Socket7

Looks interesting doesn't it? Not really? Well screw you too.

So I've had my stereo in my car for a while now, and I've been putting off doing this for a very long time because I thought it would be a complete pain in the ***; it was, but not as much as I had figured. The biggest problem was getting the test tones I wanted to use. Since I'm working with a 30 band equalizer, I wanted each tone to center around the frequency I'd be working with. For some reason, most internet test tones work on 5 to 10 Hz frequency gaps, rather then a 1/3rd octave scale. So I downloaded some tone generating software and made the tones I needed. You can download them here.

It is widely believed that in order to properly tune a graphic equalizer with this many channels that you need a $1500 Real Time Analyzer. An RTA does make the job easier, but I'd rather spend $1500 on mods, not tools, so I decided I was going to do this with a $50 Radioshack SPL meter. It takes a bit more time, but you can get just as good a picture of what's going on with your system without shelling out massive amounts of cash.

It turns out that the $50 Radioshack SPL meter has a surprisingly flat frequency response, needing only minor adjustments to bring it's measurements inline with your $1500 RTA with measurement microphone (Ok, the super low range is pretty out of whack, but good luck finding an EQ that will tune at 10-30Hz. so who cares)

Correction tables for the radioshack DB meter can be found here....
http://forum.audiogon.com/cgi-bin/fr...6&openflup&1&4
http://www.danmarx.org/audioinnovation/rsmeter.html

"But Socket!" you say. "Why would I want to go to all this effort? I've already got a killer system! I can hear it half a mile away!" Well that's good for you, but there is more to fine audio then dual 10" Subs playing at 130Db. Crossovers need to be set correctly, Levels need to be matched, and the over all response curve needs to be smoothed out to prevent unnatural peaks and holes in your frequency response that could lead to you not being able to hear your favorite part of your favorite song, or even damage your speakers.

Cars are the worst possible environments for audio work. They're noisy (acoustically and electrically), and they don't generally have good acoustical properties either. With my car sitting at idle, the noise from the car itself was around 65dB. My test tones ended up playing around 80dB, and that's a bit to close to the noise floor for my comfort. I will be repeating this test at higher volumes to see if I get more accurate results.


My testing procedure was to have the car at idle, Set the stereo to my usual listening volume, then while sitting in the drivers seat use the DB meter to measure the volume of each tone. Because of all the reflections inside a car, I found with some test tones that moving the meter a few inches in one direction or the other could lead to over a 5 dB difference in SPL! I'm still not sure what to do about this but I've got 2 thoughts, Hold the meter very still, or sweep it around and take an average. For these tests, I just tried keeping the meter in the same spot. I measured the rear speakers, The front speakers, the sub, and then everything together.

Looking at the graph, I see a couple of good things, and a couple of bad things.

The good news is that the front and rear speakers seem to be level matched well, and seem to be at the same average SPL. The sub is significantly higher then this, Which is also OK, as I have control over it's volume from the front seat, Giving me the option of blowing people out of their seats with bass, or backing off to make it match well with the rest of the system.

The bad news is that it seems my crossovers are set to low across the board. It appears the sub is starting to roll off at 50 Hz, when it should probably be closer to 80, and the same is true of my component speakers. It's also pretty clear I've got a severe hole in the 100-200 Hz range, and a peak in the 4-6.3KHz range.

There is one more piece of bad news. Shortly after making this graph, I managed to blow out my rear speakers. (probably from a combination of a low crossover point and some stupid EQ work I did before plotting this graph.)

Once I get my new speakers in, I will make and post more graphs, and the end results of my EQ fiddling. I'm doing this mostly because I need to write this all down to see it in my head correctly, and I'm posting it here in case anyone has any interest in doing it themselves or if they have any questions or suggestions. (I'm very interested to know If I'm doing anything wrong.)

And here are my new toys to install this weekend (and finally clean up my trunk)

firebirdude

iTrader: (3)

What was your total time spent for everything beginning to end including modeling up the graph?

You made the graph from scratch? I take it each symbol is were you took a measurement. Then did you just free hand the line? What's with the yellow line at 40Hz?

Also, I seem to remember folks saying that the radio shack dB meter does not accurately measure low frequencies. Folks would pull 151dB @ 51Hz on a Termlab mic during a competition, then turn around the next day and measure with the Radio Shack meter and only bump out a 135 @ 51Hz. (For example)

Certainly not knocking on any of the work you've done, just asking questions.

Also note a lot of install shops own nice RTA's and will let you use them for a price. I think i was quoted $50 awhile back. Same price I know, but it's probably a lot quicker.

EDIT: Damn it. Just read your links. lol I knew I wasn't talking out of my *** about the meter!

WikkedOne

I think I am going to take your method and do it to my car.. except I'm not going to post a pretty graph on here - out of pure emberrasment... I'm afraid to see the gap between my subs and interior speakers (100 - 200 hz ).... Thanks for the sound files that helps a lot!

firebirdude

iTrader: (3)

Bah. Many people will say that's the hardest range to conquer. You're not alone.

WikkedOne

Well - it's about what I thought... I startied with my
pink noise at about 88 DB and idle was about 68 DB - so stuff that droped that low wasn't even really registering at all

My highs are REALLY high, i think because I was holding the meter so the mic was pointed toward the tweeter...

looks like I need to do some EQ work... I'm really starting to wish I had an external EQ.. the built in EQ in the Pioneer Z series is way better than the F series.. but is still poor at best..

you can't change each channel independently.. you can only change one per section (low mid or highs) - then select a band in that range then set a wave "wide or narrow" that will also change the ones around it instead of letting you set each one to a level you want... it's a stupid, lazy design.

Socket7

Total time spent was about 2 hours, including finding a website to print the perfect size graph paper for writing down all the values, generating the tones, burning them to CD, and driving over to my work where I wouldn't bother or attract the attention of neighbors.

Once I had all the values, I plugged them into an open office spreadsheet, applied the necessary correction values for the meter to each data point, and made a graph from all the plots together. There were a couple different smoothing methods, but this one made the most sensible looking curve (the other was much much worse). I suspect the weirdness at 40hz on the rear speakers is from putting in 0 db values for the 25 and 32 hz range, but I'm not sure. I'm working it out as I go.

The local stereo shop I go to charges $60 an hour for RTA tuning. I'm sure I could get more tunes/hour by using their kit, but I already had the DB meter for other reasons, and I've already spent enough money at my local stereo shop. If I let a shop do it for me, then I'd never get to see the curves myself, and that would make me a sad panda.

For a piece of 50 dollar kit, the Radioshack SPL meter is pretty damn flat as far as response curve goes. More importantly, it's consistent, so the calibration corrections don't need to be calculated for each individual meter. I figure any inaccuracies remaining after applying corrections are small enough that they're going to be negated by environmental noise, or I can compensate for them by ear.

Socket7

100-200 hz is the bane of my existence. I'll be listening to some tight drum and bass, and the bassline hits a certain note and it's WAY quieter then everything else. I don't know if I've got waveforms canceling each other out or what, but it drives me crazy and it's what drove me to do this in the first place. I'd love to apply a half millisecond delay to the left channel and see if things get louder, but I've got no gear that can do it, and I'll be damned if I'm going to drop $$$ for yet another box to put noise in my signal path. Maybe if I could build a delay circuit for like $20....


Wikkid, Looking at your graph and mine, it seems like the hole around 100-200 and the peak at 4-8KHz could be environmental, seeing as we both have it.

mat128

Try editing your track on your computer to add a delay to the left channel then burning it. Should cost you close to nothing (worst case, a cd-r) and you would be able to verify the cancelling.

Edit: Oh and which meter did you use? digital or analog?
http://www.radioshack.com/family/ind...goryId=2032222

firebirdude

iTrader: (3)

Actually.... that might be doable. Not a terrible idea!

Socket7

Great idea! How didn't I think of that.

I could put a delay on an entire song and see if anything changes. If it solves the problem then I can put a delay on all 50gb of my mp3s! Why use a delay box at all!

I'm trying this out tomorrow when I get my new deck speakers in.

I had been trying to think of a work around for not being able to control delay in my system for ages. I feel like a dope for not thinking of adding the delay to the audio itself.

I have the digital meter. I haven't seen any information on using analog ones. they could be exactly the same or totally different.

firebirdude

iTrader: (3)

You know...... your readings are going to change slightly because you're now using different speakers.

Socket7

I was actually planning on re-graphing everything. I have all day, and several sheets of graph paper.

I'm also gonna have to do it all over again every time I make an adjustment to see the result of it, and I should be making a graph for every seat in the car (but I'm pretty sure I'm too lazy to try and trap myself in the back seat of my car).

I'm interested in seeing the results of the crossover changes i made this afternoon. midbass seems better. Not perfect at all, but better.

I did say this process is a pain in the ***, but the upside is that for no money, I'm getting practice at properly tuning a stereo, how to use Excel, and I'm working on my penmanship!

firebirdude

iTrader: (3)

This is all well and good.....and of course being the person am I, I'm thinking of doing it myself...... but in the end.... if it sounds great to you, who the hell cares right?

WikkedOne

I also used the digital meter... i am going to print out my results and play with my EQ tomorrow and try again...


oh and as for taking the data - i just made the spreadsheet on my laptop and took it out to my car and entered it in as I did it...less graph paper

i think i'm going to try again with the car off too - to see how low the dips actually go..

firebirdude

iTrader: (3)

I think I'm more interested to see how different the curves are with the car running/not running/windows up/windows down/etc.

Socket7

A tune for driving with the windows down or the sunroof open. I like that idea. I've got 8 memory slots to fill anyways...

My laptop has a 10 minute battery life these days.