making my own speakers from the ground up really makes it apparent what works and what doesn't. Passive works fine but the analysis and design phase takes so much time that I said screw it. I found passive to be a lot of work to get worse results most of the time. I find most criticisms of active the not have much weight and proponents of passive swim too much in the depths of audiophile nonsense.
I even committed the sin of running analog signal into active xover and out, sounds transparent to me.
I even committed the sin of running analog signal into active xover and out, sounds transparent to me.
I couldn't agree more.
A DSP makes it so easy for a speaker builder to eliminate phasing errors and dial in the sound. There's almost infinite tuning capability, all in real time, and it sounds fantastic. And the bass tightens up considerably with no passive inductor coil between the woofer and amp. I will NEVER go back to passive crossovers.
Parallels my experience very closely, and I'm not looking back. DSP though has been a steep learning curve, implementing in analog whether active or passive would have been much steeper. 😀
Because I am still chatting with the OP about the original post and have not wandered off to chat about other things (though I nearly joined in with the low frequency issues)? Still trying to prod him into answering his own question though without much progress. Yes I think I am in the right thread but I am beginning to wonder if I may be in the wrong forum.
Just for fun, I will share my DSP setup and tuning tweaks, which takes about 30 minutes from start to finish.
My setup likely uses lower crossovers and steeper slopes than would be typical for conventional speakers-- because I'm driving homebuilt hybrid ESLs and a pair of Ripol subs.
The setup is 6-channel stereo, driven thru a DBX Driverack Venu 360 DSP/Crossover with iPad interface.
Initial setup:
I first time-align the mid-bass woofer and stat panel using the microphone, RTA and a test tone matching the crossover frequency. I adjust the time delay until the RTA shows max output (i.e. max constructive interference), which for my speaker is a 0.33 millisecond delay on the panel. No need to time align the subs because of their longer wavelengths and placement beside/in-plane with the ESLs.
I then set the crossover frequencies (70Hz @ 48db/octave for the woofer-to-sub and 260Hz @ 48db/octave for the panel-to-woofer, using mirror-symmetric Linkwitz-Riley filters).
Then I adjust and balance the channel gains using the mic/RTA with pink noise, which puts me in the ballpark.
Next I use the the auto-tune feature, first defining a reference response curve, and the auto-tune then does three quick frequency sweeps, then automatically overlays parametric EQ's to smooth out the nasties and match the reference curve.
Final tuning:
Flat response doesn't quite suit me so I do the final tuning with the 31-band graphic, as follows:
ESLs can be a bit harsh in the upper midrange so I play the song "Holiday" by Erin Bode. Erin has a high voice that can cut diamond if the system is tuned hot-- but if I can take the edge off her voice, then nothing else I'm likely to play will sound harsh.
After Erin, I follow up with Dave Brubeck's "Take Five"-- and if Morello's hi-hat is still crisp, then I know I haven't taken too much off with Erin's track. Typically; a 3db dip at around 2.5kHz (i.e. the BBC dip) takes the edge off and gets the tuning perfect.
So that's all there is to it--- and I would use the same technique for any speaker. Of course the crossover points and slopes would be configured specifically to whatever drivers are used.
My setup likely uses lower crossovers and steeper slopes than would be typical for conventional speakers-- because I'm driving homebuilt hybrid ESLs and a pair of Ripol subs.
The setup is 6-channel stereo, driven thru a DBX Driverack Venu 360 DSP/Crossover with iPad interface.
Initial setup:
I first time-align the mid-bass woofer and stat panel using the microphone, RTA and a test tone matching the crossover frequency. I adjust the time delay until the RTA shows max output (i.e. max constructive interference), which for my speaker is a 0.33 millisecond delay on the panel. No need to time align the subs because of their longer wavelengths and placement beside/in-plane with the ESLs.
I then set the crossover frequencies (70Hz @ 48db/octave for the woofer-to-sub and 260Hz @ 48db/octave for the panel-to-woofer, using mirror-symmetric Linkwitz-Riley filters).
Then I adjust and balance the channel gains using the mic/RTA with pink noise, which puts me in the ballpark.
Next I use the the auto-tune feature, first defining a reference response curve, and the auto-tune then does three quick frequency sweeps, then automatically overlays parametric EQ's to smooth out the nasties and match the reference curve.
Final tuning:
Flat response doesn't quite suit me so I do the final tuning with the 31-band graphic, as follows:
ESLs can be a bit harsh in the upper midrange so I play the song "Holiday" by Erin Bode. Erin has a high voice that can cut diamond if the system is tuned hot-- but if I can take the edge off her voice, then nothing else I'm likely to play will sound harsh.
After Erin, I follow up with Dave Brubeck's "Take Five"-- and if Morello's hi-hat is still crisp, then I know I haven't taken too much off with Erin's track. Typically; a 3db dip at around 2.5kHz (i.e. the BBC dip) takes the edge off and gets the tuning perfect.
So that's all there is to it--- and I would use the same technique for any speaker. Of course the crossover points and slopes would be configured specifically to whatever drivers are used.
Last edited:
What you are overlooking here is the difference in capability between the filters generated by a hardware based DSP device and those generated by AudioLens / Acourate. Just for starters, check out the difference in the number of taps being used. And I’m not even going to start on the evils of the SRC in MiniDSP. I reiterate, search out and digest Mitch’s articles / YouTube vids.
Sounds like a good process 🙂
I haven't heard of that method before for getting time alignment. Is that how most people do it? I've seen some people talk about plugging an audio interface into itself or something, but I haven't learned how to do that yet.
I'm not missing that Studley.
What you are missing, imho, is that a hardware DSP device does not generate any filters.
Only FIR filter making software does that.
Yes, a PC has enormously greater capability to run larger FIR files that any know DSP hardware platform.
But you are mixing up what causes what, and how it matters when....imo.
I can assure you I am perfectly clear on how the filters get generated. I’m simply pointing out that these software / PC based DSP solutions have far greater capability than any hardware based DSP product.
Well, my DBX Venu 360 has adjustments for time alignments on each channel but, as far as I can determine, it doesn't compare and time align the drivers automatically.
One way to align drivers is to simply measure the physical offset between the drivers' voice coils, and then calculate the required time delay based on the speed of sound at sea level (1,126 ft/sec). But I'm not sure how accurate this is because I don't know whether the sound from a driver emanates from the plane of the voice coil or from some point of pressure within the cone, forward of the voice coil. Since I've never resolved that question, I prefer to base the delay on reading the max SPL of a test tone matching the crossover frequency, either by ear or by the RTA reading.
With both drivers playing the same tone; any phase error creates destructive interference that reduces the SPL. So; max SPL verifies that the drivers are exactly aligned (regardless of whether their voice coils are in the same plane).
If aligning drivers by ear, rather than reading SPL on an RTA, I find that its easier to find the exact alignment point by reversing phasing on one driver (I can flip phasing within the DBX without swapping wire connects), and then adjusting the time delay to obtain the lowest SPL (then flip the phasing back to what it should be). The ears are very sensitive to SPL and somehow it's easier to judge the lowest volume, than the highest volume.
Last edited:
Yep, with that i can easily agree.........when looking at DSP capability in totality.
Must say though, i don't think that matters as much as good implementation.
Hi Charlie. I didn't get a chance to say hi, but I saw you at the Parts Express Speaker Design Competition not long ago. I liked your speakers!
So, I've been talking throughout this thread about the MiniDSP Flex Eight, but I just came across this 8 channel DAC from topping on Apos. Topping DM7. I have just over $500 in store credit just sitting around on Apos, after an amp I bought from them fritzed out on me. Do you think that would be a suitable alternative to the Flex Eight? There's nothing really on Apos that interests me at this point, so it would be amazing to be able to be able to use my in store credit for a DSP solution, if the DM7 fits the bill for what you're talking about
I have a DM7 that I bought from Apos. I use it with Linux, and do DSP under Linux with the DM7 acting as the DAC. In that role I think it is stellar. Keep in mind that is doesn't have DSP built into it like the miniDSP FLEX series (maybe you knew that). I used a laptop and my DM7 to drive the amps in my SDC entry, which unfortunately went off the rails when I had some amp trouble. But I have been doing this software DSP for almost a decade now and love the flexibility it brings to the party.
Thanks for your response! That was unfortunate.. your right woofer went out of I recall correctly? I was on the left side so i didn't really detect the trouble. They still sounded good to my ears, even with the technical difficulties
What a funny coincidence about the DM7. Yes I'm aware it doesn't have built in DSP. What software do you use? Is it relatively straightforward to integrate with the DM7?
Expensive is not the deciding factor.
The output signal, and circuit design is.
And yes there are several different ways to design a better circuit. A simple example a simple 4 diode rectifiers, used for simplicity and budget, not performance.
It is interesting to see what these on a large scale does to it's supplied sine wave, and how much distortion a badly designed rectifier can actually cause.
A common problem with PWM DACs is switching noise, whereas a PCM dac does a much better better job at suppressing that, before any "correcting" measures, done in circuit design.
A Oscilloscope will show you differences, if your bothered to look.
And yes there is several approaches to make a "purer" sine wave from square pulses😱
Ah well, a good classic example that is common. Samsung TV's 2000-2010 vintages.
Using too small,/undersized caps, with marginal voltage ratings etc.
Millions of TVs died before or around warranty, and made you buy a new one, as fixing consumer electronics, is no longer commonplace in the "west".
Meaning they benefitted from it in sales/revenue for a time, until they got called out for it too many times.
And did corrective measures in design, basically upgrading a couple of capacitors to appropriate ones🙄
Alternatives are OktoDAC 8 and Flex HT. The later would support both internal and external DSP processing of I understand correctly...
Indeed, system gain requires attention in DSP setups. Only after reducing amp gain and adding some passive attenuation on top, I had satisfactory low level details in a Fusion amp setup...
Luckily with modern DACs, with better SNR, there is more and more headroom...