I'd like to understand how the XA100.5 can have a specified slew rate of 50v/us and yet an upper -2dB frequency of just 100KHz. Probably it is my understanding that is at fault but doesn't a slew rate of 50v/us for 100W output imply a frequency response to approx 200KHz? I base this on the peak amplitude for 100W being 40V and for a sine wave the max frequency = 50,000,000 / 2 * pi *40.
Where am I going wrong? Is it perhaps a case of the amplifier having a frequency response to 200KHz but with a filter at the input limiting the response to an input signal to 100KHz? Would this constitute an amplifier with 50V/us slew rate or would this be described as 25V/us based on what can be achieved from the (filtered) input?
Ignoring for a moment whether this really matters in practice, I remain curious both as to what this means and how it might be achieved. My prototype XA100.5 clone - well more something that shares the same general topology than a direct clone to be honest - seems unable to match this performance. I can just about stretch the response to 200KHz with sufficient feedback but then there is a peak in the response at around 114KHz, presumably due to lack of phase margin. Adding compensation to flatten the response brings the frequency response down to around 180KHz. The associated square wave response shows significant overshoot. Placing a filter at the input to roll off the high frequencies at 100KHz helps but does not completely solve the problem. The measured slew rate under these conditions is about 45V/us without the filter and 25V/us with.
Ian.
Where am I going wrong? Is it perhaps a case of the amplifier having a frequency response to 200KHz but with a filter at the input limiting the response to an input signal to 100KHz? Would this constitute an amplifier with 50V/us slew rate or would this be described as 25V/us based on what can be achieved from the (filtered) input?
Ignoring for a moment whether this really matters in practice, I remain curious both as to what this means and how it might be achieved. My prototype XA100.5 clone - well more something that shares the same general topology than a direct clone to be honest - seems unable to match this performance. I can just about stretch the response to 200KHz with sufficient feedback but then there is a peak in the response at around 114KHz, presumably due to lack of phase margin. Adding compensation to flatten the response brings the frequency response down to around 180KHz. The associated square wave response shows significant overshoot. Placing a filter at the input to roll off the high frequencies at 100KHz helps but does not completely solve the problem. The measured slew rate under these conditions is about 45V/us without the filter and 25V/us with.
Ian.
The "small signal bandwidth" and the "power bandwidth" are two total different things.
The "small signal bandwidth" is limited by stability.
But the "power bandwidth" is limited by slew rate and maximum output voltage.
They don't relate to each other directly.
Usually, when we say "bandwidth", it means the "small signal bandwidth".
Because we must designate the maximum output voltage,
Then we can calculate the "power bandwidth" by slew rate.
So, an amplifier's "small signal bandwidth" may higher or lower than its "full power bandwidth".
If an amplifier is working at a frequence higher than its "power bandwidth",
then the SID(Slew Induce Distortion) will occur because of slew rate limmit.
That means an amplifier's "small signal bandwidth" lower than its "full power bandwidth" will be better.
The "small signal bandwidth" is limited by stability.
But the "power bandwidth" is limited by slew rate and maximum output voltage.
They don't relate to each other directly.
Usually, when we say "bandwidth", it means the "small signal bandwidth".
Because we must designate the maximum output voltage,
Then we can calculate the "power bandwidth" by slew rate.
So, an amplifier's "small signal bandwidth" may higher or lower than its "full power bandwidth".
If an amplifier is working at a frequence higher than its "power bandwidth",
then the SID(Slew Induce Distortion) will occur because of slew rate limmit.
That means an amplifier's "small signal bandwidth" lower than its "full power bandwidth" will be better.
Last edited:
In the case where you have a higher slew rate than a deliberately limited bandwidth, how do you measure the maximum dv/dt? Won’t this be impacted by the mechanism used to limit the bandwidth? Is it a case of taking the dv/dt measurement without the deliberate limitation in place?
I am using the same method to measure slew rate (rise time of the leading edge of a 10KHz square wave in my case) and I see around 1.6us at full output with feedback but without any compensation and around 3us with the bandwidth limitation in place. I believe these correspond roughly to 50V/us and 25V/us in my case.
Ian.
Fascinating - I would never have thought to try something like this! Just ran some tests on my prototype and it does indeed work. Not that I doubted it would but I'm not entirely clear on how it does so. Anyway, I now have a really easy way to measure slew rate. Many thanks as always Nelson 
Ian.
Ian.
There are two ways to limit the "small signal bandwidth" of an amplifier,
One is put a low-pass filter in front of the amplifier,
The other is put a compensation capacitor in the feedback network.
Put a low-pass filter in front of the amplifier will slow down the input signal to prevent the amplifier driving at full speed.
Put a compensation capacitor in the feedback network will not slow down the input signal, so the amplifier still can drive at full speed.
That is the difference between them.
One is put a low-pass filter in front of the amplifier,
The other is put a compensation capacitor in the feedback network.
Put a low-pass filter in front of the amplifier will slow down the input signal to prevent the amplifier driving at full speed.
Put a compensation capacitor in the feedback network will not slow down the input signal, so the amplifier still can drive at full speed.
That is the difference between them.
Thanks for your input wensan and apologies for not replying to your previous post.
The low-pass filter at the input of an amplifier I can understand. High frequencies are blocked from entering the amplifier and hence cannot appear at the output.
A compensation capacitor in the feedback network is more difficult to grasp intuitively. This still acts to reduce the high frequency response of the amplifier, except in this case by increasing the feedback with increasing frequency. Although a high speed input signal can enter the input of the amplifier, the output will be attenuated which feels like a similar result to the other case. What does driving at full speed mean in this case?
Ian.
The low-pass filter at the input of an amplifier I can understand. High frequencies are blocked from entering the amplifier and hence cannot appear at the output.
A compensation capacitor in the feedback network is more difficult to grasp intuitively. This still acts to reduce the high frequency response of the amplifier, except in this case by increasing the feedback with increasing frequency. Although a high speed input signal can enter the input of the amplifier, the output will be attenuated which feels like a similar result to the other case. What does driving at full speed mean in this case?
Ian.
The slew rate limit of an amplifier is caused by the transient overload(or saturation) of the input stage.
An amplifier "driving at its full speed" means when input singal's slew rate too fast cause the transient overload of the input stage happen, the output singal of this amplifier will slew at its maximum slew rate.
An amplifier "driving at its full speed" means when input singal's slew rate too fast cause the transient overload of the input stage happen, the output singal of this amplifier will slew at its maximum slew rate.
Last edited:
Is slew rate always due to input stage limitations or is it anything that limits the max speed of an amplifier. In my case for example, the capacitance of the output MOSFETs may be more of a limit than the input stage. Either way it comes down to how fast capacitance can be charged by the available current and presumably is independent of any feedback applied.
At least I can see now why driving the input stage into saturation allows slew rate to be measured directly. This will effectively bypass any feedback and shows the limit imposed by the dominant capacitance.
I'm now also wondering how relevant slew rate is assuming it is not too slow to cause TID. It seems to me to be more of a maximum value that cannot really be attained by an amplifier in practice, or at least one wouldn't want to knowingly run an amplifier under these condtions.
Ian.
At least I can see now why driving the input stage into saturation allows slew rate to be measured directly. This will effectively bypass any feedback and shows the limit imposed by the dominant capacitance.
I'm now also wondering how relevant slew rate is assuming it is not too slow to cause TID. It seems to me to be more of a maximum value that cannot really be attained by an amplifier in practice, or at least one wouldn't want to knowingly run an amplifier under these condtions.
Ian.
Slew rate - Wikipedia, the free encyclopedia
If the (few) mathematical formulaes bother you, go straight to
the three last lines..
If the (few) mathematical formulaes bother you, go straight to
the three last lines..
That is one definition but doesn't deal with the nub of the problem. The last lines you mention say:
"Slew rate helps us to identify what is the maximum input frequency applicable to the amplifier such that the output is not distorted"
but this would appear to conflict with the mathematical statement further up which defines slew rate in terms of the max input current at saturation. An amplifier in saturation will not deliver an undistorted output signal.
Ian.
"Slew rate helps us to identify what is the maximum input frequency applicable to the amplifier such that the output is not distorted"
but this would appear to conflict with the mathematical statement further up which defines slew rate in terms of the max input current at saturation. An amplifier in saturation will not deliver an undistorted output signal.
Ian.
This figure shows the model of a amplifier with feedback.
It explains when the output current of the input stage reach its maximum value,
the output voltage of the integrator stage and buffer stage will slew with a constant slew rate.
The figures below are a simulation that demonstrates how the slew rate limmit happen.
I demonstrated these figures in Chinese forums about five years ago.
So the explanation in the figure is in Chinese.
An externally hosted image should be here but it was not working when we last tested it.
It explains when the output current of the input stage reach its maximum value,
the output voltage of the integrator stage and buffer stage will slew with a constant slew rate.
The figures below are a simulation that demonstrates how the slew rate limmit happen.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I demonstrated these figures in Chinese forums about five years ago.
So the explanation in the figure is in Chinese.
Last edited:
I use OPA549 as an example.
OPA549's slew rate limmit is 9 V/us.
OPA549's gain-bandwidth-product is 0.9 MHz.
If you use OPA549 to make a amplifier that gain is 2.
That means the bandwidth of this amplifier is 450KHz.
But can OPA549 work at such high frequence?
Not really!
If the peak output voltage of this amplifier is 30 V,
then its full power bandwidth is about 48KHz.
If the peak output voltage of this amplifier is 15 V,
then its full power bandwidth is about 95KHz.
If the peak output voltage of this amplifier is 3 V,
then its full power bandwidth is about 477KHz.
If the peak output voltage of this amplifier is 1.5 V,
then its full power bandwidth is about 955KHz.
So slew rate help us to figure out how big output signal that the amplifier can handle at high frequence.
OPA549's slew rate limmit is 9 V/us.
OPA549's gain-bandwidth-product is 0.9 MHz.
If you use OPA549 to make a amplifier that gain is 2.
That means the bandwidth of this amplifier is 450KHz.
But can OPA549 work at such high frequence?
Not really!
If the peak output voltage of this amplifier is 30 V,
then its full power bandwidth is about 48KHz.
If the peak output voltage of this amplifier is 15 V,
then its full power bandwidth is about 95KHz.
If the peak output voltage of this amplifier is 3 V,
then its full power bandwidth is about 477KHz.
If the peak output voltage of this amplifier is 1.5 V,
then its full power bandwidth is about 955KHz.
So slew rate help us to figure out how big output signal that the amplifier can handle at high frequence.
Last edited:
Wensan,
Thanks for your patience in explaining the mechanics of slew rate limiting in a traditional design, i.e. based on the model used by many op-amps. The details for a quad-symmetric amp like the XA series are different but I suspect the principle is the same given that slew rate is defined as the maximum dv/dt that can be attained under any conditions. In most, perhaps all, cases this will correspond to the input differential being saturated, i.e. where one side is running at the full value of the tail current and the other is effectively not conducting. I now also see that this value of dv/dt cannot be improved by the application of feedback.
The value of this parameter is clearly useful when designing an amplifier as per your OPA549 example. Where I am still less clear is what use this value has when specified for ready built product. If the maximum power bandwidth is also specified then the maximum effective dv/dt that can be obtained is already known. The maximum slew rate must be at least this value but could be higher if the amplifier is bandwidth limited, as most are. Under these conditions, what is the value of knowing the higher slew rate? Is it perhaps to state the maximum input signal frequency that can be applied before TID sets in?
Ian.
Thanks for your patience in explaining the mechanics of slew rate limiting in a traditional design, i.e. based on the model used by many op-amps. The details for a quad-symmetric amp like the XA series are different but I suspect the principle is the same given that slew rate is defined as the maximum dv/dt that can be attained under any conditions. In most, perhaps all, cases this will correspond to the input differential being saturated, i.e. where one side is running at the full value of the tail current and the other is effectively not conducting. I now also see that this value of dv/dt cannot be improved by the application of feedback.
The value of this parameter is clearly useful when designing an amplifier as per your OPA549 example. Where I am still less clear is what use this value has when specified for ready built product. If the maximum power bandwidth is also specified then the maximum effective dv/dt that can be obtained is already known. The maximum slew rate must be at least this value but could be higher if the amplifier is bandwidth limited, as most are. Under these conditions, what is the value of knowing the higher slew rate? Is it perhaps to state the maximum input signal frequency that can be applied before TID sets in?
Ian.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.