Some More Questions On ELS Design

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
When I have everything finished I'll start doing some measurements. This is going to be a 3 way, electronically crossed over system using a dynamic woofer up to about 250 -300 Hz, the large panels up to about 2000-3000 hz and the small ones for frequencies above that. Actual crossover frequencies and slopes yet to be determined. The crossover is a Beheringer DCX2496 and the amps will be two QSC PLX 2502's and one QSC RMX 1450. The hard part of this project is going to be the speaker frames, as they have to have a very high WAF before my better half lets them in the house.
 
Some more progress to report. All six panels have been rebuilt. The stators got three coats of polyester resin and this time I used 1" 3M tape fpr the spacing. I'm now running the bias at 8KV for the large panels and 4 KV for the small panels. I haven't checked to see how high I can run up the bias, but I'm sure there is much more in them. But no need to as they play plenty loud. I've also built a temporary frame to mount the panels for testing. I'll start tomorrow on that if I can figure out how to test??? In any case, listening to this one speaker in momo, it just sounds wonderful. Diana Krall Live in Paris is live in my shop.:) Here are some pictures:

IMG_0500.jpg


IMG_0513.jpg


IMG_0514.jpg


IMG_0512.jpg
 
That Is Fantastic !!!

I did some quick frequency response measurments of my little panels today so maybe these can be of use to you.
These were done with no eq.

The Red plot is with the microphone (Peavey PVM520 TM) about 1 inch from the diagphram and the blue line is the signal.

The Green plot is with the microphone about 18 inches away and the black line is the signal.

I am still not quite familiar with HOLMImpulse and when I implemented Highpass I didn't notice the low frequency rolloff slope

The humps I believe are from reflections from many nearby flat surfaces off of my monitors and such.

I have found by using a digital VST crossover that I can set it as low as 240hz or 250hz without any coloration's or restrictions in vocals or instruments using Deep Purple's "lazy" as a sample running it full range (no eq ).
Below 240Hz the Diagphram resonance's start to color the sound.

The second Graph was done the other day and was EQ'ed to be flat with no dedicated crossover but biamped using the filters on my mixer as the crossover with a Sony 5.25" SS-MB150H speaker box using only the woofer and duct tape over the tweeter.

The plot of this setup is shown in Red and the woofer alone is in Green.

Both graphs were done without the VST crossover.

My panel being about the same width as yours seems to have a natural roll off around 600hz to 300hz (-3db at 400hz).
I don't have it setup to do extremely accurate measurements yet as these are just some quickies that I had done just to see where I am at.

I am currently working on my transformer situation at the moment.
I have been using any where from 5Kv to 10Kv as a bias voltage.
Today I hit 102db at 1 meter using my cheapy amp on single tones with out it shutting down
This is with a 1:256 transformation ratio, 8kv of bias and a 15Vp to 20Vp in to the transformer, I forgot to note the exact figure as this is when I decided to switch amps.
Tomorrow, I will use the DC300a and see what it does as it was getting late.

I am looking forward to see some more results of your build !!!


Keep on DIYing !!!
:cheers:


jer :)


P.S. I was using a 36db and 48db slope on the VST crossover test as this allowed me to not saturate the transformers with the frequency settings this low.
 

Attachments

  • Small esl.jpg
    Small esl.jpg
    248.1 KB · Views: 214
  • With small woofer.jpg
    With small woofer.jpg
    256.6 KB · Views: 67
Last edited:
Measurements

I took some frequency response of the panels and they look pretty good to me, but I'm no expert in these matters, so your comments are appreciated.

The first graph is a measurement of the noise in the garage without driving the speakers and me being quiet.
Room.jpg


The second graph is the frequency response of the small panel alone.
SmallPanel.jpg


The third graph is the response of one of the large panels using four torroid transformers for step-up.
RightLargePanel.jpg


The fourth graph is one of the large panels driven by two oil burner transformers in parallel/series.
OilburnerTransformers.jpg


The last graph is all three panels with the large ones driven by the 4 toroids and one channel of the amp and the other amp channel driving the small panel.
Allpanels.jpg


The frequemcy response of the oil burners sucks, but it does go a lot lower in the bass and they do swing more voltage.
 
Looks like the large panels have a resonance of about 52 hz and the small one about 130 hz. The Behrenger DCX2496 crossover should be able to filter out these peaks. The response from all the panels is so smooth that I think I'll be able to use a 6 dB/octave crossover slope and set the crossover to somewhere between 600 and 1000 hz. What do you think? The other option is to use the oil burner transformers and set the ctossover to 48 dB/Octave at about 400 Hz. I guess I'll have to experiment once the frames are built and they are moved to my theater room.
 
Last edited:
Those Graphs look real nice I just did some more measurments today as well,and I got some really flat lines by placing the microphone about 2 to 4 inches from the speakers and running them fairly loud.
This helped a lot reduce the effect of the room reflections.

The phase response was flat at 85 degrees across the bandwidth too !!!

This helps to show that the accuracy of them was quite spot on and this was with the little woofer as well.

I used REQ V5 for thes tests and I would post the graphs for you but it doesn't have a export graphics function so I have to find my Screen capture program to do this.
I have them saved so I can do this later.

Measuring at a distance will show you exactly were the cancellations start to occur.
For my little panel this was about 600hz.


Being the nature of the panels being both ESL's I think that 6db slope should be good or even 12db.
But 12db to 24db may be in order to help with the low frequency saturation limits of the high frequency step up transformers.

Also to limit the high's going into to the oile burners because those will have a lot of extra capacitance and the high frequency's will cause a very low impedance to the amplifier.
Those may need even a higher slope.

I haven't spent a lot of time with the digital crossover setup yet because I have to setup my other machine as a source as the VST wave player I have works okay but I don't have a very good selection of quality samples setup yet.

So,I have been back to using the filters on my mixer for now and I have it tuned flat with little effort and except for the extreme lows, the bass sound location disappears from the two drivers and work in unison incredibly well.
Using this method the crossover falls at 1100Hz at about a 6db slope.

Some how I lucked out and that is the way it turned out.

Although I really like the lower crossover point as well the little woofer box is not tuned properly and is quite boomy.
I had to seal the port and pull out about 6db of 80hz and add everything I had at (15db) at around 40hz and this really tightened up the bass.
The only thing I do to the ESL is use a 80hz highpass cut and adjust the 12khz shelving to suit the material I am listening to at the time as the panel itself has a very flat response.

Bass is bass and the kick drum and the rest of the toms and snare have a very natural and clean snap.
Listening to the Cult's "Wild Flower" and Kiss's "Strutter" as well as The Hollies "Long Cool Women" the drums sounded like the were right here live in this room with no coloration or boomyness.
Not bad for a little 5.25" woofer in a now sealed box.

Anyway Great Job!!!
I can't wait to hear some more of your results as your panels are exactly what I have been contemplating should I finally build a bigger set.

I had just recieved as a gift a pair of 10" X 49" preforated metal that I think I will cut in half to do a pair of smaller bass panels and then add a tweeter section to use nearfield in front of my mixer.
And I am stiil going to do the tweeter/midrange strips for my big speakers aswell as this little panel proves that its width is in the proper range to complement them.

Its high frequency dispersion is not overly narrow and in some case not even noticeable very much.

You will know more once you get them into position.
As with mine sitting in front of my computer monitors I sometimes get some peaks in the 3khz to 6khz as well as a hump at 9khz to 20khz.
It becomes fatiguing at times and I have to pull those out with a filter as well.

I have discovered that my system is capable of transient peaks of at least 105db or more!

I have found that this response changes when I either move the speaker or my measurement microphone around.
So it is not the speakers as I have confirmed that with today's measurements.


Cheers !!!
:cheers:

jer :)
 
Last edited:
Another question. I keep hearing that electrostatics have to be equalized because they have a rising response with frequency. Yet, the response for my speakers I is essentially flat. Am I doing something wrong in my measurements? They were measured with the mic about 1' away from the center of the panels.
 
Another question. I keep hearing that electrostatics have to be equalized because they have a rising response with frequency. Yet, the response for my speakers I is essentially flat. Am I doing something wrong in my measurements? They were measured with the mic about 1' away from the center of the panels.

The natural near field response of an ESL is flat. It is the far field response that is rising.
1 ft. would certainly be considered closer to near field than far field.
Try listening to some music and start standing 1 ft. from them and slowly start backing away. You will notice the sound getting leaner and brighter the further you back away. Or, if you have the capability to take windowed measurements, measure the response at 1ft, 2ft, 4ft, 8ft and see the trend as you move out of the near field and into the far field.

Also, adding resistance in series with the primaries of your transformers flattening the rising response starting at a frequency determined by the value of the series resistance and the panel capacitance. You didn't post a schematic, but from the pictures it looks like you might have added some series resistance.

This post provides some further information on what other things influence the flattening of the natural rising response of an ESL:
http://www.diyaudio.com/forums/plan...rrent-vs-voltage-drive-esl-6.html#post2353285
 
Last edited:
Here is the measurement from my 1.2 X 0.25 meter panel. I measure it near field with the microphone only about 2-3 inches from the panel. I always let my ESL run full range.

I think it all depends on what kind of music you listen to. For me, I really prefer mine this way and I don't use a subwoofer or EQ.

ESLFrequenciesResponse2.jpg


Wachara C.
 
Yes,
The first picture is the nearfield response(red line) as I had already posted at about 2 inches from the diagphram using my large diagphram Peavy 520TN microphone.
The hump at 4Khz is an irregularity in the response of this particular microphone.
The only EQ used on this test was a 80Hz highpass low cut filter to keep the transformers from saturating and as well as the low end change in response of the signal level as shown by the black kine

The next two charts are done with a calibrated Dayton measurement microphone and the db scale on the left is calibrated to a brand new SPL meter as well.
These measurements were done at a distance of 1 meter.
The second chart is the raw data of the combined woofer and ESL.

The third chart shows the woofer and ESL responses separately and as well as combined.
The only EQ that was used on the ESL was a 80Hz highpass low cut filter and about 4.5db cut using the a highpass shelving EQ at 12Khz.
The EQ for the woofer was the same as previously mentioned.

You can see the rising response of the ESL panel in these charts and is the reason for the high frequency cut needed to flatten out the response curve.

After this EQing they sounded much better and the fatiguing that I had been experiencing earlier had ceased.
Although I don't mind the little bit of the brightness at times.

jer :)
 

Attachments

  • Small esl.jpg
    Small esl.jpg
    248.1 KB · Views: 201
  • raw.jpg
    raw.jpg
    265.6 KB · Views: 196
  • DESKTOP ESL RESPONSE.jpg
    DESKTOP ESL RESPONSE.jpg
    388.4 KB · Views: 196
Last edited:
The project is on hold for the time being as I'm waiting for my new Laguna tools slot mortise machine to be shipped.
Regarding one of my prior post concerning the use of oil burner transformers for the bass panels, do you think they could be made to work up to 500 hz using equalization in the electronic crossover? The bass these put out is much stronger than the torroids. Here is a repost of the frequency responce of the bass panels driven by the oil burners:

OilburnerTransformers.jpg
 
Last edited:
Regarding one of my prior post concerning the use of oil burner transformers for the bass panels, do you think they could be made to work up to 500 hz using equalization in the electronic crossover? The bass these put out is much stronger than the torroids.

Can you clarify what you mean by the bass is stronger with the oil burner transformers?
Is the overall level higher? (ie is the step-up ratio higher) or is it that the oil burner transformers have more output below 50Hz. Most likely the reason the output for the toriods falls off more quickly below 50Hz is related to their low primary inductance and the amount of resistance you have in series with them. Can you post the schematic for how you have the toroids wired up, including the series resistance?

In any case, the slope of the oil burner response is nice and smooth. I see no reason why you couldn't include a complementary shelving filter in the electronic crossover. However, this would result in reduced headroom in this frequency range.
 
The step up of the oil burners transformers (2 of them per channel) with the primaries wired in parallel and the secondaries in series, and a 1.2 ohm resistor in series with one of the primary legs is 167:1. The step up of the toroids (4 per channel) with the primaries in parallel and a .25 ohm resistor on each primary leg (8 resistors in all) and one 1.3 ohm resistor feeding four of the parallel .25 resistors, is about 150:1. Listening to both, the oil burners sound much better in the low bass than the toroids. And measurements of the frequency response seem to back this up. Here is the response of the toroids, which shows a steep drop off after resonance.

LeftLargePanel.jpg
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.