Allae Crossover "Mod"
Posted by: Greg Beatty on 29 May 2004
My Allaes have sounded great and I have no regrets of the purchase. But...
...in my room, they don't quite do the "band is in the room" thing that some other speakers I home demmed were able to do. I believe this is a quantity of bass issue as the speakers that did the "live" trick were bassier.
To find out more, I have been exploring room modes, moving the speakers around, and finally got a Radio Shack analog sound pressure level (SPL) meter. Used in conjunction with the bass, midrange, and treble warble tones on Stereophile's Test CD3 and the published-on-the-net correction values for the meter, I measured the in-room response using my Sony ES series CD player and Panasonic DVD player as sources. Both sources measured the same.
The response graphs I got were nearly identical to what Stereophile reported for the NBL. Not saying Allaes are NBLs, but the Allaes definitly have the same frequency response signature. A drop off of 3db starting at 500hz on down and a rise in the mids and treble region.
A bit more research turned up the notion of baffle step. Baffle step is drop in forward radiation of bass frequencies that "wrap around" the baffle of the speaker. The frequency of the step is determined by the width of the baffle - narrower baffles beginning the step at a higher frequency than wider baffles. So I plugged the width of my Allaes into the equation and - wouldn't you know it - the baffle step starts at 500hz!!!
The usual remedy is for the speaker's crossover to include components to compensate for baffle step. The maximum decrease in bass is 6db, which is subjectively half as loud. Naim assumes (or so I have read - moderators feel free to correct if I am wrong here) that the baffle step bass drop will be addressed by placing the speaker near or flat against a *solid* wall and the lost bass will be redirected back into the room. This is the "boundary reinforcement" that we read about from time to time.
Well guess what. I moved my Allaes flat against the wall...took measurements in different places in the room...and even placed the Allaes *in the corners* trying to get additional boundary reinforcement from the side walls.
No dice. I moved some of the resonant frequencies around a bit, but the overall 3db drop remained. Oh, and I don't know if the Allaes have partial baffle step correction and so that is why I measure -3db instead of -6db or if they have no correction and my walls are partially reinforcing the bass. In either case, bass to match the mids and treble is impossible in my room without some adjustment.
A bit more research and I find this:
How To Correct Baffle Step
which explains a very simple crossover circuit to correct baffle step. And the "mod" is in quotes because the simple circuit is placed in between the speaker wire and the crossover - no need to actually change the Naim crossover, just need to do the baffle step correction prior to the speaker.
The "mod" consists of putting an inductor and a resistor in between the speaker wire and the speaker. Dead easy.
So I have ordered two 1.5mH inductors and I already have 8 ohm, high-wattage resistors. So for less than $20 including shipping I will be able to add baffle step correction to my Allaes. The inductors are probably not very good at $5 each, but I want to try the concept before shelling out for better inductors.
I see no reason why this couldn't be done with any speaker as the circuit is standard when building crossovers.
I will write back when the baffle step correction is up and running.
- GregB
Insert Witty Signature Line Here
...in my room, they don't quite do the "band is in the room" thing that some other speakers I home demmed were able to do. I believe this is a quantity of bass issue as the speakers that did the "live" trick were bassier.
To find out more, I have been exploring room modes, moving the speakers around, and finally got a Radio Shack analog sound pressure level (SPL) meter. Used in conjunction with the bass, midrange, and treble warble tones on Stereophile's Test CD3 and the published-on-the-net correction values for the meter, I measured the in-room response using my Sony ES series CD player and Panasonic DVD player as sources. Both sources measured the same.
The response graphs I got were nearly identical to what Stereophile reported for the NBL. Not saying Allaes are NBLs, but the Allaes definitly have the same frequency response signature. A drop off of 3db starting at 500hz on down and a rise in the mids and treble region.
A bit more research turned up the notion of baffle step. Baffle step is drop in forward radiation of bass frequencies that "wrap around" the baffle of the speaker. The frequency of the step is determined by the width of the baffle - narrower baffles beginning the step at a higher frequency than wider baffles. So I plugged the width of my Allaes into the equation and - wouldn't you know it - the baffle step starts at 500hz!!!
The usual remedy is for the speaker's crossover to include components to compensate for baffle step. The maximum decrease in bass is 6db, which is subjectively half as loud. Naim assumes (or so I have read - moderators feel free to correct if I am wrong here) that the baffle step bass drop will be addressed by placing the speaker near or flat against a *solid* wall and the lost bass will be redirected back into the room. This is the "boundary reinforcement" that we read about from time to time.
Well guess what. I moved my Allaes flat against the wall...took measurements in different places in the room...and even placed the Allaes *in the corners* trying to get additional boundary reinforcement from the side walls.
No dice. I moved some of the resonant frequencies around a bit, but the overall 3db drop remained. Oh, and I don't know if the Allaes have partial baffle step correction and so that is why I measure -3db instead of -6db or if they have no correction and my walls are partially reinforcing the bass. In either case, bass to match the mids and treble is impossible in my room without some adjustment.
A bit more research and I find this:
How To Correct Baffle Step
which explains a very simple crossover circuit to correct baffle step. And the "mod" is in quotes because the simple circuit is placed in between the speaker wire and the crossover - no need to actually change the Naim crossover, just need to do the baffle step correction prior to the speaker.
The "mod" consists of putting an inductor and a resistor in between the speaker wire and the speaker. Dead easy.
So I have ordered two 1.5mH inductors and I already have 8 ohm, high-wattage resistors. So for less than $20 including shipping I will be able to add baffle step correction to my Allaes. The inductors are probably not very good at $5 each, but I want to try the concept before shelling out for better inductors.
I see no reason why this couldn't be done with any speaker as the circuit is standard when building crossovers.
I will write back when the baffle step correction is up and running.
- GregB
Insert Witty Signature Line Here
Let us know what it does to the all-important phase response... 
By all means give it a go, but I reckon your foot won't be tapping to the music as much.
Andrew
Andrew Randle
Linn Binn Sinner
By all means give it a go, but I reckon your foot won't be tapping to the music as much.
Andrew
Andrew Randle
Linn Binn Sinner
Mr. Tibbs -
The inductor and resistor will be in parallel with each other. The inductor's "resistance" is frequency dependent - providing much more resistance to higher frequencies. This means that the low freqs go through the inductor (with low resistance) and the higher freqs go through the resistor and are reduced accordingly. The value of the resistor, relative to the impedance of the speaker, determines how much reduction there is. The value of the inductor determines the frequency that the reduction starts.
Naim have assumed that the walls will provide reinforcement of the bass at and below the baffle step frequency and I have a shoddy-build American home with walls that do not do this. So some correction seems in order. Hope that helps to clarify what I'm attempting to do.
If it turns out for the worst, oh well. It will have been a fun experiment
- GregB
Insert Witty Signature Line Here
The inductor and resistor will be in parallel with each other. The inductor's "resistance" is frequency dependent - providing much more resistance to higher frequencies. This means that the low freqs go through the inductor (with low resistance) and the higher freqs go through the resistor and are reduced accordingly. The value of the resistor, relative to the impedance of the speaker, determines how much reduction there is. The value of the inductor determines the frequency that the reduction starts.
Naim have assumed that the walls will provide reinforcement of the bass at and below the baffle step frequency and I have a shoddy-build American home with walls that do not do this. So some correction seems in order. Hope that helps to clarify what I'm attempting to do.
If it turns out for the worst, oh well. It will have been a fun experiment
- GregB
Insert Witty Signature Line Here
James -
I'm certainly no expert at this (duh) and am just trying to interpret what I read and improve my hifi a bit
My understanding is that the circuit is a 6db per octave filter that works over a one octave range and then flattens out.
If 8ohm is too much, and I suspect it will be as 8ohm is for a 6db correction, I plan to add a second 8ohm in parallel to reduce the resistance to 4ohm and bring the correction to 3db, which is the measured bass loss I have.
It seems the baffle step correction is doable...even if I have the present implementation wrong. It also seems that the crossover is the place to make the correction and not line-level EQ, although that is another alternative if need be.
I will report back
- GregB
Insert Witty Signature Line Here
I'm certainly no expert at this (duh) and am just trying to interpret what I read and improve my hifi a bit
My understanding is that the circuit is a 6db per octave filter that works over a one octave range and then flattens out.
If 8ohm is too much, and I suspect it will be as 8ohm is for a 6db correction, I plan to add a second 8ohm in parallel to reduce the resistance to 4ohm and bring the correction to 3db, which is the measured bass loss I have.
It seems the baffle step correction is doable...even if I have the present implementation wrong. It also seems that the crossover is the place to make the correction and not line-level EQ, although that is another alternative if need be.
I will report back
- GregB
Insert Witty Signature Line Here
Initial Results
Well, I had a package waiting for me when I got home from work yesterday and it was the inductors. Getting them wired up with the resistors was a bit of work, owing mainly to being home with my 1-year-old. Always a good time to play with the hifi
But I couldn't wait and managed to get the parts installed. And...
...It works!!!
More bass for sure and, on quick listening, is seems to be coming in at about the right frequency. The volume control needs to go higher to get the same listening level, indicating a drop in efficiency.
More quick listening (with a teething 1-year-old wrapped around my legs) and it sounds a bit thick overall.
With one 8-ohm resistor in the circuit the increase in bass (well, decrease in everything else really) sounds a bit much. This was expected as the Allaes are 6-ohm speakers and an 8-ohm resistor would be expected to give at least 6db of bass boost. I only need 3db of boost. Of course, the speakers are near the room corners, so moving the speakers is an option.
I took measurements using my SPL meter with test tones and verified that the bass was on the warm side, with the speakers in their current position.
So at the last minute of the evening, in goes a second 8-ohm resistor in parallel with the first on each speaker. This reduces the resistance to 4-ohms.
A quick listen and sounds better. And the volume doesn't need to go as high.
This morning I popped in a CD of Dvorak's 9th symphony that I have listened to countless times in the past. In my current setup and room, I have not been able to enjoy this CD. The orchestra is too lean - it just isn't credible.
About 2/3 way into the first movement I realized I didn't want to leave. The balance was better and I was liking it!
Tonight or over the weekend I will try to get a free five minutes to make measurements on the new setup and see where I am.
And about this phase business...
Inductors shift phase by 90 degrees or so and this shift would only be to the 500hz or so and below that are going through the inductor.
How would I measure this? And what, if any, would the sonic penalty be?
- GregB
Insert Witty Signature Line Here
Well, I had a package waiting for me when I got home from work yesterday and it was the inductors. Getting them wired up with the resistors was a bit of work, owing mainly to being home with my 1-year-old. Always a good time to play with the hifi
But I couldn't wait and managed to get the parts installed. And...
...It works!!!
More bass for sure and, on quick listening, is seems to be coming in at about the right frequency. The volume control needs to go higher to get the same listening level, indicating a drop in efficiency.
More quick listening (with a teething 1-year-old wrapped around my legs) and it sounds a bit thick overall.
With one 8-ohm resistor in the circuit the increase in bass (well, decrease in everything else really) sounds a bit much. This was expected as the Allaes are 6-ohm speakers and an 8-ohm resistor would be expected to give at least 6db of bass boost. I only need 3db of boost. Of course, the speakers are near the room corners, so moving the speakers is an option.
I took measurements using my SPL meter with test tones and verified that the bass was on the warm side, with the speakers in their current position.
So at the last minute of the evening, in goes a second 8-ohm resistor in parallel with the first on each speaker. This reduces the resistance to 4-ohms.
A quick listen and sounds better. And the volume doesn't need to go as high.
This morning I popped in a CD of Dvorak's 9th symphony that I have listened to countless times in the past. In my current setup and room, I have not been able to enjoy this CD. The orchestra is too lean - it just isn't credible.
About 2/3 way into the first movement I realized I didn't want to leave. The balance was better and I was liking it!
Tonight or over the weekend I will try to get a free five minutes to make measurements on the new setup and see where I am.
And about this phase business...
Inductors shift phase by 90 degrees or so and this shift would only be to the 500hz or so and below that are going through the inductor.
How would I measure this? And what, if any, would the sonic penalty be?
- GregB
Insert Witty Signature Line Here
"James might have some suggestions for you, but if it sounds good, isn't that the bottom line?"
Of course But I'm still curious...and if there is a way to do the EQ without the phase "penalty", I would be interested. I suspect other makes of speakers have the inductor/resistor in there to correct for baffle step loss and then either are off in the phase area or correct it elsewhere in the crossover.
James - acoustic, I guess. And I guess without the equipment it isn't going to happen. I did find some lab work reported on the 'net that compared theoretical to measured inductor phase shifts. It seems the simple predicted 90 degree shift is often not what is observed.
I wonder, given all the chatter here about Naim bass, if Naim would ever consider offering a baffle step circuit as an option or even as standard with their speakers. The designs assume wall reinforcement that is often not lacking, especially in U.S. homes.
- GregB
Insert Witty Signature Line Here
Of course
But I'm still curious...and if there is a way to do the EQ without the phase "penalty", I would be interested. I suspect other makes of speakers have the inductor/resistor in there to correct for baffle step loss and then either are off in the phase area or correct it elsewhere in the crossover.James - acoustic, I guess. And I guess without the equipment it isn't going to happen. I did find some lab work reported on the 'net that compared theoretical to measured inductor phase shifts. It seems the simple predicted 90 degree shift is often not what is observed.
I wonder, given all the chatter here about Naim bass, if Naim would ever consider offering a baffle step circuit as an option or even as standard with their speakers. The designs assume wall reinforcement that is often not lacking, especially in U.S. homes.
- GregB
Insert Witty Signature Line Here
Update
No idea why this was moved to the Padded Cell, but anyway...
After living with "the mod" at 4 ohms for a week, I decided it was just *a bit* plump. So knocked it down to 3 ohms and did the wiring up in a much more secure fashion that provides better contact. Did this Friday and have had it this way over the long weekend.
Also, while doing the better wiring, I took the opportunity to plug the Nait 5 straight back into the Allaes. Had it this way for half a day as I worked on the better mod wiring (again with a 1-year-old in tow) and...
No Way!!! Sounds like the bottom of everything is just missing. Clearly not a case of initially liking the mod and, upon hearing the stock setup, preferring the stock setup. I have had that experience with several tweaks in the past and the stock setup usually wins.
Not this time...well, not in my bass sump room at least.
Popped in the 3-ohm baffle step circuit and Woo Hoo!!! Back in business. And it sounds even better with the improved connections between the components
Sounded just fabulous over the weekend. My wife and I kept digging out old CDs. Much easier to not notice the sound and just enjoy the music. I will re-evaluate in a week or so and see if 2 ohms are worth a go. The bass doesn’t need much of a boost, but it makes a significant difference.
- GregB
Insert Witty Signature Line Here
No idea why this was moved to the Padded Cell, but anyway...
After living with "the mod" at 4 ohms for a week, I decided it was just *a bit* plump. So knocked it down to 3 ohms and did the wiring up in a much more secure fashion that provides better contact. Did this Friday and have had it this way over the long weekend.
Also, while doing the better wiring, I took the opportunity to plug the Nait 5 straight back into the Allaes. Had it this way for half a day as I worked on the better mod wiring (again with a 1-year-old in tow) and...
No Way!!! Sounds like the bottom of everything is just missing. Clearly not a case of initially liking the mod and, upon hearing the stock setup, preferring the stock setup. I have had that experience with several tweaks in the past and the stock setup usually wins.
Not this time...well, not in my bass sump room at least.
Popped in the 3-ohm baffle step circuit and Woo Hoo!!! Back in business. And it sounds even better with the improved connections between the components
Sounded just fabulous over the weekend. My wife and I kept digging out old CDs. Much easier to not notice the sound and just enjoy the music. I will re-evaluate in a week or so and see if 2 ohms are worth a go. The bass doesn’t need much of a boost, but it makes a significant difference.
- GregB
Insert Witty Signature Line Here
Johan -
Thank you for the input on the brick wall idea. I have not abandoned this idea and maybe someday...but not while there is a 1-year-old who could pull the setup down on himself. Glad to hear it works
And thank you for the phase shift info. The working range is supposed to be 1 octave and then the response flattens out again (the highs go through the resistors and avoid the inductor altogether).
Alex -
Better or worst than tone controls or an EQ...I don't know.
The Good Thing is that I have it kicking in at the correct frequency and doing the job with far fewer components than an EQ. The EQ would let you adjust in real time and also adjust the mids and highs to taste. More components and more connections, but I would not claim that my solution is definitly better. I also like avoiding another box in the stack.
(I'm better off not having access to real time adjustments - too tempting!) A tone control on a receiver may be at the right frequency or might not.
And I improved the quality by connecting each comonent more ridgedly and with greater contact area than before. The early setup was twisted together to allow for easy changing of the indcutor and resistor values. Now than I am close, I don't mind doing it up better.
Doing it up better also added a nudge of bass, so I may end up trying 2-ohms after all.
Are you using Subs, BTW?
- Greg
Insert Witty Signature Line Here
Thank you for the input on the brick wall idea. I have not abandoned this idea and maybe someday...but not while there is a 1-year-old who could pull the setup down on himself. Glad to hear it works
And thank you for the phase shift info. The working range is supposed to be 1 octave and then the response flattens out again (the highs go through the resistors and avoid the inductor altogether).
Alex -
Better or worst than tone controls or an EQ...I don't know.
The Good Thing is that I have it kicking in at the correct frequency and doing the job with far fewer components than an EQ. The EQ would let you adjust in real time and also adjust the mids and highs to taste. More components and more connections, but I would not claim that my solution is definitly better. I also like avoiding another box in the stack.
(I'm better off not having access to real time adjustments - too tempting!) A tone control on a receiver may be at the right frequency or might not.
And I improved the quality by connecting each comonent more ridgedly and with greater contact area than before. The early setup was twisted together to allow for easy changing of the indcutor and resistor values. Now than I am close, I don't mind doing it up better.
Doing it up better also added a nudge of bass, so I may end up trying 2-ohms after all.
Are you using Subs, BTW?
- Greg
Insert Witty Signature Line Here
Alex -
"Somehow my answer was put beneath yours but I didn't see yours."
It took me quite some time to compose the post. My guess is the forum software takes the time stamp from when the Reply window is opened rather than when the Post Now button is pressed. So you posted while I was writing
"Strange. OK, last questions: Why the positive lead? Doesn't the negative lead carry the actual signal?"
I was just following directions in using the positive lead. As far as I know, the signal is AC (alternating current) so the idea of the signal coming from the positive or negative lead is misleading (pun intended). That is, if you hooked up your speakers backwards they would work just fine as long as both speakers were wired the same way. Some people say they can hear the difference…I doubt I could.
"Last but not least, where is the dB in the equation? More resistance is a higher compensation?"
The reference I have states that 6 db of compensation will result when the resistance equals the impedance of the speaker. The mids and highs are going through the resistors and are depressed by the resistors...so decreasing the resistance will lessen the baffle step compensation and vice versa.
"Meaning more Ohms is more bass but less treble and mids?" - Yes!
"And when calculating, don't you have to adjust your inductor to whatever changes you make to your resistor?"
I've wondered about this...but have not found it to make a significant practical difference. I have only used 1.5 mH inductors.
"Or are you not trying to do this by formula and just listening?"
I measured the frequency response of the system, inserted the circuits (one for each speaker of course) and measured again. Adjustments to the resistance are based on listening. I give it a few days so I have time to hear the tele, hear a few CDs and DVDs...
Then I decide if it is too thick sounding or still too thin and adjust the resistance. I measure after changing the resistance as well just to quantify the effect. Something I like to do (I'm a measurement freak) but that is not necessary. As they say, if it sounds good...
And, really, and equalizer would be an easier way to do this. I just like the challenge and believe that the end result may be better by using fewer components and using higher grade components than one would find in a cheap EQ.
- GregB
Insert Witty Signature Line Here
"Somehow my answer was put beneath yours but I didn't see yours."
It took me quite some time to compose the post. My guess is the forum software takes the time stamp from when the Reply window is opened rather than when the Post Now button is pressed. So you posted while I was writing
"Strange. OK, last questions: Why the positive lead? Doesn't the negative lead carry the actual signal?"
I was just following directions in using the positive lead. As far as I know, the signal is AC (alternating current) so the idea of the signal coming from the positive or negative lead is misleading (pun intended). That is, if you hooked up your speakers backwards they would work just fine as long as both speakers were wired the same way. Some people say they can hear the difference…I doubt I could.
"Last but not least, where is the dB in the equation? More resistance is a higher compensation?"
The reference I have states that 6 db of compensation will result when the resistance equals the impedance of the speaker. The mids and highs are going through the resistors and are depressed by the resistors...so decreasing the resistance will lessen the baffle step compensation and vice versa.
"Meaning more Ohms is more bass but less treble and mids?" - Yes!
"And when calculating, don't you have to adjust your inductor to whatever changes you make to your resistor?"
I've wondered about this...but have not found it to make a significant practical difference. I have only used 1.5 mH inductors.
"Or are you not trying to do this by formula and just listening?"
I measured the frequency response of the system, inserted the circuits (one for each speaker of course) and measured again. Adjustments to the resistance are based on listening. I give it a few days so I have time to hear the tele, hear a few CDs and DVDs...
Then I decide if it is too thick sounding or still too thin and adjust the resistance. I measure after changing the resistance as well just to quantify the effect. Something I like to do (I'm a measurement freak) but that is not necessary. As they say, if it sounds good...
And, really, and equalizer would be an easier way to do this. I just like the challenge and believe that the end result may be better by using fewer components and using higher grade components than one would find in a cheap EQ.
- GregB
Insert Witty Signature Line Here
Alex -
Thank you for the link. The Edge suggests that the resistor be equal to the impedence of the speaker (set R1 = 7 ohms and the recommended values for R2 will be 7 ohms for 6db of compensation). I am running at 7 ohms now. My earlier tests with 8 ohms were not correct due to one speaker being wired up improperly. I'm likely to end up at 6 or 7 ohms.
The Edge suggests a higher value inductor to have the bass lift coming in at a lower frequency. The value I use is based on 500hz as the -3db point.
You can always experiment or just get an equalizer and set it that way. It is nice, either way, to be able to measure what is happening. You can use the Radio Shack meter with test tones and a table of correction values (I have a spreadsheet for this that graphs the results as well) or an equalizer with a spectrum analyzer built in.
Let me know how it goes!
- Greg
Insert Witty Signature Line Here
Thank you for the link. The Edge suggests that the resistor be equal to the impedence of the speaker (set R1 = 7 ohms and the recommended values for R2 will be 7 ohms for 6db of compensation). I am running at 7 ohms now. My earlier tests with 8 ohms were not correct due to one speaker being wired up improperly. I'm likely to end up at 6 or 7 ohms.
The Edge suggests a higher value inductor to have the bass lift coming in at a lower frequency. The value I use is based on 500hz as the -3db point.
You can always experiment
or just get an equalizer and set it that way. It is nice, either way, to be able to measure what is happening. You can use the Radio Shack meter with test tones and a table of correction values (I have a spreadsheet for this that graphs the results as well) or an equalizer with a spectrum analyzer built in.Let me know how it goes!
- Greg
Insert Witty Signature Line Here