Ok, problem solved after some frustration and swearing 
Upgraded software in the flex eight, and re-installed drivers in the laptop.
I disabled the 'sloping curve' and did a pair of 'long gated' FR measurements at close distance just to match the gains for the amps, and a pair of at listening position. Room improvements can (should/must) start soon.. I can't do much about positioning of speakers or listener due to room/space restrictions, so I just have to make the best of it.
Upgraded software in the flex eight, and re-installed drivers in the laptop.I disabled the 'sloping curve' and did a pair of 'long gated' FR measurements at close distance just to match the gains for the amps, and a pair of at listening position. Room improvements can (should/must) start soon.. I can't do much about positioning of speakers or listener due to room/space restrictions, so I just have to make the best of it.
A little update:
On my quest for simplicity, I have tried some first order filters between woofer and mid, and it seems to work pretty well. Putting a first order LP on the woofer at abt 30Hz (experimenting with that) even seems to make a fairly good 'all in one' compromise between LT, room gain and low order XO just above 200Hz for the woofer. A little bit of baffle step and mid EQ on that, and the woofer seems pretty happy. I did the same for the tweeter, a first order HP slope around 13kHz to compensate for waveguide and a little bit of EQ on the top end to even it out. This makes an acoustic 2nd order HP slope just below 2kHz.
So, woofer and tweeter both fall into place pretty easily, but the mid takes a lot of massaging to shape, 8 biquads for it at the moment.
I find it hard to compare the sound between the different settings. It takes a couple of seconds to switch between the presets, and by then it almost sounds the same. I think I need a lot of time to get the new sound of the speakers to settle in before I can hear the nuances between the settings.
The better half listened to them yesterday, and gave her approval saying she 'could hear everything in the music', and she wanted to hear it loud, so they are accepted into the home even if they are big.
On my quest for simplicity, I have tried some first order filters between woofer and mid, and it seems to work pretty well. Putting a first order LP on the woofer at abt 30Hz (experimenting with that) even seems to make a fairly good 'all in one' compromise between LT, room gain and low order XO just above 200Hz for the woofer. A little bit of baffle step and mid EQ on that, and the woofer seems pretty happy. I did the same for the tweeter, a first order HP slope around 13kHz to compensate for waveguide and a little bit of EQ on the top end to even it out. This makes an acoustic 2nd order HP slope just below 2kHz.
So, woofer and tweeter both fall into place pretty easily, but the mid takes a lot of massaging to shape, 8 biquads for it at the moment.
I find it hard to compare the sound between the different settings. It takes a couple of seconds to switch between the presets, and by then it almost sounds the same. I think I need a lot of time to get the new sound of the speakers to settle in before I can hear the nuances between the settings.
The better half listened to them yesterday, and gave her approval saying she 'could hear everything in the music', and she wanted to hear it loud, so they are accepted into the home even if they are big.
Oh, forgot to mention.. I also tried a DSP setting for listening towards the corner/side of the room (sofa corner) where I usually sit doing stuff on the computer or just lie down to relax. I just used some delay and brought down the level of the closest speaker, and it sound surprisingly good. There is more bass towards the corner, but it also sounds better defined. I think it must be the asymmetry in the position/room/placement of speakers vs listening position that even out the bass. Anybody have similar experiences?
You could use huge coil in series with the woofer
Delay and attenuation works nice, can make nice phantom center and all, for off-center listening position. You can utilize it to have more balanced bass if you found out it works. There is two things that are affected by positioning, the stereo image and bass (modes). Both can be manipulated to a listening position with DSP. You could use head tracking and arduino to twist the stereo image to your location automatically if you wanted what ever is practical and gives most fun out of the system
Delay and attenuation works nice, can make nice phantom center and all, for off-center listening position. You can utilize it to have more balanced bass if you found out it works. There is two things that are affected by positioning, the stereo image and bass (modes). Both can be manipulated to a listening position with DSP. You could use head tracking and arduino to twist the stereo image to your location automatically if you wanted
what ever is practical and gives most fun out of the system
No big coils, I have DSP now
I remember trying multiple subs with DSP and trying MSO to even out bass in the previous apartment, but I did not like it with music transients. MSO generated some weird delays between subs, and it all sounded wrong with the kick drums and bass guitar transients. Based on that I don't think I want to make odd delays that are not related to listening distance to the speakers, but with the listening position to the side, I don't mind adjusting delays to correct for the distance to speakers, and it does seem to do good for the bass.
Overall SQ suffers a bit in the corner, but much less than expected!
I just did another simulation experiment.. I adjusted the woofer box sim to match the measured impedance response, then compared the simulated SPL vs nearfield measurements, and it turned out a measurement 15cm from the cone was almost a perfect match up to at least 200Hz, minus some wiggles. A measurement taken 'inside' the cone had a FR that looked like it was a lower tuning and higher Q, but the one taken at 15cm matched the sim. So I generated diffraction FR based on that and used it to adjust woofer LP filters a bit. Using the simulated woofer response, it was a bit boosted in the 200Hz area, so I adjusted the baffle step compensation to start lower to even that out.
Playing it now, and bass is possibly a little better. Seems simulation comes a long way for bass, and measurements are tricky, so it might even be better to rely on sims..
Well, then there is the (elephant in) the room too of course.., room gain and modes etc..
I remember trying multiple subs with DSP and trying MSO to even out bass in the previous apartment, but I did not like it with music transients. MSO generated some weird delays between subs, and it all sounded wrong with the kick drums and bass guitar transients. Based on that I don't think I want to make odd delays that are not related to listening distance to the speakers, but with the listening position to the side, I don't mind adjusting delays to correct for the distance to speakers, and it does seem to do good for the bass.
Overall SQ suffers a bit in the corner, but much less than expected!
I just did another simulation experiment.. I adjusted the woofer box sim to match the measured impedance response, then compared the simulated SPL vs nearfield measurements, and it turned out a measurement 15cm from the cone was almost a perfect match up to at least 200Hz, minus some wiggles. A measurement taken 'inside' the cone had a FR that looked like it was a lower tuning and higher Q, but the one taken at 15cm matched the sim. So I generated diffraction FR based on that and used it to adjust woofer LP filters a bit. Using the simulated woofer response, it was a bit boosted in the 200Hz area, so I adjusted the baffle step compensation to start lower to even that out.
Playing it now, and bass is possibly a little better. Seems simulation comes a long way for bass, and measurements are tricky, so it might even be better to rely on sims..
Well, then there is the (elephant in) the room too of course.., room gain and modes etc..
I just tried to load some linear phase filter biquads into the minidsp, but it did not look promising. In this case I was trying to load a LP XO between woofer and mid. I loaded the biquads for the mid, but looking at the 'EQ graph' for the channel, there was no HP on the mid at all after the LP HP biquads were loaded. Seems this is a no-go on the separate channels of the flex eight, I think I saw some FIR-button on the main input channels though..
I think I tried LP FIR on a 2-way in the past, but had problems with the sound.. listening fatigue if I remember correctly..
I think I tried LP FIR on a 2-way in the past, but had problems with the sound.. listening fatigue if I remember correctly..
I started a separate thread about treating my living room, feel free to chime in
https://www.diyaudio.com/community/threads/improving-acoustics-in-the-concrete-bunker.393051/
https://www.diyaudio.com/community/threads/improving-acoustics-in-the-concrete-bunker.393051/
A small update:
I have spent some time listening to these now. I think I prefer LR2 between mid/tweeter compared to LR4. Between mid/woofer I'm not sure. I tried LR2, LR4 and 'electrical 1st order' so the slopes are shallow at the XO point, but get steeper further away from XO. There is still some stuff to do on the integration between those I think, and also on the tuning of the woofer/bass with the room.
I think they sound better than my precious speakers in most ways. I especially like the sound of female voices on these. At the same time, I think something is happening in the lower midrange that 'enhances' the female voices, but sometimes time can sound a bit different/wrong with other stuff, or it could be my ears are just used to the sound of the previous speakers, and this is more correct.
I have been focusing on the room a bit lately, so not so much tuning going on with these right now. There is also some 'current drive' experiments lurking in the back of my head. I will definitely have to do some experimentation with that later on.
I would also like to try some different amps, especially for bass, and maybe for tweeters too.
I have spent some time listening to these now. I think I prefer LR2 between mid/tweeter compared to LR4. Between mid/woofer I'm not sure. I tried LR2, LR4 and 'electrical 1st order' so the slopes are shallow at the XO point, but get steeper further away from XO. There is still some stuff to do on the integration between those I think, and also on the tuning of the woofer/bass with the room.
I think they sound better than my precious speakers in most ways. I especially like the sound of female voices on these. At the same time, I think something is happening in the lower midrange that 'enhances' the female voices, but sometimes time can sound a bit different/wrong with other stuff, or it could be my ears are just used to the sound of the previous speakers, and this is more correct.
I have been focusing on the room a bit lately, so not so much tuning going on with these right now. There is also some 'current drive' experiments lurking in the back of my head. I will definitely have to do some experimentation with that later on.
I would also like to try some different amps, especially for bass, and maybe for tweeters too.
I re-did the LR2 tuning for these today. I have liked the LR2 version most, so I decided to concentrate on that and fine tune it some more.
I have added some more FR corrections, and went back to the approach to correct the drivers to 'flat' quite far beyond their intended band using biquads, and applied filters separately in the minidsp instead of in the biquads. The idea/reason is that I can tune the XO frequencies 'on the fly' when I listen, and maybe do some distortion measurements with multitone and adjust XO on the fly while measuring too.
I investigated the woofers a bit, since I have been hearing some 'resonance' from time to time. It became more apparent when I tried them with the TV listening to news etc with just voices. The woofers could be 'heard', they did not blend together well with the mid. When measuring I found there was a small (abt 1dB) ripple with a corresponding 'ring' in the waterfall around 270Hz. I tried playing around with the stuffing first, since it could be a standing wave top to bottom, but that did absolutely nothing, so I EQ:ed it out, and lowered the XO to 250. I was able to pry the woofer out, and the tape let go from the magnet, and glue was intact (I glued magnet to the braces with some tape in between because of a basket resonance, mentioned earlier in the thread).
I also measured the L & R speakers separately at listening position, and made some channel specific room EQ. It seems they excite some different modes in the room, so I figured I should try to EQ that individually too. Not sure if this is the correct approach though, but it seemed logical at the moment.
I re-verified all the levels and time alignment for the drivers too. Some small adjustments were needed. I still need to re-do the 'corner' listening position settings in a similar way.
I have not had much time to listen yet, so time will tell if I'm happier after this fix. There is still the 'elephant in the room' though, -room acoustics.. I'm at a bit of a standstill there, not sure what to do next, what absorption material to use etc.
I have added some more FR corrections, and went back to the approach to correct the drivers to 'flat' quite far beyond their intended band using biquads, and applied filters separately in the minidsp instead of in the biquads. The idea/reason is that I can tune the XO frequencies 'on the fly' when I listen, and maybe do some distortion measurements with multitone and adjust XO on the fly while measuring too.
I investigated the woofers a bit, since I have been hearing some 'resonance' from time to time. It became more apparent when I tried them with the TV listening to news etc with just voices. The woofers could be 'heard', they did not blend together well with the mid. When measuring I found there was a small (abt 1dB) ripple with a corresponding 'ring' in the waterfall around 270Hz. I tried playing around with the stuffing first, since it could be a standing wave top to bottom, but that did absolutely nothing, so I EQ:ed it out, and lowered the XO to 250. I was able to pry the woofer out, and the tape let go from the magnet, and glue was intact (I glued magnet to the braces with some tape in between because of a basket resonance, mentioned earlier in the thread).
I also measured the L & R speakers separately at listening position, and made some channel specific room EQ. It seems they excite some different modes in the room, so I figured I should try to EQ that individually too. Not sure if this is the correct approach though, but it seemed logical at the moment.
I re-verified all the levels and time alignment for the drivers too. Some small adjustments were needed. I still need to re-do the 'corner' listening position settings in a similar way.
I have not had much time to listen yet, so time will tell if I'm happier after this fix. There is still the 'elephant in the room' though, -room acoustics.. I'm at a bit of a standstill there, not sure what to do next, what absorption material to use etc.
I have been playing around with a few amps for bass the last couple of days. I started with a Yamaha PA amp. I have tried that for subs before, and did not find it optimal, so I wanted to try some other amp. I started with a high bias AB amp, it's a bit under-cooled though, so it kept overheating. I then remembered I had some amp modules (LJM MX50) that I thought had great bass when I tried them years ago.
My experience is that low output impedance (high damping factor) is good for bass, so I ran my LTspice sim on those, and they did have exceptionally low output impedance. I'm guessing the CFP output is better than EF in this respect. Since I will only use them for bass, I increased the loop gain a bit more (reduced degen on LTP), and adjusted compensation cap to get a nice square wave response. I swapped out the high bias boards for the MX50 and plugged them in.
I also did a 'polarity inversion' on one channel just for the fun of it, inverting polarity on one woofer channel in the DSP, and then swapping the leads to the corresponding woofer, to get the correct polarity again. Something done on some class D sub amps to prevent 'rail pumping', but should not really be needed on a AB amp, but I figured I could do it just for fun. Current draw from the power supply rails will be more even like that, so less LF modulation on the rails. Subjectively I think there is a small improvement in bass comparing to the Yamaha amp I first used. A little more controlled and maybe some more 'slam'. I have less power now, but I don't see it as an issue with my listening levels. If I want I can switch from 37V rails to 50V too, since the transformer has double taps, and the amp should survive it with 8ohm loads if not pushed too hard too long.
My experience is that low output impedance (high damping factor) is good for bass, so I ran my LTspice sim on those, and they did have exceptionally low output impedance. I'm guessing the CFP output is better than EF in this respect. Since I will only use them for bass, I increased the loop gain a bit more (reduced degen on LTP), and adjusted compensation cap to get a nice square wave response. I swapped out the high bias boards for the MX50 and plugged them in.
I also did a 'polarity inversion' on one channel just for the fun of it, inverting polarity on one woofer channel in the DSP, and then swapping the leads to the corresponding woofer, to get the correct polarity again. Something done on some class D sub amps to prevent 'rail pumping', but should not really be needed on a AB amp, but I figured I could do it just for fun. Current draw from the power supply rails will be more even like that, so less LF modulation on the rails. Subjectively I think there is a small improvement in bass comparing to the Yamaha amp I first used. A little more controlled and maybe some more 'slam'. I have less power now, but I don't see it as an issue with my listening levels. If I want I can switch from 37V rails to 50V too, since the transformer has double taps, and the amp should survive it with 8ohm loads if not pushed too hard too long.
I decided to do some experimenting with 'high impedance drive' on the mids on these today. After some experimenting I decided to run a 8,2ohm resistor and 1,5mH inductor in series with the drivers. This gave a significant (5-10dB) reduction in distortion in the whole range of the mids. I checked the distortion both with swept sine and multitone. One interesting thing was running a band limited multitone (200-2kHz) and see what happened with the 'grass' above 2k when adding a coil, easy 10dB reduction. The resistor is to also reduce distortion in the lower mids, where the coil has very little effect.
Sadly I lost my usb-stick with ALL the data for this setup -so I added the L+R on the mids, measured FR and then made new EQ to make them flat, and applied the same XO settings as before. Validated the time alignment by checking with mids inverted to see there was nice and deep dips at XO. Quick and dirty test basically.
Have not listened much yet, but of course I imagine hearing an improvement
Sadly I lost my usb-stick with ALL the data for this setup
-so I added the L+R on the mids, measured FR and then made new EQ to make them flat, and applied the same XO settings as before. Validated the time alignment by checking with mids inverted to see there was nice and deep dips at XO. Quick and dirty test basically.Have not listened much yet, but of course I imagine hearing an improvement
I did some experiments with a series resistor on the tweeter today, but with 4,7ohm in series with the 4ohm tweeter, I was not able to measure any difference, so I left them as is. I don't think it would do much good, and going up in resistor values, I would have to start compensating for the significant resonance hump in the tweeter impedance (FF removed) etc.
I did similar series resistance testing some time ago: https://audiodiyblog.blogspot.com/2022/11/distortion-exhibition-with-series.html?m=1
I used 0, 40 and 112 ohms for a 16 ohm compression driver and measured differences were very, very small.
I used 0, 40 and 112 ohms for a 16 ohm compression driver and measured differences were very, very small.
Yes, I have noticed it helps more on some drivers than others. I think there was some mention about that in Meriläinen's paper on current drive too, some drivers benefit more, depending on coil formers etc. Then there is the matter of microphones too, sometimes it can be the microphone distortion you see, not the driver. Joseph Crowe did some tests on this, and wrote about it here on the forum too.
I see you use the same mid/bass drivers as me, did you try measuring distortion on those?
I see you use the same mid/bass drivers as me, did you try measuring distortion on those?
Last edited:
Nope, I got discouraged by thin results on the CD. I could run some tests, because I think the midrange might benefit more from it. But it has much lower sensitivity, so it might not be desirable to lower it even further after testing.
Mics, especially electret mics, have distortion. But distortion adds up, so if you look at differences in measurements between resistor configurations, you should see the effect, if any.
Mics, especially electret mics, have distortion. But distortion adds up, so if you look at differences in measurements between resistor configurations, you should see the effect, if any.
Maybe just adding a coil in series could improve sound?
Not so sure about the distortion measurements adding up.. if mic distortion dominates, I think it will mask the drivers distortion, and you won't be able to tell if you lowered distortion from the driver. I have mostly done multitone measurements, and I prefer to do it like that (Link). One thing to try could be to compare a couple of tweeters at the same level, and if distortion levels are similar, IMO it's probably the mic you're seeing.
Not so sure about the distortion measurements adding up.. if mic distortion dominates, I think it will mask the drivers distortion, and you won't be able to tell if you lowered distortion from the driver. I have mostly done multitone measurements, and I prefer to do it like that (Link). One thing to try could be to compare a couple of tweeters at the same level, and if distortion levels are similar, IMO it's probably the mic you're seeing.
Well, mic should add its own distortion on top of the speakers distortion so that the measured result is the total distortion of all components in the chain. The idea here is to not measure absolute distortion, but compare results of different measurement runs, where the only changed factor is the number of resistors in the path. If there are differences caused by change in resistance, it should show up as a difference in between the measurements.
Since most mics are anyhow quite good, if the distortion of a calibrated measurement microphone (although it's a room acoustics mic) is a limiting factor, a question rises whether these added-resistance-configurations have much real world meaning at all. I'm interested in reducing speaker distortion because it's been generally thought to be the weakest link in sound reproduction chain, so I planned on getting a proper dynamic mic for this purpose, but have yet to make any acquisitions. Putting multihundredeuros to measure something that isn't a problem in the first place isn't very tempting. So I've been busy with other projects.
Since most mics are anyhow quite good, if the distortion of a calibrated measurement microphone (although it's a room acoustics mic) is a limiting factor, a question rises whether these added-resistance-configurations have much real world meaning at all. I'm interested in reducing speaker distortion because it's been generally thought to be the weakest link in sound reproduction chain, so I planned on getting a proper dynamic mic for this purpose, but have yet to make any acquisitions. Putting multihundredeuros to measure something that isn't a problem in the first place isn't very tempting. So I've been busy with other projects.
A proper dynamic mic ... is not that easy to find. You would need an omni mic with calibrated frequency response for useful measurements. Don't know any on the market, there are some omnis but you would need to do a calibration by yourself with your normal measurement mic.
Buying a good measurement microphone would be much easier.
Buying a good measurement microphone would be much easier.