Is there a link between perceived sound and output impedance of an amp?
As some may know I listen to a modified version of the great Leach amp.
I messed around with the circuit and found that I liked the sound best when the open loop gain was lowered a bit compared to the original. I’m unsure if this is linked to the fact that the output impedance of the amp rises as the open loop gain is lowered.
Could it be that people’s obsession with the sound of no feedback designs is in fact linked to a higher output impedance of these amps. I’m wondering if it is possible to make a non feedback amp with as low output impedance as one with tons of global feedback.
Your thoughts
\Jens
As some may know I listen to a modified version of the great Leach amp.
I messed around with the circuit and found that I liked the sound best when the open loop gain was lowered a bit compared to the original. I’m unsure if this is linked to the fact that the output impedance of the amp rises as the open loop gain is lowered.
Could it be that people’s obsession with the sound of no feedback designs is in fact linked to a higher output impedance of these amps. I’m wondering if it is possible to make a non feedback amp with as low output impedance as one with tons of global feedback.
Your thoughts
\Jens
Hi,
small values of resistor in series with the speaker (0r05 to 1r0) will increase the Q of the bass driver. This is very noticeable in the sound balance coming from a full range speaker.
It is an acceptable method for correcting slight Q mismatch between stereo pairs.
Resistors towards the higher end of the range will give the bass much more low end extension and can change a bass light speaker to a bass strong version.
These same effects could be mimiced ( that does not look right - mimic with an ed) by designing the amp output impedance to suit the speaker set-up.
small values of resistor in series with the speaker (0r05 to 1r0) will increase the Q of the bass driver. This is very noticeable in the sound balance coming from a full range speaker.
It is an acceptable method for correcting slight Q mismatch between stereo pairs.
Resistors towards the higher end of the range will give the bass much more low end extension and can change a bass light speaker to a bass strong version.
These same effects could be mimiced ( that does not look right - mimic with an ed) by designing the amp output impedance to suit the speaker set-up.
Not |Zout| but...
My theory-du-jour is that the size of the imaginary component of ZOut is important (perhaps more than the magnitude). Not supported by experience yet.
Blumlein, 1935, says match impedance:
http://www.doramusic.com/patents/425553.htm
My theory-du-jour is that the size of the imaginary component of ZOut is important (perhaps more than the magnitude). Not supported by experience yet.
Blumlein, 1935, says match impedance:
http://www.doramusic.com/patents/425553.htm
Dear,
If Zout increase, the dampingsfactor become way lower. So basscontrol is less. It can look like there is more bass, but it is just the out of controle resonating woofer.
Try it yourself. Take a woofer (without filter etc.) and short the two terminals. Give a solid pop with you're finger on the woofer membrame. You will hear a tight and short pop. Now short the woofer with a 1 Ohm or higher resistor and pop again. Hear the difference...
A woofer has to be shorted if the sinwave is passing the zero point. The higher the output impedance, the more the woofer is out of control on the zero point.
With kind regards,
Bas
If Zout increase, the dampingsfactor become way lower. So basscontrol is less. It can look like there is more bass, but it is just the out of controle resonating woofer.
Try it yourself. Take a woofer (without filter etc.) and short the two terminals. Give a solid pop with you're finger on the woofer membrame. You will hear a tight and short pop. Now short the woofer with a 1 Ohm or higher resistor and pop again. Hear the difference...
A woofer has to be shorted if the sinwave is passing the zero point. The higher the output impedance, the more the woofer is out of control on the zero point.
With kind regards,
Bas
The damping factor seems to be related with the speaker impedance curve. As far as understand, speaker makers tune their sound, considering the drivers impedance peaks which are common in low frequency zone, and presuming the low damping factor of the power amp (voltage source amp).
For this reason, if the amp output impedance is increasing (to a current source amp with low damping factor), more current would occur especially in the low frequency area so that the speaker might sound like the uncontrolled bass.
Recently I contacted the speaker designer of German Audio Physic, and they confirmed this. In addition, I got the same confirmation myself after having built Pass F1 Amp (of low damping factor).
Anyhow, I solved the excessive bass by attaching and adjusting a low-cut filter at the input of my pre amp. Wonderful bass!
DIY is full of fun!
Regards
jH
For this reason, if the amp output impedance is increasing (to a current source amp with low damping factor), more current would occur especially in the low frequency area so that the speaker might sound like the uncontrolled bass.
Recently I contacted the speaker designer of German Audio Physic, and they confirmed this. In addition, I got the same confirmation myself after having built Pass F1 Amp (of low damping factor).
Anyhow, I solved the excessive bass by attaching and adjusting a low-cut filter at the input of my pre amp. Wonderful bass!
DIY is full of fun!
Regards
jH
Increasing the output impedance of the amplifier is always going to change the frequency response of the loudspeaker. When a single driver is employed, both the LF roll-off due to resonance and the HF roll-off due to voice coil inductance will vary.
When a multi-way system with passive crossover is employed, crossover points will be shifted upwards or downwards and the Q of the filters will also change just by increasing the output impedance of the amplifier.
In other words: Jens, you have been probably just fiddling with your speaker system
When a multi-way system with passive crossover is employed, crossover points will be shifted upwards or downwards and the Q of the filters will also change just by increasing the output impedance of the amplifier.
In other words: Jens, you have been probably just fiddling with your speaker system
JensRasmussen said:
Hi Jens:
How did you lower the amplifier's open loop gain?
Does the lowering of the gain automatically reflects as an increase in the output impedance of the amp?
Have you measured the output impedance before and after the change?
Applied feedback affects closed loop gain and output impedance of amps but also THD and other types of distortion. No feedback designs give, IMO, the most acceptable distortion structure.
Regards,
Milan
Output impedance of the amp is placed in series with the speaker. For single speaker, it changes Qts of the speaker. This may sometimes be useful (bass speaker). For 2, 3 or 4 way speaker box with passive crossover circuits, it is highly undesirable to increase amplifier's output impedance and Zout --> 0 is ideal.
I upped the emitter-resistor value in the input differential amp to lower the open loop gain.
AFAIK negative voltage feedback always lowers the output gain. I derived it using opamps, but it should work with power amps also.
I have no equipment to measure Z-out of an amp, but according to my simulations the open loop gain is about 65dB vs. the original 70dB.
I agree that the technically ideal amp has 0 ohm output impedance, but this may not be what we perceive as the best sound. I guess that was what I tried to say in the first post.
\Jens
AFAIK negative voltage feedback always lowers the output gain. I derived it using opamps, but it should work with power amps also.
I have no equipment to measure Z-out of an amp, but according to my simulations the open loop gain is about 65dB vs. the original 70dB.
I agree that the technically ideal amp has 0 ohm output impedance, but this may not be what we perceive as the best sound. I guess that was what I tried to say in the first post.
\Jens
I can't believe that this had any effect on the output impedance that would impact the sound off any speaker and therefore I believe that your perceived effect on the sound was due to other reasons, e.g lower or higher distortion, phase margin...
For output impedance to have any noticeable effect it must be noticeable in relation to the resitive impedance of the speaker which is usually in the order of ~0.8 of the nominal impedance or for a 8 ohm speaker ~6.5ohm, to change the output impedance from say 0.01 to 0.1 ohm can then not have any effect and you need to go up in the single ohm range to notice the effect.
For your amp when you changed the open loop gain 5dB you are certainly not changing the output impedance to several ohms so therefore I believe that there is some other effect you hear.
Regards Hans
jens mentions open loop gain, if we assume ~ 30 dB closed loop gain then the excess loop gain (~="feedback") varies from 40 to 35 dB - a predictably audilbe level difference due to the gain accuracy alone, a fair comparison requires padding the higher gain channel to match closed loop gains to < 0.1 dB
a/b/x double blind testing establishes that 0.1 dB level differences are audible - the most restrictive claim I've seen is 0.1 dB level difference over more than an octave in frequency response can be detected
the "Carver Challenge" post I pointed to earlier suggests a good approach to working out the possible differences by deeply nulling the response of the original vs the modded amp (both with real speaker load) by padding frequency response and adding series impedance to the higher feedback version
http://www.diyaudio.com/forums/show...2392#post152392
a/b/x double blind testing establishes that 0.1 dB level differences are audible - the most restrictive claim I've seen is 0.1 dB level difference over more than an octave in frequency response can be detected
the "Carver Challenge" post I pointed to earlier suggests a good approach to working out the possible differences by deeply nulling the response of the original vs the modded amp (both with real speaker load) by padding frequency response and adding series impedance to the higher feedback version
http://www.diyaudio.com/forums/show...2392#post152392
http://www.diyaudio.com/forums/showthread.php?postid=152392#post152392
(i cut&pasted the wrong part)
(i cut&pasted the wrong part)
tubetvr
Definitely agreed.
There are more crucial parameters that probaly changed. I'd vote for better stability, higher open-loop bandwidth and lower TIM. Just guessing, I have a feeling, that even great designers as John Curl are sometime confused about how to scientificly explain why lower feedback tends to sound better (to some at least).
Definitely agreed.
There are more crucial parameters that probaly changed. I'd vote for better stability, higher open-loop bandwidth and lower TIM. Just guessing, I have a feeling, that even great designers as John Curl are sometime confused about how to scientificly explain why lower feedback tends to sound better (to some at least).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.