I am currently building a MTM tower with Purifi bass and Satori beryllium tweeters, discussed elsewhere here. But advice would be welcomed about a notch filter.
The bass driver is good and may not need a notch; I plan to try it with and without.
I have seen Lars's excellent article about notches for the NAA version, with a notch in series with the driver:
0.22mH, 4.2uF, 100ohms all in parallel. To be clear I call this a parallel notch.
I use 'series notch' to mean components in series, with the total cct then in parallel with the woofer.
I'm new to all this so I may be wrong, but I favour the parallel notch for two reasons.
First, Lars suggests it is low distortion.
Second, I have a low power SET amp. The series notch seems to work by taking extra current to drag the signal down, which may not be good with my low current amp?
The parallel notch seems to work by blocking the undesired frequencies, seems kinder to the amp?
Anyway, my case.
As I have low power and want high efficiency, I am looking to minimise the insertion loss of the filter, and I therefore guess at the following instead of Lar's values.
0.1mH, 15uF, 15 ohms all in parallel.
This gives just over 4kHz, which is where this driver has a peak. The peak is not great so 15 ohms resistor seems to give enough filtering. Result looks good in Xsim, as well as having less of a peak, the little that remains is now 5dB further down at about -25dB. Very small loss of sensitivity, but I think the main series inductor can be reduced by 0.1mH (to restore the same BSC) which should restore the sensitivity.
Looks good to me, but what do I know.
Any advice welcomed. The crossover in general will be discussed elsewhere.
The bass driver is good and may not need a notch; I plan to try it with and without.
I have seen Lars's excellent article about notches for the NAA version, with a notch in series with the driver:
0.22mH, 4.2uF, 100ohms all in parallel. To be clear I call this a parallel notch.
I use 'series notch' to mean components in series, with the total cct then in parallel with the woofer.
I'm new to all this so I may be wrong, but I favour the parallel notch for two reasons.
First, Lars suggests it is low distortion.
Second, I have a low power SET amp. The series notch seems to work by taking extra current to drag the signal down, which may not be good with my low current amp?
The parallel notch seems to work by blocking the undesired frequencies, seems kinder to the amp?
Anyway, my case.
As I have low power and want high efficiency, I am looking to minimise the insertion loss of the filter, and I therefore guess at the following instead of Lar's values.
0.1mH, 15uF, 15 ohms all in parallel.
This gives just over 4kHz, which is where this driver has a peak. The peak is not great so 15 ohms resistor seems to give enough filtering. Result looks good in Xsim, as well as having less of a peak, the little that remains is now 5dB further down at about -25dB. Very small loss of sensitivity, but I think the main series inductor can be reduced by 0.1mH (to restore the same BSC) which should restore the sensitivity.
Looks good to me, but what do I know.
Any advice welcomed. The crossover in general will be discussed elsewhere.
With low power SET considerations in mind your approach is sensible. Maybe some graphs or a screenshot would allow others to nod their heads in agreement.
Did you take measurements to help understand baffle step, or did you use a diffraction tool to indicate what is needed or calculator?
Did you take measurements to help understand baffle step, or did you use a diffraction tool to indicate what is needed or calculator?
Thanks Raymond.
About baffle step; I hope to get in the right ballpark by analysis as below, then tune by ear and measurement. I did that many years ago with the present DIY speakers which are Seas ER18RNX/27TDFC MTM; a MTM version of Zaph's SR71. I had help from Zaph so I had a good idea about BSC (1.2 to 1.4mH). I did some analysis which confirmed this. It worked well, I tried a range and selected 1.2mH. Very good speakers indeed; my wife says I'm mad to try anything different, won't get better sound than these; she may well be right.
Anyway, my attempt at BSC for the Purifi is based on guesswork as there is not much info around, what there is is contradictory.
The Purifi speakers are to be almost exactly the same size as the Seas ones, same position in the same room so comparison may be useful.
Comparing the Purifi and ER18RNX, the latter drops off significantly in the bass, the Purifi does not; I guess the difference to be about 3dB, suggesting the need for roughly half the BSC; about 0.6mH inductor.
Mark (islandpink) has made a very successful speaker with the drivers I am to use, but single bass driver. He gets good bass with only 0.56mH inductor, which equates to 0.28mH for my two drivers in parallel. Quite a bit different to my guesswork. However I think I've seen larger inductors used with this driver so I think my first guess of 0.6 mH is a reasonable start point and this is on order.
If anyone has experience of BSC for thise driver, I'd be interested.
About baffle step; I hope to get in the right ballpark by analysis as below, then tune by ear and measurement. I did that many years ago with the present DIY speakers which are Seas ER18RNX/27TDFC MTM; a MTM version of Zaph's SR71. I had help from Zaph so I had a good idea about BSC (1.2 to 1.4mH). I did some analysis which confirmed this. It worked well, I tried a range and selected 1.2mH. Very good speakers indeed; my wife says I'm mad to try anything different, won't get better sound than these; she may well be right.
Anyway, my attempt at BSC for the Purifi is based on guesswork as there is not much info around, what there is is contradictory.
The Purifi speakers are to be almost exactly the same size as the Seas ones, same position in the same room so comparison may be useful.
Comparing the Purifi and ER18RNX, the latter drops off significantly in the bass, the Purifi does not; I guess the difference to be about 3dB, suggesting the need for roughly half the BSC; about 0.6mH inductor.
Mark (islandpink) has made a very successful speaker with the drivers I am to use, but single bass driver. He gets good bass with only 0.56mH inductor, which equates to 0.28mH for my two drivers in parallel. Quite a bit different to my guesswork. However I think I've seen larger inductors used with this driver so I think my first guess of 0.6 mH is a reasonable start point and this is on order.
If anyone has experience of BSC for thise driver, I'd be interested.
This may not make much difference in this case, but broadly speaking you're more likely to damage a valve amp with a higher impedance load than lower.
Frequency plots of woofers only, with crossover in place. First with notch then without.
Hope this is useful.
Hope this is useful.
Re BSC: I made a mistake, I forgot that Mark uses a 4 ohm 6.5inch Purifi driver, not 8 ohm. His results with this should be similar to mine with two 8 ohm drivers in parallel. Which tends to confirm my 0.6mH as being a reasonable start point.
Hi Mike,
Just for clarity you are using the same aluminium coned driver as the application note but the 8 ohm version used in parallel to give you a four ohm target impedance? I ask as maybe a paper coned Purifi will have different break up/resonance modes.
The graph indicates that your filter works around 3-3.5Khz as far as I can tell, it seems to be effective. What does the impedance look like ?
As the application note is a massively useful piece of information with a thorough explanation of the reasoning behind it you are not tempted to do anything about the notch for the highest frequency peak and achieve minimum distortion profile. Again frequency and impedance information should let you see you are still within limits for you SET solution.
Baffle step inductor size as you say 0.6mH seems like a suitable starting point, I take it you do this by ear, and have a box of suitable inductors handy. What are the dimensions of your SR71 MTM baffle?
Just for clarity you are using the same aluminium coned driver as the application note but the 8 ohm version used in parallel to give you a four ohm target impedance? I ask as maybe a paper coned Purifi will have different break up/resonance modes.
The graph indicates that your filter works around 3-3.5Khz as far as I can tell, it seems to be effective. What does the impedance look like ?
As the application note is a massively useful piece of information with a thorough explanation of the reasoning behind it you are not tempted to do anything about the notch for the highest frequency peak and achieve minimum distortion profile. Again frequency and impedance information should let you see you are still within limits for you SET solution.
Baffle step inductor size as you say 0.6mH seems like a suitable starting point, I take it you do this by ear, and have a box of suitable inductors handy. What are the dimensions of your SR71 MTM baffle?
At resonance, the L and C (in parallel) go open-circuit, kind-of. They work together to block current from outside of their loop, so the 15 ohm resistor is where the signal gets carried and appears in series with the speaker.
This is beneficial because, as we see from the force equation
F=BIL
current is the key ingredient that the amplifier ultimately tries to control in the speaker, where B is the motor strength and L is the length of wire suspended in the magnetic field.
The speaker impedance tends to be unstable, and changes with coil position and magnetization of the magnet iron (etc.). Therefore, adding 15 ohm to the total 'R' in the
I = V / R
equation stabilises the total impedance, which then stabilises the current.
The series notch (in parallel with the speaker) doesn't do any of that. At resonance, the L+C work together to form a short circuit, so the R component shunts current adjacent to the speaker. This could be effective at reducing levels if the amplifier has high impedance, but it does nothing to stabilise current through the speaker.
Not necessarily. What you said about the parallel filter also has a way of applying to the series filter. The result is a mixed bag where some regions of the source impedance are greater and some are reduced.
This is a random example I just made with no thought put into it...
Don't panic when you look at the graphs, I simply used the mechanical properties of the purify driver included in the VituixCad driver file.
The information simply highlights the difference between the Filter in the app note and your version. It makes it easy to compare the filter characteristics thats all. No Baffle step, no diffraction and no break up included.
The information simply highlights the difference between the Filter in the app note and your version. It makes it easy to compare the filter characteristics thats all. No Baffle step, no diffraction and no break up included.
Attachments
I am using two of the paper cone PTT6.5X08-NFA-01 in parallel MTM. The baffle is just over 8 inches wide; I know, wider would be better but the resulting divorce would be a problem 🙂.
Just for info, the Salk Bepure 2 looks like a 2.5 way but is TMM with bass drivers in parallel.
I thank you all for such good info, I am learning.
Just for info, the Salk Bepure 2 looks like a 2.5 way but is TMM with bass drivers in parallel.
I thank you all for such good info, I am learning.