haha yeah an incredible steady hand 
well i did make the first CNC out of MDF wood everything drilled out of hand, then i replaced components made witht he verry same crappy machine. then after a year i thought it was time to drop some money into alumnium. still as cheap as posssible since i was still a student. i designed my machine with the ballscrews and linear bearings i already had, and only used existing almumnium profieles. since my wooden cnc could not mill out complete parts of 15 mm thick alumnium. i made a pocket into wood where i could drop in the alumnium flat stock or L profiele i ordered online , just to drill the holes and the countersunk stuff. since the cnc was crappy at best, i needed to work around it. still i could drill a pilot hole with far more accuracy as i could do by hand.
so thats how i ended up with this cnc mill. started out with 8 wooden planks , 3 all threaded rods and 3 cheap steppers and a controller board of only 40 euro. used gass hoses as coplings , steel flatstock as linear bearing with skate barings etc its funny if i look back.
well i did make the first CNC out of MDF wood everything drilled out of hand, then i replaced components made witht he verry same crappy machine. then after a year i thought it was time to drop some money into alumnium. still as cheap as posssible since i was still a student. i designed my machine with the ballscrews and linear bearings i already had, and only used existing almumnium profieles. since my wooden cnc could not mill out complete parts of 15 mm thick alumnium. i made a pocket into wood where i could drop in the alumnium flat stock or L profiele i ordered online , just to drill the holes and the countersunk stuff. since the cnc was crappy at best, i needed to work around it. still i could drill a pilot hole with far more accuracy as i could do by hand. so thats how i ended up with this cnc mill. started out with 8 wooden planks , 3 all threaded rods and 3 cheap steppers and a controller board of only 40 euro. used gass hoses as coplings , steel flatstock as linear bearing with skate barings etc
its funny if i look back.
So this is interesting - I have a large number of Acoustat Panels - any benefit of stacking them? I posted about this a long while ago (stacking Acoustat panels) and really can't remember what the verdict was - don't think I ever tried it. Although I do remember putting three acoustats panels in a triangle (edge to edge) - nothing of real benefit there from what I remember...
There is no benefit compared to the 3 panels side to side. its just that it is possible to have a smaller panel with higher output in the lower frequencys so if you can live with 3 huge panels then there is no need to go stacking . if you got the room thats the best for sure!! (if they at least dont play fullrange)
another problem is that if you stack those panels you wil have a rather big space between the 2 membrames, in my design the rear stator of the first ESL is the the front stator of the next behind it. that saves allot of space in between.
for low frequencys that is not a big deal, so if you want extra low end you could stack a few, downside resonance goes up!
so if you can live with 3 huge panels then there is no need to go stacking . if you got the room thats the best for sure!! (if they at least dont play fullrange)another problem is that if you stack those panels you wil have a rather big space between the 2 membrames, in my design the rear stator of the first ESL is the the front stator of the next behind it. that saves allot of space in between.
for low frequencys that is not a big deal, so if you want extra low end you could stack a few, downside resonance goes up!
Last edited:
got 2 stacks playing! yeaah, single panel resonance is around 400Hz 2 around 540hz 3 i dont know yet one thing i notice is a really nice benefit and im not sure if there is a theory behind it and if its even possible, but it seems to work, the thing is with one panel i can only up the HV bias to a certain point and it stucks into a stator,so i let it drain and give him al little less untill it plays normally.
But with 2 i can add considerable more bias. my theory is because the resonance goes up the membrame is more stable and can handle some more bias. Bolsert you know anything about that >?
i am a bit exhausted , bussy with these for 11 hours non stop now, i will do some propper measurements tomorow.
i will start off with a basic sigmented panel based on resitors i have that i can halve when a panel is added. Btw when i add the third panel do i substract 1/3 of the resistance i used for 2 panels?
so i start with 300k on a sigment from a single panel, i then add one panel so half the resistor value, ad another panel and then take 1/3 of the 150k ? and leave it at 100 k? Bolsert is that correct?
one thing i notice is a really nice benefit and im not sure if there is a theory behind it and if its even possible, but it seems to work, the thing is with one panel i can only up the HV bias to a certain point and it stucks into a stator,so i let it drain and give him al little less untill it plays normally. But with 2 i can add considerable more bias. my theory is because the resonance goes up the membrame is more stable and can handle some more bias. Bolsert you know anything about that >?
i am a bit exhausted , bussy with these for 11 hours non stop now, i will do some propper measurements tomorow.
i will start off with a basic sigmented panel based on resitors i have that i can halve when a panel is added. Btw when i add the third panel do i substract 1/3 of the resistance i used for 2 panels?
so i start with 300k on a sigment from a single panel, i then add one panel so half the resistor value, ad another panel and then take 1/3 of the 150k ? and leave it at 100 k? Bolsert is that correct?
Last edited:
Some last photo's on the contruction for tonight
the acid bath i used to etch the spacers, instead of milling it all. my dishwash rack with all components washed and rinsed.
glueing the spacers, with 3m 30 nF (magnepan glue)
and the glued membrames on the stators. i used a classic tape method to stretch this time, since am not sure yet what tension they should have, i just gave it as much as i dared to. on the 3um foil
for tonightthe acid bath i used to etch the spacers, instead of milling it all. my dishwash rack with all components washed and rinsed.
glueing the spacers, with 3m 30 nF (magnepan glue)
and the glued membrames on the stators. i used a classic tape method to stretch this time, since am not sure yet what tension they should have, i just gave it as much as i dared to. on the 3um foil
Attachments
This is not something I had noticed...yet
Are you playing music while you are increasing the HV until diaphragm collapses to a stator? If so, then I think your theory is reasonable. If it is happening with no music playing, I’m not sure. Do you have flashing neon lamp charging indicators hooked in line with each HV lead? It is possible that when the second diaphragm is hooked up to the opposite polarity HV that there is some increased leakage currents not experienced with just the single diaphragm hooked up that is loading the HV supply down, reducing the actual amount of charge on the diaphragms for a given setting of the HV supply.If you are wanting to keep the response the same as you add sections, you decrease the resistance by the same ratio you are increasing the capacitance. So if you use resistance R with one section, use R/2 with two sections, R/3 with three sections, R/4 with four sections, etc…
Now, a couple questions for you
1) Was there a reason you switched from the staggered hole alignment used in your original round stators shown back in post#10? Perhaps the straight alignment is easier to program in the CNC software. In any case, for the same open area percentage, the staggered hole alignment will provide a stiffer stator…if that becomes an issue.
2) Can you tell me what the center-to-center distances were for the 2mm and 1mm stator holes shown in post# 35 and #36?
Thanks for sharing!
Attachments
Since single Acoustat panels can already be driven to the point of diaphragm slapping stators there is no increased output capability at low frequencies with stacked panels. You will just be able to reach the slapping threshold with roughly half the voltage input to the panels required for overdriving a single panel.
More information on the low-frequency aspect of stacking two Acoustat Panels was posted in the other thread:
http://www.diyaudio.com/forums/planars-exotics/226890-my-acoustat-panel-experiment.html#post3309417
The high frequency aspect had not been mentioned, but it is not good news either. With two panels stacked back to back, the two diaphragms will be roughly 35mm apart. Using the spreadsheet from post # 25, you can see that the phase difference between the two diaphragms will put a huge notch in the upper midrange.
Attachments
Nice comb filtering
About bias i dont have blinkers on them, and yes i changed bias while playing music, up it until i hear the sound change and some faint crackling of discharging. discharge the whole thing by shorting bias - and + then back of and turn on the bias again. ill do a remeasurement of max bias today (not in volage dont have an meter that can get so high) i turn it to the max alloud add another stator measure SPL, then see if i can upp the bias more, and remeasure.
well about the holes i cant seem to find the file of the first 2 stators i made... but the 1 mm version has 1.9 mm from center to center.
Yes you are right about the layout of the holes, i did change it to straight, this was because it mirrored better when i rotated one stator 180 degree, in the end i did not rotate them at all so i could use the staggered manner the next time to have some more stability.
First measurements.
1 panel 2 stacked 3 stacked. fullrange instead of sigmented, i sigement after the test, im not sure why i wanted that in the test. its not needed at all.
You can see clearly see that up till 15-18khz it still adds considerable SPL. 1 to 3 stacked diference of 12dB ~!! thats pretty much. it also becomes a bit more ragged. im wondering if this is the effect of all the smal holes stacked behind each other. and maybe some resonances of the combined membrames and support dots. **** i now realize i should have spaced dots on every panels slightly different, so i can smooth out the resonances instead of amplifieng them with each stack.
next i want to emmbed some verry fine mesh in between the stacks to see if i get rid of some peaks and dips.
ans also want to look what happens when i remove the stator spacer with alle the holes for an open spacer (that mean the copper is facing one of the menbrames) but might reduce the effects of the holes. i could always revert to painting one side of the stator if needed (i doubt it)
another thing that worries me slightly is the resonace of stack 1 and 2 are like one would expect, with 3 it looks split into 2 resonances.
the resonace of the 2 stack is still visible in the 3 stack and a new one from the 3 stack a little higher
1 panel 2 stacked 3 stacked. fullrange instead of sigmented, i sigement after the test, im not sure why i wanted that in the test. its not needed at all.
You can see clearly see that up till 15-18khz it still adds considerable SPL. 1 to 3 stacked diference of 12dB ~!! thats pretty much. it also becomes a bit more ragged. im wondering if this is the effect of all the smal holes stacked behind each other. and maybe some resonances of the combined membrames and support dots. **** i now realize i should have spaced dots on every panels slightly different, so i can smooth out the resonances instead of amplifieng them with each stack.
next i want to emmbed some verry fine mesh in between the stacks to see if i get rid of some peaks and dips.
ans also want to look what happens when i remove the stator spacer with alle the holes for an open spacer (that mean the copper is facing one of the menbrames) but might reduce the effects of the holes. i could always revert to painting one side of the stator if needed (i doubt it)
another thing that worries me slightly is the resonace of stack 1 and 2 are like one would expect, with 3 it looks split into 2 resonances.
the resonace of the 2 stack is still visible in the 3 stack and a new one from the 3 stack a little higher
Attachments
Last edited:
Well i removed the perforated layer between the first back stator and the second membrame, this is the result.
Green where the 3 stators just stacked with the perforated plate in place (2 in total)
Red is without these plates, much better you can see ONE resonance just like we had with the 2 and one stack. around 560hz then when you look at the THD this improved allot.
Blue Same, but now felt is added at the front and the rings are sigemented. no Eq used yet in the crossover. again looks better less distortion and overal ok response, at 1.8 khz the drop could be corrected by an low shelf, or by adding a baffle because my guess is this is where the whole contruction does not function as a baffle anymore. maybe ill check.
glad i got rid of some humps tobad we cant lose the other holes
so in short, the machining of 4 of those perforated plates wich took around 3 hours in total, gave me some extra distortion at least i can scratch 3 hours from the next build
Green where the 3 stators just stacked with the perforated plate in place (2 in total)
Red is without these plates, much better you can see ONE resonance just like we had with the 2 and one stack. around 560hz then when you look at the THD this improved allot.
Blue Same, but now felt is added at the front and the rings are sigemented. no Eq used yet in the crossover. again looks better less distortion and overal ok response, at 1.8 khz the drop could be corrected by an low shelf, or by adding a baffle because my guess is this is where the whole contruction does not function as a baffle anymore. maybe ill check.
glad i got rid of some humps tobad we cant lose the other holes
so in short, the machining of 4 of those perforated plates wich took around 3 hours in total, gave me some extra distortion
at least i can scratch 3 hours from the next build Attachments
Last edited:
Hi,
All of your work is quite interesting. However in my opinion such small panels would suite ESL headphones better. Doubling linear dimension would give a lot more than 6dB, perhaps in the order like 12 dB, so this could be considered as an option.
For me THD figures look quite mistrusting. 35 dB below the level is not what is expected from an ESL, especially at high frequencies. If you have another source(like other ESL panel, or good quality dynamic speaker) I would suggest you cross test your measurement equipment first. Secondly it's quite weird that stacked panel shows so much higher distortion levels. Are you sure nothing is overloaded, measurements are performed at exactly same position, etc. ? It's quite possible that the reason actually something different than stacking itself.
The first thing that comes to mind is that two panels would increase load on power supply, causing the voltage to collapse. Therefore increasing voltage would only restore it to previous level. It is epecially probable if one or both panels are leaky. Have you measured exact voltage supplied by EHT after connectig one and then two panels? It could be done by using a very large HV resistor divider(e.g 10 GOhms), connected in series with DMM and measuring voltage across terminals.
Regards,
Lukas.
All of your work is quite interesting. However in my opinion such small panels would suite ESL headphones better. Doubling linear dimension would give a lot more than 6dB, perhaps in the order like 12 dB, so this could be considered as an option.
For me THD figures look quite mistrusting. 35 dB below the level is not what is expected from an ESL, especially at high frequencies. If you have another source(like other ESL panel, or good quality dynamic speaker) I would suggest you cross test your measurement equipment first. Secondly it's quite weird that stacked panel shows so much higher distortion levels. Are you sure nothing is overloaded, measurements are performed at exactly same position, etc. ? It's quite possible that the reason actually something different than stacking itself.
The first thing that comes to mind is that two panels would increase load on power supply, causing the voltage to collapse. Therefore increasing voltage would only restore it to previous level. It is epecially probable if one or both panels are leaky. Have you measured exact voltage supplied by EHT after connectig one and then two panels? It could be done by using a very large HV resistor divider(e.g 10 GOhms), connected in series with DMM and measuring voltage across terminals.
Regards,
Lukas.
Last edited:
the THD measurements needed to be taking with a huge grain of salt. since when i look at the noise level they are almost as high as the THD. my laptop makes noise and also it is measured in room half a meter away. i measured allot of panels, and thd is always in this region. not mater what i use, even my open baffle woofer is in the same region. wich ofcourse hes is not. obvious resoances still show up, as show in previous post.
what do you mean by linear dimension ?
if you mean go bigger then you missed the point. i can go big and ad another 24 dB if i want, the thing is i had a panel with a certain size in mind because i like the dispersion of it , now i got the same size panel with the same asociated dispersion but it plays 12 db louder then the original.. thats my point!. want good dispersion you have to go small, going small means a loss of spl. this is my form of solving that. making a taller panel has other problems like being big, and i dont like the fact you cant even be in the room enjoy them you need to sit in front of them at the right hight or you wont hear highs. ofcource you can build them 2 meters tall then that is solved. but i might be not the only one that dont want 2 meter high panels in there room. i wanted to make this panel to mate to a set of OB bass drivers, using a large panel screws the idea of keeping it relative small.
you know how to increase accuracy in the thd measurements ? since i have no idea where to start. i am a total newb in distortion measurement!. and how to make them proper
Yes the question about the higher bias might be the drop, i did not see any increase in spl in my latest test when upping the bias after a stack, we might forget about me saying that
Pretty loud headphones
what do you mean by linear dimension ?
if you mean go bigger then you missed the point.
i can go big and ad another 24 dB if i want, the thing is i had a panel with a certain size in mind because i like the dispersion of it , now i got the same size panel with the same asociated dispersion but it plays 12 db louder then the original.. thats my point!. want good dispersion you have to go small, going small means a loss of spl. this is my form of solving that. making a taller panel has other problems like being big, and i dont like the fact you cant even be in the room enjoy them you need to sit in front of them at the right hight or you wont hear highs. ofcource you can build them 2 meters tall then that is solved. but i might be not the only one that dont want 2 meter high panels in there room. i wanted to make this panel to mate to a set of OB bass drivers, using a large panel screws the idea of keeping it relative small. you know how to increase accuracy in the thd measurements ? since i have no idea where to start. i am a total newb in distortion measurement!. and how to make them proper
Yes the question about the higher bias might be the drop, i did not see any increase in spl in my latest test when upping the bias after a stack, we might forget about me saying that
Last edited:
Here a measurement of the same panel at different volumes (close miced with sigmentation as you can see in the drop in freq),
So because i played my sweep at a lower level for the red curve the difference between the THD and the curve is now dropped by 20 db. so im not sure how to normally read these figures but it looks to me a bit weird.
its not because i play the sweep 20 db louder i have 20db less thd.
How on earth do you read these measurements ?,
Your statement of the 35 dB below measurement level now increased to 50 db, only by adding some volume.
am i missing something here ?
Attachments
This is what Rew makes of it, although sometimes its higher sometimes its lower. my soundcard is not calibrated.
0.344 % total harmonic dist. got as low as 0.2 something, i dont know what my soundcard does, as well as my minidsp wich is connected in between as well. and the Umik 1, did use the calbritaion file ofcourse. but it does not acount for distortion
0.344 % total harmonic dist. got as low as 0.2 something, i dont know what my soundcard does, as well as my minidsp wich is connected in between as well. and the Umik 1, did use the calbritaion file ofcourse. but it does not acount for distortion
Attachments
Last edited:
Hi,
It is likely that the sound card+miniDSP+transformer contribute only small amount to distortion. The latter should be measured as some appear considerably better than others. Typical for mains toroids is between 0.1% to 0.05% THD at low frequencies and this drops at mid range and above. This is assuming that series impedance is low between transformer and amp + DC offset is no more than few millivolts. Therefore IMO passive crossovers should be avoided while using transformers.
Usually I have measured very low levels of combined distortion, less than 0.1% from few hundred hertz and up. Of course speakers are much bigger. You may also find that distortion levels somewhat depend on signal level.
My proposal for larger panels would allow considerably lower cross over point. In such a case ESL could act both as mid-range and treble unit, as opposed to currently restricted to high frequencies only. Keep in mind that most of music has peaks that range up to 300 to 400 Hz. That means above this range active EQ or higher step up could be increasingly used to compensate for lower sensitivity of panels.
Of course, it adds some complexities. With such a small D/S spacing and thin PCB laminate more supporting rings would be necessary. However you may consider increasing D/S spacing to let's say 1 mm and compensate by using higher step up & EHT voltage.
Regards,
Lukas.
It is likely that the sound card+miniDSP+transformer contribute only small amount to distortion. The latter should be measured as some appear considerably better than others. Typical for mains toroids is between 0.1% to 0.05% THD at low frequencies and this drops at mid range and above. This is assuming that series impedance is low between transformer and amp + DC offset is no more than few millivolts. Therefore IMO passive crossovers should be avoided while using transformers.
Usually I have measured very low levels of combined distortion, less than 0.1% from few hundred hertz and up. Of course speakers are much bigger. You may also find that distortion levels somewhat depend on signal level.
My proposal for larger panels would allow considerably lower cross over point. In such a case ESL could act both as mid-range and treble unit, as opposed to currently restricted to high frequencies only. Keep in mind that most of music has peaks that range up to 300 to 400 Hz. That means above this range active EQ or higher step up could be increasingly used to compensate for lower sensitivity of panels.
Of course, it adds some complexities. With such a small D/S spacing and thin PCB laminate more supporting rings would be necessary. However you may consider increasing D/S spacing to let's say 1 mm and compensate by using higher step up & EHT voltage.
Regards,
Lukas.
Last edited:
ah you mean bigger dc spacing, well problems is with the high frequencys. higher spacing say from 0.5 to 1mm stills mall but it results in me dropping one stack, because there will be out of phase material at 17-18 khz wich results in a huge drop. 3 panels was max based on 0.5 spacing. (so 2 mm membrame to membrame)
btw 0.1 percentTHD i am according to REW at 0.34% we are talking about 0.2 percent. biggest source should be the esl, but i am in no way trained or have much experience with measureing THD, so 0.344 without calibrated soundcard is prety ok for me, and i bet it should be easy to get a 0.1 or more of that number by doing it right still noone is gone hear 0.2 % to be honest. The stepup ratio i use now is already 120 and i dont use teroid power trannies, its a trannie for esl's.
and yes bigger area could be a huge problem with such small dc spacing its hard already i cant just ad some more hv or it gets sucked into a stator. and the walhalla for me at least would indeed be a slightly bigger version that goes down to around 300 or so.
but as we said thats hard with these spacings, if i increase that i need a bigger area since i cant stack 3 but only 2 so i lose almost 6 db. no free luch so it seems
i planing on winding my own trannies and i will hopefully end up with 1-150 thats should be enough when crossed at 800 or something with a slightly lower bias voltage to mate with my 2 12 inch woofers in paralel with baffle step correction. i must say the panel already can play at ears shredding levels, it really starts to ring in my ear when i measured for the THD. not nice.
but i dont want to hold back the woofers to much on the minidsp for dynamics etc
btw 0.1 percentTHD i am according to REW at 0.34% we are talking about 0.2 percent. biggest source should be the esl, but i am in no way trained or have much experience with measureing THD, so 0.344 without calibrated soundcard is prety ok for me, and i bet it should be easy to get a 0.1 or more of that number by doing it right
still noone is gone hear 0.2 % to be honest. The stepup ratio i use now is already 120 and i dont use teroid power trannies, its a trannie for esl's. and yes bigger area could be a huge problem with such small dc spacing
its hard already i cant just ad some more hv or it gets sucked into a stator. and the walhalla for me at least would indeed be a slightly bigger version that goes down to around 300 or so. but as we said thats hard with these spacings, if i increase that i need a bigger area since i cant stack 3 but only 2 so i lose almost 6 db. no free luch so it seems
i planing on winding my own trannies and i will hopefully end up with 1-150 thats should be enough when crossed at 800 or something with a slightly lower bias voltage to mate with my 2 12 inch woofers in paralel with baffle step correction. i must say the panel already can play at ears shredding levels, it really starts to ring in my ear when i measured for the THD. not nice.
but i dont want to hold back the woofers to much on the minidsp for dynamics etc
Last edited:
Hi,
If you plan to wind your own trannies keep in mind that doing it by hand is a serious challenge. In the past I have tried this, by using EI tube amp cores and self-built winder that had winding counter on it and 1-1 ratio hand roller. I have abandoned this since the amount of time required for the job was insane, each layer required very consistent and high quality winding and insulation layer.
IMO your target step-up of 1:150 is far too high for D/S spacing of 0.5 mm.
Treble unit of quad esl 57 had similar D/S and was prone to arcing if driven above something like 15 Vrms into ~1:110 trafo.
Higher power amp is a lot cheaper to make than high turns ratio step-up. Since your project is only restricted to high frequencies, core saturation is excluded from the equation. So lower turns ratio like 1:50 or 1:75 is favorable in combination with higher-voltage capable amplifier.
Edit : LM3886 amps are quite useful in driving reactive loads, considering their high peak current capability and good protection. However they fail quite fast if asked to drive an ESL transformer in bridge mode. I do not know exactly why but it looks like their protection circuitry interacts when invoked, causing failure of one of chips and sending high frequency signal to transformer, causing further damage due to over-voltage to ESL panel.
Regards,
Lukas.
If you plan to wind your own trannies keep in mind that doing it by hand is a serious challenge. In the past I have tried this, by using EI tube amp cores and self-built winder that had winding counter on it and 1-1 ratio hand roller. I have abandoned this since the amount of time required for the job was insane, each layer required very consistent and high quality winding and insulation layer.
IMO your target step-up of 1:150 is far too high for D/S spacing of 0.5 mm.
Treble unit of quad esl 57 had similar D/S and was prone to arcing if driven above something like 15 Vrms into ~1:110 trafo.
Higher power amp is a lot cheaper to make than high turns ratio step-up. Since your project is only restricted to high frequencies, core saturation is excluded from the equation. So lower turns ratio like 1:50 or 1:75 is favorable in combination with higher-voltage capable amplifier.
Edit : LM3886 amps are quite useful in driving reactive loads, considering their high peak current capability and good protection. However they fail quite fast if asked to drive an ESL transformer in bridge mode. I do not know exactly why but it looks like their protection circuitry interacts when invoked, causing failure of one of chips and sending high frequency signal to transformer, causing further damage due to over-voltage to ESL panel.
Regards,
Lukas.
Last edited:
The winding is gone be fine i got an old Micafill winding machine standing here doing nothing. yeah you might be right with the turn ratio,, reason why i opt for such high ratio is i am using class D TPA3116 amplfiers in this build they cant put out the amount my main amp can. another option would be a small gain before the Tpa3116 since its only high frequency's there is enough room left in the amplifier. i should hook the tpa3116 up tomorow see how loud it can go, on my main amp its unberable how loud it Can go.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.