You will be just fine with 6 resistors per bank for voltage and power ratings even with high level sine wave testing.
For typical music listening they will likely never even get warm.
If I recall correctly, you plan to use 115:1 step-up ratio with an amp that can put out about 30Vrms. So you would be applying about 3.5kVrms to the stators. Attached pic shows voltage and power requirements for the first 10 resistors when sine wave testing at the 3.5kVrms maximum level using the parameters you chose in Post #70.
You can see things aren’t quite as bad as you think they might be. It is not the voltage applied to one end of the resistor that matters, but the voltage difference between the two ends. Usually the voltage and power requirements are only of concern for the first few resistors in the ladder where there is significant voltage differences between the adjacent segments.
Attachments
So wouldn't the fact that folks put tape (etc) on the diaphragm that essentially makes the ESL a collection of little panels play into this?
Or to try and put it another way, if I put no spacers/support throughout a 48" H X 14" W panel I would expect the Q to be quite different than if I put spacers horizontally every 1" (reductio ad absurdum).
So shouldn't we be thinking about the "panel" sizes in terms of the distance between spacers? e.g, more realistically, 6"H X 14"W with about 8 of them stacked on top of each other.
I've been noodling this a bit as I am thinking about whether it is better to have lots of horizontal spacers or 2 vertical spacers (which have more contiguous membrane area).
Yes I should of noted that there will be 12 resistors per *electrical* segment. i.e. 6 resistors for each physical segment "side".
BTW this is what I got in your spreadsheet.....
30.05 (Vrms) input limit
Vin: 28.00 (Vrms) amplifier voltage
Step-Up: 120.0 step-up ratio
Vbias 3,800.0 (VDC) bias voltage
Line Source: Finite Type of Line Source
Vstators 3,360 (Vrms) stator-to-stator voltage
4,752 (Vpeak) stator-to-stator voltage
Two separate issues to think about here…
1) Radiation impedance: If all portions of your panel are receiving the same stator drive voltage (which it will at frequencies < fL) the radiation impedance loading on the diaphragm will be the same not matter if you use vertical spacers, horizontal, or no spacers at all. So, the resistive part of the radiation impedance that golfnut described providing the damping to lower Q will be the same.
2) Resonance frequency: For a given diaphragm tension, the resonance frequency for rectangular membranes is mainly defined by the width of the panels. For example, suppose you divide your panel vertically into three sections 45” x 4.5” and tension the diaphragm to achieve resonance of 100Hz. With the same tension, resonance would still be ~100hz if you divided the diaphragm into 10 horizontal sections with 4.5” x 14”
In summary, for a given panel:
- radiation impedance dependents only on the overall dimensions of the panel
- resonance frequency will be dependent on distance between spacers
You didn't mention what value resistors you got...
You probably already know this, but to increase voltage capability you need to put the resistors in series. For increased power capability series and/or parallel combinations can be used.
Looks good
Hi bengel
Hopefully you have the impression that this issue of radiation impedance etc is complicated - it is
To add to bolserst's comments ... think of the system as a RLC circuit
R is the radiation resistance - determined by panel size and shape
L is the airmass attached to the membrane - also determine by panel size and shape
C is the membrane compliance - determined by membrane tension, distance between spacers etc.
Q=2.pi.fo.L/R
The compliance affects the Q indirectly through the resonant frequency
fo = 1/2/pi/sqrt(L.C).
What makes it complicated is the fact that there are many different ways the membrane can move - its not as simple as a single current through an RLC circuit. One way to represent the movement is as the sum of many different modes - each characterised by separate resonant frequencies. We are generally only interested in the lowest frequency because that will have the lowest R ( as shown in the picture in previous post) and therefore the highest Q. By putting in silicon dots, bit of tape etc, you are making the compliance different for different modes. To my mind its a messy way of supressing the resonance.
Another complication is that most ESLs have multiple sections in a panel. You might think that they each have their own resonant frequency - but because they couple to each other through shared air movement, they are in fact coupled and do not behave independently. (look up coupled oscillators on wikipedia ) Again, the collective movement can be represented as a bunch of different modes. The lowest frequency mode has all of the sections resonating together. If I find time today, I'll see if I can find a good youtube video showing the effect.
Hope that's added to the fog
regards
Rod
Hopefully you have the impression that this issue of radiation impedance etc is complicated - it is
To add to bolserst's comments ... think of the system as a RLC circuit
R is the radiation resistance - determined by panel size and shape
L is the airmass attached to the membrane - also determine by panel size and shape
C is the membrane compliance - determined by membrane tension, distance between spacers etc.
Q=2.pi.fo.L/R
The compliance affects the Q indirectly through the resonant frequency
fo = 1/2/pi/sqrt(L.C).
What makes it complicated is the fact that there are many different ways the membrane can move - its not as simple as a single current through an RLC circuit. One way to represent the movement is as the sum of many different modes - each characterised by separate resonant frequencies. We are generally only interested in the lowest frequency because that will have the lowest R ( as shown in the picture in previous post) and therefore the highest Q. By putting in silicon dots, bit of tape etc, you are making the compliance different for different modes. To my mind its a messy way of supressing the resonance.
Another complication is that most ESLs have multiple sections in a panel. You might think that they each have their own resonant frequency - but because they couple to each other through shared air movement, they are in fact coupled and do not behave independently. (look up coupled oscillators on wikipedia ) Again, the collective movement can be represented as a bunch of different modes. The lowest frequency mode has all of the sections resonating together. If I find time today, I'll see if I can find a good youtube video showing the effect.
Hope that's added to the fog
regards
Rod
I got 6x5.6 kOhm for the 1st segment and 5x9.1+1x8.2 kOhm for the other segments (physical).
Thanks so much for the responses... a lot to noodle here
. A more practical question I have here is... is there a way to estimate the resonant frequency based upon the panel width, material and tension? All I have to go on right now, is stretch it to 1% to 1.5%.
Trying to think of a way to stretch it and measure the resonance before affixing to the stator panel.... (have a couple of ideas that I'm not sure will work yet).
I wanted to post a little tip that is helping save a ton of time and aggravation putting these wire stators together.
Basically, I've found that if you take the threaded rod and put a strip magnet directly under it, you can simply "thread" the wires in between the threaded rod and the magnet strip. This has the advantage of keeping the threaded rod in place and keeping the wires from jumping out of their grooves.
Once you do one end, run a bead of glue across in one or more places, let it dry, then you can "comb out" the rest of the wire lengths to get them spaced right without the wires sliding around on you.
See attached picture, hope it helps someone.
Basically, I've found that if you take the threaded rod and put a strip magnet directly under it, you can simply "thread" the wires in between the threaded rod and the magnet strip. This has the advantage of keeping the threaded rod in place and keeping the wires from jumping out of their grooves.
Once you do one end, run a bead of glue across in one or more places, let it dry, then you can "comb out" the rest of the wire lengths to get them spaced right without the wires sliding around on you.
See attached picture, hope it helps someone.
Attachments
Hi Bengel
I promised to find a video showing the different modes of oscillations for coupled oscillators - see https://www.youtube.com/watch?v=kvG7OrjBirI . Note especially that the mode where all the masses (representing the sections of an ESL panel) move together has the lowest frequency.
regards
I promised to find a video showing the different modes of oscillations for coupled oscillators - see https://www.youtube.com/watch?v=kvG7OrjBirI . Note especially that the mode where all the masses (representing the sections of an ESL panel) move together has the lowest frequency.
regards
So question on Bolserst's spreadsheet.....
Been thinking about the graph showing a -6 dB dip from 1000hz down to 250hz (and then some) for finite sources that are elevated above the ground.
What would happen to the response if the front baffle (including ES panel) were sloped something like 4 degrees? I've noticed alot of hybrid designs (see: martin logan) are sloped and I am wondering if it is to flatten out the low end dip in response.
Been thinking about the graph showing a -6 dB dip from 1000hz down to 250hz (and then some) for finite sources that are elevated above the ground.
What would happen to the response if the front baffle (including ES panel) were sloped something like 4 degrees? I've noticed alot of hybrid designs (see: martin logan) are sloped and I am wondering if it is to flatten out the low end dip in response.
![25104785491_5920004b0a_m[1].jpg](/community/data/attachments/487/487404-842a47b02c30864fc50550d54b802efb.jpg)
So question about amp inductance... do we need to take into account the inductance of the speaker wire from the amp to the transformer?
Some the coax cable I am looking at is .10uH per foot and 16.2 pF per foot... so given I need about 20 feet to reach the speakers from my amp, do we add that to the value put into the equation above? i.e. add extra inductance and capacitance to the value the transformer(s) see?
Hi Bengel
No, you can pretty much ignore the speaker cable inductance and capacitance. If you were to choose a bad transformer (eg one of the commercial ones with very high capacitance) its possible that you might notice the effect of a long cable, but I doubt it. If you have the option, chose a transformer that would suit an amplifier at the upper end of the range of output inductances - it will future proof your design and make it insensitive to speaker cables. e.g. the N=150, 200 mH, 70 pF option would be good.
It is interesting that these transformers are simpler to make than the commercial ESL transformers, because they do not to interleave the primary and secondary windings to get the leakage inductance down.
It seems to me that there are two options if you are having them custom made
Option 1: two 1:75 transformers on EI cores. Wind the primary on the bobbin first, then the secondary on top with the low voltage end closest to the primary to minimise the effect of the inter-winding capacitance. (The winding order is the opposite way around from normal power transformers). The secondary will need to be multilayer to get the requisite number of turns.
Option 2: There are a couple of slight advantages (weight and input impedance) putting the two bobbins on the same core - and C-cores or UI cores (better, if you can find them) work nicely.
If you are only after 250 Hz performance, the transformer design is a breeze - hard to go wrong. For a full range ESL working to 50 Hz the transformers requires a bit more care.
regards
No, you can pretty much ignore the speaker cable inductance and capacitance. If you were to choose a bad transformer (eg one of the commercial ones with very high capacitance) its possible that you might notice the effect of a long cable, but I doubt it. If you have the option, chose a transformer that would suit an amplifier at the upper end of the range of output inductances - it will future proof your design and make it insensitive to speaker cables. e.g. the N=150, 200 mH, 70 pF option would be good.
It is interesting that these transformers are simpler to make than the commercial ESL transformers, because they do not to interleave the primary and secondary windings to get the leakage inductance down.
It seems to me that there are two options if you are having them custom made
Option 1: two 1:75 transformers on EI cores. Wind the primary on the bobbin first, then the secondary on top with the low voltage end closest to the primary to minimise the effect of the inter-winding capacitance. (The winding order is the opposite way around from normal power transformers). The secondary will need to be multilayer to get the requisite number of turns.
Option 2: There are a couple of slight advantages (weight and input impedance) putting the two bobbins on the same core - and C-cores or UI cores (better, if you can find them) work nicely.
If you are only after 250 Hz performance, the transformer design is a breeze - hard to go wrong. For a full range ESL working to 50 Hz the transformers requires a bit more care.
regards
I am going to go with these (they are on order, or backorder more accurately). 4 per ESL Panel.
Vigortronix VTX-146-015-106 Toroidal Transformer 230V Single Primary 15VA 0-6V 0 | Rapid Online
No such luck. Response below 1kHz will be unchanged when tipping the panel back.
I think the main reason for tipping the ESL panels back is to keep the top end response relatively intact when a listener goes for a seated to standing position…so you will mainly see this feature on shorter ESLs. If the ESL panels are tall enough to extend 6” or so above the ears of a standing listener then there isn’t any acoustic reason to tilt the panel.
****
I meant to add that one way to boost the low end is with judicious use of baffle extensions to the sides of the panel.
It is a common technique used in full range ESLs and ribbons. Keeping the extensions around 25% of ESL width, you can get a nice 3dB boost at the bottom end without a notch developing in the midrange.
Some pics and details here:
http://www.diyaudio.com/forums/plan...ostatic-baffle-step-filter-6.html#post4200385
Last edited:
No, not with any degree of accuracy.
However, you can build a dummy frame the size of your panel and attach a stretched diaphragm to it. Then, play test tones thru a woofer held near the diaphragm. Sweep upward in frequency and you will have no problem identifying the resonance frequency of the diaphragm. This will get you close to the final answer. In general, using this test, I would aim for 20% - 30% higher than the target frequency to account for a few complicating factors discussed below that add mass and reduce tension.
Resonance Frequency (Fs) is proportional to the square root of (tension/mass).
Double mass and Fs goes down by factor of 1.414 = sqrt(2)
Double tension and Fs goes up by factor of 1.414 = sqrt(2)
Moving Mass:
For ESLs, the moving mass is pretty much defined by the air load (ie radiation impedance).
As discussed previously in this thread, the overall size of the ESL panel defines the radiation impedance.
1) Suppose you have a small panel that you measure Fs = 150 Hz. If you start stacking identical panels like this next to each other to form a large ESL, you will notice Fs goes down the more panels you stack together. So, you need to test full setup if you want the final answer.
2) Another complication is that adding baffle extensions will also increase the air load and lower Fs.
Tension:
For ESLs, the tension of the diaphragm is achieved by stretching or heat shrinking.
1) Invariably, tension reduces from its starting point over time. The amount is dependent on the film type and whether or not it has been heat treated. See some of the data Wrinex has posted on his experiments:
http://www.diyaudio.com/forums/planars-exotics/282010-my-take-stretch-jig-2.html#post4514003
2) When you apply bias voltage to your finished ESL panel, the diaphragm will be pulled away from its central resting position by a force proportional to its displacement from the center. In effect, a negative tension increment is added which lowers Fs with increasing bias voltage.
Measurements and details posted in this thread:
http://www.diyaudio.com/forums/plan...agm-resonance-change-hv-bias.html#post1884466
So I've been thinking about a streching table design that would allow me to do this for the past few weeks. I have an idea that might work, I'll post some photos if I decided to build it (i.e. if it is worth the trouble).
I was thinking about the mylar "relaxing" as time goes by. Looking at winerx posts, it happens a bit more quickly than I expected
. So given I am crossing at 250hz, I was wondering if I even need to bother "tuning" the Fs of the diaphragm. Just stretch it 1.25% and it should be least 1.5 octaves below. With a 24dB crossover, that should put the Fs out of the "danger zone".
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.