Current vs voltage drive ESL? - Page 6 - diyAudio
Go Back   Home > Forums > Loudspeakers > Planars & Exotics

Planars & Exotics ESL's, planars, and alternative technologies

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 3rd November 2010, 07:17 PM   #51
diyAudio Member
 
bolserst's Avatar
 
Join Date: Jun 2005
Location: Fort Worth, Texas
Quote:
Originally Posted by markusG View Post
It does make me wonder exactly what I'm seing in this grapgh and what makes it different from the spl plots I'm used to seeing. (The "flat" ones...)
The measurement I posted was taken using a HV reference probe to remove any influence of amplifier and/or transformer response.
The far-field native response of all unbaffled dipole ESLs is a rising response with increasing frequency. (+6dB/octave for point source, +3dB/octave for line source)
Anytime you see a flat ESL response, there is some equalization involved(intentional or not):

- curved ESLs spread out the highs rather than focus them flattening the high frequency response
- mass of diaphragm can flatten the HF response
- mass of air in stator holes can flatten the HF response
- any series resistance in primary or secondary circuit changes constant voltage drive to constant current and flattens the response
- any segmentation and ladder resistor network adds area with decreasing frequency to flatten the response
- leakage inductance of transformer flattens HF response
- boosting of LF response with underdamped HP filters can flatten the mid and LF response
- baffling around the edges of the ESL can flatten the mid and LF response

Also, many times measurements are not taken far enough away to be in the far-field.
As was mentioned earlier in this thread, near field response of ESL is flat.

It may help to think of the SPL response from the simulation as the maximum possible output .vs. frequency.
The desired response will be the loudest flat line you can fit underneath this maximum curve dependent on your crossover frequency.

Last edited by bolserst; 3rd November 2010 at 07:47 PM.
  Reply With Quote
Old 4th November 2010, 05:54 AM   #52
diyAudio Member
 
kavermei's Avatar
 
Join Date: Jan 2005
Location: Lokeren, Belgium
Send a message via MSN to kavermei
Bolserst, those are very interesting measurements, and remarkably close to the Walker equation. Thanks!

Markus, a crucial thing here is that these measurements were made without accounting for the influence of the transformer -- this might explain why you've not seen these frequency responses before.

Thankfully, an ESL panel has a rising frequency response! Many of the practicalities bolserst lists in the previous post tend to have some HF loss and therefore (at least partially) equalize it out. Imagine if ESL panels had a 3dB/oct *drop* in response, we would have been in a lot more trouble.

Kenneth
__________________
Never send a human to do a machine's job. --Agent Smith
  Reply With Quote
Old 4th November 2010, 07:56 AM   #53
markusA is offline markusA  Sweden
diyAudio Member
 
markusA's Avatar
 
Join Date: Dec 2009
Location: Gothenburg
Blog Entries: 1
I think you're right bolserst, it's near-field vs far-field.
  Reply With Quote
Old 4th November 2010, 10:52 AM   #54
diyAudio Member
 
Join Date: Jan 2008
Location: Toronto and Delray Beach, FL
So, we have designed wonderful panels that are close to theory when measured in the carefully defined manner defined so well above.

We mount them on a board or frame of some sort, stand them at some distance from the back wall, they form some angle between them and the listener, sit at some distance from them, in a room of some asymmetry, which has some degree of absorption.... and then we begin trying to make them sound right because there is SOME big differences between design and practice at this point in time caused by some of these factors.

Careful, theory-based design is important. It makes a good starting point for tuning systems when they get to your music room.

Of course it may be asked, if you need to do substantial re-tuning when the speakers get to your music room, how deep into theory-based design do you need to go?

So we find Markus still hanging around the library instead of in the workshop.
__________________
Dennesen ESL tweets, Dayton-Wright ESL (110-3200Hz), mixed-bass Klipschorn (35-110), and giant OB using 1960's Stephens woofer (18-35); Behringer DSP. HiFi aspirations since 1956

Last edited by bentoronto; 4th November 2010 at 10:55 AM.
  Reply With Quote
Old 4th November 2010, 11:16 AM   #55
markusA is offline markusA  Sweden
diyAudio Member
 
markusA's Avatar
 
Join Date: Dec 2009
Location: Gothenburg
Blog Entries: 1
bentoronto> I'm just hanging around the library while waiting for the buildig materials to arrive.
I have my upcoming ESL design pretty much done. I'm trying to figure out the theory because it's fun, not because I have to.
  Reply With Quote
Old 4th November 2010, 11:29 AM   #56
diyAudio Member
 
Join Date: Jan 2008
Location: Toronto and Delray Beach, FL
Beautiful! My kind of approach.
__________________
Dennesen ESL tweets, Dayton-Wright ESL (110-3200Hz), mixed-bass Klipschorn (35-110), and giant OB using 1960's Stephens woofer (18-35); Behringer DSP. HiFi aspirations since 1956
  Reply With Quote
Old 8th November 2010, 04:51 PM   #57
diyAudio Member
 
bolserst's Avatar
 
Join Date: Jun 2005
Location: Fort Worth, Texas
Quote:
Originally Posted by Calvin View Post
The Quads 57 needed a lot of U to drive the resistors that should lead to a current drive of the panels. Because of their d/s being reasonably small the Quads needed much less U when voltage driven.
The same basically applies to the Quad 63 and its heirs. Its smallish d/s needed much lower values of U if it were voltage driven and if they omitted with the crude delay line and used resistor segmentation instead (the difference would be that the impedance value and as such the need for high U-trannies would decrease considerably).
Hell Calvin,
I have not had the chance to inspect or measure a Quad 63. But, from the circuit diaphragm and details mentioned in Baxandall’s ESL paper I am confused by your comment about the step-up ratio(U) of the transformers being unnecessarily high because of the delay lines. The D/S spacing is 2.5mm. The step-up ratio(U) is 245 and Vpol = 5.25kV. This seems to fit in pretty well with the optimum values calculated by Baxandall assuming a breakdown voltage for dry air of Emax = 4kV/mm. See attached table which summarizes required optimum Vpol & Vsig for different D/S spacing. I also added columns to show the required transformer step-up ratio(U) required to obtain Vsig for amplifier outputs of 50W/8ohm, 100W/8ohm, & 200W/8ohm.

I highlight the Quad 63 values in blue. Note that it was designed for a maximum recommended input voltage of 40Vpeak = 28.3Vrms = 100W/8ohm.
Looking at this table you can also get a feel for how quickly Vsig and setup-up ratio(U) quickly gets out of hand if D/S spacing larger than a few mm is used.

I understand that if you removed the delay lines and ladder resistors the panel would be louder in the midrange, but those components are used to EQ the response flat. So, I guess you could use a smaller step-up ratio(U) transformer to get the same midrange SPL level if the delay lines and ladder resistors were removed. But, the high step-up ratio(U) is needed to provide the maximum possible SPL at low frequencies(<100Hz) where the delay lines and resistors aren’t really influencing the response. At least that is the way I understand it.
Attached Images
File Type: jpg ESL_voltages.jpg (71.6 KB, 180 views)

Last edited by bolserst; 8th November 2010 at 04:54 PM.
  Reply With Quote
Old 9th November 2010, 07:44 AM   #58
Calvin is offline Calvin  Germany
diyAudio Member
 
Calvin's Avatar
 
Join Date: Nov 2004
Location: close to Basel
Hi,

the breakdown voltage of air is at best 2kV/mm, a safe value beeing ~1.5kV/mm, which roughly corresponds to 1kVrms/mm of signal voltage.
The problem is that the the driving source only sees a small capacitance through the delay line, which is just a fraction of the complete panel area. So the impedance values are high and as such the transformation factor alas voltage demand must be high. Add to this the losses of the series resistors that shall allow for current drive conditions and You end up with the ridicolously high U of 1:270 of the ESL63. Still though the current-drive condition doesnīt apply in the low bass, since this required such extreme high resistor values that the voltage demand and transformation-factor would result in impractical values. The only thing that saves the day here is the high-Q resonance of the panel. Quad simply says...if the resonance is there anyway and we canīt get away with it, letīs use it. So in the low bass region the panel works under lossy voltage drive conditions (lossy, because the resistors are still in the signal-path). The associated falling amplitude response with decreasing frequency is then countered by the high Q of the resonance to give a quite linear response. Over all the whole is marvellously engineered ..... the only pity beeing, that its concept starts from compromised premises.
The flaws of the concept as I see it, are:
- to neglect the fact that a 1:270 transformation factor does not com without serious drawbacks. A transformer with lower transformation factor will always measure and sound better.
- to assume that a point source is advantagous in real room situation
- using a delay line instead of a chain of simple segmenting resistors, which eq the panel and lead to a similar distribution character but without the drawbacks of the delay line.
- driving fullrange, which discludes the current-drive condition over the full bandwidth
- driving fullrange and thereby accepting and fudging around with a high-Q resonance, which nobody would accept as a high quality solution if dynamic bass drivers were to use. The high-Q audibly recognizable in that it leads to a distinct soft-noted bass.
- small sized panel segments, series resistors, high-Q resonance at a low frequency and high-U transformer spoil the dynamic range.

Using simple resistor segmenting instead, the driving source sees the whole panel area, successively decreasing with frequency. So impedance values are considerably smaller, hence the voltage demand and transformation factor can be lower. With a panel of the ESL63īs dimensions one could expect values of app. 1:100. Since the first segmenting resistor feeds into the second segment the trannieīs (or sourceīs) drive voltage feeds directly into the first segment, hence no losses or increased voltage demand. The segmenting resistors serve two purposes...to eq the panelīs frequency response and to shape its distribution character. So it serves the same aims as the delay line, but without suffering from thats serious compromises.

jauu
Calvin
  Reply With Quote
Old 9th November 2010, 08:59 AM   #59
diyAudio Member
 
Join Date: Jan 2008
Location: Toronto and Delray Beach, FL
A few odd thoughts:

Any transformer seems a weak link in an ESL system (compared to a direct drive amp). But my guess is that Quad was trying to get their speakers to work with low power amps and so may have needed the high turns ratio units that Calvin indicates. That seems an odd goal today and not an inherent limitation of their highly successful segmentation design.

Yes, trafficking in resonances (like the multi-resonant tapped horn) seems low-fidelity. But the resonances you get making bass by shaking large hunk of cardboard (shaped like a cone) is altogether of a different order as shaking mylar film. Again, funny to try to get serious bass from ESLs (large Dayton-Wrights in sulfur hexafluride gas, excepted), so this is another will-of-the-wisp Quad goal that seems funny today.

I guess I prefer the "look at me" show-off quality of Martin Logans to the refined full ESL compass of the Quads. But no question that a Quad owner has a fabulous speaker pair to listen to.

We ESL fans gladly accept various shortcomings that come with ESL designs because we are so attached to the crystal clarity.
__________________
Dennesen ESL tweets, Dayton-Wright ESL (110-3200Hz), mixed-bass Klipschorn (35-110), and giant OB using 1960's Stephens woofer (18-35); Behringer DSP. HiFi aspirations since 1956
  Reply With Quote
Old 9th November 2010, 02:39 PM   #60
diyAudio Member
 
Join Date: Nov 2004
Location: Ohio
In the midst of this discussion I have a question about the idea that a fullrange ESL has to be compromised in the bass. I know and understand Ben what your saying from a layman perspective, however what explains how I'm getting great bass weight and dynamics from a single Acoustat panel wired into two stock 121A non-medallion interfaces, with the bass tap on the transformer wired for Model 3? Meaning that the bass transformer is set for 3 panels of Acoustat not one (there are multiple taps available for different panel numbers). This set up is in a modest sized living room (not small not large) and that the panels I have are totally undressed with no framing of any kind. They are propped up on sticks of wood a few feet (maybe 32-34 inches) from the back wall and 5 to 7 feet from side walls. They are not the most dynamic speakers I've ever heard but the smack of drums and weight of kick drum and the sound of both electric and acoustic upright bass instruments is totally a believable. This all coming from a single 9" x 45" Acoustat panel that is close to 30 years old being driven by a high current Harman Kardon design from the mid 90's. Really a head shaker in terms of resolution, dynamics, coherence, and the sweetest mids and highs you coulsd ever want. They are the kindest speakers to my ears I've ever had yet all the details are perfect. Every nuance is there. Yes I can drive them enough to smack the mylar right into the stators. Yes the charge will migrate if I shake them up a bit too much and volume will suffer momentarily while the panel charges even up. I can even get them to arc but they really have real bass into the 30 hz range. In my mind if an open baffle servo powered dynamic sub were introduced into the contiguous acoustic space comprising the front and rear air space of the Acoustat driver, then there would be an acoustic "helper". This dynamic sub operating 80 or lower would couple with the mylar acoustically pressurizing it and thus possibly limit the excursion of the mylar with large bass transients. The sub would necessarily have to be open baffle to fully couple with the air space. Why not do a larger panel than Acoustat 9"x 45"?
I have never heard the bass resonance or ringing of the mylar diaphragm. Is this a considerable problem with large unsegmented full range panels? Any thoughts?
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
ESL Headphone direct drive Bazukaz Planars & Exotics 22 28th April 2012 10:07 PM
CG to drive DIY ESL SM7UYJ Chip Amps 6 19th May 2009 07:23 AM
ESL Direct Drive Atom666 Planars & Exotics 6 5th October 2008 08:09 AM
mix current drive with voltage drive at LF? capslock Multi-Way 0 27th April 2006 10:17 AM
2 x LM4780 in parallel to drive ESL panel Padel Chip Amps 6 3rd December 2004 04:44 AM


New To Site? Need Help?

All times are GMT. The time now is 09:13 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright Đ1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2