@tktran said:
Dial it in with DSP and you're done. I dunno why we like to inflict pain on ourselves with passive parts on low crossover points...
I should add it’s a unique challenge (combo of an art and science) of this hobby. Go to a few events and you’ll see some impre$$ive external crossovers.
That being said, as you referenced for the ‘low crossover points’ I’ve moved to a hybrid approach as I bi-amp my speakers; actively cross between the woofer and the top end, and a passive for the mid/tweet crossover. miniDSP/Legacy Powerbloc4. It gets me to happy hour a bit earlier…
Oh no. I didn’t mean to start a crossover fight. Both have their uses.
In fact I feel like I’m going full circle. After years of having a DSP based system with 6 of amps of equal power, I’m thinking of locking it in with a passive crossover. The mid/tweeter rarely use over 10W
@Wolf said:
I would agree. You can only attenuate with DSP. You can't short out or shunt off.
Actually the fact that you can apply boost with DSP is one of its major advantages over passive networks...
You can split up passive crossovers and amplify each driver individually, allowing for different gain applied to each driver for "boost", DSP is not required for this functionality. Downside of this arrangement is requiring more amplifier channels regardless of whether the crossover is active or passive. IMO it is better to always attenuate the tweeter passively for SNR benefit.
You may have a Cabinet resonance of around about 170 Hz, which shows up at increase of all harmonics
(Pesky pink but also grey and yellow)
To verify this consider taking a nearfield measurement of the woofer or port to look for definite dip/peak around that same ~170Hz frequency.
IIME standing waves causing resonances in cabinets are the most common culprit around frequencies.
At slightly higher frequencies in the lower midrange (250 to 1000Hz), it's due to driver frame fixing to baffle resonances.
A good foam, rubber or sorbothane gasket helps.
At slightly higher frequencies still, check for loose screws (or overtightened screws)
What are we talking about when you say "suppressing a resonance at Fs"?
Ron
PS I used the MiniDSP because I was having trouble getting the woofer response dialed in with a passive crossover and my friend likes to crank things up.
You may have a Cabinet resonance of around about 170 Hz, which shows up at increase of all harmonics
(Pesky pink but also grey and yellow)
To verify this consider taking a nearfield measurement of the woofer or port to look for definite dip/peak around that same ~170Hz frequency.
IIME standing waves causing resonances in cabinets are the most common culprit around frequencies.
At slightly higher frequencies in the lower midrange (250 to 1000Hz), it's due to driver frame fixing to baffle resonances.
A good foam, rubber or sorbothane gasket helps.
At slightly higher frequencies still, check for loose screws (or overtightened screws)
Yeah, this doesn't use my usual hyper-bracing approach. I wanted it to be reproducible to a degree. I doubt anyone will build this due to cost/performance but still...
Also, no matter what I am working on I get a similar peak in distortion in my measurements, might be compounded by the room. My mic won't do adequate nearfield distortion without clipping or I would be doing that.
Agreed.
Yet while I generally use passive for mid/ tw, the HD version of the miniDSP can target to +/- 1Hz, down to -16 dB (times two if you need it at the driver and input sections) and can go as narrow as a 50 Q. Not quite grounded... but a scalpel of a tool.
@jr@mac said:
The impedance peak at Fs. They need to be squashed on a lot of drivers.
This is sort of re-iterating what RonE wrote above, but... I had to laugh when I saw you said that "DSP cannot do this" RE:
@jr@mac said:
'suppressing a resonance at Fs is not something DSP can do'.
With DSP it's a complete non-issue because the impedance of the driver vs frequency doesn't matter at all. That's one of the joys, or freedoms, afforded by DSP. I do not measure driver impedance, ever. All that matters is what the driver's acoustic output is doing, and how I can use that in my design.
The same goes for your statement that
@jr@mac said:
"They also dont stop content from exciting the driver at it's Fs - undesirable for many tweeters."
You are completely missing the point, or so it seems. The driver impedance has no effect on, or influence over, what the crossover filters are doing - zero. At Fs, or any frequency.
I don't want to get into a DSP vs passive crossover pissing war, just want to point out what I see as an incorrect assumption...
It seems to be a relative misunderstanding amongst the active/DSP users with regards to Fs resonance suppression (or even upper end breakup) and the why and how it is applied and needed. I'm really trying hard to not sound insulting or that I'm belittling the active xover designer, as I know that in particularly Charlie knows what he is doing in that regard.
1. DSP cannot suppress an impedance resonance by applying output attenuation as it typically is used. You cannot short out or shunt off a parallel compensation using DSP. This is factual.
2. This issue with DSP/active is not about xover interaction and impedance. This is about Fs compensation, or the other end of shorting out an upper breakup. This is not involving the load VS frequency rolloff and interaction. Compensation is not always about rolloff.
3. This is ALL ABOUT the impedance of the driver and the effect of the driver's inherent resonances on the driver's response and not on the xover and how it affects the response. All resonances occur with some kind or form of energy storage and a corresponding impedance bump, no matter how small. You can attenuate to your heart's desire and try and kill the issue this way, but it does not always work by attenuation. That bump is still present, and rears it's ugly head in the response because of it. In some cases you have to short out the bump to minimize its footprint and not sound buzzy, nasal, or piercing. Just attenuating it does not prevent the driver from seeing the offensive signal and self resonating to where HD is higher in the response than necessary.
How does one identify the need for a Fs resonance shunt, Wolf? (I'm trying to understand).
If one avoids passing signal containing frequencies at or near Fs to the driver (steep slopes) using DSP, does this not prevent the problem you identify?
Back on topic, not that I didn't enjoy the conversation but I really think someone should start a thread on active vs passive and duke it out over there.
@Steve_Lee said:
How does one identify the need for a Fs resonance shunt, Wolf? (I'm trying to understand).
If one avoids passing signal containing frequencies at or near Fs to the driver (steep slopes) using DSP, does this not prevent the problem you identify?
Let's say that you don't pass the frequency that is offensive, but you really can't say that. Said frequency may be -50dB down from nominal, sound about right? That frequency still marginally will excite the driver because it is present, even at -50dB, because the output at resonance is easier to create. Upper harmonics will also still be there, and likely in-band and very audible. Remember that suppression minimum is -25dB, good is -40dB, and gone is -50dB. You can roll it off, but expect peaking HD to come through.
One more thing- you have to have a voltage divider upstream for a parallel/conjugate/shunt circuit to do anything more than affect the impedance passively. If you want the frequency response benefits, there has to also be a series component before such a circuit. Such a thing as a series cap is usually used to prevent turn on pops in tweeters running active, so in thus way it can serve a dual purpose.
JR, sorry for this sidetrack, but I feel information can be a source of knowledge. This was not intended to be a 'pissing match' as far as I can see it. Maybe both parties can learn something from my go arounds with bcodemz a good 5 years ago now on this subject over on PETT should they want to delve further.
@Wolf said:
It seems to be a relative misunderstanding amongst the active/DSP users with regards to Fs resonance suppression (or even upper end breakup) and the why and how it is applied and needed. I'm really trying hard to not sound insulting or that I'm belittling the active xover designer, as I know that in particularly Charlie knows what he is doing in that regard.
1. DSP cannot suppress an impedance resonance by applying output attenuation as it typically is used. You cannot short out or shunt off a parallel compensation using DSP. This is factual.
2. This issue with DSP/active is not about xover interaction and impedance. This is about Fs compensation, or the other end of shorting out an upper breakup. This is not involving the load VS frequency rolloff and interaction. Compensation is not always about rolloff.
3. This is ALL ABOUT the impedance of the driver and the effect of the driver's inherent resonances on the driver's response and not on the xover and how it affects the response. All resonances occur with some kind or form of energy storage and a corresponding impedance bump, no matter how small. You can attenuate to your heart's desire and try and kill the issue this way, but it does not always work by attenuation. That bump is still present, and rears it's ugly head in the response because of it. In some cases you have to short out the bump to minimize its footprint and not sound buzzy, nasal, or piercing. Just attenuating it does not prevent the driver from seeing the offensive signal and self resonating to where HD is higher in the response than necessary.
OK. Thanks for your thoughts Ben. I don't find what you are saying insulting, but thanks for allowing for that possibility. I see the point you are trying to make about a network in parallel with the driver and that this is not something that can be duplicated by manipulating the input power (e.g. with DSP or with a passive series network).
Personally I don't feel that the impedance peak resulting from the mechanical resonance of the drivers (meaning at Fs) is affecting the other parts of the passband. The energy storage is not broad band, it is by definition stored in the resonance at the resonance frequency. Other frequencies away from the peak are not coupled to it except through HD and IMD perhaps.
If you can explain in some detail about how the passive shunt is doing something special that I am not seeing, please do! I would love to learn about that. I have to admit that I am just off doing my DSP thing and because of that I do not really give passive crossovers much thought, so I may be missing some key point or what not.
Comments
I would agree. You can only attenuate with DSP. You can't short out or shunt off.
InDIYana Event Website
I should add it’s a unique challenge (combo of an art and science) of this hobby. Go to a few events and you’ll see some impre$$ive external crossovers.
That being said, as you referenced for the ‘low crossover points’ I’ve moved to a hybrid approach as I bi-amp my speakers; actively cross between the woofer and the top end, and a passive for the mid/tweet crossover. miniDSP/Legacy Powerbloc4. It gets me to happy hour a bit earlier…
BoM
Not what I would call value oriented. I used primarily surplus components, or stuff I bought used.
Sounds pretty good, might be one minor tweek yet to come, however.
What are you talking about? Maybe the Tang Band lady should have picked another...
LMT
Actually the fact that you can apply boost with DSP is one of its major advantages over passive networks...
Oh no. I didn’t mean to start a crossover fight. Both have their uses.
In fact I feel like I’m going full circle. After years of having a DSP based system with 6 of amps of equal power, I’m thinking of locking it in with a passive crossover. The mid/tweeter rarely use over 10W
And I want 4 amps back for other projects!
You can split up passive crossovers and amplify each driver individually, allowing for different gain applied to each driver for "boost", DSP is not required for this functionality. Downside of this arrangement is requiring more amplifier channels regardless of whether the crossover is active or passive. IMO it is better to always attenuate the tweeter passively for SNR benefit.
1M measurements.
Looks good!
You may have a Cabinet resonance of around about 170 Hz, which shows up at increase of all harmonics
(Pesky pink but also grey and yellow)
To verify this consider taking a nearfield measurement of the woofer or port to look for definite dip/peak around that same ~170Hz frequency.
IIME standing waves causing resonances in cabinets are the most common culprit around frequencies.
At slightly higher frequencies in the lower midrange (250 to 1000Hz), it's due to driver frame fixing to baffle resonances.
A good foam, rubber or sorbothane gasket helps.
At slightly higher frequencies still, check for loose screws (or overtightened screws)
I have given a DSP system to a friend, but he's a genius at electronics. We both loved the way it turned out.
https://diy.midwestaudio.club/discussion/1678/charlies-mcm-project/p1
What are we talking about when you say "suppressing a resonance at Fs"?
Ron
PS I used the MiniDSP because I was having trouble getting the woofer response dialed in with a passive crossover and my friend likes to crank things up.
The impedance peak at Fs. They need to be squashed on a lot of drivers.
Yeah, this doesn't use my usual hyper-bracing approach. I wanted it to be reproducible to a degree. I doubt anyone will build this due to cost/performance but still...
Also, no matter what I am working on I get a similar peak in distortion in my measurements, might be compounded by the room. My mic won't do adequate nearfield distortion without clipping or I would be doing that.
Active systems don't have to deal with the interaction between the passive crossover and the driver impedance.
Yes, I know that.
They also dont stop content from exciting the driver at it's Fs - undesirable for many tweeters.
Agreed.
InDIYana Event Website
Agreed.
Yet while I generally use passive for mid/ tw, the HD version of the miniDSP can target to +/- 1Hz, down to -16 dB (times two if you need it at the driver and input sections) and can go as narrow as a 50 Q. Not quite grounded... but a scalpel of a tool.
This is sort of re-iterating what RonE wrote above, but... I had to laugh when I saw you said that "DSP cannot do this" RE:
With DSP it's a complete non-issue because the impedance of the driver vs frequency doesn't matter at all. That's one of the joys, or freedoms, afforded by DSP. I do not measure driver impedance, ever. All that matters is what the driver's acoustic output is doing, and how I can use that in my design.
The same goes for your statement that
You are completely missing the point, or so it seems. The driver impedance has no effect on, or influence over, what the crossover filters are doing - zero. At Fs, or any frequency.
I don't want to get into a DSP vs passive crossover pissing war, just want to point out what I see as an incorrect assumption...
This is not what JR and I are saying.
InDIYana Event Website
OK, sorry if I misunderstood. What are you saying exactly about eg. suppressing resonance at Fs, or if not that please let me know. Thanks.
It seems to be a relative misunderstanding amongst the active/DSP users with regards to Fs resonance suppression (or even upper end breakup) and the why and how it is applied and needed. I'm really trying hard to not sound insulting or that I'm belittling the active xover designer, as I know that in particularly Charlie knows what he is doing in that regard.
1. DSP cannot suppress an impedance resonance by applying output attenuation as it typically is used. You cannot short out or shunt off a parallel compensation using DSP. This is factual.
2. This issue with DSP/active is not about xover interaction and impedance. This is about Fs compensation, or the other end of shorting out an upper breakup. This is not involving the load VS frequency rolloff and interaction. Compensation is not always about rolloff.
3. This is ALL ABOUT the impedance of the driver and the effect of the driver's inherent resonances on the driver's response and not on the xover and how it affects the response. All resonances occur with some kind or form of energy storage and a corresponding impedance bump, no matter how small. You can attenuate to your heart's desire and try and kill the issue this way, but it does not always work by attenuation. That bump is still present, and rears it's ugly head in the response because of it. In some cases you have to short out the bump to minimize its footprint and not sound buzzy, nasal, or piercing. Just attenuating it does not prevent the driver from seeing the offensive signal and self resonating to where HD is higher in the response than necessary.
InDIYana Event Website
How does one identify the need for a Fs resonance shunt, Wolf? (I'm trying to understand).
If one avoids passing signal containing frequencies at or near Fs to the driver (steep slopes) using DSP, does this not prevent the problem you identify?
Wolf explains this perfectly.
Back on topic, not that I didn't enjoy the conversation but I really think someone should start a thread on active vs passive and duke it out over there.
I am working on the write-up for these currently.
Let's say that you don't pass the frequency that is offensive, but you really can't say that. Said frequency may be -50dB down from nominal, sound about right? That frequency still marginally will excite the driver because it is present, even at -50dB, because the output at resonance is easier to create. Upper harmonics will also still be there, and likely in-band and very audible. Remember that suppression minimum is -25dB, good is -40dB, and gone is -50dB. You can roll it off, but expect peaking HD to come through.
InDIYana Event Website
I started another thread, guys - lets move over there - I transferred some of the discussion already - thanks (sorry, JR).
https://diy.midwestaudio.club/discussion/2157/passive-resonance-suppression-of-drivers-around-fs-vs-dsp-discussion/p1
One more thing- you have to have a voltage divider upstream for a parallel/conjugate/shunt circuit to do anything more than affect the impedance passively. If you want the frequency response benefits, there has to also be a series component before such a circuit. Such a thing as a series cap is usually used to prevent turn on pops in tweeters running active, so in thus way it can serve a dual purpose.
JR, sorry for this sidetrack, but I feel information can be a source of knowledge. This was not intended to be a 'pissing match' as far as I can see it. Maybe both parties can learn something from my go arounds with bcodemz a good 5 years ago now on this subject over on PETT should they want to delve further.
InDIYana Event Website
OK. Thanks for your thoughts Ben. I don't find what you are saying insulting, but thanks for allowing for that possibility. I see the point you are trying to make about a network in parallel with the driver and that this is not something that can be duplicated by manipulating the input power (e.g. with DSP or with a passive series network).
Personally I don't feel that the impedance peak resulting from the mechanical resonance of the drivers (meaning at Fs) is affecting the other parts of the passband. The energy storage is not broad band, it is by definition stored in the resonance at the resonance frequency. Other frequencies away from the peak are not coupled to it except through HD and IMD perhaps.
If you can explain in some detail about how the passive shunt is doing something special that I am not seeing, please do! I would love to learn about that. I have to admit that I am just off doing my DSP thing and because of that I do not really give passive crossovers much thought, so I may be missing some key point or what not.
Apologies is this is too much thread hijacking.
Writeup attached. Let me know if it works.
Loaded perfectly... reading through it now. I REALLY appreciate when people do full document write-ups!