .
A strong soft clipper, takes effect last, followed by. . .
A hard clipper with a non-audio voltage setting for surge overload protect.
*Only the 3 soft clippers are doing an audio processing task.
Mild current soft clipper is set for lower voltage (the amp has almost begun to clip),
medium current soft clipper is set in the middle (the amp has begun to clip),
strong current soft clipper is set for highest voltage (the amp is clipping).
the roundness is increased--the duration of flat top is decreased
The "VR1" (compressor for television) from AudioVox, Terk, does that. It generally works but often sounds disturbing when the volume suddenly changes. It is better than either fiercely loud commercials or unintelligible voice tracks on movies, but not attractive for music.
The ClipNipper (compressor for boosting small amplifiers) does that. It is unable to reduce bass impact. The frequency response stays level. It doesn't make the typical compressor "YAAAA!" sounding effect. Maybe it doesn't make that noise because the bass is as loud as the vocals? Thanks for the clues!
Clip the amp before it may do worse. Sounds just like the amp clipping normally (maybe sounds just like a tube amp clipping normally). Removes only a little bit of peak power without affecting the normal signal range of operation. Oh, we're back on topic--I think you've just named a soft clipper.
Adjustable like this? ZR431 Zener shunt regFor the moment, I'm on this simple idea: + 2 adjustable zener Diodes. The interest is that it has no action before the threshold, because just a resistance in the signal path.
Slight problem: The voltage tolerance is so tight that there's very little curve and a lot of flat. I like the adjustable idea, but we need "more roundness, less flat" You might want to build a clipping indicator to see that real music signal is so very dynamic, that your device will be giving peaks an audible flat top more often than expected.
Can you do this multi-step approach?
A mild (tiny current) soft clipper takes effect first, followed by. . .
A medium current soft clipper takes effect next, followed by. . .
A strong soft clipper, takes effect last, followed by. . .
A hard clipper with a non-audio voltage setting for surge overload protect.
*Only the 3 soft clippers are doing an audio processing task.
Mild current soft clipper is set for lower voltage (the amp has almost begun to clip),
medium current soft clipper is set in the middle (the amp has begun to clip),
strong current soft clipper is set for highest voltage (the amp is clipping).
the roundness is increased--the duration of flat top is decreased
More limiters than easter eggs. 
A few compressors for comparison. . .
Here's Rod Elliot's spin on a compressor Audio Amp Power Limiter
I think it interesting that Rod Elliot proposes handmade DIY optocouplers. It is also possible to use the 10mm led's file the top of the led flat, glue it to a CDS cell (with clear glue) and paint the device black. You can make a Rod Elliot limiter (compressor) a ClipNipper, a lightspeed attenuator (that you can automate) or even a Volumax for your border blaster.
My ClipNipper uses a similar approach to Rod Elliot's circuit, but I found that the current reliant LED were more likely to predict what needs doing for helping small size amplifiers.
It seems to be a matter of scale:
He has 100u delay and 10k input loss. I have 22u delay and 1k input loss. Neither offer fast peak protection--LDR and cap delays allow peaks to pass prior to LDR activation. Both offer protection against constantly high current. His compressor is more dramatic (more effective = less transparent). I think he designed for big amp (he mentions 500w) while I designed for small amp (10w to 25w).
And now adding the soft clipper. . .
Neither compressor can block x-max or peaks unless set to severely reduce the amplifier's power output. That might be okay if using Rod Elliot's 500w amplifier, but it is a fail with my T-amp. Instead of setting the compressor to reduce output power greatly, I would prefer to add the fast soft clipper to block x-max and block peaks as needed to catch what the compressor didn't do.
I guess that the soft clipper's zero delay (fast attack) combined with the compressor's pattern gets very close to doing the job thoroughly. And both can be set mildly for transparency (low disturbance). I believe that what we don't want to hear is the flat top of normal clipping or hard clipper. Soft clipper also has a flat top, just not sharp edges. By adding the compressor the trailing edge is restored to almost normal for music signal and the flat top is gone.
Unfortunately, there were optos to make or buy, and that is slightly inconvenient.
Therein is the motivation to do something differently.
The Bob Cordell Super Gain Clone with Klever Kilpper example is great because, I believe that NatSemi made the unfortunate mistake of assuming that music transient peaks of significant height, don't occur during normal playback. In addition to a clever topology, he also sets the Klever Klipper to take effect before the chip amp's disturbing Spike system activates. He put a Klever Kork into NatSemi's noise generator.
Not everyone use a Klever Klipper
Some of us want more simplistic circuit.
But, with a more simplistic circuit, there's a caveat in that effective is not transparent, but transparent is a lot less effective.
And the simple options. . .
For the normal amplifier (those without inbuilt clippers), the soft clipper if set aggressively, looks just like a hard clipper except for ineffectively rounded edges (sounds like Spike) or the soft clipper if set mildly looks like a rather ineffective headroom boost (imitation tube/valve clipping in addition to regular clipping). Instead of choosing between bad or bad, what I'd rather do is have load at the input dynamically adapt to the current requirement of the given signal.--Instead of a flat top haircut, I'd like greater resemblance to normal music signal.
Done without op-amps.
Done without optocouplers.
Done without complex sensor.
Done without cutting output power severely.
At the very least, I wish to have the simple soft clipper altered to reduce the duration of the flattened portion of the signal (via curving the edges more). Without actually building compressor circuits with op-amps and/or optos, the best I can think of at the moment is multi-stage soft clipper also employing at least one current reliant diode.
And simplified works, but there's a problem.
Even with the fast circuit at the bottom of post 1, I can't quite seem to avoid a bit of treble hash and unlevel frequency response. The bass gets cut first. The switching pollutes the treble. This needs a filter, but I haven't managed it yet. It seems to need a voltage activated addition that would engage a "tone control" as needed and concurrently with soft clipper activation. Maybe each stage that switches on can also engage a really mild RC enough to affect a partial compensation?
I'm in search of something simple, but that wasn't simple.
Thanks for the Bob Cordell references. I enjoyed reading those parts of his book. On reading some of his interviews and papers I guess that he puts the Baker Clamp into the soft clipper category along with Klever Klipper. But, what I'm actually sure of is that they're mentioned together.
A few compressors for comparison. . .
Here's Rod Elliot's spin on a compressor Audio Amp Power Limiter
I think it interesting that Rod Elliot proposes handmade DIY optocouplers. It is also possible to use the 10mm led's file the top of the led flat, glue it to a CDS cell (with clear glue) and paint the device black. You can make a Rod Elliot limiter (compressor) a ClipNipper, a lightspeed attenuator (that you can automate) or even a Volumax for your border blaster.
My ClipNipper uses a similar approach to Rod Elliot's circuit, but I found that the current reliant LED were more likely to predict what needs doing for helping small size amplifiers.
It seems to be a matter of scale:
He has 100u delay and 10k input loss. I have 22u delay and 1k input loss. Neither offer fast peak protection--LDR and cap delays allow peaks to pass prior to LDR activation. Both offer protection against constantly high current. His compressor is more dramatic (more effective = less transparent). I think he designed for big amp (he mentions 500w) while I designed for small amp (10w to 25w).
And now adding the soft clipper. . .
Neither compressor can block x-max or peaks unless set to severely reduce the amplifier's power output. That might be okay if using Rod Elliot's 500w amplifier, but it is a fail with my T-amp. Instead of setting the compressor to reduce output power greatly, I would prefer to add the fast soft clipper to block x-max and block peaks as needed to catch what the compressor didn't do.
I guess that the soft clipper's zero delay (fast attack) combined with the compressor's pattern gets very close to doing the job thoroughly. And both can be set mildly for transparency (low disturbance). I believe that what we don't want to hear is the flat top of normal clipping or hard clipper. Soft clipper also has a flat top, just not sharp edges. By adding the compressor the trailing edge is restored to almost normal for music signal and the flat top is gone.
Unfortunately, there were optos to make or buy, and that is slightly inconvenient.
In a broader definition, I see a few limiters in use: Transformer throttling, CRC resistor, some types of amplifier compensations, current dumper cable, flexy bridge rectifier (KBPC1610, Stealth), SMPS safety limiting, LM3875 LM3886 NatSemi Overture Spike System screeching harder limiter, TDA7293 TDA7294 soft limiter (decreases bass). Well, I think we all have limiters to some extent, but they might not be as attractive as possible.
Therein is the motivation to do something differently.
The Bob Cordell Super Gain Clone with Klever Kilpper example is great because, I believe that NatSemi made the unfortunate mistake of assuming that music transient peaks of significant height, don't occur during normal playback. In addition to a clever topology, he also sets the Klever Klipper to take effect before the chip amp's disturbing Spike system activates. He put a Klever Kork into NatSemi's noise generator.
Not everyone use a Klever Klipper
Some of us want more simplistic circuit.
But, with a more simplistic circuit, there's a caveat in that effective is not transparent, but transparent is a lot less effective.
And the simple options. . .
For the normal amplifier (those without inbuilt clippers), the soft clipper if set aggressively, looks just like a hard clipper except for ineffectively rounded edges (sounds like Spike) or the soft clipper if set mildly looks like a rather ineffective headroom boost (imitation tube/valve clipping in addition to regular clipping). Instead of choosing between bad or bad, what I'd rather do is have load at the input dynamically adapt to the current requirement of the given signal.--Instead of a flat top haircut, I'd like greater resemblance to normal music signal.
Done without op-amps.
Done without optocouplers.
Done without complex sensor.
Done without cutting output power severely.
At the very least, I wish to have the simple soft clipper altered to reduce the duration of the flattened portion of the signal (via curving the edges more). Without actually building compressor circuits with op-amps and/or optos, the best I can think of at the moment is multi-stage soft clipper also employing at least one current reliant diode.
And simplified works, but there's a problem.
Even with the fast circuit at the bottom of post 1, I can't quite seem to avoid a bit of treble hash and unlevel frequency response. The bass gets cut first. The switching pollutes the treble. This needs a filter, but I haven't managed it yet. It seems to need a voltage activated addition that would engage a "tone control" as needed and concurrently with soft clipper activation. Maybe each stage that switches on can also engage a really mild RC enough to affect a partial compensation?
I'm in search of something simple, but that wasn't simple.
Last edited:
You got-it.
Can-you precise your goal, danielwritesbac ?
You do not allow on a high end system that clipping never occurs. A simple clipping indicator would be sufficient to warn-you that you have to reduce your level for no distortion, don't you think ?
Of course, this implies that your speakers are efficient enough for high SPL, and your amp powerful enough.
Rare clippings, only on the top of half a period on the attack of an instrument, are not really a problem. It will gives-you the feeling of more "punch", and your brain will not have time to analyses the distortion as so. I doubt that softening of the clipping would change the landscape in a drastic way. If you use commercial recordings, (CDs) the limitation of those rare peaks are already done during mastering in a transparent way (digital limiters).
So, did you want a soft clipping for frequent overdrive of your amp ? (no fidelity expected )
Means less dynamic and more average level ?
A compressor will give-you more agreeable results, i think. That you can bypass when you want a linear reproduction.
PS: About adding attack on an instrument with a compressor, that can be done only with a multi track recording, where all the instruments are separated, so you can apply this to one single instrument alone.
Last edited:
You got it. My amp isn't powerful enough. It isn't 768 watts, and I don't have rare high efficiency speakers either.
I disagree, partially. Clipping indicators, from Rod Elliot and others, show that clipping is very, very frequent. I agree that it is short duration.
If this meets a typical aggressive soft clipper, it gets worse distortion, and I don't want that.
Decreasing the distortion for a soft clipper means decreasing the current of the soft clipper (decreased effectiveness). That's good! Then it has a mild, low distortion, very gentle, effect on the signal. That's not very helpful, but it is a lot less harmful.
I think "a lot less harmful" is the useful place to start.
After that point, there's a game of increasing effectiveness without increasing distortion. For that, I propose to set an additional mild, low distortion, very gentle soft clipper at slightly higher signal voltage. The signal has again been curved gently--yes it has now been curved gently twice.
EDIT:
I'd like to see really gentle times three different stages, and all low distortion and adjustable.
A soft clipper means you'll loose more linear power than with a "hard" one. Because if it as action in the last 3Dbs (half of power), the last half of your power will make distortion to the signal, whatever this signal would have been clipped or not.
And if it has less margin action, i believe the difference will not be obvious.
I don't know if i make this clear... my poor English...
But you didn't had answering my question about your context. High end system ?
What is your amp ?
Why are-you limited with power of your amp ? Can't you use a more powerful amp ?
And if it has less margin action, i believe the difference will not be obvious.
I don't know if i make this clear... my poor English...
But you didn't had answering my question about your context. High end system ?
What is your amp ?
Why are-you limited with power of your amp ? Can't you use a more powerful amp ?
The amp is whatever I happen to be making at the moment.
A non-defective soft clipper will give you a headroom boost.
Perhaps the topic is not clear until after building a few options for comparison?
No !
if the Vpk is limited by the Rail supply and the losses through the output stage, then implementing a soft clip circuit will reduce the headroom before the distortion sets in, i.e the clip starts earlier than if the soft clip circuit were not fitted.
The current setting is overly aggressive if soft clipper is causing brick wall errors like that.
Peak height and duration are individually adjustable.
A "mild" soft clipper is for pinching the top of square wave to decrease duration of clipping peaks, without cutting the height of the peak.
It is a compressor's poor cousin.
Decreasing the duration of the peak looks good for output device SOAR.
Decreasing the duration of the peak allows faster power supply recharge.
Long term test almost done:
Since Thursday, May 03, 2012, according to the date stamp locked into the forum attachment, the ltp soft clipper has been in active service protecting real amplifiers without reducing useful headroom. It will have 6 months active service by November 3, 2012.
P.S.
Thank you for the helpful post that illustrates the typical soft clipper problem; however, there are many choices and I'm focusing on uncommon soft clippers that can't brick wall within the useful range of the amplifier. We might agree that the amplifier's normal clipping pattern is usually better than adding a problem.
You have so many, many choices for soft clipper,
And their qualities vary. Mostly awful. Few good.
Last edited:
Soft Clipper in common language is a device that ABSOLUTELY MUST: Provide the more favorable clipping pattern of a tube amplifier along with the implied headroom increase. Anything that fails the expectation is necessarily considered poor quality and possibly something to avoid.
Is there, perhaps some alternative and highly malicious definition specific to engineer-speak? Is the term "soft clipper" used as an "inside joke" to catch the unwary? Why is everyone promoting slightly softened circuits overly similar to hard clipper behavior? Does talk of soft clipper automatically incur hazing? What's going on? And, where is the quality control? Did I accidentally request something unexpectedly bad?
Is there, perhaps some alternative and highly malicious definition specific to engineer-speak? Is the term "soft clipper" used as an "inside joke" to catch the unwary? Why is everyone promoting slightly softened circuits overly similar to hard clipper behavior? Does talk of soft clipper automatically incur hazing? What's going on? And, where is the quality control? Did I accidentally request something unexpectedly bad?
Last edited:
That is where lie our reactions: we think clipping has never to occur in a hifi system.
When highest quality 400W amps (like hypex N400) are available and affordable, (and some other, a little less good for 1/10 of its price), we tend to look, instead, at this kind of solution.
Anyway, we are curious to heard from your experiments.
But, once again, if for some reason, i had to respond at that kind of technical requirement (increase the average level with not excessive distortion), i will look limiters side.
Last edited:
Right on!! Thank you very much!!! I have just now realized my error and have finally understood Sreten's warning.
Thank you for this warning.
I have finally understood that the layman's terms of tube/valve amp soft clipping are not done with a typical soft clipper. The layman's language and engineer's language are almost opposites. And I have finally understood that typical soft clippers may be quiet protectors but aren't desirable audio effects. I will again attempt something better for the thread's intro.
Thank you for this post!
It is a light shining in the darkness.
Apparently Bob's re-spelling of Klipper is useful for describing a non-typical clipper (I'm surprised to discover that a clipper that benefits audio is not typical), and the re-spelling has some humor content that I wasn't aware of until now. Now, that name "Klever Klipper" would have been quite funny if I'd learned the meaning of it a bit earlier.
Thanks again for the excellent point of reference. I am quite grateful. Thank you for the headroom information too. That was an excellent clue.
The point at which I'd like to cause clipping is 4v input to a power amp--a non-audio signal. And, I wish to gently squeeze (compress?) but not clip audio signals.
Tube amps use very high tension rails. Audio signal peak voltages are far from them.
The distortion increases slowly and with a regular curve as the level increase. In fact there is no clipping.
Same thing with an analog magnetic tape.
An analog mixing desk has 24db of headroom, a professional digital recorder 15 or 18db.That is to protect the signal from any clipping during recording and mixing processes.
But, during mastering, all undesirable peaks are limited carefully, with less sound degradation as possible. Those peaks you wanted to get rid off by a soft clipper.
After that, the result of a soft clipper will be: more average distortion at high levels, less disagreeable than natural clipping. Chose your poison.
The distortion increases slowly and with a regular curve as the level increase. In fact there is no clipping.
Same thing with an analog magnetic tape.
An analog mixing desk has 24db of headroom, a professional digital recorder 15 or 18db.That is to protect the signal from any clipping during recording and mixing processes.
But, during mastering, all undesirable peaks are limited carefully, with less sound degradation as possible. Those peaks you wanted to get rid off by a soft clipper.
After that, the result of a soft clipper will be: more average distortion at high levels, less disagreeable than natural clipping. Chose your poison.
Need to find a little diode.
This is going much better. Thanks!
Much like this:
(a) An ineffective soft clipper followed by,
(b) an ineffective soft clipper followed by,
(c) a regular soft clipper.
Parts problem, capacitance and voltage:
At line level, it may be possible use 1n5819 0.15v rectifier schottky but that could be coarse. Where can we find the low capacitance 0.1v current reliant diode? Does 1n5819 have a baby sister (smaller, lower capacitance)?
.
This is going much better. Thanks!
Less disagreeable than natural clipping would be a non-standard soft clipper circuit or a standard compressor. We can do both: compress it before hit it. And for the purpose of this thread, that's what I'm looking for.
Much like this:
(a) An ineffective soft clipper followed by,
(b) an ineffective soft clipper followed by,
(c) a regular soft clipper.
Code:
(c) -
(b) / \
(a) / \
(a) \ /
(b) \ /
(c) -At line level, it may be possible use 1n5819 0.15v rectifier schottky but that could be coarse. Where can we find the low capacitance 0.1v current reliant diode? Does 1n5819 have a baby sister (smaller, lower capacitance)?
.
Cordell's "Klever Klipper" works like a charm. Properly adjusted (tracking rails etc), it kicks in just below output stage clipping to exactly avoid that clipping which tends to be "sticky" (amp takes quite some time to recover from clipping) and sounds much more annoying than a briefly compressed, then "chopped off" peak but still properly amplified.
Key point is that the amp never ever opens its feedback loop -- with the exception of current overload "clipping" which is a different animal, much harder to tame.
One can adjust the "softness" of the clipper by letting it work at lower (large transition) or higher signal amplitudes.
Soundwise, any soft-clipper will steal some volts of undistorted headroom but I found Cordell's pretty well sounding for clipped LF (<500Hz) signals, sort of like the speaker itself is running out of steam, not the amplifier, unless you punish it with gross input overload. With occasional short HF (>2kHz) overload spikes the clipping was effectivly undetectable to my ears.
Key point is that the amp never ever opens its feedback loop -- with the exception of current overload "clipping" which is a different animal, much harder to tame.
One can adjust the "softness" of the clipper by letting it work at lower (large transition) or higher signal amplitudes.
Soundwise, any soft-clipper will steal some volts of undistorted headroom but I found Cordell's pretty well sounding for clipped LF (<500Hz) signals, sort of like the speaker itself is running out of steam, not the amplifier, unless you punish it with gross input overload. With occasional short HF (>2kHz) overload spikes the clipping was effectivly undetectable to my ears.
Thank you sir!!! Awesome!!!
I'm trying to figure out the circuit locations of all of the audio effect.
Comparison:
A slow attack compressor (clipnipper in my signature line) can't stop bass impact; however, a fast soft clip circuit does remove biggest signal first, as quickly as possible. They can be used together if level frequency response is desired.
Either changes power supply workload.
Speculation guesswork at the power circuit effect:
The slow attack compressor decreases the duration of workload and allows the power supply reservoir to recharge faster so the amplifier's time spent defenseless is reduced. The fast soft clipper prevents peaks from discharging the power supply, and gives one the chance of buying a transformer big enough to prevent discharge so that amplifier is never defenseless (true if there's umbilical cable or diode or capmulti, operating the amplifier board somewhat lower voltage than power board). Lastly, the possible combination of compressor and soft clipper first reduces workload duration and then also blocks peak discharge.
Very Klever resource management!! But some of what we hear is surely power circuit effect.
Assumptions:
By putting the sensor at amplifier input, and by reducing power supply workload duration and peaks, maybe (maybe) usable headroom for music is not reduced. A current reliant diode sensor might make a circuit that guesses reasonably, in which case there's no need to connect power noise to amplifier input. Details: Zener, Bat, and 1n4148 are useful for voltage precision; however, the curve of current reliant 1n5819, MR and LED are more attractive for compressor effect.
I'd like to try for a no op-amp, no opto, simplified solution at line level.
Symmetric versus Dual Asymmetric:
And we can do "forced symmetry" which is good for a little boost. While researching the Volumax type compressors, a radio station transmitter headroom boost (also works on audio power amp), I also read that the major flaw was the necessity of selecting announcers and DJ's with more symmetrical voices, but DJ's with asymmetrical voices were fired or paid much less (because too quiet). Later on, the asymmetrical voice guys got their jobs back when forced symmetry compressor circuits (Dual Asymmetric, one for upswing another for downswing) were added at the mic preamps. After that, it was quickly discovered that normal music is an asymmetrical signal.
But someone forgot. . .
On checking out the waveforms of quite a few tracks with the computer wav editor, it shows that most of the tracks have asymmetric peaks.
Given the typical recording studio output (containing asymmetric clips), I propose to "upgrade" our project to dual-asymmetric sensor/detector, so that it is unnecessary to harm both sides of the waveform every time. The upgrade is one detector for upswing and a second detector for downswing. In my observation, it is unlikely that both detectors will activate at precisely the same time during normal music signal playback. If the circuit is fast enough, dual asymmetric looks doable.
Question:
Is Bob Cordell's compress+soft-clip circuit a Symmetric or Dual Asymmetric?
Last edited:
Much closer
The BAT85, BAT86 has low capacitance for low noise. On researching headphone amp clippers, the 1N34A and LED have mainly 2nd order harmonic like a tube clip.
For experiment, I used:
1k5 series loss at input to help form the voltage divider.
10k-BAT86clipper-680n as the main shunt.
10k-BAT86clipper-BAT86clipper added shunt.
This successfully constrained the TV to reasonable output.
I'm looking forward to trying out some 1N34A germanium.
The BAT85, BAT86 has low capacitance for low noise. On researching headphone amp clippers, the 1N34A and LED have mainly 2nd order harmonic like a tube clip.
For experiment, I used:
1k5 series loss at input to help form the voltage divider.
10k-BAT86clipper-680n as the main shunt.
10k-BAT86clipper-BAT86clipper added shunt.
This successfully constrained the TV to reasonable output.
I'm looking forward to trying out some 1N34A germanium.
Attachments
Last edited:
I am entering this thread hoping I find a good solution to avoid power burnout due to "sudden" input overload.
I have built a very clever power amp that is DC coupled but seems very sensitive to input transient bursts.... I burned one channel just by unplugging input rca during operation.
Need some input clamp that does not affect signal in any way.
here you have the schematic ... fell free to launch any ideas that will be soon built in. Results will be reported.
I have built a very clever power amp that is DC coupled but seems very sensitive to input transient bursts.... I burned one channel just by unplugging input rca during operation.
Need some input clamp that does not affect signal in any way.
here you have the schematic ... fell free to launch any ideas that will be soon built in. Results will be reported.
Attachments
Last edited:
Clipping input:
At location "C4" put a series pair of FR diodes "facing up" and another series pair of FR "facing down" so there's 4 diodes total. A series pair of FR = 1v (and you'll need two series pairs--one pair is backwards polarity). OR, series pair of 1n400x is 1.3v.
Measure and adjust the value of C4 smaller so that the total of parasitic (diode body) and normal (c4) capacitance of that location is the same 1n as previously (or similar).
You will then get normal operation when the input is 1v or less, or some distortion if the input is more than 1v. Add an additional FR for upswing and an additional FR for downswing, if you want 1.5v (which will also have lower capacitance--diodes in series is like little capacitors in series, less capacitance that way).
If the FR diode body capacitance is still too high, you can switch to bat86 diodes instead (0.3v each), but signal schottky types are more frail and thus the series resistance, R6, may need to be greater to help protect them.
You can probably use up to 330R at R6 without affecting the audio. In that case, 2 pair 1n4148 clippers (0.7v+0.7v=1.4v clipper parallel with C4) is an easy option.
Usage, for example, computer music source, when the computer switches on or awakes from standby, the ranchy blast from its output, sometimes 12v, will be cut off to 1v by your input hard clipper.
*alternative to the input hard clipper protection is an active circuit, such as Cordell's Klever Clipper and many many other options. But none are as simple as antiparallel diodes with the voltage threshold set higher than desirable signal. These don't work very well alone, but can work rather well as the shunt leg of a voltage divider (your R6 is the other piece).
Clipping amplification:
But if you had needed to clip the amplification, then you'd need a much different circuit, such as an ltp soft clipper.
For this, change R22 to a 2k multi-turn trimmer in parallel with an 18k resistor, resulting in the original value of 1.8k. Next attach series two antiparallel pairs of BAT86 (0.3v + 0.3v = 0.6v clipper with very low capacitance) to the wiper of the 2k trimmer, and next, a 1k multi-turn trimmer (as a variable resistor) goes between the BAT86 clipper and the positive input. The 2k trimmer is clip voltage, the 1k trimmer is clip current. This describes an LTP soft clipper. Startup positions of the trimmers is half (wiper at centerpoint).
Usage, for example, a surge high enough to cause the outputs to stick (or severe hfe droop, or other non-linear range), has INSTEAD, clipped the inputs, which can recover instantly and rather easily (there's a big difference in the current of the clip--tiny at inputs or huge at outputs).
*I have not used the ltp soft clipper on a dc/mixed coupled amplifier; however, if there's trouble, it is possible to put a capacitor series to the 1k trimmer so that the clipper won't switch on for DC. If you undersize the cap, it will ignore some of the lowest bass pitches and appear to function as a rudimentary form of compressor instead of the intended function.
**The LTP soft clipper circuit will relocate your imaging linearity sweet spot to take effect during somewhat louder volume playback than previously, just as the LTP soft clipper has begun to catch a few peaks infrequently, is just about when the imaging will kick up to bigger and the tone will foobar slightly in identical proportion to the imaging boost. To swamp a bit more of that, you may want to double up your power decouplers, C5, C8, 100u||100u (200u, an orthodox value anyway).
***Alternative to the ltp soft clipper circuit is: baker clamps can be used instead.
I did use both the input clipper and the LTP clipper on my circlophone. Those two clippers do different tasks.
P.S.
Also suggest changing C3 to approximately 2uF, although 1uF may do fine or better; however, the currently 50uF input coupler could have too much discharge force, like a tiny tazer.
P.P.S.
I suppose there's some 100u in your parts box. Quadruple parallel 100u (400u low loss) would not sound bad if put in series between R22 and ground (400u as a coupler). It would be a worthy experiment to see if ac coupling will yield a more sturdy amplifier with louder and more lifelike dynamics (as expected).
At location "C4" put a series pair of FR diodes "facing up" and another series pair of FR "facing down" so there's 4 diodes total. A series pair of FR = 1v (and you'll need two series pairs--one pair is backwards polarity). OR, series pair of 1n400x is 1.3v.
Measure and adjust the value of C4 smaller so that the total of parasitic (diode body) and normal (c4) capacitance of that location is the same 1n as previously (or similar).
You will then get normal operation when the input is 1v or less, or some distortion if the input is more than 1v. Add an additional FR for upswing and an additional FR for downswing, if you want 1.5v (which will also have lower capacitance--diodes in series is like little capacitors in series, less capacitance that way).
If the FR diode body capacitance is still too high, you can switch to bat86 diodes instead (0.3v each), but signal schottky types are more frail and thus the series resistance, R6, may need to be greater to help protect them.
You can probably use up to 330R at R6 without affecting the audio. In that case, 2 pair 1n4148 clippers (0.7v+0.7v=1.4v clipper parallel with C4) is an easy option.
Usage, for example, computer music source, when the computer switches on or awakes from standby, the ranchy blast from its output, sometimes 12v, will be cut off to 1v by your input hard clipper.
*alternative to the input hard clipper protection is an active circuit, such as Cordell's Klever Clipper and many many other options. But none are as simple as antiparallel diodes with the voltage threshold set higher than desirable signal. These don't work very well alone, but can work rather well as the shunt leg of a voltage divider (your R6 is the other piece).
Clipping amplification:
But if you had needed to clip the amplification, then you'd need a much different circuit, such as an ltp soft clipper.
For this, change R22 to a 2k multi-turn trimmer in parallel with an 18k resistor, resulting in the original value of 1.8k. Next attach series two antiparallel pairs of BAT86 (0.3v + 0.3v = 0.6v clipper with very low capacitance) to the wiper of the 2k trimmer, and next, a 1k multi-turn trimmer (as a variable resistor) goes between the BAT86 clipper and the positive input. The 2k trimmer is clip voltage, the 1k trimmer is clip current. This describes an LTP soft clipper. Startup positions of the trimmers is half (wiper at centerpoint).
Usage, for example, a surge high enough to cause the outputs to stick (or severe hfe droop, or other non-linear range), has INSTEAD, clipped the inputs, which can recover instantly and rather easily (there's a big difference in the current of the clip--tiny at inputs or huge at outputs).
*I have not used the ltp soft clipper on a dc/mixed coupled amplifier; however, if there's trouble, it is possible to put a capacitor series to the 1k trimmer so that the clipper won't switch on for DC. If you undersize the cap, it will ignore some of the lowest bass pitches and appear to function as a rudimentary form of compressor instead of the intended function.
**The LTP soft clipper circuit will relocate your imaging linearity sweet spot to take effect during somewhat louder volume playback than previously, just as the LTP soft clipper has begun to catch a few peaks infrequently, is just about when the imaging will kick up to bigger and the tone will foobar slightly in identical proportion to the imaging boost. To swamp a bit more of that, you may want to double up your power decouplers, C5, C8, 100u||100u (200u, an orthodox value anyway).
***Alternative to the ltp soft clipper circuit is: baker clamps can be used instead.
I did use both the input clipper and the LTP clipper on my circlophone. Those two clippers do different tasks.
P.S.
Also suggest changing C3 to approximately 2uF, although 1uF may do fine or better; however, the currently 50uF input coupler could have too much discharge force, like a tiny tazer.
P.P.S.
I suppose there's some 100u in your parts box. Quadruple parallel 100u (400u low loss) would not sound bad if put in series between R22 and ground (400u as a coupler). It would be a worthy experiment to see if ac coupling will yield a more sturdy amplifier with louder and more lifelike dynamics (as expected).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.