Not at all and I truly appreciate the counsel!
Totally agreed on the room impacts and will have to try and identify which are room related and which are driver related. For instance, I'm not sure if the 1.5-5k peak is due to the shallow slopes or room reflections. This I perhaps can test by playing with the xo point and slopes, but not sure on the others.
If there's suggestions on using REW data or perhaps measuring technique or reading I can do on the topic, I'd love to get up to speed!
My hope was that I could come up with a minimal set of filters for a pllxo or active circuit using op amps.
If it's a better path measuring the woofer impedance and getting it going in XSIM, I'd be happy to follow that process as well, but not familiar territory for me.
Good post, Sam. It's a big topic but you seem to have the right idea.
A crossover simulator is a good tool but it won't show you what to do. Vituixcad has tried to incorporate some of the acoustic side by allowing polar measurements to be used. As useful as this can be, it still doesn't tell you what to do, not in the direct sense.
I wouldn't normally push a person toward acoustics first before they get a feel for other things, but when you feel ready for it you might begin by looking at waveguides and/or multi-subs. Not because they are the ideal place to start, but because they are a topic many begin with (that's also a caution that there's both good and not so good information there).
A crossover simulator is a good tool but it won't show you what to do. Vituixcad has tried to incorporate some of the acoustic side by allowing polar measurements to be used. As useful as this can be, it still doesn't tell you what to do, not in the direct sense.
I wouldn't normally push a person toward acoustics first before they get a feel for other things, but when you feel ready for it you might begin by looking at waveguides and/or multi-subs. Not because they are the ideal place to start, but because they are a topic many begin with (that's also a caution that there's both good and not so good information there).
Regarding impedance measurements. The technique is not unique to REW. You create a condition of constant current so that the impedance can be read as a voltage. You have to protect your interfaces from electrical damage, arrange a useful signal to noise ratio for your result and calibrate the output to a useful level so you can read off values. You then trace the output or export it to a file.
I like this video Generating Simple Frequency Response Correction Filters using Room EQ Wizard
Regards,
Dan
Regards,
Dan
Down to two filters and switching between auto EQ's adjustment and the two curves is indiscernible at the seating position with LR24.
Low Shelf and suckout for the midrange peak. The dip in the low end seems to be a room node, so ignoring it.
Baffle is 21", so went with a +9db shelf with 3db from a knee at the calculated 217hz and seems to work quite well.
A -7db notch filter with a Q of 1.7 @3.5khz seems to flatten out the top end nicely.
May be able to flatten it a bit more by adustign the notch filter upwards, but not hearing improvement with AutoEQ on (last pic shows it's recommended filters, which are fairly shallow corrections).
Auto EQ adjustments on top of the 2 filters are fairly minor and only very slightly audible. Not discernable improvement to my ears.
Going to try to put my analog active crossovers in unity gain and create some passive line level filters to see if those do the job. I'll try first with the Kenwood external EQ, though it only has 12db of boost/cut range, but I may be able to add 6db on the low end with it's BSC adjustment pot.
I have a spare +/- 20V rail balanced dc power supply in the crossover case and a few spare op amps, in the event I find that active filters work much better here.
Low Shelf and suckout for the midrange peak. The dip in the low end seems to be a room node, so ignoring it.
Baffle is 21", so went with a +9db shelf with 3db from a knee at the calculated 217hz and seems to work quite well.
A -7db notch filter with a Q of 1.7 @3.5khz seems to flatten out the top end nicely.
May be able to flatten it a bit more by adustign the notch filter upwards, but not hearing improvement with AutoEQ on (last pic shows it's recommended filters, which are fairly shallow corrections).
Auto EQ adjustments on top of the 2 filters are fairly minor and only very slightly audible. Not discernable improvement to my ears.
Going to try to put my analog active crossovers in unity gain and create some passive line level filters to see if those do the job. I'll try first with the Kenwood external EQ, though it only has 12db of boost/cut range, but I may be able to add 6db on the low end with it's BSC adjustment pot.
I have a spare +/- 20V rail balanced dc power supply in the crossover case and a few spare op amps, in the event I find that active filters work much better here.
Shifted the notch to 3.9K and a decent midrange with just the 2 filters. Hopefully the passive eq works/sounds as good. Doing some calc's and will start a new thread.
After a few hours of listening. Quite good, but a little bright for my tastes. Seems I like a bit of a slope on the 500Hz to 10kHz range. Used a third filter in that region with and reduced the downward gain on the 3.5khz suckout.
In DSP, I have 8 parametric adjustment downstream of the xo point for each output and 8 on the input, so 24 for a 2 way setup with no insertion losses since it's done by calculations in the DSP, but wondering in about analog circuits.
It seem with active vs. passive, the number of filters is less of an issue in active analog vs passive, allowing for not just the cross and driver corrections, but room effects and smaller bumps, no insertion loss problem. I see the value of DSP and active vs passive clearly for complex designs and higher orders.
My process has been to make all the filters on the inputs and when I like them/simplify them, move them to downstream portion and re-run the sweeps to see the effect and autoEQ filter recommendations on top of that and re-tweak each curve one at a time, each time focusing on different parts of the FR that need additional adjustment.
For pllxo I'm assuming the insertion losses rule out more than one filter on each leg? While I can goose my preamp up to 20V, I'd rather not do that and trying to keep it simple to learn.
While I'm still a ways away from a build, some Questions to guide my DSP experimentation and listening tests...
1. Is there any general thought on the number of EQ corrections in active analog circuits? I'm assuming more components=more noise so it's a bit of a balance? Of do you go all out, because the impact from that is negligible with today's better opamps?
For potential pllxo --
2. For the EQ components of a crossover is it theoretically better to do these before the xo circuit high pass/low pass or downstream? Or is it more of an art/science in balancing/tuning the potential insertion losses en route to the drivers in pllxo? In my case, one is obvious in that one spans both HP and LP, so must be upstream, but wondering about the others.
3. Wondering if two filters with less attenuation are better than one filter with a deeper attenuation as far as insertion loss? In my case two -3db filters vs one 8db drop for the 500Hz to 10kHz. Would filters with a wider Q have more insertion loss than narrow filters? Or is this something that I'll just need to measure because of all the variables?
In DSP, I have 8 parametric adjustment downstream of the xo point for each output and 8 on the input, so 24 for a 2 way setup with no insertion losses since it's done by calculations in the DSP, but wondering in about analog circuits.
It seem with active vs. passive, the number of filters is less of an issue in active analog vs passive, allowing for not just the cross and driver corrections, but room effects and smaller bumps, no insertion loss problem. I see the value of DSP and active vs passive clearly for complex designs and higher orders.
My process has been to make all the filters on the inputs and when I like them/simplify them, move them to downstream portion and re-run the sweeps to see the effect and autoEQ filter recommendations on top of that and re-tweak each curve one at a time, each time focusing on different parts of the FR that need additional adjustment.
For pllxo I'm assuming the insertion losses rule out more than one filter on each leg? While I can goose my preamp up to 20V, I'd rather not do that and trying to keep it simple to learn.
While I'm still a ways away from a build, some Questions to guide my DSP experimentation and listening tests...
1. Is there any general thought on the number of EQ corrections in active analog circuits? I'm assuming more components=more noise so it's a bit of a balance? Of do you go all out, because the impact from that is negligible with today's better opamps?
For potential pllxo --
2. For the EQ components of a crossover is it theoretically better to do these before the xo circuit high pass/low pass or downstream? Or is it more of an art/science in balancing/tuning the potential insertion losses en route to the drivers in pllxo? In my case, one is obvious in that one spans both HP and LP, so must be upstream, but wondering about the others.
3. Wondering if two filters with less attenuation are better than one filter with a deeper attenuation as far as insertion loss? In my case two -3db filters vs one 8db drop for the 500Hz to 10kHz. Would filters with a wider Q have more insertion loss than narrow filters? Or is this something that I'll just need to measure because of all the variables?
Consider the gain structure with respect to S/N, along with the practicality of implementing the filters. Otherwise you may be worrying about something that isn't a problem..
Naturally it is easier to do them once, ie upstream.. however there are numerous categories for EQ, stages at which they are needed to be adjusted and schemes for finishing their implementation which make it difficult to answer this.
Indeed, dropping the HF -7db and I really only need to adjust baffle step and the bit of peak on the horn center, so back to quite a good sound and sounding much better than the auto EQ that pushes it to flat from 500hz.
This allows for + 6db/6dbslope on the BSC to 217Hz and the notch filter is -3db at 3.7Khz with a Q of 2.04. Playing with a notch filter (pink peak, for which tried various Q widths) for the 500 to 1k region to take out the dip doesn't seem to do much audibly when turned on and off listening to music. Playing with filters in the 40-500 range does increase the bass response, but feels bloated, so assuming it's room related, but will confirm with individual speaker sweeps in REW.
Since the Driverack is setup more for room measurements, with a mininmum of 2 sweeps, it's easier to do that with REW. I can do an individual driver, with REW
Sound is very much to my liking so will keep listening, tweaking.
This allows for + 6db/6dbslope on the BSC to 217Hz and the notch filter is -3db at 3.7Khz with a Q of 2.04. Playing with a notch filter (pink peak, for which tried various Q widths) for the 500 to 1k region to take out the dip doesn't seem to do much audibly when turned on and off listening to music. Playing with filters in the 40-500 range does increase the bass response, but feels bloated, so assuming it's room related, but will confirm with individual speaker sweeps in REW.
Since the Driverack is setup more for room measurements, with a mininmum of 2 sweeps, it's easier to do that with REW. I can do an individual driver, with REW
Sound is very much to my liking so will keep listening, tweaking.