Congrats Fluid on getting sound out of your towers!
Yep, you jump up and start to head-bang or boogie, and the magic is gone!
Yep, you jump up and start to head-bang or boogie, and the magic is gone!
Popcorn time is over enjoy
...Yep, you jump up and start to head-bang or boogie, and the magic is gone!
For that reason can share for a few weeks had for exercise had run below flat baffle within 1/8 wavelength as is ctc data they use in Synergy concept, startet up using 8th order LR linear phase but for latest few days upped that to difficult 16th order because that into XDir show unique good verticals power response when within 1/8 wavelength, and can say it works so magic is there at whatever height listened, drawback for this particular setup is HD raise for 10F when asked to cover down at a 180Hz XO point.
Attachments
Last edited:
Because there actual was a bit difference in XSim simulation with 5 real world TC9 measured free air impedance used in post 352 simulation when either wired as "5x parallel connected drivers wired as 5 series" or "5x series connected drivers then put in 5 parallel" have this better overview to share.
Added same color to the drivers at left side as they represent in curves at right, and at right all 7 impedances have now their corresponding phase curve presented.
Added same color to the drivers at left side as they represent in curves at right, and at right all 7 impedances have now their corresponding phase curve presented.
Attachments
That's a good idea. I do want to measure them outdoors first if I can to see the difference the felt has in measurements.
Congrats Fluid on getting sound out of your towers!
Yep, you jump up and start to head-bang or boogie, and the magic is gone!
Thanks!
Popcorn time is over enjoy
For that reason can share for a few weeks had for exercise had run below flat baffle within 1/8 wavelength as is ctc data they use in Synergy concept, startet up using 8th order LR linear phase but for latest few days upped that to difficult 16th order because that into XDir show unique good verticals power response when within 1/8 wavelength, and can say it works so magic is there at whatever height listened, drawback for this particular setup is HD raise for 10F when asked to cover down at a 180Hz XO point.
Could you not do better by putting them on the same baffle so the distance between the two drivers is smaller allowing you to either push up the crossover or relax the order? Due to the two box construction it seems there is a bigger gap than needed.
In my measurements there is big difference between free air and in cabinet. The free air seems to sum like in XSim but when put in a box they don't join as neatly with the wiring scheme I used. Balancing the Fs peak of the drivers to be more equal between the sub enclosures worked at the peak but it moved the discontinuity to a different location.
Maybe things will change after more regular usage but I have a feeling there will still be a dip of some kind. The good thing is that I couldn't hear anything in particular that suggested a real problem. Frequency response and distortion measurements might show something else.
Good spotted
if i was less lazy and bringing them close on same baffle frq point could be upped to better 240Hz point, being mostly a exercise speaker to learn from it did its purpose and showed me the better acoustic performance when power response around XO point gets very close to perfect and room reflections from all directions starts to have same tonality.Imagine as suggested take some listening time period and enjoy while they break in a bit and then look at it later, its interesting building stuff to follow thanks for that
Learned back in school time that current goes from positive to negative and if that still holds true driver one in a series row of five will be the first to relative start push out and its negative electric pole at that moment is steady and stiff, but because the series share same volume acoustics push and pull to the other cones and that maybe varies potential on negative electric pole for driver one starting those varying bumps at Fs peak, and then comes driver two/three/four and five and the intended stiff electric pole potentials for all drivers is now like a elastic.
Last edited:
I think unequal current sharing or some similar mechanism is at play and I have inadvertently made the cabinet quite susceptible to it by having 5 seperate enclosures that contain the series drivers. It would seem on reflection that it may be better to have them all in one and damp the connections like wesayso or to do like Owen and wire the drivers in the sub enclosure in parallel. Even though it looks a bit ugly on the impedance so far it doesn't sound like a freakshow so it's more a curiosity than an issue to be resolved at this point.
I have been thinking again about an analogue bass boost circuit and I think it is a waste of time unless I use an analogue gain control.
The DAC I have earmarked for this is the Sabre 9018 with digital volume control. If I boost analogue after the DAC to maintain bit resolution I need to use analogue volume otherwise all I end up doing is attenuating digitally to get the volume I want so I end up right back where I started except perhaps with a bit extra noise from the op amp stage.
I need to do some tests to see how much voltage I need out of the Najda to get to too loud as the First One has about a 1.5V input sensitivity. I was testing the other day at 1.5V from the Najda and I remember being pretty close to full volume. It could well be that the output is well below the nominal value due to the amount of input attenuation I need to avoid clipping. Or it may be as Owen suggested that a bigger amp is needed to get these up to too loud
If all drivers were exactly identical, the difference would probably be null. However, if certain drivers had different impedance at various frequencies, then parallel sets of 5, strung in series, the individual drivers of each parallel set would take on different currents, affecting the overall currents in the four remaining drivers of each set. With 5 series sets connected in parallel, each set has exactly the same current running through all 5 drivers, and the current through each set would be an average of all 5 drivers within each set. So a set of 5 drivers in series, connected 5x in parallel, might have less variation between a set of 25 drivers with slightly different characteristics, resulting in smoother sound.
Another advantage of stringing 5 drivers each in series, with each set connected in parallel, is simplicity of the wiring. If the drivers are arranged with all terminals to one side, then each set of 5 drivers connects top to bottom, and a single red and single black wire connect each set. The black wire needs to be tapped 5 times and the red wire needs to be tapped 5 times, such that only the drivers on the far ends of each series set need two wires connected to the end contacts. Compared to 5 drivers in parallel, each set series connected, each driver's plus terminal connects to the plus terminal of the driver above and below it, resulting in more splices and more connections.
So each set of 5 drivers connected in series, 5x sets in parallel, would be simultaneously easier and more convenient to wire up, and have more balanced response between all 25 slightly unequal drivers in the array. My 25 cents...
Another thing to consider, it was mentioned that another member said that series connected drivers should not be housed in the same enclosure. Is there some reason for that? Do the drivers influence each other when all drivers get the same current? I do know from experience with my childhood toys that two electric DC motors behave strangely when connected in series. Normally, when you "stop" a motor, it's current increases immensely, and current reaches a minimum point when the motor freely spins. For my series motor experiment, both motors (which were otherwise identical) spun at about 50% speed, each passing half of the voltage supply. Adding resistance to one motor would cause it to stop, in which case the second motor spun faster. The spinning motor would bear the entire voltage across it's terminals, and the current would be below the startup threshold of the stalled motor, such that the stalled motor would sit motionless until some source of resistance was applied to the spinning motor.
Why am I bringing this up? DC rotational motors are obviously different beasts than a speaker cone, which is technically a linear motor. However, if the DC motors can behave in anomalous ways when connected in series, it is plausible that an array of linear motors such as speaker drivers may influence one another, through the backwave pressure inside the enclosure. So do strange effects occur when speakers connected in series share a single enclosure? I have no idea. But based upon the DC motor experiment, it may be plausible. Then again, this effect may be negligible if all series drivers have the same wind resistance on them. Obstructing one of the drivers might produce more output from the others even with the same current running through them all, as the DC motor experiment demonstrated, but if all drivers are doing the same work and have the same wind resistance, any loading effects may be negligible.
Back to my above example of 5 drivers connected in series per set, 5 sets connected in parallel, the most obvious arrangement for connecting them would be side by side, which places each series connected set within the same acoustic chamber. It definitely leaves one to a moment of pause if this is really the optimal solution.
Last edited:
@stardust4ever well I don't know about that...
I had begged BYRTT to supply me his X-Sim simulation so I could check it on a driver level to find out what is happening.
(Yes I am that crazy that I would consider doing a complete re-wire)
If I plot 5 drivers in X-Sim in series I get this:
This is with 10 watt applied to see the graph more clearly.
S1 to S5 are in series and form one of the 5 strings of drivers connected in parallel.
If I do the same plots of Owen's parallel drivers:
Here S1 to S5 are parallel and form one of the strings of drivers in series.
Again 10 watt applied to the complete array.
I think Owen (OPC) might be the brightest one here.
Let's look at the watts for both of these cases:
First in series:
versus in parallel:
This is only a first look but I genuinely think Owen was on to something here.
Lets pump up the watts, plot it at 100 watt into the array!
First the series connected drivers:
Then the parallel connected drivers:
Thanks to BYRTT for supplying the sims, to OPC for opening my eyes, to bwaslo for the amazing X-Sim program and fluid for the hospitality on this thread.
I had begged BYRTT to supply me his X-Sim simulation so I could check it on a driver level to find out what is happening.
(Yes I am that crazy that I would consider doing a complete re-wire
)If I plot 5 drivers in X-Sim in series I get this:
This is with 10 watt applied to see the graph more clearly.
S1 to S5 are in series and form one of the 5 strings of drivers connected in parallel.
If I do the same plots of Owen's parallel drivers:
Here S1 to S5 are parallel and form one of the strings of drivers in series.
Again 10 watt applied to the complete array.
I think Owen (OPC) might be the brightest one here
.Let's look at the watts for both of these cases:
First in series:
versus in parallel:
This is only a first look but I genuinely think Owen was on to something here.
Lets pump up the watts, plot it at 100 watt into the array!
First the series connected drivers:
Then the parallel connected drivers:
Thanks to BYRTT for supplying the sims, to OPC for opening my eyes, to bwaslo for the amazing X-Sim program and fluid for the hospitality on this thread
.Attachments
Hi Wesayso,
Amazing work here! I'm very glad you took the time to run this as I just didn't have the time to run the sims and to get such a concise summary together.
Also, to be clear, I'm definitely not the smartest one here... I've just been lucky enough to run into these things throughout my career, and to learn from people far smarter than I could ever hope to be
In addition to this, your arrays kick my array's butt!
I also didn't want to rain on fluid's parade as I know how much work it is to get the arrays wired up, and he hit a pretty big milestone getting both arrays up and running.
Some background on this:
My first job out of school was at Paradigm and the first project I worked on was the 3kW class D amplifier used in the "Sub 1" and "Sub 2" products. A quick google search will show you what these look like, but they are basically hexagons where 6 individual drivers share a common volume, and are all driven by a single amplifier (Sub 2) and two amplifiers (Sub 1).
The prototype version of this product used a series arrangement for the drivers (I think it was 2 ohm drivers wired 6 in series for a 12 ohm final impedance to get maximum output from the amplifier). This version literally tore itself apart, completely snapping a cone in half on one of the drivers. I didn't understand this at all, and asked my manager at the time what on earth was going on. He explained that when they are series wired, the individual drivers are not necessarily at the exact same position, and if one gets out of phase enough with the others, the remaining 5 drivers will destroy it. In addition to the potential for damage, it also has a huge negative impact on output for obvious reasons. In the end, the drivers were changed to be 70 ohms and all wired in parallel. This arrangement worked perfectly, and is what is shipping today. At the time I didn't really understand it, and just wrote it off as "oh, that's cool.. don't wire drivers sharing a common volume in series". I was more of an amplifier guy than a speaker guy back then
Later in life, at Blackberry, when I was spending more time doing microacoustics, I asked our local physics/audio guru guy if he had ever heard of this, and he said it made no sense... both series and parallel wiring should not make a difference. I insisted that it did, and he was interested enough that he built an entire model (at the time in our custom BlackBerry acoustics simulation tool) and sure enough, the cone position of parallel wired transducers remained very close, where the cone positions of the series wired devices were all over the place.
To be honest, I still don't fully understand precisely what the mechanism is here, but I think the answer would be obvious if you used a Klippel to plot cone position, terminal voltage, and current as a sine wave on a scope while sweeping frequency. Below resonance cone position should be proportional to voltage, at resonance I don't really know, and above resonance things should go back to voltage proportional until breakup occurs, then all bets are off. I get the feeling that in the above scenario, voltage will track much more closely to cone position than current will throughout the entire frequency range, but in particular, at low frequencies and at resonance where it matters most when several drivers are sharing the same volume.
If there is any way to plot excursion of the 5 drivers while accounting for the shared volume, that would be the best way to show the impact of this.
At the end of the day, this is definitely not a minor detail! You will get significant benefits from only running parallel drivers in a shared sealed volume. My apologies to fluid for being the bearer of bad news
Regards,
Owen
Amazing work here! I'm very glad you took the time to run this as I just didn't have the time to run the sims and to get such a concise summary together.
Also, to be clear, I'm definitely not the smartest one here... I've just been lucky enough to run into these things throughout my career, and to learn from people far smarter than I could ever hope to be
In addition to this, your arrays kick my array's butt!
I also didn't want to rain on fluid's parade as I know how much work it is to get the arrays wired up, and he hit a pretty big milestone getting both arrays up and running.
Some background on this:
My first job out of school was at Paradigm and the first project I worked on was the 3kW class D amplifier used in the "Sub 1" and "Sub 2" products. A quick google search will show you what these look like, but they are basically hexagons where 6 individual drivers share a common volume, and are all driven by a single amplifier (Sub 2) and two amplifiers (Sub 1).
The prototype version of this product used a series arrangement for the drivers (I think it was 2 ohm drivers wired 6 in series for a 12 ohm final impedance to get maximum output from the amplifier). This version literally tore itself apart, completely snapping a cone in half on one of the drivers. I didn't understand this at all, and asked my manager at the time what on earth was going on. He explained that when they are series wired, the individual drivers are not necessarily at the exact same position, and if one gets out of phase enough with the others, the remaining 5 drivers will destroy it. In addition to the potential for damage, it also has a huge negative impact on output for obvious reasons. In the end, the drivers were changed to be 70 ohms and all wired in parallel. This arrangement worked perfectly, and is what is shipping today. At the time I didn't really understand it, and just wrote it off as "oh, that's cool.. don't wire drivers sharing a common volume in series". I was more of an amplifier guy than a speaker guy back then
Later in life, at Blackberry, when I was spending more time doing microacoustics, I asked our local physics/audio guru guy if he had ever heard of this, and he said it made no sense... both series and parallel wiring should not make a difference. I insisted that it did, and he was interested enough that he built an entire model (at the time in our custom BlackBerry acoustics simulation tool) and sure enough, the cone position of parallel wired transducers remained very close, where the cone positions of the series wired devices were all over the place.
To be honest, I still don't fully understand precisely what the mechanism is here, but I think the answer would be obvious if you used a Klippel to plot cone position, terminal voltage, and current as a sine wave on a scope while sweeping frequency. Below resonance cone position should be proportional to voltage, at resonance I don't really know, and above resonance things should go back to voltage proportional until breakup occurs, then all bets are off. I get the feeling that in the above scenario, voltage will track much more closely to cone position than current will throughout the entire frequency range, but in particular, at low frequencies and at resonance where it matters most when several drivers are sharing the same volume.
If there is any way to plot excursion of the 5 drivers while accounting for the shared volume, that would be the best way to show the impact of this.
At the end of the day, this is definitely not a minor detail! You will get significant benefits from only running parallel drivers in a shared sealed volume. My apologies to fluid for being the bearer of bad news
Regards,
Owen
I'm always thrilled to learn a new lesson . There's a enough "mystic" in this audio hobby anyway. I'm glad we got to do this sim exercise even if it does mean I'm now curious and crazy enough to think of re-wiring my arrays. My guess is that together we can advance this hobby trough this kind of sharing.
Your explanation does start to make more sense to me. Re-wiring will be quite a mayor surgery for me. I'll have to plan that job in a period where I'll have a couple of days available and all the needed supplies at hand.
So glad I made my arrays serviceable right now, even though it will be no fun to separate my double baffle with the butyl rope stuck in between.
Did you ever share your impulse response? I can't remember. A raw IR response would do (showing the first 20 ms). Just curious to see if there's differences to be seen there.
. There's a enough "mystic" in this audio hobby anyway. I'm glad we got to do this sim exercise even if it does mean I'm now curious and crazy enough to think of re-wiring my arrays. My guess is that together we can advance this hobby trough this kind of sharing.Your explanation does start to make more sense to me. Re-wiring will be quite a mayor surgery for me. I'll have to plan that job in a period where I'll have a couple of days available and all the needed supplies at hand.
So glad I made my arrays serviceable right now, even though it will be no fun to separate my double baffle with the butyl rope stuck in between.
Did you ever share your impulse response? I can't remember. A raw IR response would do (showing the first 20 ms). Just curious to see if there's differences to be seen there.
Lurker on the thread here
+1! But how do you connect 25 TC9 drivers in parallel? Owen worked at Paradigm and it sounds like they custom made the subwoofer voice coil impedance 35ohms, then put 6 in parallel?
Owen, did the sim at Blackberry assume identical drivers?
Because I can see how slight tolerances could cause the problems you saw at Paradigm. We know these drivers differ in Mms, Re, Fs, sensitivity... A volume manufacturer could match, but that's not going to work for some who buys a few dozen from PE.
Who knows X well enough to see what Akabak thinks?
Because I can see how slight tolerances could cause the problems you saw at Paradigm. We know these drivers differ in Mms, Re, Fs, sensitivity... A volume manufacturer could match, but that's not going to work for some who buys a few dozen from PE.
Who knows X well enough to see what Akabak thinks?
I've send out a cry for help to xrk971. X did do some sims for me back when I had very similar results to fluid's. I hope he has some free time to fact check the sims we see here.