My DIY stats

[j.postema]

Hello,

here some pictures of my DIY electrostatic loudspeakers. First picture is made before painting them :)

https://s32.postimg.org/fymvssvg5/P1010258.jpg

https://s21.postimg.org/ee4m3cixz/P1010015.jpg

Regards

[walpurgis]

Those look great Johan. Nice job! :)

[Ali Tait]

Yes indeed, nice job. How do they sound?

[Macca]

They look good. Bit of subwoofery in the background there? What is the whole set up going to be?

[walpurgis]

They are rather reminiscent of Audiostatic speakers to me. No bad thing.

[j.postema]

@Ali Tait: even though I used very simple segmentation they still needed equalizing to sound right. They sound detailed with a tight bass. Stereo imaging is good in my opinion. They play (at least for me) loud enough. (I don't enjoy playing very loud so no problems with the stats :) ) Friends of mine seem to enjoy them too - always when they visit me the stats play their favourite music (electronic) so they have a social function as well.

@Macca: Thank you. It's always a bit messy in my living room - it's an empty pc case in the background. After a few weeks of questioning I decided to not add an electrodynamic (sub) woofer to the stats. (I can do that later) Although there is some lacking of lower frequency output I think the quality is still very good. ("tight") I decided to accept less lower range frequencies and spend the money and time on new, bigger stats rather than adding woofers to the existing ones. The new stats will be over four times bigger in radiating surface so I hope I will manage to get a good result.

@walpurgis: You are right, personally I like the Audiostatic design.

[Macca]

Forget the subs, just go bigger. I like your thinking.

[jandl100]

Yes, Audiostatics!

I had a pair - superb except for the bass that seemed to slope off slowly below about 200Hz. :scratch:
Quite impossible to sensibly integrate with a subwoofer. I kept them for a year or so, so the bass was a limitation I could live with.

My Audiostatics had the bolt on "wings" that were supposed to help with the bass. Tbh, they didn't really!

But yes, increasing the panel size should help sort that.

Great project, well done, sir! :thumbsup:

[Ali Tait]

Yes agree, very hard to marry a panel to a cone sub sucessfully. Better building a bigger panel in the first place.

[jandl100]

Yes agree, very hard to marry a panel to a cone sub sucessfully. Better building a bigger panel in the first place.

Actually, because of the very slow bass roll-off with the Audiostatics, the subwoofer integration thing was way worse than usual, even for stats.

[Barry]

Panel speakers are dipoles, sub woofers have a monopole polar distribution, so it is very difficult to integrate the two. The Bowers & Wilkins DM70 electrostatic hybrid suffered from this problem.

I once heard Audiostatic speakers at an audio show (can't remember the model) and was very impressed. Curiously the room was empty, so I could sit in the optimum position, have the system to myself and spent some time enjoying the music. In fact music was the sole topic of discussion with the Audiostatic rep - always a good sign.

Congratulations Johan on your DIY electrostatics. I knew a collegue who built himself a pair of electrostaic speakers based on the Ed Saunders series of articles in the Audio Amateur magazine. They weren't easy to build, but he persevered and eventually was successful. Me - I'm happy with my Quad '57s, but I take my hat off to you for building your own. Well done!

Did you wind your own transformers or did you use commercially available items? What value of turns-ratio did you use?

[j.postema]

Hello I would like to thank everyone for the kind words!

I'm currently in the process of building the new, bigger stats, but I'm only at about 1/4 progress. I will soon post a photo :) In the meantime, I enjoy the current stats, they still do well.

Regards

[j.postema]

Hi everyone,

although I didn't visit this place for a while I still like it :) In the meantime I made some progress on my new stats: tomorrow I plan to attach the membrane to the first new stat. I feel excited as it took me a few months to build just one stat, but I think it's rather big - although I'm getting used to it's size. It's still "do-able" in my opinion. It's just a little smaller than an average door. I will test one stat first before building the second one. As soon as the first stat is assembled I will post a photo!

[walpurgis]

We will certainly be looking forward to seeing your work Johan.

[struth]

Seen a big one before but not heard it. told they can sound super.. good luck, and some pictures will be appreciated

[jandl100]

This is a very interesting project - I love big stats.

Keep the photos coming! :thumbsup:

[j.postema]

Hi,

about two weeks ago I finished my first bigger stat. It was interesting to turn up the volume and it just sounded well :) No technical problems. So I as very happy, also initial sound quality impression was good. But after some extensive listening I concluded that bass was not tight and mid range frequencies sounded a bit smeared / dull.

I think I need do damp resonance frequency and I think (hope) that wil solve the problems. But I'm still quite satisfied with the restult. The photo shows both the first project (small stat) and the new project (big stat). I'm expecting to finish the second big stat in just a few days so I can listen in stereo. (at the moment I combine the small and big stat for stereo)

https://s18.postimg.org/vwezmvo3d/P1010340.jpg

[walpurgis]

You do some great looking work Johan.

[j.postema]

Thanks!

[User211]

If you have panel resonances you cannot avoid you can just notch them out digitally with a high Q value and a deep (say 10DB) cut. Provided you know at which frequency(s) they occur at. The idea is to apply a very narrow deep attenuation at the target frequencies.

Truth be told there are plenty of speakers that resonate badly at certain frequencies panels or not, that could almost certainly benefit from this.

Do you have any construction pics? Where did the plans come from?

[j.postema]

If you have panel resonances you cannot avoid you can just notch them out digitally with a high Q value and a deep (say 10DB) cut. Provided you know at which frequency(s) they occur at. The idea is to apply a very narrow deep attenuation at the target frequencies.

Truth be told there are plenty of speakers that resonate badly at certain frequencies panels or not, that could almost certainly benefit from this.

Do you have any construction pics? Where did the plans come from?

Hi thanks for your suggestions. How would you prefer to notch out the high Q resonance(s)? I've been thinking about getting a DSP but it is adding another digital device, sometimes with proprietary software, so I think something DIY op-amp based sounds better to me. (although a DSP gives more flexibility and options) I think correcting the source signal rather than the loudspeaker is a preferable idea as damping screens always affect high frequencies.

There's some good news too: yesterday I finished my second new loudspeaker and I didn't expect an improvement in sound quality but to me it seems these smeared mid / high frequencies and the un-tight lower frequencies have disappeared. The dull, disappointing sound isn't there anymore. I was a bit disappointed as it took many hours of labor, although it was like meditation to me, those repetitive tasks. It got me away from my computer screen which is a good thing :) Maybe just 1 loudspeaker was exiting room modi, or the small and big stat were out of phase?

The plans are made by myself although I read many books and articles about esl theory and construction. The idea is to divide the stator in multiple segments to tune frequency response and imporove high frequency dispersion and maybe amplifier load gets a little easier too. The dimensions were determined by outer dimensions of a door. The stators contain over 1 km of wire.

Here are some photos I took during construction:

https://s29.postimg.org/w9bh7vv8z/P1010290.jpg (https://postimg.org/image/w9bh7vv8z/)

https://s29.postimg.org/fwwdl1alv/P1010312.jpg (https://postimg.org/image/fwwdl1alv/)

https://s29.postimg.org/gl583z9bn/P1010314.jpg (https://postimg.org/image/gl583z9bn/)

https://s29.postimg.org/4uramlgj7/P1010317.jpg (https://postimg.org/image/4uramlgj7/)

https://s29.postimg.org/gnp3qtcz7/P1010309.jpg (https://postimg.org/image/gnp3qtcz7/)

https://s29.postimg.org/nlt7wrb3n/P1010341.jpg (https://postimg.org/image/nlt7wrb3n/)

[User211]

Assuming you use a laptop and a DAC and have a USB test mic...

You could try https://sourceforge.net/projects/peace-equalizer-apo-extension/

So install Equaliser APO then the Peace GUI. This is basically a configurable graphic equaliser that will allow you to insert a tight notch anywhere. You can also use it just for fun and games. I do.

So each vertical slider is frequency configurable. Set one to your target frequency, then use a high Q value to make a narrow notch, and use the slider to take, say 10 DB off it.

What's good is that this works for all laptop audio software e.g Spotify, TIDAL etc, JRiver.

It's only good if you know your target frequency(s). Do that by doing slow frequency sweeps in suspect areas.

For instance, if you hear a resonance when playing music, you could analyse it using this:

http://www.nch.com.au/wavepad/fft.html

If you do that I find you can quite easily "see" the instruments playing just by looking at the pattern in the temporal FFT. So if you know when and where the resonance occurred you can deduce the frequency area.

The use the same s/w to do a sweeping sine wave based tone rising in frequency from the bottom of the suspect frequency range to the top, SLOWLY, so you hear it when the tone breaks up. So if you suspect 385Hz, do it from 360 to 410Hz, say. Note at what range within the sweep causes an audible problem, then design a notch to get rid of it.

Of course you could stone sweep from 20 to 20KHz, but it has to be loud to hear resonances, you have to sweep really slowly, it's annoying and soon enough you will want to kill yourself.

[User211]

BTW I use a notch with the above s/w at 385Hz very, very occasionally. Why? The Apogee mid range/treble ribbon has a problem in the sense that it is split into two sections. In the longer section, the frequency that corresponds to its length means that high energy levels cause it to oscillate wildly (thanks, Morrissey!:D). You can physically see the ribbon go nuts, and hear it.

I noticed this issue listening to an Ocarina being played back. I've also heard it with some brass stuff. It is very rare that it actually causes a problem. TBH the crossover could quite easily fix this by being steeper than it is as the x-over frequency is around 500Hz.

I wonder if Apogee actually were ever aware of it for that reason. It is pretty simple maths. I'm sure they must of been. But it is so narrow band I guess they just decided to stick with mild slopes on the mid/treble ribbon.

[User211]

BTW I never noticed the resonance until I was sent an Ocarina track by the Apogee restorer, who in turn got it from the ribbon manufacturer. Determined to find out what was going on, I devised the above method.

It is good fun when you get into it!:)

[User211]

Another thing you need to be sure of is that any resonances you think you hear aren't present in the source. Pretty good way to check that out is to use a decent pair of headphones as a double check. You've built a massive pair of headphones anyway in a way:)

[walpurgis]

Another thing you need to be sure of is that any resonances you think you hear aren't present in the source. Pretty good way to check that out is to use a decent pair of headphones as a double check. You've built a massive pair of headphones anyway in a way:)

Yes. That's good tip.

[j.postema]

Thanks for all your good advice! I will try the FFT analyzer and equalizer.

[j.postema]

Just af few days ago I received the last 32 resistors for the second stat for electrical segmentation of the stators. Last two nights I compared my DIY panels to a set of Final 0.3 hybrid stats that were stored in my house, they belong to a family member. I'm happy to conclude that my DIY stats sound better than the Finals: clearer, more detailed, tighter bass. The Finals sound a little dull and hollow compared to the big stats. But the Finals are still quite good in my opinion given their much smaller size. I'm not sure if they are very good reference speakers, I've read some posts about problems with wrong filters, dips in frequency etc., cheap woofer together with some very positive reviews. But for now I'm happy with my new stats. Still not sure how they compare to other DIY stats and commercial stats like Martin Logan, Acoustat, Quad, Sanders Sound etc. and to magnetic loudspeaker systems.

I'm considering to build a (magnetic) woofer to supplement for low frequencies (below 100 Hz). I think it will be a sealed or transmission line woofer. Still not sure, but to me it sounds interesting. I'm wondering if those lower frequencies will add valuable musical information to the overall sound?

Picture below: comparing the Final 0.3 hybrid stats to the new DIY panels (white stat in middle was not connected)

https://s29.postimg.org/izmkymyaf/final_en_grote_elektrostaat.jpg

By the way, I tried to install Windows Vista on an old pc to run the EQ software, but it failed as it was too slow. I will retry soon with my fast pc.

Regards

[jandl100]

Awesome! :respect: They look fantastic.

[User211]

They do look great - didn't realise just how much bigger they were than the previous pair.

Be curious to see an FR plot with the mic at the listening position.

Dipole subs would be a good idea I think. Since the main speaker is a dipole, I'd be really tempted to maintain that aspect all the way down.

How about an OB woofer with some huge drivers? 15 inchers... get yerself some proper radiating area...:D

[j.postema]

Thanks for your kind words. Yes I wanted a "no compromise" loudspeaker just to see what is possible, especially for low frequency capabilities.

I think you are right about dipole / OB woofers, although I'm a bit afraid about sensitivity and maximum output. But maybe I will just give it a try, I think it's nice to not have cabinet coloration.

I added frequency response images (brown line is freq. response) together with distortion graphs. The first picture is measured at listening position (microphone about 3 meters from the loudspeaker), but unfortunately only at about 70 dB output and with an external equalizer.

After that I did several new measurements at 1 meter distance and higher output (picture #2 and #3), without external equalizer. There is some heavy distortion below 200 Hz and I think this is caused by undamped resonance: when I attach a damping screen at the back stator this distortion is much lower. I'm using a UMIK-1 microphone so I have no idea about distortion of this microphone, so these graphs have to be taken with a grain of salt.

https://s27.postimg.org/irrgn6ws3/distortion.png

https://s28.postimg.org/eiiifkodp/image.png

https://s28.postimg.org/5ow7c7qst/image.png

[User211]

The Umik mic is very good I use one. I assume you applied the correction file for your particular mic before doing the FR tests.

Will PM you in a couple of days or so.

[j.postema]

Hi, good to hear that the microphone is very good :) Yes I applied the correction file.

Regards

[j.postema]

I tested Equaliser APO and Peace GUI and I didn't manage to observe any audible sound quality degradation. First I played a record from my Raspberry pi + hifi DAC+ and at the same time I started the same record on the windows pc, Peace GUI turned off, same sound level. There was no audible difference in sound. After that I turned on Peace GUI, still no difference between both systems. Then I measured both fundamental resonant frequencies using REW and added a notch filter for each individual channel. After that still no degradation of sound quality of the Windows + Peace GUI system compared to the Raspberry. So I'm convinced now that these software equalizers are quite good, at least for the low cost!

[User211]

A cost of zero. The Peace GUI looks dated but it is functional. But the best bit is it isn't application specific - it works with all media player software.

Once you've used a test mic and know what is wrong with your system FR wise, I have found it to be the best method of correction. I vastly prefer it to the likes of DIRAC.

You can do things with it that aren't practical in a passive x-over. That's cool.