I did a simulation once of the Zeta 2 amplifier made by Lars back in 2000's and the result was really weird.No matter the level the cup of the sine signal was perfectly round circle.As I built it in 2003 I knew its sound was not compressed or distorted but the simulations showed that perfectly soft circle rounded signal instead of a sine wave. I still have a feeling that amp does something special that no other class ab did...and it's folded cascode , no clipper anywhere...Still wonder why Lars never wanted to present it his diyaudio stream.He had quite an audience at the time.
One can modify to watch the amplifiers output as described earlier making the amplifier a sort of power compressor
Just how smart is chat gpt?
In analog audio processing, soft clipping can be achieved by using a circuit that includes a non-linear device such as a diode or a triode tube. One common method is to use a diode clipping circuit, which consists of a diode placed in series with the audio signal. As the amplitude of the signal increases, the diode begins to conduct and the excess signal is "clipped" off, creating a soft clipping effect. Another method is to use a triode tube clipping circuit, which uses a vacuum tube to add harmonic distortion to the signal as it approaches the maximum level. It's worth noting that these analog circuits can be relatively complex and may require expertise in circuit design and construction.
Most professional audio amplifiers feature built-in compressors that are activated when the output signal approaches the clipping point. These compressors typically reference the difference between the supply voltage rails, V_supply and the output signal, V_out, and trigger when the output voltage (V_out) approaches the rail voltage (V_supply) by comparing the relationship between V_supply and V_out with a threshold value (V_threshold) using a mathematical function such as:
if |V_out| >= V_supply - V_threshold then activate the compressor
This mechanism is similar to clipping circuit LEDs, but it also includes an input signal trimming function, which can be achieved by either cutting the input signal or reducing the gain of an input buffer stage using mathematical functions such as:
V_out = V_in * Gain
Where Gain is a parameter that can be adjusted to reduce the output voltage.
Additionally, compressors can also monitor the input signal in relation to a set input voltage reference, V_ref. This approach allows the compressor to operate independently of the back emf control mechanism, which prevents clipping of the negative feedback signal by comparing the relationship between V_in and V_ref with a threshold value (V_threshold) using a mathematical function such as:
if |V_in| >= V_ref + V_threshold then activate the compressor
Amplifiers that employ separate voltage and current gain stages must also include protection diodes between the voltage gain stage and the current gain stage's power supply to avoid hard saturation of the output stage, by allowing current to flow only in one direction.
When executed properly, this technique can result in a "fatter" sound when the amplifier is driven to its limits. However, if not properly implemented, attempting to overdrive the amplifier can result in undesirable distortion.
In summary, the use of built-in compressors and protection diodes are essential for preventing clipping and distortion in professional audio amplifiers, and when implemented correctly, can result in a more pleasing and dynamic sound.
In analog audio processing, soft clipping can be achieved by using a circuit that includes a non-linear device such as a diode or a triode tube. One common method is to use a diode clipping circuit, which consists of a diode placed in series with the audio signal. As the amplitude of the signal increases, the diode begins to conduct and the excess signal is "clipped" off, creating a soft clipping effect. Another method is to use a triode tube clipping circuit, which uses a vacuum tube to add harmonic distortion to the signal as it approaches the maximum level. It's worth noting that these analog circuits can be relatively complex and may require expertise in circuit design and construction.
Most professional audio amplifiers feature built-in compressors that are activated when the output signal approaches the clipping point. These compressors typically reference the difference between the supply voltage rails, V_supply and the output signal, V_out, and trigger when the output voltage (V_out) approaches the rail voltage (V_supply) by comparing the relationship between V_supply and V_out with a threshold value (V_threshold) using a mathematical function such as:
if |V_out| >= V_supply - V_threshold then activate the compressor
This mechanism is similar to clipping circuit LEDs, but it also includes an input signal trimming function, which can be achieved by either cutting the input signal or reducing the gain of an input buffer stage using mathematical functions such as:
V_out = V_in * Gain
Where Gain is a parameter that can be adjusted to reduce the output voltage.
Additionally, compressors can also monitor the input signal in relation to a set input voltage reference, V_ref. This approach allows the compressor to operate independently of the back emf control mechanism, which prevents clipping of the negative feedback signal by comparing the relationship between V_in and V_ref with a threshold value (V_threshold) using a mathematical function such as:
if |V_in| >= V_ref + V_threshold then activate the compressor
Amplifiers that employ separate voltage and current gain stages must also include protection diodes between the voltage gain stage and the current gain stage's power supply to avoid hard saturation of the output stage, by allowing current to flow only in one direction.
When executed properly, this technique can result in a "fatter" sound when the amplifier is driven to its limits. However, if not properly implemented, attempting to overdrive the amplifier can result in undesirable distortion.
In summary, the use of built-in compressors and protection diodes are essential for preventing clipping and distortion in professional audio amplifiers, and when implemented correctly, can result in a more pleasing and dynamic sound.
Power compressors are also common in audiophile circles, basically most amplifiers that use a modified feedback mechanism to an already linear amplifier could be considered a power compressor for example Hegel and I guess the new accuphase models
Last edited:
Years ago, I put a simple LED+LDR circuit around the monitor amp at a roller-rink. The effect was that as the DJ pushed up the mixer output, it just didn't get any louder, but there was no distortion. A shunt LDR has the right attack-recover time constants and ~no distortion curve. OnAudio: LOL! way to make a simple thing sound real complicated. You must be in sales?
Nice. LDRs are used in well regarded compressors (called opto compressors, in recording studios). Not that I like to listen to overprocessed music, nowadays.
edit: https://www.diyaudio.com/community/threads/chat-gpt-ideas-about-amplifier-design.394541/
He asked the question about soft clipping to an AI (chat gpt). Apparently these things have reached a level of usability, for some tasks, for some people. The problem will be when people in power who are not intelligent start to take AI too seriously.
edit: https://www.diyaudio.com/community/threads/chat-gpt-ideas-about-amplifier-design.394541/
Last edited:
Was this you or the AI? See, you must now always state who is "thinking"... You have been associated with a chat app.... or should we from now on regard all your posts coming from the machine?
Do tell.
//
Ya, LED, LDR, bridge and 3 resistors per channel. Just fixed parameters, no adjustments.
Sorry, I see that you are just posting chat GPT output. And it makes perfect sense that it would parrot sales material available on the internet, much of which is pseudo-technical.
I would use the blue curve from page 1 (or something very similar) for a signal splitter for a full-range active speaker.
Basic recipe:
1) Active 1st order XO separates high and low frequencies. (Higher order filters would also be possible, but only if they can be summed back together with correct phase.)
2) Soft clipping is applied to bass-only.
3) Low frequencies (pre-distorted) + high frequencies (clean) are recombined.
4) Output to full-range active speaker. (OR: output to pro-woofer, etc.)
Discussion:
The trade-off is some harmonics in the bass, where the bass peaks are deliberately flattened a little bit. The benefit would be cleaner treble with less IMD because the highs now have more overhead to "ride on top of" various drum beats or other bass notes without distortion from momentarily hitting speaker limits such as Xmax or excessively high voltage and current.
It would probably / certainly have to be tuned by ear, as the whole point is to balance 2 different distortion mechanisms against one another to find a happy mid-point. And -- for a pre-amp style device -- I would treat the whole system as an active speaker, so hot-swapping speakers or amplifier gain levels would be disallowed to prevent the soft-clipping from accidentally kicking in at the wrong gain level.
Basic recipe:
1) Active 1st order XO separates high and low frequencies. (Higher order filters would also be possible, but only if they can be summed back together with correct phase.)
2) Soft clipping is applied to bass-only.
3) Low frequencies (pre-distorted) + high frequencies (clean) are recombined.
4) Output to full-range active speaker. (OR: output to pro-woofer, etc.)
Discussion:
The trade-off is some harmonics in the bass, where the bass peaks are deliberately flattened a little bit. The benefit would be cleaner treble with less IMD because the highs now have more overhead to "ride on top of" various drum beats or other bass notes without distortion from momentarily hitting speaker limits such as Xmax or excessively high voltage and current.
It would probably / certainly have to be tuned by ear, as the whole point is to balance 2 different distortion mechanisms against one another to find a happy mid-point. And -- for a pre-amp style device -- I would treat the whole system as an active speaker, so hot-swapping speakers or amplifier gain levels would be disallowed to prevent the soft-clipping from accidentally kicking in at the wrong gain level.
There's a possible approach using the TAS5825 Advanced Features. It can automatically reduce its gain as a function of power supply voltage. The intent is to avoid clipping as a battery discharges. Allow the power supply to have a load-line (voltage decrease with current increase) and it'll automatically compress the input amplitude to fit whatever powersupply voltage value it has to work within. Supposedly.
It also has a few other tricks (Thermal, speaker excursion) that act to digitally reduce the input level. Just sayin a lot of this has been thought through by TI and their implementations may be useful. Unsure of the details, such as does the powersupply ADC have its own pin, or is it connected to Vcc internally to the chip.
It also has a few other tricks (Thermal, speaker excursion) that act to digitally reduce the input level. Just sayin a lot of this has been thought through by TI and their implementations may be useful. Unsure of the details, such as does the powersupply ADC have its own pin, or is it connected to Vcc internally to the chip.
Here's an idea of mine, shown with opamp but applicable to any feedback amp I think:
Emulated:
The onset of soft clipping is set by the zener voltages of D8 ad D17 (here 12V), and more sets of anti-parallel diodes and resistors increase the headroom. I'm using 1N4148's which are limited to 75V, but their low capacitance should reduce the effect of the network on the feedback compensation behaviour. Halving the resistors at each stage seems to give a reasonable waveform, each stage contributes to the roundness of the waveform:
Emulated:
The onset of soft clipping is set by the zener voltages of D8 ad D17 (here 12V), and more sets of anti-parallel diodes and resistors increase the headroom. I'm using 1N4148's which are limited to 75V, but their low capacitance should reduce the effect of the network on the feedback compensation behaviour. Halving the resistors at each stage seems to give a reasonable waveform, each stage contributes to the roundness of the waveform:
Last edited:
That's remarkable, I've been working on something similar, but for a different purpose. For what it's worth, the Diodes, Inc 1N4148 has reverse rating of 100V.