Home Theater Forum and Systems banner

1 - 6 of 6 Posts

·
Registered
Joined
·
509 Posts
Discussion Starter #1
Hi all, I kinda 'promised' this in my build thread, and I see that there is a discussion happening at the moment about full range equalisation, and it made me get off me backside and finally post this how to.

Well, not really a how to as such, as that implies that I am right!! Nahh, but it is how I do it and I like the results.

Just for background. I have always been plagued by a bit of 'harshness' at times. I use a deqx as the crossover etc for my system, and one of it's functions is that you can use the remote to dial in eq points on the fly. So at those times of harshness, I used to cut back (what I thought to be) the offending area.

I had always been fine with eq in the bass region, as are most here, yet had 'philosophical' objections to the use of eq in the full range, mainly based on things of the 'where is the correct point, left ear, centre of forehead, right ear?' type. Because, as we know, the position of the mic makes a huge impact in the upper frequencies.

One night I had a brainwave, and what follows (with hopefully not too boring explanations) is what I found.

Ok, to start, (and hoping I can remember how to load graphs....been a while) you may have to take this on faith, but as I use the deqx I know the speakers are flat and accurate. For the sake of illustration, I just now threw the mic vaguely in front of the right speaker to show my points, but I am in the middle of renos so there are ladders etc everywhere, so 'ignore' the roughness of the graph.

for thread right raw measurement.jpg

Ok, managed to do that without too much drama, unsmoothed and 1/3 on top of it, right speaker (woofers turned off, won't bother explaining why and we are talking full range anyway)

So (for me at least) that was the starting point, that I know the speakers are flat. Hence, at the LP, any non smoothness can only be due to the room. Stating the obvious I know, but hey. (I emphasise smoothness to avoid confusion with 'flat at the LP' which can imply equal sound levels at all frequencies at the LP)

Here is the bass measurement from LP

for thread, unsmoothed to 200.jpg

But this is what we are interested in.

for thread three measurements across lp.jpg

These are the three measurements across the LP, 1/3 octave. Not a pretty sight!! I usually take (say) five measurements across the LP, but for the illustration now I only did three.

As I say, the room is a mess with stuff everywhere (am building huge bass traps actually!!) so tbh it does not reflect the normal situation. Still, if you can imagine it, the harshness (say) I used to cut back with the remote was around 2k (what most think of as 'bright' etc they usually blame the tweeter....nahh, safe bet number one, it is never the tweeter, always the midrange. You can basically take that to the bank)

Now, this is the 'breakthrough' in my thinking I mentioned before....the 'peak' I used to cut back to cure the 'brightness/harshness' was only a peak because of the room induced dips before it.

Did that make sense?? I found that when I filled in the valleys to make a smooth graph (as we should have because the speakers are flat) it cured the 'brightness'.

So, my goal in doing this is not to 'tailor' the graph so much (tho of course that is perfectly ok if it is your taste) but to simply restore it to what 'should' be it's natural shape, 'perfectly' smooth with the natural high end tweeter roll off.

(it would be a mistake to have 20k at the same level as 2k as 200...ie flat at the LP across all frequencies, that would be horribly bright indeed).

So, heeding the natural high end roll off as natural, my intent is solely to smooth the graph.

Ok, we have the three (or five or whatever) measurements across the LP...and when we look we can see what stays constant and what changes.

I then load them all into the average tab, and apply FULL smoothing to them, and then average them all.

This is for a number of reasons. We only want gentle correction, massaging. Earlier I spoke of philosophical objections, this was one of them. It is a fact that you cannot use drc to do full range correction per se, one glance at an unsmoothed graph will show you why. AND that correction is so intimately connected to that one solitary point in space that anywhere else is very wrong..left ear right ear stuff.

Once we have switched to 1/1 smoothing then we are only doing gentle corrections, and strangely enough it 'forces' you into use only very broad filters! (that's a good thing in case you were wondering)

So go back to the filter adjust tab, and use the eq filters window to work out what filters to apply to the averaged graph.

for thread use filters to predict average response, make 'normal'.jpg

Again, this is only for the sake of illustration, but here I applied the filters to the average, with (as explained) the sole aim of gently massaging it back to a smooth natural shape.


I now (of course) load those filters into the deqx (you would use whatever unit you have) and redo the measurements across the LP. You will probably find the measured response to not accurately track the predicted response, due no doubt to the fact we have fully smoothed them prior to working out the filters, and any slight change in the frequency in a fully smoothed graph can make a large difference, and the mic will not be in the same spot the second time around.

Whatever, as said before, this is just gentle massaging so it does not matter too much.

So here are the three measurements across the Lp after filters applied

for thread. measured response after correction across lp.jpg

Even now we can see an obvious improvement in the collection of graphs.

So for the final 'check' those measurements are again loaded into the average tab at full smoothing and of course averaged.

for thread the average of the meaured response after correction.jpg

Bingo, and it very closely matches the predicted response earlier. So we have used very broad gentle filters to restore the correct balance between the frequencies, and have not limited ourselves to one position in space, presumably the more measurements across the lp (you can go up and down, back and forth as well) will make any correction less severe and more gentle, and more broadly applicable.

Bear in mind that as we have two ears, they can cross compensate a lot with measured discrepencies from ear to ear, so we really do only want gentle corrections.

I have also found that (depending on your chair) by measuring in free space (ie move the chair away) you are more and more only addressing the room. It is your call whether or not you feel you need to correct from the reflections straight off the chair. I find it better to move it away.

I find that tonally it all works much better, more relaxed and less strident (you may have different symptoms of course, as I say I was troubled by 'harshness')

Funnily enough, the real cure for my harshness was not 'cutting back the offending high' but 'filling in the valleys that caused the offending high'.

That was the big counter intuitive surprise for me.

By all means try it for yourself, I make no guarantees that you will like it as much as I did, and if any try it report back the findings good and bad.

I will not be offended!!
 

·
Banned
Joined
·
22,577 Posts
Very interesting write up, although I do have a few questions.

What exactly are you measuring? Each speaker independently or two speakers together... or all speakers (is it a two channel system or multi speaker theater system)?

I know you show the 3 lp measurements before and then after, but have you thought about measuring each lp and showing the before and after for each lp? 3 separate graphs of two measurements each to show how each lp changed from unequalized to equalized?
 

·
Super Moderator
Joined
·
9,261 Posts

Great write up, Terry! I found your comments about moving the chair to be particularly interesting.

Yeah, that 1-6 kHz region, the ear is really sensitive there. If that's messed up things can sound pretty nasty.

Where those three LP measurements like, left ear location, right ear, and between?

Now, this is the 'breakthrough' in my thinking I mentioned before....the 'peak' I used to cut back to cure the 'brightness/harshness' was only a peak because of the room induced dips before it.

Did that make sense?? I found that when I filled in the valleys to make a smooth graph (as we should have because the speakers are flat) it cured the 'brightness'
Yup, makes perfect sense. Glad to see you were able to figure that out. :T

Once we have switched to 1/1 smoothing then we are only doing gentle corrections, and strangely enough it 'forces' you into use only very broad filters! (that's a good thing in case you were wondering)
Yeah, it’s best not to get too narrow with the filters, but I don’t think I’d smooth beyond 1/6 or 1/3-octave for the purposes of deciding how to equalize. Smooth it too much and you can end up with filters too broad, so broad that they’re encroaching into areas beyond the peak or valley, where there is no problem. That said, at the end of the day, the goal is an improvement in sound quality, and if that’s what you got, then :T

Regards,
Wayne
 

·
Registered
Joined
·
509 Posts
Discussion Starter #4
Very interesting write up, although I do have a few questions.

What exactly are you measuring? Each speaker independently or two speakers together... or all speakers (is it a two channel system or multi speaker theater system)?
Thanks sonnie, good point that I should have made clear.

I am strictly a 2ch guy...rather slit my wrist than waste time watching a movie! ha ha. I did notice (praps need to read it a bit more closely) that a lot of the discussion in the other thread revolves around the mutli channel bit and how and what you measure, maybe that is why I did not pay too much attention??

So, 2 ch only, but even then is it not true that the front left and right (excluding centre for now) the most important part of ht??? If so, then I would certainly do the same steps to the two front mains.

I start with the baseline of the deqx having done all it's crossover and driver corrections, so that is covered for me. IF however I were using a deq (say) and came over to a blokes house, then I would have to do those preliminary steps.

I would use the mic about a metre away (or a distance sufficient that the drivers 'gel') and correct that speaker to flat, using the auto eq function to get it close, then check and tweak that with rew and any last corrections needed.

Left and right seperately measured and corrected independantly. Prob only from 200 up say, you don't want to start skewing individual speaker response due to the room (one in a corner one in free space). I'd correct the bass as usual from the lp with both channels driven.

Once that is done, then you are 'at the same stage' as I am when I went thru the above.

I know you show the 3 lp measurements before and then after, but have you thought about measuring each lp and showing the before and after for each lp? 3 separate graphs of two measurements each to show how each lp changed from unequalized to equalized?
Ahh, just had a thought. When I said three across the lp, you being a HT guy prob thought three seats in the theatre, I meant 'not limited to one point in space meant to represent the 2 ch listening position', so my measurements are something like one foot left of the left ear, six inches left of the left ear, centre of forehead, six inches right of right ear etc etc.

Cannot see why for HT measuring from different seats could not work tho, that may even be better as then you are only correcting the changes, nit the things that stay constant.

Given what I just said, then you can see in the before and after graphs that I did post the before and after correction for each of those three points, it's just that it is not three seats rather three different positions.

Great write up, Terry! I found your comments about moving the chair to be particularly interesting.

Yeah, that 1-6 kHz region, the ear is really sensitive there. If that's messed up things can sound pretty nasty.

Where those three LP measurements like, left ear location, right ear, and between?
Yeah, you managed to work that out even tho I was very unclear on it. I did not move the chair for that quick write up, but it IS interesting indeed to see how much the chair actually mucks things up! Without it the graph really smooths out. The lesson is that (2 ch g=uy only etc) I can imagine a good stereo upgrade would be a chair that does not reflect so much junk back at you.

On the list....prob a better upgrade than new amps or cables that most audio*&^%% rely on.

Yup, makes perfect sense. Glad to see you were able to figure that out. :T
A big win when I finally spotted it. I HAD tried full range eq before, and not liked the results. So I don't remember now just why I decided to give it a go again, but previously I had always arbitrarily limited myself to rules like 'only up to 800 hz' or things like that.

Must have been a hangover from my earlier assumptions about full range eq (and you know what happens when we assume right??) Anyway, once I realised all I had to do was massage the graph back to it's natural shape, and not have arbitrary frequency limits, then it suddenly all clicked.

Yeah, it’s best not to get too narrow with the filters, but I don’t think I’d smooth beyond 1/6 or 1/3-octave for the purposes of deciding how to equalize. Smooth it too much and you can end up with filters too broad, so broad that they’re encroaching into areas beyond the peak or valley, where there is no problem. That said, at the end of the day, the goal is an improvement in sound quality, and if that’s what you got, then :T

Regards,
Wayne
I tried that too Wayne, and TBH I didn't find your fears to be significant in the real world. Having said that, none of what I wrote is prescriptive, each can modify to suit and if they feel 'no more than 1/6' then go for it. Try it and see basically.

For many, 'transparency' or other audio^%$#@# buzzword is paramount. I strongly suspect that they could have confused a too narrow filter (arising from using 1/6 say) and attributing that very 'sharp' filters characteristics to 'transparency' or whatnot...falsely concluding something?? just theorising on the fly here.

They do not show up in my reply as I type, but (I think) if you have a look at the before and after graph of any one of the measurements at 1/3 (after doing it at 1/1) there is not too much evidence of the wide filter mucking up stuff outside it's intended range.

Ha, I will prob be proven wrong when I get to see them again after posting!! heh heh

I have only one other report on whether this works than myself, a friend who has had in the past expensive Moon pres, cdps etc etc and now just runs bog standard electronics etc with his Dynaudio C4s (I think, the really tall multi driver ones) and eq's, believe it or not, with the deq.

He reported back very enthusiastically the results, so two out of two right now. Very small sample size, which is why any reports (good or bad) back into the thread is of course welcome.

(he is now quite proud about how inexpensive his front end is....it feels a relief for him to have broken out of the audio&*^@! regarding all this stuff about electronics)

Want good sound?? 'Fix' the speakers, and fix the room. That's it. ALL else absolutely pales by comparison.

I mentioned room treatment in my build thread, have started on that too (hence the construction site I mentioned earlier in my room)

I did a quick test of fixing room reflections by simply hanging raw batts around the room.





You have absolutely NO idea of the improvement such cheap and simple measures brought. Need to pretty it all up of course, but it just goes to show that if you are serious there is no stopping anyone doing simple tests and experiments.

The improvement would make a 'front end overhaul' hang it's head in it's pathetic expensive shame.

the real point (apart from showing how audio^[email protected]#%'s who argue stuff all day long on forums are barking up the wrong tree)??

What I have outlined here is simply the last 'icing on the cake', it is NO substitute for good and proper room treatment, and conversely with good and proper room treatment these (now more subtle) changes can be appreciated all the more.

At least for HT guys, proper room treatment seems to be more done and accepted than it is for the hi-end 2ch crowd. AND eq is more acceptable.

The number of absolutely horrid expensive 2ch systems around because these important things are dismissed...you have no idea.
 

·
Registered
Joined
·
509 Posts
Discussion Starter #5
another point that just occurred to me, it is often mentioned by those 'against' eq that by simply moving the mic 'as little as two inches' really changes thins, hence do not rely on/use eq. (usually retailers of room treatments!!)

MOST eq is done in the bass region, and what should be immediately apparent is that the graphs below 200 hz change hardly at all. So that statement is an incorrect application to the wrong area.

Above 200 hz, then yeah, moving the mic a little produces big changes, 'sort' of proving that eq does not work on the full range.

that is the reasoning behind why I take multiple measurements, smooth them and average them. We are removing all the little uncorrectable changes from s slight change in position, and reducing the changes to the big gross 'globally' applicable ones, and gently smooth them.
 

·
Banned
Joined
·
22,577 Posts
When I said three across the lp, you being a HT guy prob thought three seats in the theatre, I meant 'not limited to one point in space meant to represent the 2 ch listening position', so my measurements are something like one foot left of the left ear, six inches left of the left ear, centre of forehead, six inches right of right ear etc etc.
Actually when I am measuring for my HT room, I do similar... left ear, right ear. I only eq for the main listening position. I am using Audyssey though, so it does equalize each individual speaker as well. I also did a response graph some while back and posted it here at the Shack... how at arm's length vs. ear, there was very little variation in the response... naturally that may not be the case for all rooms.


Given what I just said, then you can see in the before and after graphs that I did post the before and after correction for each of those three points, it's just that it is not three seats rather three different positions.
What I meant here is if you can post a before an after for each lp, it would be more meaningful. I see the before and afters, but they are not grouped together. You have one chart for befores and one chart for the afters. Can you post one chart for the first lp that included before and after on one chart, then the next lp before and after, etc, etc. If you save those .mdat files for the graphs above, you can simply merge them and tick the boxes for the ones you want to show and not show on the graph.
 
1 - 6 of 6 Posts
Top