Started to experiment with parallel notch filter for the midrange, with the parts I have on hand (hence the messy job 
).
I am using the calculator on the HiFi Audio Design site to come up with values for L1, C2 and R3 (as in the schematic below) as a starting point. However, the calculator calls for the "Speaker DC Resistance [ Re ]" and since electrostatic panels are basically capacitors, I presume their DC resistance would be ∞/infinite. I wonder if anyone could shed some light on how such formulae should be used for capacitive drivers?
So I used Rz (resistance at the target frequency) instead of Re (DC resistance):
The reactance of a 5.34uF capacitor at 1450Hz is 20.5 ohm
The reflected resistance of the two 27k4 resistors at the primary is (27k2 +27k2) / (100 x100) = 5.44 ohm
So the overall impedance seen by the amp at the transformer's primary is 20.5 ohm + 5.44 ohm; roughly 26 ohm.
When I tried the first calculator, Components Parallel Notch Circuit, on the page however, it turned out that different values of Re only changed the max. Damping [ At ]. The L, C, R values for the notch filter appeared to be calculated only with the frequencies chosen.
Initial measurements showed that the notch filter did work. I used 9.5 ohm for R3 and the peak was reduced by around 2dB. I need more than 6dB of reduction so need to source suitable resistors. The width of the notch also needs further adjustments with different values for L1 and C2.
).I am using the calculator on the HiFi Audio Design site to come up with values for L1, C2 and R3 (as in the schematic below) as a starting point. However, the calculator calls for the "Speaker DC Resistance [ Re ]" and since electrostatic panels are basically capacitors, I presume their DC resistance would be ∞/infinite. I wonder if anyone could shed some light on how such formulae should be used for capacitive drivers?
So I used Rz (resistance at the target frequency) instead of Re (DC resistance):
- Say the frequency peak to reduce is 1450Hz
- The panel capacitance is 534pF
- The transformer turn ratio is 1:100
- Two 27k2 (27k4 was a typo in the schematic) resistors are in series with the panel
The reactance of a 5.34uF capacitor at 1450Hz is 20.5 ohm
The reflected resistance of the two 27k4 resistors at the primary is (27k2 +27k2) / (100 x100) = 5.44 ohm
So the overall impedance seen by the amp at the transformer's primary is 20.5 ohm + 5.44 ohm; roughly 26 ohm.
When I tried the first calculator, Components Parallel Notch Circuit, on the page however, it turned out that different values of Re only changed the max. Damping [ At ]. The L, C, R values for the notch filter appeared to be calculated only with the frequencies chosen.
Initial measurements showed that the notch filter did work. I used 9.5 ohm for R3 and the peak was reduced by around 2dB. I need more than 6dB of reduction so need to source suitable resistors. The width of the notch also needs further adjustments with different values for L1 and C2.
i would say it has a impedance, you just have to measure it with the esl connected. you can use REW to measure that with your soundcard. but lookup impedance measurement REW there is a guide somewhere how to do it. it only needs a reference resistor i believe of 50 cents. if you want to be sure
Still a long way to go with the crossover equalization and getting there very slowly. I am starting to get how difficult it is to design and build electrostatics that measure flat and sound smooth (the latter doesn't necessarily follow the former), and why Sanders would choose to go the DSP route for their model 10e.
Dug up a pair B&K Sound 3636 autoformers from my Altec period and used them in place of the R in the LCR notch filter. Hopefully the sound will be good enough for the holiday season!
Dug up a pair B&K Sound 3636 autoformers from my Altec period and used them in place of the R in the LCR notch filter. Hopefully the sound will be good enough for the holiday season!
Some updates:
- I will start new experiments only after I have sourced and applied the silkscreen back damping as recommended by Sheldon and Gary. IIRC, the silkscreen material is the type used on the ESL63 and might work more effectively for the Acorn.
- The stock crossover design is apparently the result of Rob's careful tuning, which has taken into consideration all the panels' intrinsic traits. I think it should work well for most users. My experience has been that altering part of the design would often upset the balance of the whole to some degree.
- In my current implementation (the frame, panel clamping and back damping, etc.) and listening space, and within the areas where the speakers can be positioned, the measured frequency response curve always shows a relatively deep and wide dip around the 400 to 500Hz region, and one or a couple high peaks around the 1k to 3kHz region. These anomalies can be easily heard with some of my favorite albums such as Deodat de Séverac: Melodies for Voice and Piano by Anne Rodier & François-Michel Rignol, and Passion Ysaÿe by Rachel Kolly. That's why I have to get them addressed!
- I am considering to move the crossover over to the HV side of the audio transformers, per the Quad ESL. This means using resistors instead of inductors and makes it much easier to experiment with different values. I also feel that the crossover dealing directly with the panel without being translated by the audio transformer might be beneficial, though I may be wrong. Rob is trying to get the relevant parameters of the audio transformers for me to do the initial calculation.
If I were trying to do a passive crossover of much complexity on the input side, I would be measuring impedance at the audio transformer input, then using that data in a crossover simulator to dial in the circuit. It makes understanding what's going on and what doesn't work like I thought much easier.
If I'm correctly reading what you've done so far and intend, simple resistors on the high voltage side don't seem like they're going to do what you're after. You'll need additional components to do a notch, and those will have to be high voltage rated. High voltage parts are getting rarer all the time. Resistors alone are most commonly used to make a low-pass filter on the high voltage side.
You may want to have a look at this thread also:
https://www.diyaudio.com/community/threads/explaining-the-quad-esl-57-crossover.224615/
Many thanks mattstat for the suggestions.
I thought about actually measuring the impedance at the audio transformer input side with REW, but was a little concerned that mis-operations might result in damages to the MacBook used, which is also my work computer. I now plan to use my audio PC with a different, dedicated system SSD for the job.
I did, however, achieve the intended results to some degree with the notch filter by guesstimate and on-spot parts swapping. I was limited by the inductors I had on hand though, and acquiring more values just for experimenting wasn't practical. Size of the components is also a concern, as inductors and capacitors can quickly take up more space than what is available. Around the same time I happened on the ESL 57 crossover thread you mentioned, and began to wonder about the possibility of moving the crossover to the HV side.
Luckily we have a local source here in Taipei who has nearly all the available values, from 100k up to 68M ohm, in stock for the Vishay VR68, which is rated 10kV. 6kV-rated capacitors also seem readily available from 5pF up to 6,800pF.
BTW I found a site, esl.drakkapps.com, with lots of interesting information on the '57. This page details the ins and outs of the audio transformer.
I thought about actually measuring the impedance at the audio transformer input side with REW, but was a little concerned that mis-operations might result in damages to the MacBook used, which is also my work computer. I now plan to use my audio PC with a different, dedicated system SSD for the job.
I did, however, achieve the intended results to some degree with the notch filter by guesstimate and on-spot parts swapping. I was limited by the inductors I had on hand though, and acquiring more values just for experimenting wasn't practical. Size of the components is also a concern, as inductors and capacitors can quickly take up more space than what is available. Around the same time I happened on the ESL 57 crossover thread you mentioned, and began to wonder about the possibility of moving the crossover to the HV side.
Luckily we have a local source here in Taipei who has nearly all the available values, from 100k up to 68M ohm, in stock for the Vishay VR68, which is rated 10kV. 6kV-rated capacitors also seem readily available from 5pF up to 6,800pF.
BTW I found a site, esl.drakkapps.com, with lots of interesting information on the '57. This page details the ins and outs of the audio transformer.
Some thoughts to share while waiting for the silk screen and high voltage resistors:
What has happened so far only reminds me again of the importance of a beautiful midrange. For me this means no valleys in the all-important 300 to 1kHz range, and no narrow-band peaks higher up. I really don't mind the very high HF roll off, off-axis or on-axis, not because I couldn't hear up to that high, but because I find live music at concert halls or Jazz clubs feeling a bit serene above say 10k, 12kHz, and that doesn't detract from the beauty at all--quite often it's the contrary.
Now I recall the one (and quite likely the only one) type of speakers I have heard and found to better the Quad ESL in reproducing the violin: The Spendor BC1 and several other speakers in the same BBC LS3/6 lineage. I am going to set up my old BC1s as a reference for tuning the Acorn, and also the '57s once they're rebuilt. The BC1 is also outstanding with human voice and piano. Lets hope I won't end up sticking with the BC1 and delegating the '57s and Acorns to my bedroom and attic.
On the tactical front, it occurred to me that I can actually use the Quad's crossover as a starting point for adopting the Acorn's crossover on the HV side. Maybe I should try the '57's audio transformer on the Acorn to see how it turns out. As long as the input signal is kept low, there shouldn't be any catastrophic damages even if I make a couple false moves...
What has happened so far only reminds me again of the importance of a beautiful midrange. For me this means no valleys in the all-important 300 to 1kHz range, and no narrow-band peaks higher up. I really don't mind the very high HF roll off, off-axis or on-axis, not because I couldn't hear up to that high, but because I find live music at concert halls or Jazz clubs feeling a bit serene above say 10k, 12kHz, and that doesn't detract from the beauty at all--quite often it's the contrary.
Now I recall the one (and quite likely the only one) type of speakers I have heard and found to better the Quad ESL in reproducing the violin: The Spendor BC1 and several other speakers in the same BBC LS3/6 lineage. I am going to set up my old BC1s as a reference for tuning the Acorn, and also the '57s once they're rebuilt. The BC1 is also outstanding with human voice and piano. Lets hope I won't end up sticking with the BC1 and delegating the '57s and Acorns to my bedroom and attic.
On the tactical front, it occurred to me that I can actually use the Quad's crossover as a starting point for adopting the Acorn's crossover on the HV side. Maybe I should try the '57's audio transformer on the Acorn to see how it turns out. As long as the input signal is kept low, there shouldn't be any catastrophic damages even if I make a couple false moves...
Interesting ideas.
Not sure if these questions/comments belong here, or in your other thread on the '57s.
Are the '57s tweeter section run full "midrange" and up? I had heard that this was the case and there was just a first order resistor and panel capacitance filter to the outer sections of this panel.
What is the mid-tweeter section high pass filter? I understand that because of the natural rising response (with frequency) of the panel, this acts as a quasi first order high pass filter.
In order to limit diaphragm excursion, I assume there is an additional degree of high pass filter, is this correct?
Direct use of '57s interface needs to consider differences in capacitance between Acorn and '57 panels, right?
Cheers
Grant
Not sure if these questions/comments belong here, or in your other thread on the '57s.
Are the '57s tweeter section run full "midrange" and up? I had heard that this was the case and there was just a first order resistor and panel capacitance filter to the outer sections of this panel.
What is the mid-tweeter section high pass filter? I understand that because of the natural rising response (with frequency) of the panel, this acts as a quasi first order high pass filter.
In order to limit diaphragm excursion, I assume there is an additional degree of high pass filter, is this correct?
Direct use of '57s interface needs to consider differences in capacitance between Acorn and '57 panels, right?
Cheers
Grant
To quote bolserst from the thread Explaining the Quad ESL-57 Crossover (post #5):
(Tweeter panel Crossover) "The two 560pF caps and 270K resistor were added for production models after serial number 16800. The purpose was to shelve down the drive voltage at lower frequencies by 6dB to avoid arcing the tweeter panel when used with amplifier power > 15W, which was the power level originally designed for."
(Mid panel Crossover) "The two 150K resistors provide current drive for the mid panel flattening the response as was described in the description of the bass panel crossover. The 560pF cap in parallel with one of the 150K resistors provides a bypass for this resistance so the right amount of damping is applied to the 24kHz tweeter panel crossover resonance. Without it, the response of the tweeter panel would peak up at the top end."
The capacitance of Acorn's bass panels is 560pF whereas the '57's is 400pF; Midrange, 534pF and 200pF respectively.
Off topic but listening to the Spendor BC1--no stands yet, so they just sit on some 20cm-tall hard rubber blocks now--is a revelation. The timbre of human voices and all kinds of instruments just sounds so right, and the music so beautiful.
I happened upon this used pair of BC1 for sale years ago. Although they were not in tip-top condition, all drivers did work and the serial number suggested that they were from 1984, only some 750 pairs apart from the last pair of BC1 built in 1994. Recalling what REG (Robert E. Greene, my favorite audio writer, who used the ESL63 for many years) had written about the BC1 over the years (even though he never wrote a review of the BC1), I acquired them on a whim but somehow never really set them up properly in my system. Now I know why REG eventually switched to the BBC LS3/6 type (including a pair of BC1) as his main speakers.
I happened upon this used pair of BC1 for sale years ago. Although they were not in tip-top condition, all drivers did work and the serial number suggested that they were from 1984, only some 750 pairs apart from the last pair of BC1 built in 1994. Recalling what REG (Robert E. Greene, my favorite audio writer, who used the ESL63 for many years) had written about the BC1 over the years (even though he never wrote a review of the BC1), I acquired them on a whim but somehow never really set them up properly in my system. Now I know why REG eventually switched to the BBC LS3/6 type (including a pair of BC1) as his main speakers.
@gnnett Wondering how your Acorn build is coming along. Would love to know your impressions if they're already up and running.
At this stage I am seriously considering going down the DSP route as @cab has suggested, perhaps with a DSPeaker Anti-Mode x4, though that would be an investment with no guaranteed return. However the Acorn has so much transparency, clarity and presence to spare that any loss in these areas from adding DSP won't be a concern, especially if we get correct timbre reproduction/tonal balance in exchange. I am still going to try silk screen damping and crossover on the HV side (where much lower current is involved I presume) first though.
Meanwhile, I have been enjoying the supremely musical Spendor BC1 so much that I am contemplating some easily reversible tweaks such as felt on the front baffle and super tweeter protection/diffusion brass grille, à la the LS3/5a. Funnily I have found myself always spending my free time listening to music instead with the BC1...
At this stage I am seriously considering going down the DSP route as @cab has suggested, perhaps with a DSPeaker Anti-Mode x4, though that would be an investment with no guaranteed return. However the Acorn has so much transparency, clarity and presence to spare that any loss in these areas from adding DSP won't be a concern, especially if we get correct timbre reproduction/tonal balance in exchange. I am still going to try silk screen damping and crossover on the HV side (where much lower current is involved I presume) first though.
Meanwhile, I have been enjoying the supremely musical Spendor BC1 so much that I am contemplating some easily reversible tweaks such as felt on the front baffle and super tweeter protection/diffusion brass grille, à la the LS3/5a. Funnily I have found myself always spending my free time listening to music instead with the BC1...
Last edited:
My apology for the lack of updates.
I realized that I needed a speaker system with "correct" tonal balance and harmonic content as the reference in order to carry out the equalization work/mod on my Acorns. Of course "correct" is a questionable concept in audio playback so here it just means what feels to me the closest to my prior experience of live music in various concert halls and clubs.
After rediscovering the BC1 I've managed to acquire a pair of its successor, the SP1, which does even better in reproducing the violin, the piano and human voice--exactly what my implementation of the Acorn has difficulties reproducing well. So my free time during this past month was mostly spent aligning the SP1s (voice coils and cabinets, etc.). It will be some time before I start working on my Acorns again.
I realized that I needed a speaker system with "correct" tonal balance and harmonic content as the reference in order to carry out the equalization work/mod on my Acorns. Of course "correct" is a questionable concept in audio playback so here it just means what feels to me the closest to my prior experience of live music in various concert halls and clubs.
After rediscovering the BC1 I've managed to acquire a pair of its successor, the SP1, which does even better in reproducing the violin, the piano and human voice--exactly what my implementation of the Acorn has difficulties reproducing well. So my free time during this past month was mostly spent aligning the SP1s (voice coils and cabinets, etc.). It will be some time before I start working on my Acorns again.
Well it could be IM distortion but I’d be surprised if there were significant levels of that given that the Acorn 11 has separate panels for bass, mid and treble.
Thanks @mattstat and @Studley!
For me, part of electrostatics' attraction is their low-ish distortion, and the Acorn appears to be at least as good as any other electrostatics in this regard. Being dipole and usually of a planar configuration also contribute to the sound characteristics of electrostatics, but dipole planar speakers have their (often severe) problems. I am quite curious about how the Klippel system takes and sums up measurements with planar, linear source and MTM/WMTMW speakers. Speaking of distortion, it happens that I will soon have the chance to evaluate a prototype pair of Purifi Ushindi equipped speakers...
Back to my Acorns--the peaks in the higher range were the worst offenders and the hole in the mid bass didn't help either. Together they severely limit the types of music/recordings that could be enjoyed. Unfortunately, unlike with traditional dynamic type speakers, I don't have the knowhow or required parameters for using my simulation software with the Acorn to speed things up. I do, however, have a theory and some ideas that can only be tested via experiments. Just need more free time!
For me, part of electrostatics' attraction is their low-ish distortion, and the Acorn appears to be at least as good as any other electrostatics in this regard. Being dipole and usually of a planar configuration also contribute to the sound characteristics of electrostatics, but dipole planar speakers have their (often severe) problems. I am quite curious about how the Klippel system takes and sums up measurements with planar, linear source and MTM/WMTMW speakers. Speaking of distortion, it happens that I will soon have the chance to evaluate a prototype pair of Purifi Ushindi equipped speakers...
Back to my Acorns--the peaks in the higher range were the worst offenders and the hole in the mid bass didn't help either. Together they severely limit the types of music/recordings that could be enjoyed. Unfortunately, unlike with traditional dynamic type speakers, I don't have the knowhow or required parameters for using my simulation software with the Acorn to speed things up. I do, however, have a theory and some ideas that can only be tested via experiments. Just need more free time!