First measurements I dare to post!

[entrecour]

I have been playing with REW on and off for quite a while now and I have finally got to the stage where I dare post something. I am sure I could get some valueable pointers as to wether I am heading in the right direction.

My setup is a HK AVR330, BDF2496, SVS SB12-Plus, and 5 x PMC DB1+ in a standard 5.1 HT setup based around a HTPC.

I am using an ecm8000 microphone for measurements and feeding the test tones directly into the analogue input of the AVR330 set to stereo, surround off, no DSP.

I have attached the measurements including the waterfall chart. All comments appreciated. I don't understand why the waterfall doesn't extend past 100Hz.

There were several stages where I wasn't quite sure if I was doing the right thing:-

First of all I played a film with a lot of LFE (U571) to increase the signal level to the BFD from the AVR330 so the BFD signal meter was all green and occasionally orange. That mean increasing the sub channel in output in the AVR330 to max (+10) when in Dolby Digital surround mode.

However wwith the soundcard output was connected via the amp (stereo, surround off) I had to decrease the subwoofer channel output to stop clipping in the BFD. (With the AVR one has to set speaker levels for each surround mode.)

After taking the measurements, setting the target level, and programming the filters generated I reset the subwoofer channel output to +10 and rebalanced the speakers, using the subwoofer gain control when it came to the sub.

So how did I do?

[brucek]

All comments appreciated. I don't understand why the waterfall doesn't extend past 100Hz.

Because you have the graph axis on it set to LIN (linear) and not LOG (logarithmic). The FreqAxis icon button in the top right corner of REW will change that. Also use the same scale for the waterfall as your frequency response charts for easier comparisons. (45dB-105dB and 15Hz-200Hz)

So how did I do?

The filter number 3 is really the only one you probably require. Once you're past the crossover area, the filters obviously become less effective since the sub level is dropping and the mains are taking over. The best bet is to equalize the peaks before the crossover (or around 100Hz) and then add the mains and remeasure and see if you really require filters up above 100Hz.

A tip. There is a filter sort button (from low to high) on the filter panel. Very convenient to have your filters go from 1,2,3,4,5 instead of 3,2,1,4,6,5

brucek

[entrecour]

Thanks brucek,

Here are the new graphs, the corrected curve using just filter 3, and the waterfall plotted on a log freq axis. The waterfall looks a lot uglier - but I am not sure how to interpret it. Any pointers?

Thanks.

[brucek]

Basically you can consider it a frequency response graph with the added axis of time.

Drag the waterfall slider back to 1 slice. Look familiar? It's the response graph. Now move to 2 slices. You're now looking 10msec later in time at the level of the signal (if window set to 300msec). By the time you get to 30 slices, you're looking at 300msec later. If any frequency is still at an audible level you'll still be hearing it - considered perhaps not good..

brucek

[thewire]

That looks pretty good IMO. Perhaps it could be better but still good.

[entrecour]

Thanks brucek and thewire for your comments.

This time around I redid all the measurments (so they may not look identical).

I set one filter at 58Hz and then redit the measurement with main speakers and sub (I balanced both to approx 75dB).

I set the target level by setting speaker type to Full Range.

With my mains I still have a peak at 58Hz so I suppose this is a room effect, or?

Then I have some some sharp dips at 76Hz, 100Hz, and 177Hz. Are these anything to worry about?

Should I filter the peak at 130 Hz?

I have attached the waterfall - to your expert eyes does it indicate problems that I should look to remedy?

[entrecour]

Edit: I can't upload the waterfall as at 222K it exceeds the 200K jpeg limit so I have zipped it.

[brucek]

I set the target level by setting speaker type to Full Range.

No, when you add the mains and are trying to examine the crossover, you simply take the measure from 0-200Hz as if it was the sub only, so you use the sub pink noise and type as sub.

With my mains I still have a peak at 58Hz so I suppose this is a room effect, or?

Well, almost everything is a room effect from modal resonances to reflections etc.....

Then I have some some sharp dips at 76Hz, 100Hz, and 177Hz. Are these anything to worry about?

Sharp dips are usually left alone (and are generally inaudible). If they're around the crossover you can try and eliminate them with the phase control (when both sub+mains are playing).

Should I filter the peak at 130 Hz?

No.

Anyway, you have a classic example of how correcting a single peak can really help out. Look at your peak at 58Hz. You have correctly set the level matching between the sub and mains as a result of the peak at 58Hz. It overpowers and results in one-note bass. You can't turn up the sub or it will sound overpowering.

But, if you remove that peak, then you are able to wholesale turn up your bass level, and as a result, the frequencies between 20Hz and 45Hz that were previously hidden, are now revealed and you feel like you just got a new sub......

brucek

[entrecour]

Ok, I'll do another measurment with sub pink noise,etc.

What's the best way for me to remove that peak now? As I have one filter set already at 58Hz if I automatically add another filter at 58Hz will the BFD sum them correctly?

[brucek]

What's the best way for me to remove that peak now? As I have one filter set already at 58Hz if I automatically add another filter at 58Hz will the BFD sum them correctly?

I'm confused, are you saying that the sub+mains graph was taken with a BFD filter in place for the sub and that the peak was gone completely with the sub only and is back just as strong when you took the sub+mains measure? Let me see the sub only with the filter in place (you only show the probable outcome in the graph you have posted).

brucek

[entrecour]

Yes, that's what's happened unless I screwed up somewhere!

Here is the sub only plus the one filter

[brucek]

Interesting.

What is your crossover frequency?

Can you take a mains only measure from 0-200Hz and post it? (just shut off the sub).

brucek

[entrecour]

I normally use small but with this amp when in stereo, no surround, no dsp mode the speakers can only be set to large. However there is a X-Over mode which is set at 80Hz and has the exactly the same measurment signature.

That peak is at exactly the same freq as for the subwoofer i.e. 58Hz.

[entrecour]

I redid the mains only measurment with the amp set to stereo, DSP ON, SMALL

This produced a very different result... still a peak at 58Hz though

[brucek]

OK, you've definitely got a large resonant peak at 58Hz. You may have to over compensate your equalizing a bit with the subs 58Hz filter. Hopefully you can lower it somewhat below the target and then the sum of that and the mains peak will be a better result. I would also increase the level of the sub somewhat in relation to the mains so that the level around the 30Hz region was at least equal to the level of the mains. To get a better feel for the overall level of the mains, take a measure of the mains only from 100Hz-1000Hz (use full range type here) and then turn on smoothing ~1/3 octave and see what the overall level is and then bring the sub up to that level at least. I think your sub is a bit low.

brucek

[entrecour]

Thanks for the pointers.

Here are the new measurments
The first measurment is sub only, no filters.

The 2nd measurement is sub & mains with 1 filter at 58Hz. I ended up using the auto filter generated by REQ (58.35Hz, Gain -12, BW 0.25). Main speakers set to SMALL.


Do you think the sub is high enough now. I have clipping in the measurments if I turn it much higher.

Thanks!

[brucek]

Yeah, that's starting to look a lot better. I might add a filter at ~45Hz and turn the sub up a touch more.

BTW, you won't have clipping in the measurements if you simply go through the Check Levels and Calibrate SPL routine of resetting REW back to 75dB.

brucek

[entrecour]

Looking much better me thinks :-)

[entrecour]

And here is the corresponding waterfall chart. I increased the Time Range to 900ms, and kept the Window to 300ms. Looks like there are some strange things going on around 25Hz!

Should I be looking at room treatments (perhaps to address the slight dip at 36Hz) or are these kind of delays normal?

[brucek]

Looking much better me thinks

You're still a bit high at ~50Hz, but yeah, it looks fine now. Just drop the filter(s) in that area about a couple dB.

No amount of treatment will affect as low as a 36Hz dip. That has to be attacked by positioning usually and then sometimes you have to live with it if positioning isn't possible.

And here is the corresponding waterfall chart. I increased the Time Range to 900ms, and kept the Window to 300ms. Looks like there are some strange things going on around 25Hz!

Well, since the ringing at 25Hz appears to start at a much lower level than your signal, I would suggest it's room noise. This is where it doesn't hurt to use the Spectrum Analyser without a signal and look at your rooms background problems. I can identify my furnace and fridge etc.

I wouldn't look at a waterfall much more than a 600ms time axis.

Anyway, you haven't said if it sounds any better?

brucek

[entrecour]

Anyway, you haven't said if it sounds any better?

Ha! Good point! It sound a lot better, principally (as you mentioned earlier) it has allowed me to increase the sub level. There may be more room for increase there still. More listening required before I can give a final verdict.

However I have noticed some stubborn mechanical vibrations in the room as I am doing the sweeps (one long side wall is virtually all windows with venetian blinds) and I think a wall or the floor is vibrating somewhere. So probably some tweaking to do still - that's what I was hoping the waterfall might pinpoint.

I also have to check what the background room noises might be.

[brucek]

So probably some tweaking to do still - that's what I was hoping the waterfall might pinpoint.

The best way to find those pesky vibrations is to use REW's sine-wave signal generator and checkbox the 'frequency tracks cursor' option and move the cursor slowly around the low frequency area on the REW screen and find the noises - kinda fun.

Pictures are the worst I find. I put little pads in the bottom corners of the picture where it contacts the wall. Fireplace doors are horribly noisy also.

Hehe, this is what happens when you reveal the super low frequencies. It adds all sorts of vibration problems. It's like when you first get a high end system and then you hear all the horrible recordings you own.

I also have to check what the background room noises might be.

Yeah, below is a quick spectrum snapshot I took one day to see my furnace noise. I also picked up my PC soundcards 60 cycle hum and its harmonics at 120Hz, 180Hz etc, and also interesting to see my RPTV CRT horizontal oscillator at 15.750KHz in the room. The furnace was the noisy one. Turn off the furnace and it goes away. I was getting a weird ringing out with my waterfalls at about 30Hz. I thought it was my system and the room.


Spectrum Analyser - no sine wave stimulus


brucek

[entrecour]

Those are great tips.

What settings did you use for the Spectrum Analysis, anything different to the defaults?

[brucek]

What settings did you use for the Spectrum Analysis, anything different to the defaults?

See this post...

brucek

[entrecour]

Thanks, that was interesting reading.

I also noticed this

The RTA also allows you to change filters and watch the RTA screen for the real time changes. It's also really useful to adjust phase on a subwoofer for the best crossover response. You simply watch the RTA screen as you dial the phase control. A lot better than taking a bunch of measures to accomplish this task.

I only wished I had read this a few days ago when I went through 8 seperate phase measurements! :-)

I found it very difficult to choose the "best" phase. What should you aim for - the one that has the highest overall volume or the one that is the smoothest?