Crossover for Electrostatic hybrid

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
At the moment I have my homemade electrostatic speakers working well. However, I need to involve a magnetic speaker to supply the low end (~250hz and lower). As anyone can see, I need a crossover. Should I use a passive crossover after the amplifier, or should I use an active crossover and do some bi-amping? Right now I am using a Yamaha A-27 (100 WPC) to run the electrostatics, but I recently purchased a power amp, Yamaha MX-830 (190 WPC), that I plan on using to drive the inefficient ESLs. I haven't decided on what preamp to buy, but I was looking at a Yamaha C-4 (I'm sticking with Yamaha so my system looks a little coherent). How feasible is it to make a crossover right after the preamp that will send the mid/highs to the MX-830 and the lows to the A-27? Can I send the output of the preamp into the A-27 without problems with impedance matching? The A-27 is an integrated amplifier, not a power amp like the MX-830. Any thoughts or designs are welcome.
Thanks ahead of time.
As anyone around here will tell you, I'm a big fan of active crossovers. Removal of the passive crossover makes things a lot more efficient (insertion loss into filter networks can be horrendous), allowing all the power from the amp to get to the driver.
It also allows for more predictable crossover points & slopes, more easily changable levels in order to match drivers, etc. etc. etc.
The downside, of course, is money.
There have been scads of discussions on crossovers. Rummage through them and decide whether you want to build your own or buy a kit.


I have found that to get the best integration between the bass and panel you need a low a crossover point as possible.
Grey is totally correct about electronic x-overs.

There are several issues you need to consider.

1) Determine the low frequency resonance of the panel and try to keep it as far away from the x-over as possible.

2) The choice of slopes of the x-over. I usuall like 6db on the high pass and 18db on the low pass. (Similar to the Dalquist x-over). This allows a single capacitor for the high pass, if you know the input impedence of the amplifier.

3) If the amplitude of the low frequency resonance is high enough this can be corrected with filter (trap).

4) If dipole cancellation is a problem this can be corrected with shelving filter. You will be amazed what you can do with an electronic filter/x-over, that is not possible with a passive filter.

Another filter to try is the subtractive filter(see the Pass Labs web site)which sums to perfect phase and I have also have had good results with the Linkwitz-Riley filter.

The resonant frequency of my speakers is about 125hz. Where should I put the X-over? Is one octave enough room? Someone once told me that I should keep the crossover well below 500hz and the dangerous mid-range. I will look up those crossover designs that you've given me, Jam, and see what works best.
I have the capabilities of an active crossover. However, I plan on giving a pair of these speakers as a gift to someone without the means for an active crossover. It's hardly a gift if the recipient has to go out and buy a control amp and a couple of power amps. :) So, I need to have a passive crossover design as well. Where should I go to get one? Also, it'd be interesting to directly compare the active/passive crossovers. The difference is most notable when heard with one's own ears.
Thank you very much for your input.
You might want to consider a digital crossover - Sanders, formerly innersound, now uses only digital crossovers, and heres why - you can do very steep slopes. This may be less important for the woofer, but with the panel, it allows you to set the crossover much lower without exciting the resonance. Going from say 250 hz to 450 hz can make a world of difference in the overall sound, and you will only be able to do that with a 48db crossover. Keep in mind that your panels will probably become substantially more power hungry as the crossover frequency drops. One unit thats quite popular for this is the Behringer DCX2496, and its not too pricey. It can even take a digital input, which allows you to have a digital domain crossover, which is way cool.
Active is the way to go. *One reason being the difficulty of configuring a
passive that works well because the stat's impedance varies with the
frequency (it's like trying to hit a moving target). *Other reasons are
higher efficiency and the ability to easily adjust the crossover frequency
to suit your system. *If you are using two different power amps, you would
need to wire them "horizontally", with one amp driving the woofers
exclusively and the other driving the stat panels exclusively. **If you are
using matching power amps, you have the option to "vertically" bi-amp the
system, with one amp assigned to each speaker - - one channel drives the
woofer and the other channel on the same amp drives the stat panel. *As a
general rule, it takes more power to push bass; so, with matching amps
horizontally wired, the bass amp would be over-utilized while the
mids/treble amp would be under-utilized. *Music typically changes from
moment to moment-- *sometimes bass-heavy and sometimes mids/treble heavy.
As most amps have two channels drawing from a common power supply,
"vertical" bi-amping is best because whatever power not being momentarily
drawn on one channel is then available to the other channel :) **Since
re-wiring my amps from "horizontal" to "vertical", I find that my system
sounds more powerful because all power from both amps is now available.

well beeing a bad boy I ´d say "what the hell, he´s using Yamaha amps!"
Yamaha builds really fine Bikes....I love my Fazer....but amps for an ESL...I doubt it?? But then that´s my perdonal taste :cool:
I´ve written some thoughts including sims and measurements on passive crossover designs for hybrids.
Anyway, if You can allow the complexity of active circuitry You should go for active. If done correctly it clearly outperforms passively filtered hybrids. And if You want it to sound really good, then forget about OP-amp-Xovers, but use JFET or Tube gear instead and eventually get Yourself a decent Tube amp to drive the ESL :p
Also forget about 6db/18dB rules...they might to a certain degree apply to dynamic drivers, but ESLs are a totally different thing. The best Xover slope is the one that fits the demands best, not someone´s dogma.
The intended Xover frequency and the driver´s behaviour define the the required slope of the Xover. My experience shows that for example a 6dB (electrical) Xover hardly ever works well with ESLs. You almost awways need to equalize the acoustic phase cancellation, which is easiest done with a 2nd order Q>1 Highpass. Most probabely You will need to notch the ground resonance (100-150Hz). Now the panel exhibits a sharp drop off below the resonance. Together with the electrical filtering the acoustical response will then depend on the CR-freqency. If the crossover point is well above the resonance the acoustical slope will be closer to the 2nd-order HP-filter character. If the CR-frequency is closer to the panel´s resonance the acoustical slope will be much steeper.
A second notch is often needed around 1kHz-4kHz where a lot of panels -especially unsegemented ones- exhibited a slight midrange-peak.
All this could be done and has been done passive as well as active.
Since most basses run linear to much higher freqs, the bass´s acoustical slope is nearly equal to the electrical filter slope. To achieve symmetrical flanks the bass typically needs high-order filters, in the range 3rd-6th order. Typically this requires large and costly passive is Your friend here too ;)
If the needed effort is too high for You, You might think of digital filtering.
Though some think of digital as the holy grail of sonics, it surely isn´t...not necessarily. Since all DSP-based CRs I know of (wether they be IIR or FIR) use OP-amps in their analog circuitry their sonic performance isn´t much better than that of an OP-amp based analog filter. If they are of the much cheaper IIR-type they are just easier to handle, more versatile and require no knowledge in electronics, but behave exactly as an analog Xover. Only FIR-Xovers would allow for more, but those are still quite costly and often come with user interfaces wchich are not easy or comfortable to deal with.....and they come with OP-amps.

Since most amps offer a 47kOhm input impedance, You can run the output of a decent Xover without any problems into this amp. The advantage of using one integarated amp in one channel is that different gains can easily be set to level the volumes of bass-channel and ESL-channel and thereby omitting with level settings within the Xover.

Draft Calvin

Sanders' cookbook is badly outdated and I publicly nominate you to replace it. Seriously, if you were to compile and edit your prior posts alone, you'd be halfway home on a cookbook of your own.

Sorry about the public nomination but you're the heir apparent, like it or not, and I had to do it :D
Yeah, Calvin you are truly an inspiration.

About the X-overs... I have some insane ideas.
How about hooking up a MiniDSP with dedicated dacs and tube buffered outputs? S/PDIF in and no Op-amps's at all in the analogue signal path. :D

Digital filtering and no need for a pre-amp any more.

(I'm kind of half way there myself with my current project.)

thanks alot for a praise I think I don´t deserve, because what I´m writing is neither unknown nor new, nor is the level of advancement high. It´s just looking at measurement curves, some basic applied electronics, usage of very good gear and my ears.

What Markus suggests could be the next step towards sonic Nirvana ;-) Since mosts users use digital audio equipment, the straightforward way would be to stay in the digital domain as long as there is signal manipulation required -apart from simple amplifying. This requires less analog and AD/DA-stages involved with signal handling.
Staying in the digital domain also typically offers more flexibility and comfort.
The really only drawback I see is the excessive usage of OP-amps inthe analog stages simply out of convenience and cost and some marketing-efficient measurement curves.
I had tested my system with digital IIR-filters, a SOTA-digital IIR-FIR filter with room correction and the JFET-buffer based analog filters. If the listening room didn´t require extensive treatment, the analog filter always was slightly preferred, followed by a combination of IIR-filtering together with FIR-post filtering/room correction.
In the development of our HardDiscRecorders and the DAC I tested several I/V converter stages and learned that a lot of HiFi associated attributes can be found in OP-amped gear -like resolution, control, speed, etc- but lifelikeness, emotion and authenticity were unknown to OP-Amp stages. In direct listening comparisons it took usually just seconds to identify and dismiss the OP-amp stages.
So I assume -and soon will test and know- that the digital filter including FIR and IIR- equipped with discrete (feedbackless) I/V converters and analog Buffer stages will be the cream of the crop.

Last edited:
Dan, I can help you with a passive filter if you make a impedance (with phase) file for the Esl.

Or else, try a 12 dB filter with 100 uf in series and 5 mH + 12 ohm in parallell with drivetransformer input.

For the bass, start with 10 mH iron core in series with bassdriver(s), 2 ohm in series with 1000 uF paralleled.
Last edited:
Just poking my head up from weeks of mic testing:

Consider yourself lucky to have the basic freq compass anywhere near flat. Can't say that I (or Toole-like people) have heard big phase impacts esp. with ESL dipoles and I include almost any twist of phase except for wiring speakers out of phase*.

So for crossover design, you want the most compass from the ESL but not to mess too much with the principle resonance. I wouldn't lose sleep over the fine-points but I would iron-out problems with a parametric equalizer working with a simple crossover like JonasKarud suggests. Simple.

Without again pooh-poohing the fine points, the place to address cross-over issues is before you buy the SaranWrap. If you want more than just ESL tweeting, maybe the effort should go into keeping your crossover way low, into a less sensitive range than you can use with your 125 Hz panels (which are pretty low). And then maybe have a compromise fried-egg tweeter over 6 kHz. Sounds strange?

About amps, ESLs can be really piggish loads. While I am skeptical of mega-power for cone drivers, hundreds of watts of power available into 2 ohms isn't a bad thing to have.

Beats me why people cautious about every source of distortion known to human-kind are trusting when it comes to that magic word: digital. Going into and out of digital is absolutely monumental processing and Calvin is wise to urge not going in and out of digital too often. Just because computers do arithmetic flawlessly, doesn't mean the same for audiio.

*Looking at that this week with a mic... ooops, I better not say something too provocative.
Last edited:
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.