This came to mind since driver distortion has gotten more attention lately, in combination with more people (including me) moving to DSP-based active XO, so in many (some?) cases drivers are connected directly to amp outputs. Consequently mids and tweeters can see DC from the amps too.
I imagine it's a non-issue with woofers and 'normal' offset voltages that are within maybe 10mV range, since woofers have been DC-coupled for for a long time in passive crossovers, and it would likely have been mentioned and studied. I'm thinking tweeters and mids could be more sensitive though? Is there a reason to run a series cap on the tweeter more than for fault protection? Also, all amps are not equal, and some may have larger swings in DC offset with temperature etc.
Any insights are welcome!
I imagine it's a non-issue with woofers and 'normal' offset voltages that are within maybe 10mV range, since woofers have been DC-coupled for for a long time in passive crossovers, and it would likely have been mentioned and studied. I'm thinking tweeters and mids could be more sensitive though? Is there a reason to run a series cap on the tweeter more than for fault protection? Also, all amps are not equal, and some may have larger swings in DC offset with temperature etc.
Any insights are welcome!
Ehm, often, very often you see filters before the tweeters and mids (woofers like them also) and the capacitor is part of the high pass network
Filters select the correct bandwidth for each speaker and also slopes, EQ...
Edit: oh, I see, DSP is mentioned in the beginning
I wouldn't run a series capacitor on drivers for anything but DC protection in a direct coupled DSP-based system.I imagine it's a non-issue with woofers and 'normal' offset voltages that are within maybe 10mV range, since woofers have been DC-coupled for for a long time in passive crossovers, and it would likely have been mentioned and studied. I'm thinking tweeters and mids could be more sensitive though? Is there a reason to run a series cap on the tweeter more than for fault protection?
I would far prefer a speaker protection circuit to a capacitor.
The cost of an external speaker protection circuit (for amps without one built in) are less than a high voltage, high quality audio series capacitor.
Regarding the potential distortion issue caused by DC offset:
10mv or 0.01v would be ~0.0000125 watt into an 8 ohm driver.
If all that DC power was converted into sound pressure in the form of harmonic distortion, it would be more than -86dB SPL (sound pressure level) below one watt level. Up to one watt, any tweeter of ~86dB 1w/1m the distortion level would be below 0dB SPL, around the minimum level that can be heard by 18-25 year old normal hearing people in the 2-5Khz range.
Quiet listening rooms are ~30 dBSPL, levels below that are "masked", difficult to hear.
Harmonics -60dB below the fundamental frequency are .1% harmonic distortion.
The higher the drive level, the less any distortion % would be with a fixed level of DC offset.
Levels below -60dB are generally below the signal to noise ratio, you would hear amp/preamp "hiss" or room noise more than distortion.
Like amplifier noise, that fixed distortion would (could) be more audible with compression horn drivers which are much more sensitive than the above example, though I doubt 10mv DC offset could be noticed.
One side of a Hafler DH-200 amp (200w@8ohms) driving my Tannoy PBM 6.5 speakers had DC voltage on one channel that slowly grew from a small fraction of a volt to 5 volts over the course of 20 years.
5 volts into the “100watt 8ohm” woofer's DC resistance of around 4 ohms is only 6 watts, but enough to cause the woofer's voice coil to burn as the DC held the voice coil in a fixed position, allowing little cooling as the speaker sat powered, but idle.
I never noticed any distortion, but one day noticed the left woofer was not working..
Art
Hi, see Klippel Controlled Sound for what voice coil offset mean in general:
https://www.klippel.de/products/klippel-controlled-sound/stabilization.html
DC would not offset voice coil in rest, but I think it will affect the operating point of AC, so comparable to the link above in that sense. If the transducer has developed issue and is not centered, small DC offset might help Or make things worse. Some offset conditions are always present, like suspension stiffness being stiffer to other directin than other and so on, a lot of variables. I think question is what is too much then. If system is never performing near it's limits, then small DC offset isn't that audible I think. You'd hear pop/click with power cycle though.
Not sure how much tweeter is affected, but I'd assume it would similarly somewhat reduce max output and alter distortion characteristics, just like with woofers.
edit. mo Klippel stuff https://www.klippel.de/know-how/mea...splacement-dynamic-offset-the-voice-coil.html
https://www.klippel.de/products/klippel-controlled-sound/stabilization.html
DC would not offset voice coil in rest, but I think it will affect the operating point of AC, so comparable to the link above in that sense. If the transducer has developed issue and is not centered, small DC offset might help
Or make things worse. Some offset conditions are always present, like suspension stiffness being stiffer to other directin than other and so on, a lot of variables. I think question is what is too much then. If system is never performing near it's limits, then small DC offset isn't that audible I think. You'd hear pop/click with power cycle though.Not sure how much tweeter is affected, but I'd assume it would similarly somewhat reduce max output and alter distortion characteristics, just like with woofers.
edit. mo Klippel stuff https://www.klippel.de/know-how/mea...splacement-dynamic-offset-the-voice-coil.html
Last edited:
Nelson Pass did that. At about 20th minute of video, there are actual measurements.
https://www.diyaudio.com/community/threads/outside-the-box.402722/
Hmm, yeah and mentions it is at the driver resonant frequency, and that there is second harmonic shown up (in back EMF essentially).
I'm not pro at this, but trying to stitch pieces of info together: I think suspension non-linearity makes 2nd harmonic, and suspension is involved mostly on the resonant frequency, so perhaps what NP shows in the video is what suspension offset does, and little bit of DC just "biases" the voice coil (suspension) better, just like with the KCS above, shifts operating point of the driver so that distortion reduces. KCS does this automatically, where NP has done it manually, and is somewhat dependent on driver I assume. I mean, perhaps not all drivers benefit? Although all specimen NP showed in the video did, with various amounts. Perhaps it's always fine, most drivers being very similar in their suspension always have the suspension stiffening to same direction? Anyway, interesting stuff, especially if the little DC allowed for "better" amplifier stage, good stuff for DIY folk.
edit.
thinking further, I think suspension non-linearity could dominate (acoustic) distortion around Fs, but not above where motor non-linearities start to dominate just because suspension is dominated by the mass. I think air in the box is also part of the suspension, and mostly involved on the lows. If the NP experiment was repeated with full bandwidth distortion data, I bet THD would drop on the lows only, while it could go higher on highs. If motor properties, like Bl(x) or Le(x) is offset above the black line, then the bias would increase distortion on above the resonant frequency. To reduce distortion whole bandwidth, all Kms(x), Bl(x) and Le(x) need be offset to same direction.
Here is BL offset of the same driver as Kms offset above.
Now, which one to compensate for, suspension or motor offset, or all of them? Depends if the driver is mainly used around it's resonance, or above. On fullrange driver system better get driver where everything offsets to same direction. Anyway, perhaps DC is best tuned observing full bandwidth distortion measurement.
ps. looking at VC test bench speakers, some have asymmetry offset on the same side, while some others don't. Here is fullrange driver that likely benefits from DC on the whole bandwidth. But would this need negative DC?
https://audioxpress.com/article/faitalpro-2fe24-pro-sound-2-5-full-range
I'm not pro at this, but trying to stitch pieces of info together: I think suspension non-linearity makes 2nd harmonic, and suspension is involved mostly on the resonant frequency, so perhaps what NP shows in the video is what suspension offset does, and little bit of DC just "biases" the voice coil (suspension) better, just like with the KCS above, shifts operating point of the driver so that distortion reduces. KCS does this automatically, where NP has done it manually, and is somewhat dependent on driver I assume. I mean, perhaps not all drivers benefit? Although all specimen NP showed in the video did, with various amounts. Perhaps it's always fine, most drivers being very similar in their suspension always have the suspension stiffening to same direction? Anyway, interesting stuff, especially if the little DC allowed for "better" amplifier stage, good stuff for DIY folk.
edit.
thinking further, I think suspension non-linearity could dominate (acoustic) distortion around Fs, but not above where motor non-linearities start to dominate just because suspension is dominated by the mass. I think air in the box is also part of the suspension, and mostly involved on the lows. If the NP experiment was repeated with full bandwidth distortion data, I bet THD would drop on the lows only, while it could go higher on highs. If motor properties, like Bl(x) or Le(x) is offset above the black line, then the bias would increase distortion on above the resonant frequency. To reduce distortion whole bandwidth, all Kms(x), Bl(x) and Le(x) need be offset to same direction.
Here is BL offset of the same driver as Kms offset above.
Now, which one to compensate for, suspension or motor offset, or all of them? Depends if the driver is mainly used around it's resonance, or above. On fullrange driver system better get driver where everything offsets to same direction. Anyway, perhaps DC is best tuned observing full bandwidth distortion measurement.
ps. looking at VC test bench speakers, some have asymmetry offset on the same side, while some others don't. Here is fullrange driver that likely benefits from DC on the whole bandwidth. But would this need negative DC?
https://audioxpress.com/article/faitalpro-2fe24-pro-sound-2-5-full-range
Last edited:
No way around it
All amplifiers have DC offset.
Just some designers have goals to have very very low offset.
Some basic topologies will have low offset as well.
Over time or thermal range, when amplifier gets hot
DC offset can increase.
But it is below the accepted levels.
And accepted levels are grossly high.
But otherwise completely fine.
And as said even dirt basic amplifiers
will have very very low offset.
Much below the accepted levels.
The answer is = it doesn't matter
And much more challenges to building
a speaker system. To overthink things
people swear about mystical capacitors in the signal path.
But normal midrange ,tweeter distortion at high listening levels
and the puff puff of junkie little slot ports is far far far above
absolutely non audible DC offset. Even more audible than old fashioned
cap coupled amplifiers.
There is many ways to improve and achieve hifi.
worrying about DC offset is not one of them
1 watt say at 2.83 volts is far far far higher for the voicecoil
than .001 volts or .01 volts it is below the decimal point
And the number below the decimal = 0 And most listen over a Watt of power.
IT does = Nothing
20 volts probably wake up a speaker lol, but 0 volts ...No
All amplifiers have DC offset.
Just some designers have goals to have very very low offset.
Some basic topologies will have low offset as well.
Over time or thermal range, when amplifier gets hot
DC offset can increase.
But it is below the accepted levels.
And accepted levels are grossly high.
But otherwise completely fine.
And as said even dirt basic amplifiers
will have very very low offset.
Much below the accepted levels.
The answer is = it doesn't matter
And much more challenges to building
a speaker system. To overthink things
people swear about mystical capacitors in the signal path.
But normal midrange ,tweeter distortion at high listening levels
and the puff puff of junkie little slot ports is far far far above
absolutely non audible DC offset. Even more audible than old fashioned
cap coupled amplifiers.
There is many ways to improve and achieve hifi.
worrying about DC offset is not one of them
1 watt say at 2.83 volts is far far far higher for the voicecoil
than .001 volts or .01 volts it is below the decimal point
And the number below the decimal = 0 And most listen over a Watt of power.
IT does = Nothing
20 volts probably wake up a speaker lol, but 0 volts ...No
Last edited:
Yep, perhaps not too audible and concerning. Interesting stuff nevertheless, connects to all kinds of things at least in theory. Take for example people who say they hear absolute polarity of speaker, what if it's due to the DC from amp making the speaker distort more or less depending whether the woofer offset gets better or worse? Perhaps one doesn't hear the polarity as such, but effect of DC on acoustic distortion, perhaps the additional distortion makes more dynamic sound, or less dynamic sound. This would mean, that if one hears absolute polarity perhaps the drivers on ye speakers aren't that good. Or, perhaps they just hear the polarity change I don't know. Anyway, interesting stuff. ps. this could be tested switching polarity either on the input or output of power amp.
Last edited:
Just assumptions, and actual phase would make sense with crossover design.
Polarity is Polarity
The phase of the speaker completely changes many times over the bandwidth.
And isn't important till you get to the crossover point.
Depending on crossover frequency and phase response of the drivers being
summed, you change polarity one way or the other. very simple
Luckily in the modern world the actual behavior of speakers and sound
is rather predictable. And done with math.
Software does most that math for you. Then displays the behavior.
Understanding crossover design and phase response of a driver
and the effects of the baffle. Is much easier to see and understand now
in software. It makes the experience very visual and understood
by computer aided design. Once you have walked through a basic 2 way
in simulation. Center to Center spacing, piston diameter and phase response in crossover
points all make more sense. No need to assume or imagining what it might do.
Just observe what it actually does, build your circuit and enjoy the listen.
Many things not understood and intimidating can be learned faster.
Polarity is Polarity
The phase of the speaker completely changes many times over the bandwidth.
And isn't important till you get to the crossover point.
Depending on crossover frequency and phase response of the drivers being
summed, you change polarity one way or the other. very simple
Luckily in the modern world the actual behavior of speakers and sound
is rather predictable. And done with math.
Software does most that math for you. Then displays the behavior.
Understanding crossover design and phase response of a driver
and the effects of the baffle. Is much easier to see and understand now
in software. It makes the experience very visual and understood
by computer aided design. Once you have walked through a basic 2 way
in simulation. Center to Center spacing, piston diameter and phase response in crossover
points all make more sense. No need to assume or imagining what it might do.
Just observe what it actually does, build your circuit and enjoy the listen.
Many things not understood and intimidating can be learned faster.
Oh for sure. While the properties you mention are the most important ones, and do not change no matter which way the speaker wires are hooked to an amplifier. Hypothesis is the DC stuff in opening post and in Pass video should show up in distortion plot and give more or less distortion depending which way the speaker wires are hooked, assuming:
Well, if distortion was audible anyway anyway, interesting stuff to wonder about.
edit.
anyone have the Faital Pro 2fe24 at hand? There seems to be quite crazy offset, both Kms and BL on positive side
https://audioxpress.com/article/faitalpro-2fe24-pro-sound-2-5-full-range
The article lists a lot of commercial products the driver is used on. Perhaps the products have figured this out and accompanied some DC offset in their DSP? Anyone have any of them and dare to play around? I speculate sound of these products could be a lot better if the DC was used to make the offsets less, perhaps they already do, or perhaps little improvement doesn't mean anything as they likely sound quite distorted anyway, and people just don't care. Unfortunately I don't have any of them... nor time to play around
- driver(s) on the speaker have meaningful offset in their non-linear parameter(s)
- the amplifier has meaningful DC on the output
Well, if distortion was audible anyway
anyway, interesting stuff to wonder about.edit.
anyone have the Faital Pro 2fe24 at hand? There seems to be quite crazy offset, both Kms and BL on positive side
https://audioxpress.com/article/faitalpro-2fe24-pro-sound-2-5-full-range
The article lists a lot of commercial products the driver is used on. Perhaps the products have figured this out and accompanied some DC offset in their DSP? Anyone have any of them and dare to play around? I speculate sound of these products could be a lot better if the DC was used to make the offsets less, perhaps they already do, or perhaps little improvement doesn't mean anything as they likely sound quite distorted anyway, and people just don't care. Unfortunately I don't have any of them... nor time to play around
Last edited:
The speaker moves back n forth regardless.
But a non inverting amplifier should have positive DC offset
and a inverting amplifier would negative DC offset.
Speaker polarity just designates forward or rearward movement.
The overall phase changes over the bandwidth.
Speaker distortion will always be 1% to 3% with direct radiators at listening levels.
Depending on listening level , and all speakers have linear rating
to 70% of magnet BL when they then reach 10% distortion.
So basically any speaker can be driven to 10% distortion or further.
So the significance of 2.83 volts at 1 watt compared to .001 volts of DC offset
is such a waste of time to even ponder.
If .001 volts makes a difference to a speaker its = junk
if reversing polarity improves distortion its = junk
With any crossover in a 2 or 3 way system.
One of the drivers is highly likely to be reversed polarity.
And will need to be to make the system work.
Its just not even worth thinking about or testing in any way shape or form.
All amplifiers have DC offset and all speaker systems
will be reversed or non reversed at some point. Nothing you can
do about it. Can block DC with a capcitor. But then we walk into
the mystical magical world of magic distortion caps. So really a waste of time.
Interesting I agree, but not important to ponder what .001 volts does when the speaker
operates at 2.83 volts or higher. much much higher in high SPL
Sure you can Add the offset. Test a speaker at 2.830 volts then test it at 2.831 volts
But a non inverting amplifier should have positive DC offset
and a inverting amplifier would negative DC offset.
Speaker polarity just designates forward or rearward movement.
The overall phase changes over the bandwidth.
Speaker distortion will always be 1% to 3% with direct radiators at listening levels.
Depending on listening level , and all speakers have linear rating
to 70% of magnet BL when they then reach 10% distortion.
So basically any speaker can be driven to 10% distortion or further.
So the significance of 2.83 volts at 1 watt compared to .001 volts of DC offset
is such a waste of time to even ponder.
If .001 volts makes a difference to a speaker its = junk
if reversing polarity improves distortion its = junk
With any crossover in a 2 or 3 way system.
One of the drivers is highly likely to be reversed polarity.
And will need to be to make the system work.
Its just not even worth thinking about or testing in any way shape or form.
All amplifiers have DC offset and all speaker systems
will be reversed or non reversed at some point. Nothing you can
do about it. Can block DC with a capcitor. But then we walk into
the mystical magical world of magic distortion caps. So really a waste of time.
Interesting I agree, but not important to ponder what .001 volts does when the speaker
operates at 2.83 volts or higher. much much higher in high SPL
Sure you can Add the offset. Test a speaker at 2.830 volts then test it at 2.831 volts
I wonder if you checked out the Nelson Pass video above? he demonstrates DC up to 200mV to get 2nd harmonic reduction up to 50%+ on some drivers, at Fs. I'm trying to think why is it.
I agree very little DC makes no difference, but if the DC was deliberately adjusted to calibrate a driver then it could, as NP data in the video demonstrates. If you check out the Klippel data I've attached (or any other) it's obvious many drivers seem to have voice coil rest position offset regarding to symmetry of the non-linear parameters. Calibrating offset of the voice coil using DC could increase output before a driver hits the 10% distortion. Just like you'd bias your transistors to have most linear swing both ways before too much distortion. No?
I agree very little DC makes no difference, but if the DC was deliberately adjusted to calibrate a driver then it could, as NP data in the video demonstrates. If you check out the Klippel data I've attached (or any other) it's obvious many drivers seem to have voice coil rest position offset regarding to symmetry of the non-linear parameters. Calibrating offset of the voice coil using DC could increase output before a driver hits the 10% distortion. Just like you'd bias your transistors to have most linear swing both ways before too much distortion. No?
Last edited: