Think you have good source using John L. Murphy's sharings.
Allow me comment i think the grade of performance this nice build perform using miniDSP will be some compromised going analog.
For the analog setup two OPA could be spared and their always questionable signature ruled out in do the XO as PLLXO. Its not the XO themselves think those can perform as well as miniDSP being analog passive/analog electric/PLLXO. Its the Linkwitz transform circuit and the calibration of on axis acustic center the two drivers between by analog allpass filter that can't keep up to precision when done in digital domain. OPA in TINA simulation if cheap or expensive configured as buffer or gainstage or XO can keep pretty high end performance for HD and S/N specs but it goes wrong with Linkwitz transform for those two specs and could degrade especially the midbass/midrange quality RS225-8 adds up in this FAST system. And analog allpass filters add for 1.order 180º and 2.order 360º phase turn which then ruin the flat phase presented so far for build. These two filters think unbeatable when done in digital domain where impedance and weird component values wont rule in same way for linkwitz transform and where pure delay is solution to calibration of on axis acustic center the two drivers between.
I'm personal impressed your build trying to make a simple full ranger higher performing in a relative fair simplest config and you documented data looks really great, this inspires me to build this one with digital domain filters. The digital filters performance could maybe be taken even farther than miniDSP choosing a solution where one can upgrade on DAC specs and if using computer as transport as i do get a solution where its galvanic isolated from analog domain.
Look forward when second box dried up if Zoom H4 recording stereo set is possible and shared or what your subjective comment would be
.
Allow me comment i think the grade of performance this nice build perform using miniDSP will be some compromised going analog.
For the analog setup two OPA could be spared and their always questionable signature ruled out in do the XO as PLLXO. Its not the XO themselves think those can perform as well as miniDSP being analog passive/analog electric/PLLXO. Its the Linkwitz transform circuit and the calibration of on axis acustic center the two drivers between by analog allpass filter that can't keep up to precision when done in digital domain. OPA in TINA simulation if cheap or expensive configured as buffer or gainstage or XO can keep pretty high end performance for HD and S/N specs but it goes wrong with Linkwitz transform for those two specs and could degrade especially the midbass/midrange quality RS225-8 adds up in this FAST system. And analog allpass filters add for 1.order 180º and 2.order 360º phase turn which then ruin the flat phase presented so far for build. These two filters think unbeatable when done in digital domain where impedance and weird component values wont rule in same way for linkwitz transform and where pure delay is solution to calibration of on axis acustic center the two drivers between.
I'm personal impressed your build trying to make a simple full ranger higher performing in a relative fair simplest config and you documented data looks really great, this inspires me to build this one with digital domain filters. The digital filters performance could maybe be taken even farther than miniDSP choosing a solution where one can upgrade on DAC specs and if using computer as transport as i do get a solution where its galvanic isolated from analog domain.
Look forward when second box dried up if Zoom H4 recording stereo set is possible and shared or what your subjective comment would be
.
Byrtt,
I see your point about time delay being a degradation in analog domain vs digital domain but not sure why you think Linkwitz transform in analog world is worse than digital world as phase impact is the same - or is it not? From what I see in miniDSP, setting up the LT involves editing the biquad polynomials.
I see your point about time delay being a degradation in analog domain vs digital domain but not sure why you think Linkwitz transform in analog world is worse than digital world as phase impact is the same - or is it not? From what I see in miniDSP, setting up the LT involves editing the biquad polynomials.
Seems right in principle phase and frequency response Linkwitz transform circuit analog or digital domain is same but because the analog one add a rumble filter (12dB HP) a probably minimal difference would be there but lets ignore it. Its a couple years ago i did sim in TINA that filter and have to look else on old backups or something to find the files but from memory this special circuit degrade performance a OPA can perform. As written in previous post its the HD and S/N specs that goes skyrocket compared to the high specs one can reach using OPA to make buffers gainstages or XO. Into TINA under "Analysis" run the "Fourier Analysis" and the "Noise Analysis" on a buffer or gainstage and see how transperant (subjective view) or objective good specs a OPA can perform, but from my tests when placing a LT circuit around the OPA those two specs skyrocket. Some would say these less good specs is okay at low bass frequencyes but as in this FAST system RS225-8 blend up high in frequency i mention it, had probably been better if RS225-8 had XO 80-100Hz being real sub. Now hope I'm not wrong and made wrong sims in past and will try either find the old files or run some new setups and get back with results.
Find it genious combine great group delay spec good old simple sealed box and transient perfect good old BW1 with two modern relative wideband good performance drivers, repair low end extension which is often lacking for sealed verse other alignments and acoustic on axis offset drivers between done with modern DSP. This have given on paper a very high performer, a system cost including miniDSP and four amps that even low end income people by save up can reach to get into this quality domain which i suppose it is. Fell in love this build forgive me that and my concern when i see something i think will be compromise compared grade or level it is at now. In long run hope impedance correction filters discussed over wesayso line arrays thread will be documented improvement and if added your nice build take it even further up making "Ref Monitor" yet more meaningfull.
Find it genious combine great group delay spec good old simple sealed box and transient perfect good old BW1 with two modern relative wideband good performance drivers, repair low end extension which is often lacking for sealed verse other alignments and acoustic on axis offset drivers between done with modern DSP. This have given on paper a very high performer, a system cost including miniDSP and four amps that even low end income people by save up can reach to get into this quality domain which i suppose it is. Fell in love this build forgive me that and my concern when i see something i think will be compromise compared grade or level it is at now. In long run hope impedance correction filters discussed over wesayso line arrays thread will be documented improvement and if added your nice build take it even further up making "Ref Monitor" yet more meaningfull
.
Last edited:
You are probably right about how all this works so well with DSP XO why mess with it? At $150 including amps the miniDSP active route is tough to beat given the flexibility.
I think I am just curious to try a passive solution to see how it may sound. My oil filled 50uF 370V motor run cap just arrived today. I just need to wind my own inductor from old magnet wire and I can maybe give it a try. No time alignment though.
I think I am just curious to try a passive solution to see how it may sound. My oil filled 50uF 370V motor run cap just arrived today. I just need to wind my own inductor from old magnet wire and I can maybe give it a try. No time alignment though.
In general, what is lost (or gained?) when re-implementing something like this as a traditional two-way (I.e. passive speaker-level XO)?
IIRC, the mini DSP takes an analog signal, does ADC, DSP, then DAC, right? Depending on how heavily you weight the contribution of the electronics to overall SQ, the nanoDSP might step things up a notch, since IIRC it only does DSP (input and output are digital). But then you need two DACs, which could be pricy depending on the DAC.
At any rate, that's why I was hoping for a passive speaker-level XO, for the simplicity of plug'n'play... But that probably introduces some other compromise(s) of which I'm unaware...
IIRC, the mini DSP takes an analog signal, does ADC, DSP, then DAC, right? Depending on how heavily you weight the contribution of the electronics to overall SQ, the nanoDSP might step things up a notch, since IIRC it only does DSP (input and output are digital). But then you need two DACs, which could be pricy depending on the DAC.
At any rate, that's why I was hoping for a passive speaker-level XO, for the simplicity of plug'n'play... But that probably introduces some other compromise(s) of which I'm unaware...
Why not?
In a FAST, with the drivers having a small C-C and the 1/4 wl of the XO point larger, a 1st order XO -- passive or active -- should be pretty close to time aligned. The wl at the XO is much larger than any driver emission centre being not perfectly lined up.
dave
Have a laugh admit i'm green at this passive stuff although very interested and a problem is i can't try it out real world in present i only have the 10F8424G00 and probably all the XO components so lack of RS225 and the box design is problem until later when financed. Been a little easier if we lived closer each other then passive version probably had been rocking at some stage by now
.Passive XO is constructed in XSim and my logic said me if i design same slopes passive as xrk971 satisfied use in miniDSP we must get close, so bass slope is nearly exactly same for both versions, the passive just have less low end extension in lack of LT circuit, and for 10F the passive one is padded little more than miniDSP and gives think a pretty close to miniDSP sum. It should be said XO is set at electric 350Hz but acoustic slopes cross at 730Hz probably because 10F is padded down at least the 6dB baffle step loss. Because this passive version have nearly same slopes as miniDSP expect it needs nearly same delay to perform same acoustic IR/SR/FR/PHASE.
So hope the cost put to get a 57uF capasitor and a 3,64mH inductor is not lost plus make me unpopular and will give nearly same performance as the miniDSP once subtracted we don't have the delay for correcting on axis acoustic center between the two drivers, we don't have the LT circuit low end extension, and lastly the two drivers is no more seeing directly into to a low z amp and the electric motion dampening effect (elastic) this have in that RS225-8 looks thru a 350mOhm inductor and 10F have a padding network and a cap in the route.
If passive XO needs delay as miniDSP setup did to calibrate on axis acoustic center the two drivers between, a electric solution is add a 1.order allpass but its drawback is its a 180º phase turn circuit adding not all 180º but some to the system nice flat phase gained by use of BW1 XO and it ruin some the perfect transient response. As pointed out by xrk971 instead of electric delay it could be done with a separately mounted Dagger that could be moved back physically to get the delay. Regarding delay or not if planet10 have tweaks he very welcome share
.Think DSP active route is tough to beat given the flexibility, also see in picture 2 plot the measured acoustic deviation compared textbook is so easy to correct by EQ and then the square waves previous shown would probably show considerly improvement even we still in minimum phase domain.
Attachments
LR2-BW1-LR2 XO Test
I stuck the Heil AMT on top of the speaker and loaded up the4-way XO plugin into miniDSP. I set the XO at 1200Hz with LR2 for the RS225 and AMT and BW1 for the 10F at 1200Hz (HPF and LPF). It took -20dB of padding on the AMT to level match it with the rest of the system (that is a lot of efficiency we are throwing away). The time alignment, even with digital delay is tricky.
Here is photo of setup:
Here is the minDSP XO screen shot:
Here is measured acoustic XO, appears closer to 1.8kHz:
The impulse has a pre-ringing but is clean otherwise:
Here is the high resolution 4ms gated freq response. Not sure why there is the oscillation in the AMT response:
Listening to it, there definitely is more "air" on the top, although my ears have a hard time discerning it sometimes. Violins with classical music sound especially good. But, the 10F/RS225 2-way still sounds darn good to my ears.
Attached is a sound clip, change .asc extension to .mp3 to listen.
I stuck the Heil AMT on top of the speaker and loaded up the4-way XO plugin into miniDSP. I set the XO at 1200Hz with LR2 for the RS225 and AMT and BW1 for the 10F at 1200Hz (HPF and LPF). It took -20dB of padding on the AMT to level match it with the rest of the system (that is a lot of efficiency we are throwing away). The time alignment, even with digital delay is tricky.
Here is photo of setup:
Here is the minDSP XO screen shot:
Here is measured acoustic XO, appears closer to 1.8kHz:
The impulse has a pre-ringing but is clean otherwise:
Here is the high resolution 4ms gated freq response. Not sure why there is the oscillation in the AMT response:
Listening to it, there definitely is more "air" on the top, although my ears have a hard time discerning it sometimes. Violins with classical music sound especially good. But, the 10F/RS225 2-way still sounds darn good to my ears.
Attached is a sound clip, change .asc extension to .mp3 to listen.
Attachments
Last edited:
This was a special "hole filler" xo that is supposed to be time and phase accurate. It uses two LR2 on the woofer and tweeter with regular polarity so they cancel and make a notch. The mid is the put in the hole to fill it up with a BW1 slope. If done right it is linear phase throughout entire bandwidth.
Passive 1st order BW XO test
I received my oil-filled 50uF 360V industrial motor run capacitor yesterday so I assembled a 1st order Butterworth XO per design from Byrtt: 56uF cap and 4.6R padding resistor pair for 10F, and 3.64mH coil for RS225 as low pass. I added a couple of existing 3.3uF caps to the motor cap for 57uF, and unrolled a 2.5mH + 2.5mH coil to get the 3.64mH inductor. All I had on hand were two 5R resistors. Here is the temporary XO setup outside on the back of the speaker:
And here is the measurement at 0.5m and 0.71v (equiv to 2.83v at 1m), and below it the predicted response of the XO that Byrtt provided me using member Bwaslo's Xsim software and measured impedance curves (.ZMA) and raw frequency .FRD curves. Byrtt adjusted for the 5ohms vs 4.6ohms and had to add 3in of time delay between the 10F and RS225 to compensate for the lack of the mechanical time alignment. With the proper delay the mid range bump would go away and the step response would be better. This is very good agreement between the model and measurement:
Thanks to Byrtt for the XO design! It sounds very nice but bass is reduced of course due to no Linkwitz transform. This is where ported bass reflex would help. Well this was my first all passive cross-over and with Byrtt's help I had a chance of making it work. Otherwise miniDSP is the only way I could have done this before. The $9 oil filled industrial motor cap seems to work well and costs less than the circa $25 "audio grade" caps.
I received my oil-filled 50uF 360V industrial motor run capacitor yesterday so I assembled a 1st order Butterworth XO per design from Byrtt: 56uF cap and 4.6R padding resistor pair for 10F, and 3.64mH coil for RS225 as low pass. I added a couple of existing 3.3uF caps to the motor cap for 57uF, and unrolled a 2.5mH + 2.5mH coil to get the 3.64mH inductor. All I had on hand were two 5R resistors. Here is the temporary XO setup outside on the back of the speaker:
And here is the measurement at 0.5m and 0.71v (equiv to 2.83v at 1m), and below it the predicted response of the XO that Byrtt provided me using member Bwaslo's Xsim software and measured impedance curves (.ZMA) and raw frequency .FRD curves. Byrtt adjusted for the 5ohms vs 4.6ohms and had to add 3in of time delay between the 10F and RS225 to compensate for the lack of the mechanical time alignment. With the proper delay the mid range bump would go away and the step response would be better. This is very good agreement between the model and measurement:
Thanks to Byrtt for the XO design! It sounds very nice but bass is reduced of course due to no Linkwitz transform. This is where ported bass reflex would help. Well this was my first all passive cross-over and with Byrtt's help I had a chance of making it work. Otherwise miniDSP is the only way I could have done this before. The $9 oil filled industrial motor cap seems to work well and costs less than the circa $25 "audio grade" caps.
Attachments
Last edited:
Sound clips in stereo
These are the 2-way setup (miniDSP active XO and Linkwitz transform) and were recorded from listening position at about 70 inches away with speakers spaced 70 inches apart. I really like how they sound. The detail and bass articulation is excellent. Imaging/soundstage are very good as well. The balanced sound works well with all genre's. I have listened to some classical and that is excellent as well.
Change .asc to .mp3 to listen
These are the 2-way setup (miniDSP active XO and Linkwitz transform) and were recorded from listening position at about 70 inches away with speakers spaced 70 inches apart. I really like how they sound. The detail and bass articulation is excellent. Imaging/soundstage are very good as well. The balanced sound works well with all genre's. I have listened to some classical and that is excellent as well.
Change .asc to .mp3 to listen
Last edited:
