... and can only be approached with minutes of accommodation time for your nerves/muscles to adjust in a anechoic chamber - not exactly the level that can be heard while listening to music
still its not impossible to design and build amps where some errors (such as load current in ps cap “quality” sensitivities) are lower than any of the listed noise floors - including a margin for "sub threshold" detection of correlated sounds
to expand on Mooly - a useful way to think about it is that the power supply is inside the feedback loop for global negative feedback amplifiers - except for input stage common mode errors - diff pair, ccs tail source, bootstrap cascode, balanced source impedance and these errors can be very low
still its not impossible to design and build amps where some errors (such as load current in ps cap “quality” sensitivities) are lower than any of the listed noise floors - including a margin for "sub threshold" detection of correlated sounds
to expand on Mooly - a useful way to think about it is that the power supply is inside the feedback loop for global negative feedback amplifiers - except for input stage common mode errors - diff pair, ccs tail source, bootstrap cascode, balanced source impedance and these errors can be very low
Last edited:
Can you point me to a power amp or an active preamp that doesn't use feedback of any kind.
Member
Joined 2009
Paid Member
There is a return path for the AC current from the output, through the load and back through the PSU caps to the power devices that produced the output. In this regard the caps are in series with the load for AC signals. Nobody said DC ?
No no no
The caps are not in series with the load in the way you mean.Example... take a Class B amp (Class B so it draws little power for this experiment).
The amp is DC coupled so reproduces a 1 hz squarewave perfectly. The load is say 8 ohms and the output a couple of watts.
The power supply has 10,000 uf caps... even these would cause visible LF distortion if they were in series with the load, but they don't, because they aren't. The voltage across the load is a perfect squarewave.
Now swap those 10,000 uf caps for 470uf. The output is still a perfect squarewave.
Why... because of the action of negative feedback, and the points of reference used for the signal input and outputs. The PSU caps aren't in the equation.
Non feedback amp as bobo asked... if it's an emitter follower buffer it's a stage with 100% local feedback. The emitter just reflects the input, the PSU supplying the power... doesn't matter if there is 1 volt or a 100 volts of ripple on the supply due to the size of the PSU caps, as long as there is sufficient voltage the output follows the input approximately, which is all a non global feedback system can do.
The non feedback amp doesn't exist... what I think bobo is asking is what about a non global feedback amp.
An amp like that still has normal feedback for the VAS, it's just that the outputs run as followers outside the global loop.
If we didn't use feedback how could we set the gain of anything.
from all the above ...this comment i will go for +1000 that is an imortand factor i think ....
in an upgrade or a design it will better to notice things like that and select parts that will not effect your signal ...
as about output capacitors in amplifier i have no comment ...it something i never done never did and never will in the feature
Member
Joined 2009
Paid Member
I still don't get it and I fear it's important.
Here's a simulation, I like to see where current flows by using Spice. Do you use Spice ?
The attached file shows a simple EF output, no global feedback. A square wave input and a square wave output is seen. The current through the psu caps is a matching square wave.
Here's a simulation, I like to see where current flows by using Spice. Do you use Spice ?
The attached file shows a simple EF output, no global feedback. A square wave input and a square wave output is seen. The current through the psu caps is a matching square wave.
Attachments
We certainly have an interesting selection of responses!
For reference the paper cited by jcx in post #13 is this one (I think):
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.27.6813&rep=rep1&type=pdf
Some capacitor coupled amps seem generally well regarded... Sugden, Nytech, FirstWatt, Naim headphone amp, etc..
In their well-regarded older single ended amplifiers Sugden used an output capacitor that was exactly the same size as their power supply capacitor. It may be that they just had a lot of them, or maybe there was some special reason for the match?
Has anyone changed an output capacitor and then asked a third party to judge the difference (i.e. blind)?
Has anyone changed an output capacitor (pre or power amp) and then measured any difference in the characteristics of the equipment?
For reference the paper cited by jcx in post #13 is this one (I think):
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.27.6813&rep=rep1&type=pdf
Some capacitor coupled amps seem generally well regarded... Sugden, Nytech, FirstWatt, Naim headphone amp, etc..
In their well-regarded older single ended amplifiers Sugden used an output capacitor that was exactly the same size as their power supply capacitor. It may be that they just had a lot of them, or maybe there was some special reason for the match?
Has anyone changed an output capacitor and then asked a third party to judge the difference (i.e. blind)?
Has anyone changed an output capacitor (pre or power amp) and then measured any difference in the characteristics of the equipment?
Member
Joined 2009
Paid Member
The transformer is only 'on line' when the rectifier diodes are conducting, which is not very often. As far as I can see, at signal frequencies, the PSU caps are the only item in the circuit. The PSU caps are much lower resistance than the transformer at DC based on published information, and at AC the transformer impedance doesn't get any better.
I am somewhat reluctant to become involved in discussions around here any more, but here goes. The transformer secondary is in parallel with the filter caps. That is why very low frequencies, even DC can be reproduced.
For your sim, try using a real amp type power supply and look at the currents through the caps and transformer secondary.
Member
Joined 2009
Paid Member
Steve, don't be shy...
I've added rectifier diodes and AC voltage sources in place of transformers each with a 500mOhm internal winding impedance (resistive). What I see hasn't changed. The AC signals all goes through the psu caps. What I can also see are the rectifier charging spikes also going through the psu caps, as expected.
I've added rectifier diodes and AC voltage sources in place of transformers each with a 500mOhm internal winding impedance (resistive). What I see hasn't changed. The AC signals all goes through the psu caps. What I can also see are the rectifier charging spikes also going through the psu caps, as expected.
Do I use Spice you mean ?... (the clue is in my signature)
I can't think of an easy way to explain it, but lets try.
Think of an opamp in place of your power amp. DC coupled, non inverting, gain of 10 and supplying a 600 ohm load. The simplest circuit you can imagine.
That circuit responds down to DC, can we agree on that.
So +0.1 volt at the input delivers +1.0 volts across the load. We apply a squarewave of 1 hz at -/+ 0.1 volts to the input. Again the output is a perfect squarewave of -/+1.0 volts.
So the input is going 0.1 volt pos and the output is going to 1.0 volts pos, input goes 0.1 volt negative, output goes 1.0 volts negative.
So far so good.
The voltage that the opamp can output comes of course from the PSU.
As long as that voltage is sufficient to power the circuit there is no problem.
Where that opamp get's it's power, and it's relationship to the load current is causing the confusion.
Imagine it was from batteries, two 9 volt ones in series with the centre tap as ground.... just as everyone does with two PP3's running a CMOY etc... which would have been a good example, anyway, as the output goes positive current is drawn from the "top battery". As the output goes negative current is drawn from the "bottom" battery.
So are the batteries in series with the load ? Not exactly. If they were in series exactly the same current would flow through them, but it doesn't. The current in the battery is the sum of all the current needed by the amp.
If you added another load to the battery (a second channel) the current through the load of the first channel is not affected by what that second one does.
If they were by definition "in series" they would.
The opamp keeps the output correct (a 1 hz squarewave) through the action of feedback... as long as the supplies are there.
So what about an AC supply, tranny, rectifier, caps etc.
It's exactly the same... the amp gets it's power from "across" the reservoir caps... it is just a supply for the amp to work with. The load current is of course drawn from that supply, it modulates it, but that current isn't what's flowing through the load.
If you had large reservoir caps and snatched the wires off the transformer, the amp would still work down to DC... it doesn't need a DC path in the PSU.
Has all that made it harder to understand
Just try and visualise it, draw an opamp, it's supply, and see how it's the action of feedback and using correct grounds that determines the output.
Try and see where the load current is "referred" too (the star point) and how what goes on the PSU is irrelevant to the load current.
Nearly got it
Yes.
Yes.
That's the bit that confuses
The load current is returned to a reference point (star ground). From that point (the star) back to the PSU is irrelevant. The current that flows in that connection isn't just in series with the load, it's the sum of all currents flowing into the star point.
The quality of the caps, the value of the caps doesn't matter, the amp is maintaining a perfect output signal (because of feedback).
Think about it... if the PSU caps were "in series" with the load (speaker) you would hear all the 100/120 hz charging pulses as the caps charge from the transformer.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.