Because my power amplifier has a lot of gain (30dB) and I don't live in a stadium the effective range of the volume control is about 1/6 of the total range.
This is very annoying and that is why I am considering using a gain of about 3 (10 dB) for the amplifier I'm going to build.
The input stage is similar to the one in this image:
http://www.diyaudio.com/forums/attachment.php?s=&postid=351305
Because there is 20 dB less attenuation necessary in the pre amp to have the same output level I think the sound will improve.
But, the higher feedback might have a negative impact.
Are there more advantages/disadvantages?
This is very annoying and that is why I am considering using a gain of about 3 (10 dB) for the amplifier I'm going to build.
The input stage is similar to the one in this image:
http://www.diyaudio.com/forums/attachment.php?s=&postid=351305
Because there is 20 dB less attenuation necessary in the pre amp to have the same output level I think the sound will improve.
But, the higher feedback might have a negative impact.
Are there more advantages/disadvantages?
a well designed negative feedback amplifier will be likely to oscillate if you simply reduce the closed loop gain by changing the feedback ratio, almost certainly if you are contemplating over 10 dB change - the classic negative feedback system perscription is for 60 degree phase margin and 10 dB gain margin
thanks guys,
@audiofan,
Would a simulation show this clipping? Not that I have a simulator, but it woiuld be a good reason to get one
@jcx,
It is not an existing design, I'm going to mix ideas from several stages into 1 project. When you state '10 dB gain margin' do you mean that you shouldn't change a existing design's gain margin by more than 10dB or minimum gain should be 10dB?
@Bill,
I don't want a divider at the input because that will still lead to a 20dB attenuation which might cause some musical information to disappear below the noise level.
@audiofan,
Would a simulation show this clipping? Not that I have a simulator, but it woiuld be a good reason to get one
@jcx,
It is not an existing design, I'm going to mix ideas from several stages into 1 project. When you state '10 dB gain margin' do you mean that you shouldn't change a existing design's gain margin by more than 10dB or minimum gain should be 10dB?
@Bill,
I don't want a divider at the input because that will still lead to a 20dB attenuation which might cause some musical information to disappear below the noise level.
If you start from scratch, there's no reason not to do it. It's not a trivial task to design a power amp from the ground-up, but I'm sure you're aware of that.
If you need 20dB less closed loop gain you may be able to design your power amp with less stage(s) and/or less gain, and still have acceptable linearity. That would make the power amp simpler, possibly more wideband with less distortion rise with freq etc. It's a quite attractive idea.
I'm working (turning it over in my head, some sketches so far) on something like that to use with my Behringer DCX2496. Currently I use 30dB attenuators, which is a waste.
You are right that this attenuation before the power amp decreases the power amp S/N ration.
Alternatively, if you currently use an active preamp with gain, getting rid of the preamp just using a potmeter has its own attractiveness.
Jan Didden
If you need 20dB less closed loop gain you may be able to design your power amp with less stage(s) and/or less gain, and still have acceptable linearity. That would make the power amp simpler, possibly more wideband with less distortion rise with freq etc. It's a quite attractive idea.
I'm working (turning it over in my head, some sketches so far) on something like that to use with my Behringer DCX2496. Currently I use 30dB attenuators, which is a waste.
You are right that this attenuation before the power amp decreases the power amp S/N ration.
Alternatively, if you currently use an active preamp with gain, getting rid of the preamp just using a potmeter has its own attractiveness.
Jan Didden
Phase margin = 180 – phase lag @ freq where open loop gain = 1
~= how much extra phase shift would be needed to cause the loop to oscillate
Gain margin = open loop gain (loss) where open loop phase lag = 180 degrees
~=how much the open loop gain could increase before the loop oscillates
These are 2 points that you can see on a Bode plot that indicate how close to oscillation your feedback amplifier compensation is, a better approach for viewing negative feedback amplifier stability uses the Nyquist plot where you can see the gain/phase line relation to the instability region
Why not just avoid the region of instability by wider margins? Because negative feedback amplifier design wants to use the most loop gain at low frequencies for best accuracy and distortion reduction and physics limits how fast that gain can be reduced as frequency increases and you encounter the limiting speed device’s unavoidable phase shift in your design (usually the output transistor) – high loop gain and wide region of high gain mean you are going to get as close to the instability region as is practically safe
In practice audio amplifiers usually give up some potential open loop gain by using dominant pole compensation and have to set the gain and phase margins overly conservatively (larger dominant pole time constant) to accommodate output device variation
~= how much extra phase shift would be needed to cause the loop to oscillate
Gain margin = open loop gain (loss) where open loop phase lag = 180 degrees
~=how much the open loop gain could increase before the loop oscillates
These are 2 points that you can see on a Bode plot that indicate how close to oscillation your feedback amplifier compensation is, a better approach for viewing negative feedback amplifier stability uses the Nyquist plot where you can see the gain/phase line relation to the instability region
Why not just avoid the region of instability by wider margins? Because negative feedback amplifier design wants to use the most loop gain at low frequencies for best accuracy and distortion reduction and physics limits how fast that gain can be reduced as frequency increases and you encounter the limiting speed device’s unavoidable phase shift in your design (usually the output transistor) – high loop gain and wide region of high gain mean you are going to get as close to the instability region as is practically safe
In practice audio amplifiers usually give up some potential open loop gain by using dominant pole compensation and have to set the gain and phase margins overly conservatively (larger dominant pole time constant) to accommodate output device variation
@jcx
If I understand your explanation correctly the gain margin is the difference between the open loop gain and the actual gain achieved with feedback.
In my view a lower gain would mean a bigger margin.
What the lower gain does to the phase shift is unclear to me, maybe this would be a problem?
@janneman
Are you also thinking about designing/building a low power class a amp?
For me the power amp is my 1st real diy project, if this is working out ok i plan to replace the pre amp.
If I understand your explanation correctly the gain margin is the difference between the open loop gain and the actual gain achieved with feedback.
In my view a lower gain would mean a bigger margin.
What the lower gain does to the phase shift is unclear to me, maybe this would be a problem?
@janneman
Are you also thinking about designing/building a low power class a amp?
For me the power amp is my 1st real diy project, if this is working out ok i plan to replace the pre amp.
squadra said:
Gain margin is as you say. Lower closed loop gain gives you larger gain margin, but that means you still have a lot of gain at the freq where the phase gets toward 180 degr shift, so that means oscillations. So, all things remaining equal, if you increase feedback to get lower closed loop gain you *may* get instability and oscillations. That's why those gain clones need a minimun closed loop gain, which is quite high, for stability: they have relatively a lot of phase shift starting already at quite low frequencies.
Jan Didden
complicated question there Jan,
I think your concept of fewer stages is rooted in the idea that each stage adds phase shift and the fewer the better – but a question is whether the added stages can be much faster than the output device, if so extra stages don’t add significant phase shift and it should be possible to make good use of the extra gain (at low frequencies)
Common audio power amp practice however frequently uses VAS/Driver Qs only a little faster than the best output Qs since unity gain outputs require full amp voltage swing and high voltage transistors are naturally slower, Ed Cherry recommends CE outputs with large V gain and then the rest of the amp can operate at low voltages with very fast low voltage small signal Qs (or even op amps, today’s better op amps are faster than many common small signal Qs)
There are many tradeoffs in multistage amplifiers, especially with respect to stability and recovery transients in the face of clipping/slew/current limiting those limits, and typical transistor speeds probably leave us with the present loose rules that you can observe in current amplifier practice: 3 stages of amplification is the default, 1 or 2 for high frequency response(at reduced accuracy), 4 for low frequency precision
(all depending on how you count such tight local feedback stages such as a cascodes or triple CF or Darlington outputs)
squadra,
the gain margin is how much the open loop amplifier gain can increase (assuming the phase shift doesn’t change) before the amplifier will become an oscillator
the gain may be larger than the designed gain from component tolerance or for example h_fe rise with temp, or gm/ft changing with operating current
keeping feedback relations straight is tough, but lower closed loop gain means less feedback network attenuation which gives higher loop gain (loop transmission) and the closed loop gain intercept with the open loop gain curve moves higher in frequency, closer to the point where open loop phase shift exceeds 180 degrees and reducing gain margin (if could explain this clearly in a paragraph I should be a EE prof, not just an engineer – Gerald Graeme’s op amp books have an interesting graphical feedback analysis approach based on the Bode plots, perhaps you can find some of his old articles online for a slightly different view than the typical textbook)
If you aren’t pushing the envelope in this project then adjusting the dominant pole (usually Miller Cap across the VAS) should let you use lower closed loop gain with little difficulty
I think your concept of fewer stages is rooted in the idea that each stage adds phase shift and the fewer the better – but a question is whether the added stages can be much faster than the output device, if so extra stages don’t add significant phase shift and it should be possible to make good use of the extra gain (at low frequencies)
Common audio power amp practice however frequently uses VAS/Driver Qs only a little faster than the best output Qs since unity gain outputs require full amp voltage swing and high voltage transistors are naturally slower, Ed Cherry recommends CE outputs with large V gain and then the rest of the amp can operate at low voltages with very fast low voltage small signal Qs (or even op amps, today’s better op amps are faster than many common small signal Qs)
There are many tradeoffs in multistage amplifiers, especially with respect to stability and recovery transients in the face of clipping/slew/current limiting those limits, and typical transistor speeds probably leave us with the present loose rules that you can observe in current amplifier practice: 3 stages of amplification is the default, 1 or 2 for high frequency response(at reduced accuracy), 4 for low frequency precision
(all depending on how you count such tight local feedback stages such as a cascodes or triple CF or Darlington outputs)
squadra,
the gain margin is how much the open loop amplifier gain can increase (assuming the phase shift doesn’t change) before the amplifier will become an oscillator
the gain may be larger than the designed gain from component tolerance or for example h_fe rise with temp, or gm/ft changing with operating current
keeping feedback relations straight is tough, but lower closed loop gain means less feedback network attenuation which gives higher loop gain (loop transmission) and the closed loop gain intercept with the open loop gain curve moves higher in frequency, closer to the point where open loop phase shift exceeds 180 degrees and reducing gain margin (if could explain this clearly in a paragraph I should be a EE prof, not just an engineer – Gerald Graeme’s op amp books have an interesting graphical feedback analysis approach based on the Bode plots, perhaps you can find some of his old articles online for a slightly different view than the typical textbook)
If you aren’t pushing the envelope in this project then adjusting the dominant pole (usually Miller Cap across the VAS) should let you use lower closed loop gain with little difficulty
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.