Say you want an amp that produces 100W into 8 ohms over the audio band. That requires a certain slew rate.
It is irrelevant how the input signal to that amp is produced, electronically, a piano, muted trumpet, what have you.
The determining spec is 100W into 8 ohms, which determines the output level (28Vrms). The max frequency (20kHz) determines the slew rate required for that: 2*pi*f*Vpk-pk which is about 10Megavolts per second, or 10V/us.
Jan
SO, if 6 dB increase of output power = double voltage output > then the example of a 100W scenario vs 400W = only 20V/us ?
[ but RMS V of sine & square are not the same ]
[ but RMS V of sine & square are not the same ]
The equation I showed wants signal amplitude peak-to-peak.
The output slews from min to max.
Quizz: assuming a sine wave signal, is the slew rate constant over the wave form?
If not, where on the wave form is the slew rate the highest?
If you really understand slew rate, it's a no-brainer.
Jan
The output slews from min to max.
Quizz: assuming a sine wave signal, is the slew rate constant over the wave form?
If not, where on the wave form is the slew rate the highest?
If you really understand slew rate, it's a no-brainer.
Jan
Last edited:
A low pass filter is an important part of amplifier compensation IMV. Although most amplifier designers now have a handle on compensation, slewing distortion etc, it wasn’t always the case. Most modern power amps use degeneration, but if it’s not enough and the compensation loop is heavy handed, it is possible with fast rise/fall times to get the front end pair to switch ie one device fully off, the other conducting all the tail current. If that happens, the amplifier slews as it will be running open loop. In a bad case it will ram up against one of the rails, in a not so bad case, it recovers before that but both are highly objectionable. A front end low pass filter can be set so the signal rise time is never fast enough to cause this problem. When I test for this (inLTspice), I feed a very fast rise time signal in and look at the LTP collector currents. If they look like they are straying too close to the non linear operating regime, I apply some bandwidth filtering on the input - usually set at 500 or 600 kHz.
That said, modern designs almost universally degenerate the input pair to ensure the linear operating region is 0.5 to 1V rather than a few mV in the undegenerated case and the unity loop gain frequency (ULGF) is set at ~2MHz because modern devices feature fTs 10x or better than legacy devices like the 3055/2955. Further, modern source material is heavily bandwidth limited so high slew rates are not possible.
However, I still check for fast rise/fall time performance and slew rate because why accept good when perfect is possible.
YMMV
😊
That said, modern designs almost universally degenerate the input pair to ensure the linear operating region is 0.5 to 1V rather than a few mV in the undegenerated case and the unity loop gain frequency (ULGF) is set at ~2MHz because modern devices feature fTs 10x or better than legacy devices like the 3055/2955. Further, modern source material is heavily bandwidth limited so high slew rates are not possible.
However, I still check for fast rise/fall time performance and slew rate because why accept good when perfect is possible.
YMMV
😊
IMHO ...
Power amp input filtering need not go anything above 80Khz > F3.
[ input filtering is goood ]
Power amp input filtering need not go anything above 80Khz > F3.
[ input filtering is goood ]
Last edited:
Yes, if you set them to the same pk-pk value. Obviously.
The RMS value of a square is half the pk-pk value.
The RMS value of a sine is 1/(sqrroot of 0.5*pk-pk value) iirc, please correct me if I am wrong.
Jan
Last edited:
I think we are talking slew rate, not power.
Slew rate is essentially the time derivative of the signal, so it obviously depends on the shape of the signal.
Slew rate in a specific point on a wave is the slope of a straight line through that point.
And yes that varies with the point on the wave. The steeper the slope, the higher the slew rate.
Max slew rate is at the zero crossing, because at that point the slope is steepest.
It follows that at the top and bottom of a sine wave the slew rate is zero, because the slope is zero.
Jan
And yes that varies with the point on the wave. The steeper the slope, the higher the slew rate.
Max slew rate is at the zero crossing, because at that point the slope is steepest.
It follows that at the top and bottom of a sine wave the slew rate is zero, because the slope is zero.
Jan
Attachments
Last edited:
There seems to be a confusion over the relationship between the time domain (rate of change) and the frequency domain (bandwidth). If you have a signal that is band limited and amplitude limited, then its rate of change is also limited.
That is, if you want to reproduce the audible part of the spectrum with a given peak value, you only need to reproduce the rate of change of the highest frequency spectral component (=20kHz sine) with that peak value. It makes no difference if that's drums, cymbals, piano, trumpet, synth or a sine or square wave. As shown in my earlier post in this thread, with a typical implementation, you'd need the slew rate limit quite a bit higher to minimize distortion, but that is a technicality.
Conversely, if you have some slew rate limit and wish to stay below it, you need to limit either the bandwidth, or the amplitude, or both. That's what the input LPF is for.
I am a Vietnam vet and am familiar with a 105mm howitzer that had 6400 mil capability. It had a hexagonal baseplate with holes at the apexes through which large metal stakes could be driven to fasten the hex plate to the ground. The center of the plate had a ball socket, and the howitzer had a matching ball mounted on its bottom. You set the howitzer into the socket and a latch could be closed to prevent separation. A rubber tire on the rear of the howitzer, operated with a hand crank, would allow the howitzer to be quickly aimed in any direction. The mil was used as the unit of measure for leftward or rightward barrel motion because cutting the full circle into 6400 pieces rather than 360 pieces provided enough aiming resolution so that a fraction of a unit (the mil) was not needed. With 360, a decimal point would be needed to get the needed rotational resolution, and a misplayed 'dismal' point could result in a catastrophic aiming error
My experience was with the Hawk air defense system and later Patriot. I remember that 800 mil elevation was 'safe elevation' of the launcher (45 degrees up of course). IIRC the launcher had a slew rate of probably 2400 mil/sec in azimuth and a bit more in elevation because in the vertical direction it was fairly balanced.
It became operational in iirc 1955, and is now seeing a new life in Ukraine these days, although in heavily improved form.
The old analog computers all replaced by digital and with extensive DSP power in the radar receivers.
https://armyrecognition.com/militar...defense-vehicles/hawk-mim-23-united-states-uk
Jan
It became operational in iirc 1955, and is now seeing a new life in Ukraine these days, although in heavily improved form.
The old analog computers all replaced by digital and with extensive DSP power in the radar receivers.
https://armyrecognition.com/militar...defense-vehicles/hawk-mim-23-united-states-uk
Jan
l
Where does that supposed 20 kHz maximum frequency come from? If you don't want to make any assumptions about the signal, it could very well contain ultrasonic components. If those drive the amplifier into slew rate limiting, the resulting intermodulation components may well be audible.
Then again, FM radio has a power bandwidth of at most 3183 Hz, and most music can be broadcast over FM radio with only little high-frequency limiting.
Where does that supposed 20 kHz maximum frequency come from? If you don't want to make any assumptions about the signal, it could very well contain ultrasonic components. If those drive the amplifier into slew rate limiting, the resulting intermodulation components may well be audible.
Then again, FM radio has a power bandwidth of at most 3183 Hz, and most music can be broadcast over FM radio with only little high-frequency limiting.
I was under the impression that FM audio bandwidth extended to 15kHz, and it also had to support a 19kHz chopper signal for stereo decoding at the receiver.
I just used the generally used highest audio bandwidth.
I was explaining the mechanism, not the absolute values.
If you want to use 30kHz Marcel, go for it! 😎
Jan
It does, but the power bandwidth is either 2122 Hz or 3183 Hz due to pre- and de-emphasis. That is, FM can handle audio signals between 3183 Hz and 15 kHz, but only at lower levels than low-frequency signals.