[ToBeFrank]

I've got my low frequency area pretty good. I'm using the BFD and Audyssey afterwards. However, above 100Hz the level is down 8-10dB. Audyssey is setting the trims in the receiver so I don't understand why it would be down. Any ideas or solutions to fix this? The crossover is set to 80Hz. The graph is 1/3 octave smoothed.

 

[laser188139]

I chased this question for weeks, learning several small things along the way. So I may be able to give you a few suggestions.

1) You are using a Radio Shack meter to calibrate the SPL level and measure the frequency response. This is said to have a ±2dB error, although I have seen a thread citing even larger variation among RS microphones. The Audyssey retail microphone is also said to have ±2dB error. So you could see normal differences up to 4dB without anything out of the ordinary happening. The only way to improve upon this is to buy a calibrated microphone, then you will have better control on the frequency response during the measurement process.

2) REW drives one or two channels depending on how you connect it to the AVR. If you are driving two front channels with a single monaural signal, you will see only +3dB of gain due to destructive interference between the speakers. But the two channels fed to a single sub will show +6dB of gain. So if you are driving two front channels you should expect to see the sub elevated by at least 3dB because of the measurement procedure. You can avoid this by taking separate, single channel, measurements of the sub with the left front, and then the sub with the right front.

3) You can also see variation if your REW microphone placement does not match the position of the tip of the Audyssey microphone, both in height and in the horizontal plane. Audyssey actually tries to set the levels based on its evaluation of all the measurement positions -- in some sense an average of the six or eight positions. So it could be looking at a different value than you are seeing using REW at only one position. You can try taking six (or eight) measures in exactly the same position as you used during Audyssey setup and average these, to somewhat replicate its evaluation procedure.

If you visually add +3dB to the curve in the region of the fronts, it appears as if the levels of the peaks in that range are close to the level in the sub range, and the remaining differences could be attributed to variation between the two microphones. But you will have a better feeling for this if you take the separate, single channel measures, and compare those. Best not to recalibrate or move the microphone between the measures; you would like to see the sub curves from the two overlay each other cleanly.

Good luck,
Bill

 

[ToBeFrank]

2) REW drives one or two channels depending on how you connect it to the AVR. If you are driving two front channels with a single monaural signal, you will see only +3dB of gain due to destructive interference between the speakers. But the two channels fed to a single sub will show +6dB of gain. So if you are driving two front channels you should expect to see the sub elevated by at least 3dB because of the measurement procedure. You can avoid this by taking separate, single channel, measurements of the sub with the left front, and then the sub with the right front.

Ah ha, that explains 3dB of it. I've now run the left and rights separately and the graphs are attached. You can see the humps at 327Hz and 405Hz. Is Audyssey setting the level based on those humps rather than the average level? If so that would explain it.

Audyssey actually tries to set the levels based on its evaluation of all the measurement positions -- in some sense an average of the six or eight positions.

I was under the impression Audyssey set the distances and levels only from the first measurement in the listening position. I've asked Chris at Audyssey for clarification on this.

If the humps I have are causing the problem, any ideas how to reduce them? I don't think room modes should be the cause at those frequencies so I'm not sure if more treatments would help (I've got quite a few right now).

 

[Nuance]

Even with the added 3dB to the speakers, the subs are still running hot, for what its worth. Many people like that, and you might too, so if it sounds good just enjoy.

Room treatments might help the peaks and valleys above 100Hz or so, sure. Your response below the crossover looks good, by the way.

 

[glaufman]

I'm not sure why exactly, but having made this mistake before, I've noticed that when driving with both channels, the increase in sub output can be significantly more than the +3/+6dB theoretical. I imagine this has to do with peaks/dips and room-boundary gains at specific frequencies, but I've confirmed on more than one occasion where it looked like the sub was running 10dB hot, that my partner-in-crime had plugged in both channels again and removing one brought the levels right in line.
I can't speak to your Audyssey questions, but as to the peaks 300-400Hz, what you do with them will depend on what they're attributed to. A little experimentation can go a long way here...

 

[laser188139]

... You can see the humps at 327Hz and 405Hz. Is Audyssey setting the level based on those humps rather than the average level? If so that would explain it.
...
I was under the impression Audyssey set the distances and levels only from the first measurement in the listening position. I've asked Chris at Audyssey for clarification on this.

If the humps I have are causing the problem, any ideas how to reduce them? I don't think room modes should be the cause at those frequencies so I'm not sure if more treatments would help (I've got quite a few right now).

In terms of setting the level, Audyssey uses a band limited calculation. I believe Chris has written that it is 500Hz-2kHz, or something like that.

Sometimes Chris has written that the first measurement determines level and distance, but after a few of these he has corrected himself and written that only distance is calculated based only on the first.

The displayed levels around 1kHz-2kHz are surprising, given what I understand about the levels. But as you say, this may be related to the peaks below that. You could compare Audyssey Off and On measures on each front separately. It could be that Audyssey is attacking even larger peaks before equalization, and the side effect of pushing the peaks down is that the levels above that are lowered, too. If your AVR offers preouts for the fronts, you can measure these directly (again without a mic file) and compare Audyssey Off to Audyssey On to see directly where it is applying its changes. This can be particularly interesting, in that if Audyssey is apply no change in some range, but your measurement is showing it as significantly lower, then this could indicate a difference between your RS meter mic and the Audyssey mic.

Are you calibrating the REW mic process to the REW sub tone or the mains? I've found more accurate results using the main speaker tone for calibration, then leaving everything at this calibration while measuring the mains and subs. This is partly a matter of perception, whether the graph suggests the main is low or the sub is high. But I have found the RS meter shows a more stable level when measuring the main test tones. Pink noise in the sub range shows noticeable variation, and the RS meter does not have a Very Slow setting as REW does.

Bill

P.S. As Greg wrote, a little experimentation can go a long way. I would add, a lot of experimentation can go even farther.

 

[tonyvdb]

Just out of curiosity where do you have your crossover set at?

 

[laser188139]

Frank, when I was investigating this problem, REW with the RS meter was showing me a very high peak at the low end, even louder than what I was hearing. So I decided once to try using the RS meter, but instead of using the generic RS calibration file, I tried it with no mic calibration file but with the C weighted SPL meter box checked.

In this graph, the red line is the measurement using the generic RS meter calibration file, and the green curve is identical system measured as a C weighted SPL meter with no calibration file. Trusting that Audyssey was doing a reasonable job, I concluded that my RS meter differed from the generic file, giving me a misleading curve. (This was borne out when I later purchased a calibrated microphone.)

As your two curves look similar, with a definite upward slope toward the low end, you might be seeing something similar just not as extreme as I did. You could try taking the pair of measures again, configured as a C weighted SPL meter with no calibration file. The results will probably look different, and reality might be somewhere in between.

Good luck,
Bill

 

[Tufelhundin]

I've got my low frequency area pretty good. I'm using the BFD and Audyssey afterwards. However, above 100Hz the level is down 8-10dB. Audyssey is setting the trims in the receiver so I don't understand why it would be down. Any ideas or solutions to fix this? The crossover is set to 80Hz. The graph is 1/3 octave smoothed.

View attachment 20404

Intresting, your graph is not too different than mine "see Audyssey thread in AVR/AMP Section." It would seem Audyssey is boosting around that freq. I did use REW with different crossover settings from 120hz down to 40hz...it gets really bad lower than 80hz and 120hz seemed to help with that huge climb aroun 100-120 hz.



Blue crossed over at 120hz / green at 80hz crossover

 

[ToBeFrank]

In terms of setting the level, Audyssey uses a band limited calculation. I believe Chris has written that it is 500Hz-2kHz, or something like that.

Yep, found Chris's post on this by googling for "Audyssey band limited level".

Sometimes Chris has written that the first measurement determines level and distance, but after a few of these he has corrected himself and written that only distance is calculated based only on the first.

Got a reply from Chris and he said the distance and levels are set from the first measurement position.

The displayed levels around 1kHz-2kHz are surprising, given what I understand about the levels. But as you say, this may be related to the peaks below that. You could compare Audyssey Off and On measures on each front separately. It could be that Audyssey is attacking even larger peaks before equalization, and the side effect of pushing the peaks down is that the levels above that are lowered, too. If your AVR offers preouts for the fronts, you can measure these directly (again without a mic file) and compare Audyssey Off to Audyssey On to see directly where it is applying its changes. This can be particularly interesting, in that if Audyssey is apply no change in some range, but your measurement is showing it as significantly lower, then this could indicate a difference between your RS meter mic and the Audyssey mic.

Great idea and I should have thought of that. Attached are the Audyssey filters. After seeing them I'm leaning towards the RS being inaccurate. So I think I'm going to get a calibrated ECM8000. My next question then is can I use a cheap preamp or mixer with phantom power and get good results? Or will I end up making it inaccurate again? Any suggestions on what to use with the ECM8000 that doesn't cost a lot?

Are you calibrating the REW mic process to the REW sub tone or the mains? I've found more accurate results using the main speaker tone for calibration, then leaving everything at this calibration while measuring the mains and subs. This is partly a matter of perception, whether the graph suggests the main is low or the sub is high. But I have found the RS meter shows a more stable level when measuring the main test tones. Pink noise in the sub range shows noticeable variation, and the RS meter does not have a Very Slow setting as REW does.

I used the sub tone. I'll give it a shot with the mains tone.

Just out of curiosity where do you have your crossover set at?

80Hz. I think I've got the crossover region nailed as I can move the crossover through the 80-120Hz range and it makes virtually no difference on the frequency response.

Left channel:

Right channel:

 

[laser188139]

... Great idea and I should have thought of that. Attached are the Audyssey filters. ...

Those curves are interesting, in that the mains are much flatter, more even, than my preout curves. That probably reflects your room treatments -- I don't have any -- so it gives me hope for what I might achieve when I take that step. The curves also make obvious the 2kHz "BBC" dip in the Audyssey reference curve; this should disappear if you take the same measurement with Audyssey off.

... So I think I'm going to get a calibrated ECM8000. My next question then is can I use a cheap preamp or mixer with phantom power and get good results? Or will I end up making it inaccurate again? Any suggestions on what to use with the ECM8000 that doesn't cost a lot? ...

I bought one of the older Xenyx 502, then found I needed the Xenyx 802 to have phantom power. As you can see in the description of REW connections, the newer Xenyx 502s have phantom power. So you can opt for a new 502, or a used 802 off ebay.

 

[ToBeFrank]

Those curves are interesting, in that the mains are much flatter, more even, than my preout curves. That probably reflects your room treatments -- I don't have any -- so it gives me hope for what I might achieve when I take that step. The curves also make obvious the 2kHz "BBC" dip in the Audyssey reference curve; this should disappear if you take the same measurement with Audyssey off.

Sorry, I didn't annotate the graphs well enough. The flat curves in the graphs are the measurements with Audyssey off (not the target curves) so you can see what the Audyssey filters are doing.

I bought one of the older Xenyx 502, then found I needed the Xenyx 802 to have phantom power. As you can see in the description of REW connections, the newer Xenyx 502s have phantom power. So you can opt for a new 502, or a used 802 off ebay.

I was just doing some searching and saw the 802. I think I'll just go ahead with that one since it's only $15 more than the 502. I'm going to get the ECM8000 from Cross Spectrum. Then I think I'll be set for all the measurements I can handle.

 

[ToBeFrank]

That probably reflects your room treatments -- I don't have any -- so it gives me hope for what I might achieve when I take that step.

BTW, you'll really love your sound when you add treatments. My motivation for adding them was dialogue was hard to understand during movies. Now the dialogue is nice and clear and the bass is punchy. To be honest I probably don't even need to do any more measuring and adjusting as I think my sound is spectacular, but being an engineer I'm a bit of a perfectionist. :nerd:

 

[tonyvdb]

80Hz. I think I've got the crossover region nailed as I can move the crossover through the 80-120Hz range and it makes virtually no difference on the frequency response.

Ok, Have you tried turning the phase knob taking readings at every adjustment? That dip at the 100Hz area cold very well be your sub reacting with your mains at the crossover point as the crossover is not a brick wall and will still produce sound at above 100Hz just not at a very loud level.

 

[ToBeFrank]

Ok, Have you tried turning the phase knob taking readings at every adjustment?

Yes. I did the left and right separately and used the phase setting that was the best compromise between the two.

That dip at the 100Hz area cold very well be your sub reacting with your mains at the crossover point as the crossover is not a brick wall and will still produce sound at above 100Hz just not at a very loud level.

Are you talking about the dip at 150Hz? I have a lengthwise axial mode that is causing that. I'm still working on it as my initial attempt at treatments for it didn't work.

 

[ToBeFrank]

Well, I'm glad I got the calibrated ECM8000. It appears my RS SPL meter isn't very accurate and explains why I was seeing the 8-10dB drop after 100Hz... I don't actually have a drop! Here is the comparison: