" Full Range " Dayton RS100S project

[jdc0589]

Put this on PE as well.

Purchased two Dayton RS100S-8's before I knew a thing about speaker building (dont know much more now.. ).
They are in a 3 chamber bass reflex enclosure VERY similar to Wolf's in the PE project showcase (fractionally smaller).
These will get ditched at some point in a few years after I have built another few projects, but they serve as pretty good computer speakers for the moment.

These drivers are small enough I feel like a 2-chamber design would do just as well, but the port response in the 3rd chamber is unusually flat at the peak, and blends well with the other two (don't have those graphs uploaded).

There are no physical filters in place at the moment, all fine tuning has been done on the TERRIBLE 5-band eq on my 30 year old amp; with which, adjusting the 1khz slider doesn't necessarily mean the adjusted peak will be within 250hz of 1khz....

What do you think? Be as mean as you want, I just want some real feedback.


These are on axis measurements.



***************************************UPDATE***** ***************************
And....here is the new graph :T
Thats a +/-1.5dB tolerance.

[Jed]

For no crossover that's pretty dang respectable. Can you do anything about that peak at 1.7k? It might create a bit of harshness.

[jdc0589]

For no crossover that's pretty dang respectable. Can you do anything about that peak at 1.7k? It might create a bit of harshness.

That little blemish you have commented on is a source of never ending frustration. It can sound a tiny bit harsh at low volumes before bass can develop with poorly recorded material, but with good recordings it sounds a little bright, but pretty good overall.
For most material that isnt too bass heavy, my ear prefers a +2db boost to the lower octave and +1db to the upper octave, which helps audibly mellow the 1.7khz spike. But, I doubt you would notice it in the first place at moderate volume with good media.

There is no hope of smoothing it out with my god forsaken 5-band eq. The closest control points I have are '1khz' (which really means more like 750hz) and 4khz (which seems pretty accurate per tests).
Unless I can find a 32-band eq via sandbox- software (so I can use arta's internal 1 channel measurement/generation), it isnt getting any better without physical notch filters.

On the other hand, if anyone knows of a cheap 32-band parametric eq, I could be tempted, doubt it would be cheap though.
Some company SERIOUSLY needs to produce an affordable digital eq that can take frequency response input (list form, frd file, whatever) and automaticly render and implement an equalization curve ( I guess it would just be the inverse of the response curve?). Does something like this even exist at any price?

[jdc0589]

FYI, someone over on the PE forums was kind enough to take a raw response graph I posted (no eq), produce an frd file from it somehow (something called SPL trace?), make a zma from the dayton supplied pdf, and feed it through PCD to see what the component filter network and resulting response curve would look like.

The results: 7 components, and the FR looked almost like a carbon copy of the graph in my first post. Havent priced it out, but that would probably be fairly expensive in relation to my total current investment.

[davey_m]

Some company SERIOUSLY needs to produce an affordable digital eq that can take frequency response input (list form, frd file, whatever) and automaticly render and implement an equalization curve ( I guess it would just be the inverse of the response curve?). Does something like this even exist at any price?

If you are using a computer you can add a convolver plug-in to some media players like Foobar2000. Then you can implement a FIR correction filter to correct the driver response. Unfortunately, ready-made software packages to create the correction filter are a bit expensive:

http://www.acourate.com/

http://www.juicehifi.com/index.html

These are fully featured applications with driver correction, crossover and room correction capabilities. Maybe a bit of an overkill for you.

Another approach would be to just try room correction at the listening position. Fortunately, there's an open source application available:

http://drc-fir.sourceforge.net/

When implementing room correction via a music player like Foobar2000, you will of course only correct for two stereo channel music replay. I don't know if it's possible to correct video or multichannel replay via Windows media player, but there are convolver plug-ins for Windows Media Player.

Windows Vista is said to have a room correction capability, but I don't know too much about it.

It is also possible to create a stand-alone digital I/O driver correction equalizer with the following:

* Low power fanless Linux PC like the VIA series. Can be packaged in a compact box
* Acourate software for measurement and filter calculation
* The open source convolver BruteFIR
* ALSA compatible sound card

Look at the Acourate site for more details. Such a device matches your requirement exactly, but it is a high end solution more often used with hi-end hi-fi. But on the other hand it can be expanded to include room correction, individual driver correction and FIR based active crossovers.

[jdc0589]

Cant figure out how to delete posts..

[jdc0589]

Davey....you and some other guy on PE are my new best friends.
Here are the results of the days work:

Raw Driver Response, NO adjustment/filter/eq of any kind



Filter Response: Did this manualy since the auto functions wouldnt work....



Corrected Frequency Response:



Since I cant get any of the auto analysis features to work, I went through by hand and cant find a peak/valley variation more than 1.4dB.
To answer the inevitable question...YES, it sounds as good as that graph looks!

[mpotoka]

Maybe I missed it--what exactly were you using to get that response curve? Was it a software eq?

[MuaDibb]

jdc0589, I see you're using ARTA, are you measuring using the FR1 or FR2, and where are you setting the delay?

[jkrutke]

Corrected Frequency Response:



Since I cant get any of the auto analysis features to work, I went through by hand and cant find a peak/valley variation more than 1.4dB.
To answer the inevitable question...YES, it sounds as good as that graph looks!

That's pretty flat. Just for giggles, you should try this: 3dB/octave rise starting at about 6kHz to about +5dB at 20kHz; 3dB/octave cut starting at 200 and down about 6dB at 60Hz. Should compensate a bit for the driver's off axis droop on the top end, and a little less emphasis on the "mud range" down low. Let us know how that works if you try it.

[jdc0589]

Maybe I missed it--what exactly were you using to get that response curve? Was it a software eq?

Are you familiar with FR compensation for microphones?
Its the exact same concept. Yes, all done via software, but the key is having an media player that can implement the correction curve (mine is in the form of a .wav file).

jdc0589, I see you're using ARTA, are you measuring using the FR1 or FR2, and where are you setting the delay?

Im still new to all of this, and while ARTA is really easy to use, I havent figured all of it out yet,
I used FR1 for everything, couldnt get any kind of correct graph with fr2. I really dont remember on the delay setting in arta, I know its 0 by default with FR1 measurements though.

That's pretty flat. Just for giggles, you should try this: 3dB/octave rise starting at about 6kHz to about +5dB at 20kHz; 3dB/octave cut starting at 200 and down about 6dB at 60Hz. Should compensate a bit for the driver's off axis droop on the top end, and a little less emphasis on the "mud range" down low. Let us know how that works if you try it.

Thanks! I will try it out in a few days. I dont really need it in my current listening position (near field, speakers 2 feet from eachother on either side of the computer monitor), but it will likely be necessary when I move them back into my dorm tomorrow.

[jdc0589]

Well, im not very happy with the quality of the upper octave using foobar and its associated convolver plugin. Some high quality recordings sound perfect, but compressed mp3's seem to develop some annoying 'static' in the upper octave (mainly with symbols, etc...).

Im trying a plugin for windows media player, which seems better, except I havent gotten it to work yet...


***Another EDIT:
WMP 9/10 just blows apparently...downgraded to media player classic, and my filters work perfectly on all file types :T

[JonP]

I'm also wondering what you used to do the correction. Kinda looks like the screen to REW on those FR responses... but its hardware only with that one.

[mpotoka]

Are you familiar with FR compensation for microphones?
Its the exact same concept. Yes, all done via software, but the key is having an media player that can implement the correction curve (mine is in the form of a .wav file).

I am familiar--I was just wondering what software you used. I noted you are now using MPC (were using Foobar) but what did you use to create/implement the .wav file?

[jdc0589]

I'm also wondering what you used to do the correction. Kinda looks like the screen to REW on those FR responses... but its hardware only with that one.

I am familiar--I was just wondering what software you used. I noted you are now using MPC (were using Foobar) but what did you use to create/implement the .wav file?

Ill go ahead and spill the beans in detail:

REW was intended to be used with hardware :W . You can export filter impulse response files in wav format or the settings in text format :T .
I took initial measurements with ARTA, and since I dont have the full version had to export the FR in text form. Loaded that in to REW, did some random stuff with the filter and measured the real results with arta to make sure REW's estimates were accurate (which they were), created the filter curve by hand (the auto features sucked), exported filter impulse response as a wav file, voila. Please note, this is NOT the best way to make a filter curve. In the next week or two, I may write a simple program in Java that will open a frequency response text file, calculate the target DB lvl, and generate another text file containing the perfect equalization curve. This can be imported in to REW as an impulse response file, viewed and compared to the FR file to make sure it worked out, and then exported as a wav for use with a convolver. I can not believe REW didnt implement the type of calculations necessary to generate a eq curve for a near perfectly flat FR, its so easy, I know exactly how to do it.

FOOBAR
Used it with foobar and its convolver plugin first. It sounded wonderful with good recording in UNCOMPRESSED format, although im sure it would have been fine with a flac or something similar. But, I do have some mp3 files, and it did an absolutely TERRIBLE job applying the convolver with whatever decoder foobar uses for mp3's. Most of the issues were with the upper octave; symols took on a static like sound.

EDIT: ....Ive gone back and discovered that somehow, when playing many compressed files, the audio output was....clipping...straight out of foobar. Never had this happen before. All you have to do to fix it is lower the foobar internal volume control to 1/2 or so. Weird....

WMP 9/10
So, I switched to a multiplatform convolver that con be used with Media Player Classic, Media Player 9/10, ZoomPlayer Professional, and a few others. Tried it with WMP 10 first, and once again it was WONDERFUL with content on physical CD's. However, Either you just cant get in to the advanced setting of WMP or I cant figure out how, but the convolver would not apply at all to whatever decoders iare used for non-CD files (this could be something weird on my machine, but I doubt it). PLEASE let me know if you know what the issue was, I would much rather use WMP, as I love the interface.

ZoomPlayer Pro
Next thing I tried was the free trial of ZoomPlayer professional, which is by far the most advanced of the options, and ONLY option if you want to be able to use the convolver while watching videos (maybe WMP too). But, setup was very complicated and the new version of ZoomPlayer had a different interface than the one pictured in the convolver setup guide.

Media Player Classic
So, I moved on to Media Player classic and was about ready to give up. I download it (2.0mb), and am surprised to see you dont even have to install this thing... The interface is terrible, you cant create a library to play files from quickly and easily, it doent have very fast load times, and I was generally un-impressed....then I tried it. It took all of 1 minute to install and configure the convolver, and it works perfectly with ALL FILE TYPES ;x( ;x( ;x(. I still hate the interface; but at least it works.

Im perfectly willing to answer any question, just let me know!
Also, if you would be interested in that simple program to generate very very good eq/filter curves, make it known so I can see how much interest there is.

[jdc0589]

I am familiar--I was just wondering what software you used. I noted you are now using MPC (were using Foobar) but what did you use to create/implement the .wav file?

Well, I kind of feel like an idiot all of a sudden. Turns out the problem with foobar was....clipping?!?!?! :huh:

yea....you heard me right. For some very odd reason, no matter what output device I use, with the volume slider in foobar all the way up, the windows main volume at like 1/3, and my amp at 1/20, the foobar output was somehow clipping... I dont understand, I have NEVER had any kind of audio/media player do that before. But it makes total sense as to why it wasnt doing it with the few CDs I was listening to, which require a much higher volume to get to the same output as many of my mp3's... Somehow the mp3 files (which somehow had a really high native volume) were causing the audio to clip straight out of foobar...

This is really weird... any input/ experienced anything like this?
The good news is I can use foobar now.

[Dennis H]

Turns out the problem with foobar was....clipping?!?!?!

That's to be expected if your EQ curve is using any boost rather than all cuts. It's trying to boost loud passages above digital 0 so you need to reduce the volume in the digital domain to keep everything below 0 dBFS. At least that's my guess, I have no idea how Foobar's convolver works.

[Dennis H]

In the next week or two, I may write a simple program in Java that will open a frequency response text file, calculate the target DB lvl, and generate another text file containing the perfect equalization curve.

That would be easy in Excel. Pick a reference level below your measured curve and subtract the measured curve from that so your EQ curve is all negative numbers. Include the min function to set any positive EQ number to 0 -- like where the bottom end dives and you don't want to boost that. Something like =min($b$2-b6,0) where b2 is the reference and b6 is the dB you want to convert.

[jdc0589]

That's to be expected if your EQ curve is using any boost rather than all cuts. It's trying to boost loud passages above digital 0 so you need to reduce the volume in the digital domain to keep everything below 0 dBFS. At least that's my guess, I have no idea how Foobar's convolver works.

I understand that part, but the weird thing is my curve was almost ALL cut/pads accross the board. The only areas I had to use a little gain was no more than 1dB. While some areas had MASSIVE padding.

[jdc0589]

That would be easy in Excel. Pick a reference level below your measured curve and subtract the measured curve from that so your EQ curve is all negative numbers. Include the min function to set any positive EQ number to 0 -- like where the bottom end dives and you don't want to boost that. Something like =min($b$2-b6,0) where b2 is the reference and b6 is the dB you want to convert.

Same idea, but I think it will be easier in a stand alone utility.
Plus, I can knock it out in a day in Java.

[mpotoka]

Well I will have to think about this. I was actually considering doing an array of RS100s for my 2.1 TV room. My HTPC is basically the pre/pro--the only input I have is my Wii. What I am wondering is if I can use a solution like this to apply to everything that my computer outputs---CDs, DVDs, my OTA HD via Vista Media Center, and my Wii input. Any suggestions on how to accomplish that? Perhaps I need to ask over in the HTPC forums on AVS...

[jdc0589]

Well I will have to think about this. I was actually considering doing an array of RS100s for my 2.1 TV room. My HTPC is basically the pre/pro--the only input I have is my Wii. What I am wondering is if I can use a solution like this to apply to everything that my computer outputs---CDs, DVDs, my OTA HD via Vista Media Center, and my Wii input. Any suggestions on how to accomplish that? Perhaps I need to ask over in the HTPC forums on AVS...

I would check over there. You could start to have issues with a software solution like I am using, as it takes some time for the processing, which could cause alignment issues with the video. If you are interestedin this, I would look into a hardware solution, there is a semi-cheap behringer that can do pretty much the same thing.

[Dennis H]

Same idea, but I think it will be easier in a stand alone utility.
Plus, I can knock it out in a day in Java.

Well, you have to remember I'm well known for my laziness. A day in Java vs. a minute in Excel....

[jdc0589]

Here are some really terrible pics. I need to put another coat or two of paint on them, but they get moved around so much right now I'm just going to wait. Couldn't find the digital camera, so these are cell phone pics for the time being. They completely disappear in any low light areas, pretty neat. Finish is a 'blackboard' paint. You can actually write on it with chalk.
This isn't where they normally are, was home for Christmas break.

[mordensiur]

Maybe foobar problem can be solved with Replygain. When you cut some db problem should be out. Maybe some limiters ( " sofl clipping limiter") could help.