Hi folks,
Background:
Have been working on a DIY project to replace the DSP functionality with analog, because I am OCD about SQ, and was able to hear immediately noticeable differences in Analog vs. Digital crossovers. Easily A/Bd at the same time as I was able to run the left channel through 1 crossover and right through another and easily level match using REW.
I've been using a Driverack PA2 with Altec Model 14s, but felt I could hear that was a definite difference in mid-bass texture/detail, soundstage and crispness of highs. Using the Driverack, I was able to follow the lead of the Audiophile Nirvana guys (https://www.audiophilenirvana.com/a...ws/altec-lansing-model-14-part-4-ultimate-14/) and get fantastic results.
2019 thread of that effort: https://audiokarma.org/forums/index...h-driverack-pa2-project-and-questions.870225/
With addtional tweaking I laned on LR 24 for the sub/mids and LR 18 for the mid/high, levels matched using the DSPs gain controls on the outputs and further EQ using the AutoEQ as a starting point. Was also able to perform time alignment.
I tried upgrading to a Venu360 and the new minidsp flex 8 (not balanced, unfortunately, but was still decently quiet), but that wasn't really much better. Digital signals are more noticeably impacted detail wise and analog signals seemed to lose some life. Tried a few pre-built pro analog crossovers and they were much better, but had a bit of noise and also I didn't like that the levels and in some cases crossover frequencies could easily be altered by an errant finger. The card versions are much better in this regard.
One thing's for sure... You can pretty much ignore the snr figures published by minidsp. Upon closer inspection they show the digital input to digital output and digital input to analog output figures, but never seem to put the analog in/analog out figures, lol. It was actually the worst of the bunch. Awesome for home theatre though, but the UI on the Harman dbx products is far superior and those include RTA and signal generators built in, so no having to setup and flipping back/forth to REW for anything except time alignment.
Results so far:
In the end, I settled on the Xkitz- 3-way FET opamp based balanced boards and linear ps, as seen here. Even with external EQ, quieter, more detailed, bigger soundstage, way better bass/mid-bass textures, clearer highs.
I used and old, broken SD dvd player as a donor for the power receptacle and case , with 3d printed front/back panels (still need to tweak the makeshift back) and just purchased a few standoffs to make this very budget friendly. This was a drastic improvement over the DSP xovers and quieter and more detailed than the pro analog crossovers. Completely silent, might be the slightest combo of buzz/hiss if it's very, very quiet and I wedge my head into the CD horn. Has never been this quiet. Balanced cables helped a lot here, but these xovers really cleaned it up.
So far I've been able to accomplish the following.
-crossover points
-levels
-eq (using a Kenwood GE-7030, absolutely fantastic!), with the exception of a dip at the mid/hi crossover
-phase adjustment to eradicate nulls and reduce dip at 1500hz xover point (changed wires at mid/tweeter as required using REW, sub has selectable phase 0, 90, 180, 270)
compromise...
-Only LR4 24db slopes available on the diy crossover.. I knew that and trying to be budget friendly and can't justify the Marchand crossover expense, particularly when this one gets me 90% there)
What remains is
1. CD equalization - the EQs 12db range handled the rest of the frequency response, but doesn't have enough differential to raise this dip enough.
2. Time alignment (not too worried about this as the mid/hi are under 2ms, and the sub is about 18ms, but moveable and a lot of is ringing from the 22' ceiling and gobs of windows)
So the questions:
I have a dip at the 1500 Hz region (and 10 to 20Hz needed to be lifted, the EQ took care of that) that I want to take care of.
1. Can I alter the circuit by inserting a capacitor to adjust for the CD horn, as is done in passive networks? (passive example here: https://techtalk.parts-express.com/forum/tech-talk-forum/1410233-cd-horn-passive-equalization)
2. Would it be better to add a circuit? (plenty of room in the case to add a circuit board... something like this? .... https://sound-au.com/project173.htm, but this seems to have minimal impact on the 1.5K region. Also wondering, that even so, might the overall flattening of the HF range help raise it enough for the external EQ to do the rest)
2. As a 3rd alternative, would I be able to easily alter the circuit to have perhaps an LR2 12db or LR3 18 slope for the tweeter low pass?
3. Should I forget about trying to make this work? I hate to do this as I'm almost there and like the price! Of course, if there's other similarly priced active crossover diy boards, I'm still in the 45 day trial period.
Trying to keep it budget friendly/diy friendly. Retired, so I have time.
Glad to post REW or Driverack analysis screenshots, if it helps.
Other info for context:
Thank you for any and all advice, suggestions or comments!
Background:
Have been working on a DIY project to replace the DSP functionality with analog, because I am OCD about SQ, and was able to hear immediately noticeable differences in Analog vs. Digital crossovers. Easily A/Bd at the same time as I was able to run the left channel through 1 crossover and right through another and easily level match using REW.
I've been using a Driverack PA2 with Altec Model 14s, but felt I could hear that was a definite difference in mid-bass texture/detail, soundstage and crispness of highs. Using the Driverack, I was able to follow the lead of the Audiophile Nirvana guys (https://www.audiophilenirvana.com/a...ws/altec-lansing-model-14-part-4-ultimate-14/) and get fantastic results.
2019 thread of that effort: https://audiokarma.org/forums/index...h-driverack-pa2-project-and-questions.870225/
With addtional tweaking I laned on LR 24 for the sub/mids and LR 18 for the mid/high, levels matched using the DSPs gain controls on the outputs and further EQ using the AutoEQ as a starting point. Was also able to perform time alignment.
I tried upgrading to a Venu360 and the new minidsp flex 8 (not balanced, unfortunately, but was still decently quiet), but that wasn't really much better. Digital signals are more noticeably impacted detail wise and analog signals seemed to lose some life. Tried a few pre-built pro analog crossovers and they were much better, but had a bit of noise and also I didn't like that the levels and in some cases crossover frequencies could easily be altered by an errant finger. The card versions are much better in this regard.
One thing's for sure... You can pretty much ignore the snr figures published by minidsp. Upon closer inspection they show the digital input to digital output and digital input to analog output figures, but never seem to put the analog in/analog out figures, lol. It was actually the worst of the bunch. Awesome for home theatre though, but the UI on the Harman dbx products is far superior and those include RTA and signal generators built in, so no having to setup and flipping back/forth to REW for anything except time alignment.
Results so far:
In the end, I settled on the Xkitz- 3-way FET opamp based balanced boards and linear ps, as seen here. Even with external EQ, quieter, more detailed, bigger soundstage, way better bass/mid-bass textures, clearer highs.
I used and old, broken SD dvd player as a donor for the power receptacle and case , with 3d printed front/back panels (still need to tweak the makeshift back) and just purchased a few standoffs to make this very budget friendly. This was a drastic improvement over the DSP xovers and quieter and more detailed than the pro analog crossovers. Completely silent, might be the slightest combo of buzz/hiss if it's very, very quiet and I wedge my head into the CD horn. Has never been this quiet. Balanced cables helped a lot here, but these xovers really cleaned it up.
So far I've been able to accomplish the following.
-crossover points
-levels
-eq (using a Kenwood GE-7030, absolutely fantastic!), with the exception of a dip at the mid/hi crossover
-phase adjustment to eradicate nulls and reduce dip at 1500hz xover point (changed wires at mid/tweeter as required using REW, sub has selectable phase 0, 90, 180, 270)
compromise...
-Only LR4 24db slopes available on the diy crossover.. I knew that and trying to be budget friendly and can't justify the Marchand crossover expense, particularly when this one gets me 90% there)
What remains is
1. CD equalization - the EQs 12db range handled the rest of the frequency response, but doesn't have enough differential to raise this dip enough.
2. Time alignment (not too worried about this as the mid/hi are under 2ms, and the sub is about 18ms, but moveable and a lot of is ringing from the 22' ceiling and gobs of windows)
So the questions:
I have a dip at the 1500 Hz region (and 10 to 20Hz needed to be lifted, the EQ took care of that) that I want to take care of.
1. Can I alter the circuit by inserting a capacitor to adjust for the CD horn, as is done in passive networks? (passive example here: https://techtalk.parts-express.com/forum/tech-talk-forum/1410233-cd-horn-passive-equalization)
2. Would it be better to add a circuit? (plenty of room in the case to add a circuit board... something like this? .... https://sound-au.com/project173.htm, but this seems to have minimal impact on the 1.5K region. Also wondering, that even so, might the overall flattening of the HF range help raise it enough for the external EQ to do the rest)
2. As a 3rd alternative, would I be able to easily alter the circuit to have perhaps an LR2 12db or LR3 18 slope for the tweeter low pass?
3. Should I forget about trying to make this work? I hate to do this as I'm almost there and like the price! Of course, if there's other similarly priced active crossover diy boards, I'm still in the 45 day trial period.
Trying to keep it budget friendly/diy friendly. Retired, so I have time.
Glad to post REW or Driverack analysis screenshots, if it helps.
Other info for context:
- Asked maker of xkitz, if about this several days ago, and no response yet. Also asked about timing circuits and looking at Rod Elliots pages, but it's a bit above my mental pay grade.
- Voltage readings of outputs at crossover in volts AC are.100 sub, .035 mid, .006 tweeter to get these leveled when fed sine waves for peak frequencies for each output (I believe it was 40hz, 800hz and ~4khz)
- Crossed over at 80Hz for the mid-woofer/subwoofer (Mid: 200wpc rms amp with 150mV sensitivity, Sub: 1200wpc rms Velodyne, unknown input sensitivity, but leveling puts this about 3x the AC output voltage of the crossover LF outputs)
- Crossed at 1500Hz for the midwoofer/horn (Horn: 70wpc rms, 150mV sensitivity)
- Crossovers tried: digital- driverack PA2, Venu360 (locally available at guitar center!), MiniDSP flex 8 (unbalanced) Analog: BSS360, Rane AC23B, Ashly XR2001, xkitz - DIY
- Subwoofer has it's own equalization and phase adjustments.
- crossover circuit below:
Thank you for any and all advice, suggestions or comments!
Last edited:
Well, great analogue active crossovers cost $$ and/or impossible to find, e.g. Bryston 10B(discontinued), Krell, Pass, Discrete Marchand (custom, I have one), etc.
I have/tried most of them and also have Driverack PA2 in house, but I ended up talking to Marchand and custom ordering a discrete active crossover to my specification from them. I too seem cursed with ability to tell when dsp is used. I looked at xkitz, too, but from my experiments, my system did not sound as good with the common LR24.
Glad to hear I'm not the only 'analog nut job' lol!
I also fundamentally hated that there was ADC/DAC conversion up to 3 times for digital. (dac, xover, and the sub is DSP controlled for analysis microphone/eq/volume/remote)
I also fundamentally hated that there was ADC/DAC conversion up to 3 times for digital. (dac, xover, and the sub is DSP controlled for analysis microphone/eq/volume/remote)
Changing a fourth-order Linkwitz-Riley filter made with second-order Sallen and Key sections into a second order Linkwitz-Riley filter involves cutting out one section and changing a few component values in the other section.
Third-order Linkwitz-Riley filters don't exist, but I've been told you can approximate them anyway.
If you want to make a peak at 1500 Hz, it is probably handiest to build a separate circuit for that. What kind of peak do you need? How high, how wide, or what pole and zero locations or polynomial coefficients if you happen to know those?
I would expect an all-pass filter in a three-way Linkwitz-Riley crossover to make the phase shifts equal, but maybe that's taken care of by the woofer's phase alignment thing.
Third-order Linkwitz-Riley filters don't exist, but I've been told you can approximate them anyway.
If you want to make a peak at 1500 Hz, it is probably handiest to build a separate circuit for that. What kind of peak do you need? How high, how wide, or what pole and zero locations or polynomial coefficients if you happen to know those?
I would expect an all-pass filter in a three-way Linkwitz-Riley crossover to make the phase shifts equal, but maybe that's taken care of by the woofer's phase alignment thing.
For the CD equalisation you might find adding a passive 1st order xo (probably somewhere between 5kHz - 10kHz) to the horn may give enough tilt to get into the range of the Kenwood EQ.
Rob.
Rob.
only good dsp I've found so far was just running the filters in my computer. All the hardware I tried had some negative impacts to the sound.
I'll have to find a new multichannel dac cause everything is currently locked to my desktop.
I'll have to find a new multichannel dac cause everything is currently locked to my desktop.
Last edited:
Let's hope so. I did make the inquiry on the 25th, but no response yet as I think it's one man shop and a side hustlle.
Thanks SpeakerBob, I was wondering about that but didn't think to have a seperate one to get the full range to the amp.
My pre-amp does have 2 outputs, though one is DC.
Sam
I was going to do just that using the foobar plug in, but I have an extensive vinyl and CD collection. Too many cds to rip! Could do it by attrition I suppose, but still a gap for my reel to reel tapes and vinyl.
I love this idea. Briefly though about just addressing the dip, but was perhaps being lofty trying not to use the EQ as much for the high end as well and maybe fell down the CD equalization rabbit hole a bit.
From the REW measurement it's about 8-9db (without EQ) spanning from 1k to 3k (with a 2-3db bump in the middle-ish). About 4-5db equalized. Thinking I should address the uneq'd measurement.
I don't know about the pole and, zero and polynomic coeefficents, unfortunately. Pretty new to REW as well, but will try and figure that out, if that info is there.
edit... corrected levels
I found the foobar plugin terrible and it didn't work at all.
EQ APO does it better, if you ever choose that route. Loads IR's as well.
I think I do, if I understand what you mean.
Initially, I measured each driver individually with the only the corresponding amps activated on one channel with the mic a few inches from the center of each driver. (no EQ, not yet level matched when I did these)
Is that what you mean? (sorry for the crappy pic, haven't figured out how to export REW from one computer to another yet)
Or do you mean, sending measuring each driver w/o the crossover? That I didn't do, but there may be some measurements online, I'll see what I can find.
Just food for thought:
In my opinion the main music sources may be worth considering when deciding for analog or digital x-over.
If mainly digital sources are used DSPing comes in handy if one ist ready to use decent DACs after the digital x-over. If analogue sources are less often used or of secondary importance, a decent ADC will do a good Job.
If mainly analogue sources are used, an analogue x-over, accepting the drawback of less easy EQ possibility and x-over flexibility, ist probably the best solution as it avoids ADConversion. For secondary digital sources a less costly DAC can feed into the chain with good performance.
I mean: DSP is not just "bad" and a reproduction chain decision can be maid to represent the source priorities.
In my personal chain dig. sources are mainly used and phono is a secondary source, so I use miniDSP for processing (avoiding their DAC stages 😉 ) and have two Stereo DAC boards built into each of my 4-way speakers.
Greetings,
Winfried
In my opinion the main music sources may be worth considering when deciding for analog or digital x-over.
If mainly digital sources are used DSPing comes in handy if one ist ready to use decent DACs after the digital x-over. If analogue sources are less often used or of secondary importance, a decent ADC will do a good Job.
If mainly analogue sources are used, an analogue x-over, accepting the drawback of less easy EQ possibility and x-over flexibility, ist probably the best solution as it avoids ADConversion. For secondary digital sources a less costly DAC can feed into the chain with good performance.
I mean: DSP is not just "bad" and a reproduction chain decision can be maid to represent the source priorities.
In my personal chain dig. sources are mainly used and phono is a secondary source, so I use miniDSP for processing (avoiding their DAC stages 😉 ) and have two Stereo DAC boards built into each of my 4-way speakers.
Greetings,
Winfried
Ps.. should have also said, my goal is a harman type house curve and here's some information that may or may not be helpful.
Mid-woofer TS parameters.
Tweeter specs: https://greatplainsaudio.com/wp-content/uploads/2017/07/902-8B-16B-01.pdf
Here's the frequency response of the horn from the above document. The MR944A on the bottom left is the model I have.
Perhaps the tweeter section can cross over at a below 1500 and the mids staying the same? Don't know enough about xover circuits to know if this is possible.
A passive crossover design for the Model 14, in case it help to have the stock-ish setup. (the person who created this, Richard Saunders added a zobel network, seems to have have vanished online. Hopefully not for bad reasons)
Mid-woofer TS parameters.
Tweeter specs: https://greatplainsaudio.com/wp-content/uploads/2017/07/902-8B-16B-01.pdf
Here's the frequency response of the horn from the above document. The MR944A on the bottom left is the model I have.
Perhaps the tweeter section can cross over at a below 1500 and the mids staying the same? Don't know enough about xover circuits to know if this is possible.
A passive crossover design for the Model 14, in case it help to have the stock-ish setup. (the person who created this, Richard Saunders added a zobel network, seems to have have vanished online. Hopefully not for bad reasons)
Thanks wgh52,
Will look into this more. I had read about someone using something like the Okta-8 channel DAC, but seemed to add a lot of complexity and perhaps require a computer?
ps.. Is the crossover portion of the minidsp in the analog domain? So, if timing/eq not used, it's analog?
Ah, I did see a video on EQ APO when watching youtube REW videos!
Will keep that in mind if I go that route.
Did find this bit of trivia on the an website from the part 1 of the article via 1st link in post 1...
"Because the CD horn’s high frequencies are more spread out over its coverage pattern, they appear attenuated relative to other horns. The CD horn requires an equalization boost of approximately 6 dB per octave with a filter knee centered between 2 and 4 kHz (depending on horn design) in order to sound neutral and balanced. Most manufacturers of active electronic audio crossovers responded to this requirement by adding an optional CD EQ boost filter or high frequency shelf filter. Further refinements of the filtering process are available in DSP-based crossovers such as the dbx DriveRack PA2."
"Because the CD horn’s high frequencies are more spread out over its coverage pattern, they appear attenuated relative to other horns. The CD horn requires an equalization boost of approximately 6 dB per octave with a filter knee centered between 2 and 4 kHz (depending on horn design) in order to sound neutral and balanced. Most manufacturers of active electronic audio crossovers responded to this requirement by adding an optional CD EQ boost filter or high frequency shelf filter. Further refinements of the filtering process are available in DSP-based crossovers such as the dbx DriveRack PA2."