diy esl suggestions thread
Im going to attempt a DIY ESL build! Ive searched and even bought Roger Sanders book, but still have some specific questions.
Im looking to build an extremely high output 250/300hz-20khz+ (may augment with a line source super tweeter if ESL performance is abyssmal and price permits) ESL. I have the space for ~16" x 7'.
What is the best compromise diaphragm thickness for my specific passband?
What coating should I use on the mylar?
Im aware that several manufacturers use double sided tape provided by 3m to act as the spacer and glue. Is this a good alternative to epoxy? Is the bond strong and not decrease substantially over time?
What amplifier should I use? Would an extremely high power Pro-amp work? or should I consider something along the lines of an Adcom?
What should I look for in a high quality transformer? Roger Sanders mentions 8kv is a good start for a midrange ESL.
Re: diy esl suggestions thread
I'm not sure what you mean by extremely high output? I have a set of 18" x 46" (active area) panels and I can debone chickens with them. A super tweeter may be problematic due to the very different dispersion patterns. If you don't segment them, your dispersion will be similar to a laser, then crossed over to something that is way more dispersive. You shouldn't have any problem running your panels up to 20K unless you use very thick diaphragms.
I drive my panels to very loud levels with a 20 watt tube amp. If I step up to a citation II, that 60 watts will make my ears bleed.
Think about it this way, my panels have 4748 cm^2 effective cone area. That's one ******* HUGE woofer, and an insanely large midrange and tweeter.
Take what roger says with a grain of salt. It's a very accessible book, but I don't agree with everything he says. He seems to be obsessed with loudness (which seems to line up well withy our goals). I don't agree with him crossing over to the panels so high. I've tried it on my panels, I it has that same "phasey" sound that some of the Martin Logan speakers have. The lower I cross my panels over the better it sounds. I cross them at 130 Hz.
Re: Re: diy esl suggestions thread
Would it be a good idea to seperate the drivers into two seperate towers? I could build the ESL seperately and then construct a 7' H-Frame for the RS390HFs. How do I calculate the resonance of the h-frame and the effective baffle width (so I can calculate dipole rolloff)? Should I mount the drivers opposite (ie normal/inverted/normal/inverted) to cancel vibrations and lower even order distortions with no audible penalty (ie diffraction issues caused by basket/magnet obstruction)
I could probably cross lower, but that would drastically limit maximum SPL (4x displacement requirements, less surface tension), it would come dangerously close to the resonance of the diaphragm, which I am supposed to avoid (? I am supposed to cross before the resonance or can I cross during the resonance?), and it would put a strain on the transformer (i read that even the best ones can only cover 7 octaves)
The urethane foam which supposedly resists rotting only comes in 45 mil and 30 mil. Would 45 mil be large enough for such a low crossover point and the issue of acoustic coupling? The idea of 4 RS390HFs operating next to each of these is quite worrisome...
Does mylar come in sheets smaller than 3.8 microns? say 2 microns? I have a friend who works at a tooling shop and has access to thousands of catalogs where I can purchase all of these parts. What are specific things I should look for if I was looking for a product superior to mylar yet lighter and more durable (such as specific tensile strength, weight, etc etc).
Is the double sided tape superior to plexi+epoxy?
reactive loads? hmmm... (im not very good with electronics). Can you explain something specific (that may be provided by the manu. if I looked hard enough?) I heard they loved extremely high voltage rails...but arent so concerned with anything else.
Re: Re: diy esl suggestions thread
With 'phasey', do you mean colouration? Becasuse that's what I can hear with some of the ML. The enclosure is not resonance-free and is directly connected with the ESL panel.
There may be a severe trade off with a crossover at a low 130 Hz because you're running the ESL down to a region where phase cancellation eats away bass.
What I mean by phasey is a weird midrange sound like the woofer and the panel are both contributing, but are offset in phase too much and there is this muddled portion of the spectrum. Not really a coloration, but a odd funkyness that I can't quite describe.
My panels are 20" wide, so my pressure rolloff starts just around 700 Hz. I compensate for that with a shelving EQ.
What crossing over that low does is limit my max SPL due to the shelving EQ pushing diaphragm excursion up.
You aren't planning to do a passive crossover are you? If you are doing an active crossover, you can use a shelving EQ to fix the pressure rolloff. Otherwise you are going to either segment your panel to fix the pressure feed back 6 dB per octave rolloff, or you'll have to knock it down with a passive XO and that will kill your efficiency. IF you use an active EQ, it really doesn't matter what the efficiency is. that just effects the needed power.
Im using a combination of DCX2496 + DEQ2496 for crossover/equalization
famous last words: " I want to built an ESL...."
Welcome to the world of sorrow and distress ;-)
Well it´s not that bad, in fact it could be relatively easy, if You follow exact guidelines of working concepts. Surprisingly most guys starting such a project begin to think of tweaking before they already have a decent clue what a ESL is and which results one can reach with it..or not! :rolleyes:
First of all forget about extreme output. With regard to maximum SPL ESLs will not be the prime choice. Still though You can get more than enough high output for home uses with medium-sized panels (think of something of the size of the older ML Sequel). In fact even homecinema-stuff like the ´Terminator-theme´ or maximum dynamics like the CincinnattyPops playing ´RobinHood´ or ´Straussfest´ won´t stress a good panel much and the sonic performance is still far ahead of any dynamic speaker :D
The older MLs have two major probs. A terribly bad audio-tranny and a rather badly integrating bass solution, but they could play quite loud, because the conceptual basics were done right. Amongst those is a quite highish crossover-freq.
Using good trannies and dipoles for the bass can solve the sonic problems of such a concept (as ML shows with its magnificent Statements), or as it is done here in quite a similar way.
I´d recommend to study how the Statement is built because here imho everything is done as it should be done to get excellent sonic results as well as high output (ok, You might argue about diaphragm thickness and if it weren´t advantageous to drive this beast completely active). :devily:
What are the major points to look at?
Use a sufficiently large panel--10"x50" (25x125cm) or more even if You want to cross over as low as 250Hz. Not only does this raise the SPL but Distortion will be lower too.
Use as small the d/s distance as possible --above 250Hz 1/25"...1/16" (1.0...1.5mm) is ok.
Use very high mechanical tension of the diaphragm that places the ground resonance roughly one octave below the crossover-freq. High tension minimizes the effect of acoustic coupling with the bass (membrane can be blown into the stator by strong bass waves)
(ML reaches a ground resonance ~ 250Hz with 1/2mil (12µm), I reach 170Hz with 1/8mil (3,5µm). My recommendation is a membrane thickness between 3µm and 6µm. Thinner won´t withstand the high tension and thicker may result in a loss of detail.
Use genuine Mylar (Type S or C) or Hostaphan (RE) for the diaphragm. Those have proven to withstand the high tension over years.
When using sheet metal stators think of bowing them. This will stiffen and stabilize the stators considerably and it looks pleasing because it keeps the panel thickness very small without the need of additional bracing. With wire stators the design will be thicker and You don´t need to bow those. Using electrical segmentation You can influence the distribution character of the panel , as well as bowing broadens the dispersion slightly with sheet metal stators.
Use thin wire or thin sheet metal (~1/25"). Thicker material may restrict the bandwidth too much.
Use sheets with small holes and high openness (i.e. RV3/4) and insulate well (PET-epoxy powder, PA-powders like Rilsan or Corvel or laquer with PU-laquer as usable for staircases, or PVC for wire insulation). If possible try to round the hole edges by sanding or etching. This will reduce the electrical stress (ageing) and the risc of flashovers.
As spacers You can use the soft foam tapes of 3M --Yes the VHB-types are expensive, but they´ll save a lot of time and Your nerves) and I recommend not to use other brands.
Some epoxy-glues will work. While they resist a pull very strongly, they might fail under shearing forces, which is our case.
Too I don´t like the idea much to have a diaphragm pulled taut under high tension in a stretching frame for hours till the epoxy is completely hardened. But they may be prime choice with flat wire stator frames where the spacer is an integral part of the frame itself and where only a very thin layer of bonding material is needed. With those kind of frames I got good and stable glueing results alternatively with doublesided sticky tape from TESA for carpets.
The soft foam tapes immideately form a strong bond. There is this one risky moment when You actually place the membrane on the tape, because You have just one try to make it good! (You´ll most probabely need a helping hand here)
Because of the tape´s softness You have to take measures against the strong shearing forces of the membrane. I´ve done this (again have a closer look at ML) by pulling the membrane around the stator and glueing it to the backside of the stator too with a thin doublesided tape. This way the tension is preserved forever.
Searching the forum You´ll get answers about which coating works. Some of them can be bought, some of them can be mixed by yourself, some are expensive and some come nearly free, some are transparent, others not. Have a look and draw Your choice ;-
There is a remarkable difference between panels which are driven completely as one (sheet metal stators) and those that are segmented electrically (wire stators) and that is its capacitance. Capacitance increases with larger panel area and smaller d/s.
Sheet metal stators have a constant capacity over the frequency band (that can be calculated quite exactly from the dimensions) while electrically segmented stators have a varying and smaller capacitance. Since You have to keep the speakers impedance in mind (as specified in the amps manual) the transformation factor U depends on the design of the panel. As a guideline: the higher the capacitance of the panel, the lower the U. While medium to large sheet metal-panels will be happy with Us of ~1:50...1:75, most segmented wire stators can be driven with Us of >1:100.
Increasing the transformation factor rises the output voltage, but it costs bandwidth (and its not good for the sound and dynamics either) and puts more stress on the insulating materials (safety!!).
If You design for low Us You might even use standard offtheshelf toroidal power trannies which give excellent results (better than most ´specialized and expensive trannies btw). I use a pair of 230V/6V types per panel and got measurably and audibly better results than other trannies I tested so far -apart from those wonderful but expensive Plitron/Amplimo.
They won´t have enough inductance to drive a fullrange ESL properly but for hybrids with an elevated crossover point they are totally fine.
damn, just 10.000 sign allowed :bigeyes:
As a partner to a good panel a dipole bass/midbass can be a very fine choice.
The MLs sonic problems result from the inferior matching of panel and bass. Responsible for this are the different steepness of the acoustical filter character and strong differences between the distribution character of panel and bass. Seamless integration can be reached by solving those two probs. The first is a matter of designing the right crossover and the second is a matter of choosing the right working principle for the (mid)bass.
Again have a look at the pic above or the ML Statement. Both concepts work with a line of 6.5" drivers in a dipolar cabinet (U-shape with ML, A-shape with mine). So the dipolar linesource the panel forms is augmented by the dipolar linesource the 8 dynamic drivers form.
When You use a shaped cabinet You get a resonance effect that depends on the dimensions of the cabinet. I doubt that You´ll get enough bandwidth to reach >300Hz in good quality with drivers /cabinets larger than 8". I don´t know the RS390 driver, but the number indicates a 15” driver.
It will be hard if not impossible to reach the desired high crossover freq without running into resonance problems. I wouldn´t run such big drivers far higher than ~100Hz.
In this case I´d rather put the drivers in closed compartments and drive them actively as real Subwoofers below ~80Hz and would add a linesource of smaller 5”-7” in a dipolar cabinet, that fill the range between ~80hz and 300Hz. But heyyyy doesn´t that look similar the Statement again?? :D Did I already say I that I like the Statement as a very remarkable design? :D
Driving the ESL is not easy. The problem with most amplifiers is, that they are not designed to work stable at such complex loads. The impedance an amplifier sees when driving a loudspeaker consists of an mainly resistive (ohmic) part and a reactive part. Only the resistive part is used to create sound (real power) while the reactive part represents power that the amplifier must supply for, but that is simply wasted. This reactive power ´oscillates´ between amp and speaker just putting additional load on the power supply of the amp and generating additional heat. The phase angle (electrical phase measurement of the speaker) gives an idea what to expect. A phase angle of 0° means only real power (pure resistance), a angle of eg 45° means that real and reactive power are of same magnitude. The complete power the amp has to deliver is the vectorsum of both. To deliver lets say 10W real (+10W reactive) the amp has to supply ~15W. Most dynamic speakers stay within a phase-band of +-45°, so a slightly oversized (1.5 times) powersupply and heatsink is still ok. A ESL on the other hand is nearly a pure capacitance, meaning an phase angle of theoretically 90°. In this case there is only reactive power and no real power at all. In praxi (with the help of the tranny) the phase angle is lower than 90°. So there is a part of real power that is useful, but it is relatively small, compared to the reactive power. So the amp has to have very oversized powersupplies and heatsinks to be able to supply reasonable amounts of real power to the ESL and the huge amounts of reactive power. The problem gets worse the better the tranny´s quality is. A good lowloss tranny like the Plitron/Amplimo generates phase angles up to 85° and resistive values of less than 1Ohm with an sheet metal stator, while a rather bad quality EI-Type tranny gives 70° and >1Ohm. So the better tranny is the worse one for the amplifier.
The huge phaseshift generates a second problem for all amplifiers with global feedback. The transient response is spoiled and in many cases the amp may even start to oscillate, thereby heating up and up till some fuse blows or the amp dies.
The ´faster´ the amp is, the greater the problem is. Amps without global feedback like some SEs (wether tube or sand) are usually fine as are generally tube amps.
The upcoming d-amps are probabely the best choice, since they show themselves quite unimpressed with complex loads and can drive even huge panels exceptionally well (especially at elevated SPLs). The costs for a good D-amp are comparably lower than for any other technology. So have a look what is going on there ;)
The RS390HF have extremely low inductance (1.0mH if I recall) and a crazy resonance peak right around 1600-1800hz (the peak starts to slope slowly form right around 1.2khz and then things start to get nasty around 1.5khz). The third harmonic of this peak (assuming 1600hz) is 533hz and distortion will start to rise >400hz. I will be crossing nearly an octave below this, so non-linear distortion should be no issue.
Also heres a review of the HF series http://www.mfk-projects.com/rs_12_hi-fi.htm
quoted from the review
Anyway, one amazing thing about this woofer is its ability to perform well up to several hundred Hz. This is a little unusual for this kind of heavy high excursion driver. I guess if you are not too critical you could use this driver to 1Khz but personally I wouldn't use any 12" driver that high except in certain pro-audio applications. I would feel very comfortable however using this driver to 300 Hz. In most of the applications I have planned for it I will not use it above about 140 Hz but that is more because this is what the application calls for rather than any concern with using this driver higher in frequency.
I think this driver will be a first rate performer in both di-pole applications and box applications. It seems about as good as the XLS in the mid bass region with some frequencies possibly better on the XLS and others better with the RS. Generally the two are close in the mid bass region. The RS is better at very low frequency and will also be better than the XLS at frequencies above 100 Hz. The RS is also the quieter of the two making it a much better choice for any dipole application than the XLS.
Linkwitz uses XLS drivers in his orion design, and these drivers are claimed to perform better. Are you confident they could not reach 300hz?
Is there a difference in perceived detail between 3.8 micron and say 6 micron?
Also, what amplifier would you recommend that is <$600? Im perfectly willing to buy used (as long as its in good condition) on ebay or one of the forums.
What specific tape would you recommend (part numbers and a website would be quite helpful)
Where would I buy those transformers? (part numbers/website would also be very helpful)
- 2 x XLS-12 in H-dipole bass (HxWxD 70x36x40 cm)
- 8 x ARN 6" midrange element dipole line array, 150x25cm panel
- 2x100xm feet long ribbon tweeter
The ribbon was on the woofer and midrange line array was as close next to others as it can be. There was ~10cm air gab between. I used Behringer DCX-2496 active filter to set sharp crossover filters.
The room was quite big: I listened 2m away from the system. Walls were at least 1m apart from the speaker and I had 3 meter space behind me.
Crossover frequency 1 kHz was pretty good for the tweeter. But even then you could really hear that the sound came from 2 different direction: the tweeter and the midrange tower. And there was "only" ~30 cm between center of elements. Can you squeeze that with ESL? No. So do not use ESL as a tweeter only. You propably cannot place it close enough to any midranges.
Crossover frequency 80Hz was the best I could use between woofer and midrange tower. Above 100Hz midrange and woofer separated too much. Once again, I heard something I cannot explain better than being 2 different and annoying sources. No good. I can use 12" XLS up to 300Hz alone but it just don´t work with other dipole sources !!! I bet this is the case also with 10" woofers. It´s not matter of frequency response of a single element but how far these elements are from each others and what their polar responce is. You propably can never make ESL + dipole woofer as a midrange sound good if you set a crossover frequency above 100Hz.
Maybe if you listen 10 meters away? :)
Just my opinion based on subjective listening tests with dipole elements.
|All times are GMT. The time now is 12:16 PM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Copyright ©1999-2016 diyAudio