sss said:does feedback causes the overshoots?
Yes. I had a schematic for one HP scope. It contains 10kHz squarewave generator. In this generator there was no any feedback, and it produced extremly perfect squarewave
sajti
sss said:so what kind of amp do they use at scopes input?
it should amplify / attenuate all types of signals
The trick is bandwidth.
Remember a perfect square wave of frequency f is a sum of several sine waves: 1 x f + 1/3 x 3f + 1/5 x 5f + 1/7 x 7f.... up to infinity. To correctly amplify a mixture of sinewaves with frequency up to infinity, you need of course an infinite bandwidth.
There's nothing special about the input amplifier of a scope, except that it is a very sophisticated and very fast one.
In your case, where you limit yourself to squarewaves only, you could use a fast comparator at the final stage to square things up.
yeah but those fast amplifiers also can amplify "slow" signals with no distortion - thats specialOliverD said:There's nothing special about the input amplifier of a scope, except that it is a very sophisticated and very fast one.
OliverD said:
There's nothing special about the input amplifier of a scope, except that it is a very sophisticated and very fast one.
And it use only local feedback and/or compensation to reach the upper frequency limit. In that HP they use JFETs differential stage cascoded with another JFETs, and degeneration resistors...
sajti
how does local feedback (and not global feedback) help to reach the upper limitsajti said:And it use only local feedback and/or compensation to reach the upper frequency limit.
OliverD said:
The trick is bandwidth.
Remember a perfect square wave of frequency f is a sum of several sine waves: 1 x f + 1/3 x 3f + 1/5 x 5f + 1/7 x 7f.... up to infinity. To correctly amplify a mixture of sinewaves with frequency up to infinity, you need of course an infinite bandwidth.
There's nothing special about the input amplifier of a scope, except that it is a very sophisticated and very fast one.
In your case, where you limit yourself to squarewaves only, you could use a fast comparator at the final stage to square things up.
It is not so much the bandwidth when creating a square wave as much as how fast can you saturate a particular transistor. The "infinate" odd harmonics will be created by a transistor switch as it saturates. It is called symetrical distortion and is a horrible thing in audio amps but for recreating a square wave, it is great. What frequency does this square wave have to be? If it is slow, then this is an easy task. It is when you want to switch a transistor on and off very fast that specifics come into play. Speed of particular device, gain, input capacitace, miller capacitance and such. If a fast square wave is desired, you had better have a scope to work this one out.
cunningham said:
It is not so much the bandwidth when creating a square wave as much as how fast can you saturate a particular transistor. The "infinate" odd harmonics will be created by a transistor switch as it saturates. It is called symetrical distortion and is a horrible thing in audio amps but for recreating a square wave, it is great. What frequency does this square wave have to be? If it is slow, then this is an easy task. It is when you want to switch a transistor on and off very fast that specifics come into play. Speed of particular device, gain, input capacitace, miller capacitance and such. If a fast square wave is desired, you had better have a scope to work this one out.
For a good discussion of very high speed comparators (such as Elso uses in the Kwak Clock - AD8561) search under comparator on BOTH the Analog Devices and Linear Technology websites. In particular, you will find the following PDF from Linear to be worth reading:
http://www.linear.com/pdf/an72f.pdf
You will find several great circuit ideas in the above PDF.
The PDF for the AD8561 comparator shows how to apply hysteresis:
http://www.analog.com/UploadedFiles/Data_Sheets/764437034AD8561_0.pdf
Discussion of "hysteresis" on the AD9850 high speed comparator (the AD9850 is a 125MHz DDS chip :
http://www.analog.com/Analog_Root/s...1%3D150%26level2%3D173%26level3%3D187,00.html
sss said:i dont think in freq generators the transistor is saturates because u can change the amplitude ,or maybe its done with special amp
I designed a circuit to drive a stepper motor that saturated the output and the voltage rails driving them were adjustable for amplitude. Steppers are slow freq. though.
Bifurcated answer
This question is really two, so here are the two answers:
1. Square wave generators use saturated outputs. Since you can't drive any closer to the rail than saturation, there is essentially no overshoot. Variable levels are accomplished by a variety of means, but rarely by linear amps.
2. Linear, non-saturated amps like scope inputs and well designed audio amps use only local feedback. Global feedback has too much delay going around the loop, so intermediate stages and transient signals cause overshoot, clipping and other audible artifacts.
Ears that are well attuned to acoustic music (rather than reproduction) can hear the artifacts of nearly all compensation, so you have to keep the signal pure at all nodes and times.
Note that scope amps are usually rated at 1% or more non-linearity. Audio amplification is one of the most demanding fields of circuit design since they must simultaneously exhibit extroardinary linearity, unconditional stability, wide bandwidth, low noise and time-accurate transient response. There is no equivalent for this combination in any other electronic field - RF, digital, DSP, video, etc.
This question is really two, so here are the two answers:
1. Square wave generators use saturated outputs. Since you can't drive any closer to the rail than saturation, there is essentially no overshoot. Variable levels are accomplished by a variety of means, but rarely by linear amps.
2. Linear, non-saturated amps like scope inputs and well designed audio amps use only local feedback. Global feedback has too much delay going around the loop, so intermediate stages and transient signals cause overshoot, clipping and other audible artifacts.
Ears that are well attuned to acoustic music (rather than reproduction) can hear the artifacts of nearly all compensation, so you have to keep the signal pure at all nodes and times.
Note that scope amps are usually rated at 1% or more non-linearity. Audio amplification is one of the most demanding fields of circuit design since they must simultaneously exhibit extroardinary linearity, unconditional stability, wide bandwidth, low noise and time-accurate transient response. There is no equivalent for this combination in any other electronic field - RF, digital, DSP, video, etc.
Re: Bifurcated answer
that might be news to the guys who design ultrasound, the folks who work up at Groton CT and the SIGINT folks around Washington.
AcuVox said:Note that scope amps are usually rated at 1% or more non-linearity. Audio amplification is one of the most demanding fields of circuit design since they must simultaneously exhibit extroardinary linearity, unconditional stability, wide bandwidth, low noise and time-accurate transient response. There is no equivalent for this combination in any other electronic field - RF, digital, DSP, video, etc.
that might be news to the guys who design ultrasound, the folks who work up at Groton CT and the SIGINT folks around Washington.
I don't know what slew rate do you need, but you can achieve approx. 10 ns fall- and risetime with 74HC04. For amplifying an AD or TI current feedback amplifier could be fine. (Eg. AD8055 is not so expensive, and quite fast.) You should use low impedances and short traces.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
- Home
- Amplifiers
- Solid State
- how to make a clean square wave?