Bob Cordell said:
Right, but here we have more than one device with multiple reactive feedbacks.
Right, RC is enough for 90 degree shift.
So, in some cases Zobel load is desired. My point is, it is not necessary in all cases.
So, what is the reason for such termination of the far end of the cable?
Bob Cordell said:
Hi Bob;
your amplifiers already sound more like tube amplifiers, thanks to FETs their output resistance non-linearities cause less audible distortions if to compare with ordinary typical opamps with BJT emitter followers. Highly variable very complex impedance of speakers are less reflected on voltage amplification stage.
Some prefer sound that sound more real, i.e. that has less errors that are audible while have more of inaudible errors. Speaking of electrical damping, you can damp main resonance of the whole cone electrically damping the motor, but you can't damp electrically higher frequency resonances of cones, spiders, frames, boxes. But the whole cone may be well damped by damping volume resonances in the cabinet.
Adding a resistor you load an amp on less complex load decreasing distortions, also an output impedance has higher constant part so it's variable part that is already smaller is less reflected on power delivered to the speaker.
I believe that amps designed by John Curl also can't benefit from resistors between their outputs and speakers, because John designs for enormous output current capabilities. I borrowed his approach when instead of usual current limiting for protection of output transistors implemented toothcrushing current boosting by additional FETs.
RFI
Hi Bob,
I prefer to terminate the " transmission line" at the amp side with a cap with no series resistor. In this way, RFI is better blocked. As for HF, this side is shorted, so the other side has to be terminated (to prevent reflections or high impedance peaks). For DIYs no problem. For an amp manufacturer however, who doesn't know what kind of speakers will be used, the situation is quite different. What should he do?
Cheers,
Edmond.
PS: Why are not all speaker boxes well terminated at HF? Just laziness, or is it ignorance?
Bob Cordell said:
Hi Bob,
I prefer to terminate the " transmission line" at the amp side with a cap with no series resistor. In this way, RFI is better blocked. As for HF, this side is shorted, so the other side has to be terminated (to prevent reflections or high impedance peaks). For DIYs no problem. For an amp manufacturer however, who doesn't know what kind of speakers will be used, the situation is quite different. What should he do?
Cheers,
Edmond.
PS: Why are not all speaker boxes well terminated at HF? Just laziness, or is it ignorance?
KSTR said:
I assume we are talking about R300, L301 and C304 and then I did misunderstand you since they are actually all three in parallel. I thought the R and C were in series, so thanks for posting the image.
To: John Curl
John,
How would you feel about putting a neat 11nF capacitor across the speaker terminals of your Parasound amplifier for the purpose of reducing RFI ingress?
Brian
John,
How would you feel about putting a neat 11nF capacitor across the speaker terminals of your Parasound amplifier for the purpose of reducing RFI ingress?
Brian
Re: To: John Curl
Trolling again? Without adding the rest of the Zobel network, this is a really stupid question.
traderbam said:
Trolling again? Without adding the rest of the Zobel network, this is a really stupid question.
Re: RFI
Hi Edmond,
I agree, it would be nice if every loudspeaker had a zobel network right at its terminals to assure that it stayed a resistive load out to very high frequencies. I think there are some people out there who market such a solution. Alas, most loudspeaker manufacturers seem to care little about the impedance they present to the source, much less a subtlety like the nature of the impedance well above the audio band.
What can you say when you have speakers that dip to 3 ohms at their cabinet tuning frequency and perhaps at other frequencies, while at the same time they may peak to 40 ohms or more at a low-frequency resonance and at one or two crossover frequencies?
Anyway, I shy away from going all the way to a capacitor that makes a dead short at very high frequencies (or attempts to and ends up having resonance regions and inductive regions) in favor of at least guaranteeing that the termination is lossy at the only end I can control - the source end. You are in good company, however, as I believe Bryston puts a 0.05 uF cap right across the speaker terminals.
Cheers,
Bob
Edmond Stuart said:
Hi Edmond,
I agree, it would be nice if every loudspeaker had a zobel network right at its terminals to assure that it stayed a resistive load out to very high frequencies. I think there are some people out there who market such a solution. Alas, most loudspeaker manufacturers seem to care little about the impedance they present to the source, much less a subtlety like the nature of the impedance well above the audio band.
What can you say when you have speakers that dip to 3 ohms at their cabinet tuning frequency and perhaps at other frequencies, while at the same time they may peak to 40 ohms or more at a low-frequency resonance and at one or two crossover frequencies?
Anyway, I shy away from going all the way to a capacitor that makes a dead short at very high frequencies (or attempts to and ends up having resonance regions and inductive regions) in favor of at least guaranteeing that the termination is lossy at the only end I can control - the source end. You are in good company, however, as I believe Bryston puts a 0.05 uF cap right across the speaker terminals.
Cheers,
Bob
Eva-PowerWavebourn said:
I think that if adding some series resistance to the output of an amplifier made it sound subjectively better, such a change would work just as well on John's amplifiers. I am also an advocate of very high output current capability, and I don't think that such a capability makes an amplifier less suitable for the addition of a very small series resistance in the speaker line if that makes things sound better.
One thing that we might want to keep in mind is the kind of amplifier the speaker designer used when he voiced his speakers and tweaked his crossovers. If he used a tube amplifier, in particular, the speaker might indeed have a reason to sound better with a little bit of resistance added.
On the other hand, if he used an amplifier like John's with a very high damping factor, it seems less likely that adding a resistor would make them sound better (unless the listener happens to prefer the coloration introduced over a reproduction that is more faithful to the way the speakers were voiced by the designer).
Cheers,
Bob
(today)
Bob Cordell said:
I mean unreasonable current capabilities in terms of more absolutely objectively constant output resistance on reasonable currents. Electromagnetic relay may deliver enormous current, but output resistance will be very variable. 😀
Right; some speaker designers assume virtual amps with zero output resistance designing crossovers. Should we design amps for such virtual reality?
My point is, I've found that for optimal result (fooling imaginations) mechanical resonances should be damped mechanically, but electrical distortions should be taken care of electrically. God's for God, Cesar's for Cesar.
Now, I have to return back to my class A+C+C hybrid prototype: very linear amp made of very non-linear devices.
Talk to you later!
🙂
Anatoliy
In that Bryston schematic the 100nF capacitor is not
placed in the best location, the amplifier is easier to disturb that way. External load capacitances and inductances can easily resonate with the internal 100nF capacitor. Placing a separate RC network before the RL provides superior immunity against any load that becomes capacitive or resonant at RF.
BTW: Sometimes I feel that linear amplifier designers fear inductors, they are seldom seen on class AB amplifiers, but they are actually very useful sometimes. Inductors are your friends 😀
placed in the best location, the amplifier is easier to disturb that way. External load capacitances and inductances can easily resonate with the internal 100nF capacitor. Placing a separate RC network before the RL provides superior immunity against any load that becomes capacitive or resonant at RF.
BTW: Sometimes I feel that linear amplifier designers fear inductors, they are seldom seen on class AB amplifiers, but they are actually very useful sometimes. Inductors are your friends 😀
Eva, that's right.
I am interested in John Curl's view on putting a cap across the speaker outputs, especially one as high as 100nF.
I'm sure Bryston have a reason to do this; I would like to know why because it has undesirable side effects.
Hanging a large weight on the output is the opposite of what the amplifier designer wants. The amplifer output wants to be conveyed with as little mass as possible in the audio band. Adding more reactance to an already reactive situation is not normally helpful, for reasons already raised. Providing damping is.
A 100nF cap will sink some 10% output current at 20kHz.
Bryston's 7B/14B amps claim slew rates of 120V/us. Given that the speaker output is separated from the amp output by a 2uH inductor in parallel with 5 ohms, I estimate that the slew rate at the amp output would need to exceed 600V/us to produce 120V/us at the speaker terminals.
John, I know you are an ardent slewist
and so I would expect you to shun the whole idea. What do you think about it?
I am interested in John Curl's view on putting a cap across the speaker outputs, especially one as high as 100nF.
I'm sure Bryston have a reason to do this; I would like to know why because it has undesirable side effects.
Hanging a large weight on the output is the opposite of what the amplifier designer wants. The amplifer output wants to be conveyed with as little mass as possible in the audio band. Adding more reactance to an already reactive situation is not normally helpful, for reasons already raised. Providing damping is.
A 100nF cap will sink some 10% output current at 20kHz.
Bryston's 7B/14B amps claim slew rates of 120V/us. Given that the speaker output is separated from the amp output by a 2uH inductor in parallel with 5 ohms, I estimate that the slew rate at the amp output would need to exceed 600V/us to produce 120V/us at the speaker terminals.
John, I know you are an ardent slewist
and so I would expect you to shun the whole idea. What do you think about it?
I have put a capacitance directly across an output, on occasion. Many people here have little or no idea what causes power amp instability, or what is really necessary to fix it.
Actually, I look at it as pole splitting, and you want to keep the poles from each other. Sometimes just adding a cap across the load splits the poles and the amp will remain stable with any additional cap load. Most of the time, I just use a Zobel network with 10 ohms and a quality .1uF cap.
Actually, I look at it as pole splitting, and you want to keep the poles from each other. Sometimes just adding a cap across the load splits the poles and the amp will remain stable with any additional cap load. Most of the time, I just use a Zobel network with 10 ohms and a quality .1uF cap.
When we rate slew rate, we mean 'intrinsic' slew rate, rather than 'extrinsic' slew rate.
This can be confusing when you have output coils and input filters added, that appear to slow down the rise time to below the rated slew rate. It must be remembered that it is NOT the absolute rate of change that is important, but the effortlessness in which it allows high frequency signals up to the maximum design bandwidth to pass.
I think of it sometimes in auto terms. Why would they make a car that can do 180 mph and then rev limit it to 155 mph? Why, in fact would they even make it capable of 180 mph? It is because that up to 150 mph or so, it will be rather effortless in operation, and not at its design limit. It is the same with slew rate.
This can be confusing when you have output coils and input filters added, that appear to slow down the rise time to below the rated slew rate. It must be remembered that it is NOT the absolute rate of change that is important, but the effortlessness in which it allows high frequency signals up to the maximum design bandwidth to pass.
I think of it sometimes in auto terms. Why would they make a car that can do 180 mph and then rev limit it to 155 mph? Why, in fact would they even make it capable of 180 mph? It is because that up to 150 mph or so, it will be rather effortless in operation, and not at its design limit. It is the same with slew rate.
When the 100nF capacitor is placed after the inductor without no RC before the inductor, a combination of RLC load values exists that makes the amplifier oscillate. This external RLC will resonate with the internal 100nF and parasitic inductance resulting in a high impedance peak as seen from the amplifier side. If you adjust the external RLC values so that the system becomes tuned close to the unity-gain crossing region, BANG!
It is possible with some amp designs to have a critical load cap value that is rather small. Then, if you load the output with a larger value cap, but not so large that it current limits the amp, then you get unconditional stability. I have done this in a real power amp design.