DIY Subwoofer newbie

This weekend i finally had the time to make a more thorough calibration.

Mains + Sub (Pre-EQ)
SUB_MAINS_zpsbwlwg8ql.jpg



REW Filters Calculation ...


Sub PEQ
filters_sub_zpsvmau6bfm.jpg



PEQ Filters
Filters_Sub_peq_zpskwunhfds.png





LEFT PEQ
Filters_left_zpsv0pj1kt7.jpg



Left Filters
Filters_left_peq_zpspys7jgcm.png



Right PEQ
filters_right_zpsfgb9lsx0.jpg



Right Filters
Filters_right_peq_zpsnbmzo3ea.png



After applying the filters ...


Sub (140Hz XO) + Left vs Left (un-eq)
Sub_Left_zpsa2eim8uu.jpg


Sub (140Hz XO) + Right vs Right (un-eq)
Sub_right_zpshkbzrvua.jpg



The Gain on the American Audio VLP 2500 (CV 3200 "clone-if-existed") is set on -3.

Any comments, improvements i could make, etc?

 
It's generally recommended to use var smoothing for peq albeit it could be argued it smooths too much on the top end.

I think you're results look pretty good. The jump at around 200hz looks quite sudden on both channels. Has the sub channel got a 120hz lpf or is that full range? The target line for the sub is with a lpf which you obviously don't track higher up. You could probably do with more filters than three which will help the 100-200 range. Mini Dsp 2x4 might improve things a bit more.

How does it sound?
 
85dB range on the y axis is quite large, 60 is more common.

FDW is best for looking at full range measurement, you want to eq the direct signal after all. 1/6 smoothing on a long window will produce suboptimal results. Use no smoothing for subwoofers.

Posting the mdat somewhere makes it easier to comment tbh.
 
What is the Gain level you are using on the CV5000?
Mine is set to (I think...) the -15 detent on the attenuation knobs (so about 10 o'clock), and -13dB on the AVR subwoofer trim, but that may change when I finally pull my finger out and do some EQing with the miniDSP that is still sitting in the box... lol

TBH it might need to be -17 but we had friends' kids round and I think they've fiddled, so I really need to re-run Audyssey to check!
 
Mine is set to (I think...) the -15 detent on the attenuation knobs (so about 10 o'clock), and -13dB on the AVR subwoofer trim, but that may change when I finally pull my finger out and do some EQing with the miniDSP that is still sitting in the box... lol

TBH it might need to be -17 but we had friends' kids round and I think they've fiddled, so I really need to re-run Audyssey to check!

9 oclock on the cv5000 should be about right with between -10 to -5db on your avr trim! Even if your response curve was perfectly flat from 10hz through to the xover point and the cv5000 at say 11oclock!! It would start to sound boomy again.
 
The jump at around 200hz looks quite sudden on both channels. Has the sub channel got a 120hz lpf or is that full range? ... You could probably do with more filters than three which will help the 100-200 range. Mini Dsp 2x4 might improve things a bit more.

How does it sound?

Yes, i'm not entirely understanding the "jump" around 200Hz.

The sub channel is at full range (250Hz on UMC-200).

Yes, due to the enormous output of the sub on the 100-200 region, it would help having more filters, moreover due to the limitation of the UMC-200 of -15 value to cut.

I'm waiting to check the new offerings form miniDSP Dirac, and also to check how the new plugins will be applied to the nanoAVR DL. If they allow that everything is correctly set (bass management, delays, levels, etc) using the nanoAVR DL, i might consider it in the future.

It sounds WONDERFULL :)

The dryness and the impact of the bass form this sub is amazing.
Yesterday i watched some bass heavy scenes like the Sonic canon on Incredible Hulk, and WOW, it is everything i hoped that it would be. I cant even imagine what it will be when i add the second subwoofer :)
 
Use no smoothing for subwoofers.
Posting the mdat somewhere makes it easier to comment tbh.

You are correct, and i usually don't use smoothing, but to be honest i forgot ...

Please download the mdat file here.

I would be very grateful if you had the time to advise me on the best way to correct the response, so please accept my thanks in advance :)
 
Mine is set to (I think...) the -15 detent on the attenuation knobs (so about 10 o'clock), and -13dB on the AVR subwoofer trim ...

Initially i leveled all the speakers and sub, using the method described by @Markus here.
This procedure uses REW RTA to match the subwoofer 63Hz response to the other speakers 1kHz response:
rta-periodic-pn.png


Using this procedure, the amplifier gain was in the -10db setting.

But, this level did not satisfied in regular listening sessions, so i upped to the -6 setting.

After setting the PEQ on the UMC-200, in order to obtain a type of House Curve between the Low and the High frequencies, i upped the amplifier level to -3 (Processor level is at the 0 setting).

I'm not entirely sure if there is any inconvenient of having the setting as high (at least comparing to the -10 original value).
 
I'm not entirely sure if there is any inconvenient of having the setting as high (at least comparing to the -10 original value).
bear in mind that you shouldn't set gain according to volume, you set gain to deliver a clean signal through your signal chain. The point where volume comes into play is that your listening habits (volume) dictate the peak signal that your signal chain needs to pass cleanly.
 
^^
I'm not sure i've understood you correctly :)

Based in what you've said, what to determine the gain value to use?

And, once selected a gain value, if i need to increase the subwoofer "gain", should i use the Processor Gain settings?
 
9 oclock on the cv5000 should be about right with between -10 to -5db on your avr trim! Even if your response curve was perfectly flat from 10hz through to the xover point and the cv5000 at say 11oclock!! It would start to sound boomy again.
I think we're both talking around the same point - in that the AVR and sub amp trim need to balance to produce the correct levels.

What I've done is attempt to maximise the gain at the amp end and then minimise the gain at the AVR end, which if I understand correctly, means that there is less chance of clipping being sent to the amp, and therefore a cleaner signal through the chain.

I could turn the amp attenuation up (i.e. turn the dial down to -20 or whatever) and then increase the sub trim in the AVR from -13 to -10, to match your suggestion, but I think my way is the 'better' method??

@mattkhan needs to confirm my (mis)understanding :laugh: lol
 
Based in what you've said, what to determine the gain value to use?
An example to illustrate

Let's say you have a processor that is rated to produce 1Vrms at "rated output" (means a single channel playing a 0dBFS signal). You send such a signal down the lfe channel and your sub amp sees 1Vrms at its input.

You then low pass a single main channel and add that to the LFE to produce a mono SW feed. You play the same 0dBFS content down both channels at the same time. LFE is 10dB stronger than a main channel so you just increased the output by ~2.4dB which means your sub amp now sees a peak level of ~1.32Vrms.

Add 4 more main channels for 5.1 and you just added another ~5.8dB so your sub amp now sees a peak level of about 2.58Vrms.

Now let's say your processor can't do this (sum channels) without you creating headroom by reducing the levels (trims or MV) by -10. Your peak *clean* output is now down to 0.816Vrms.

Now let's say your sub amp has an input sensitivity of 1.5Vrms. If you actually need that to drive your sub to the output you want then you have a problem with respect to delivering sufficient clean signal. OTOH say you had an input sensitivity of 0.7Vrms then you need to trim a bit more somewhere to avoid clipping.

I don't know what the numbers are for your signal chain but hopefully this illustrates what gain structure is and why it is influenced by target volume levels but should not be used to control the volume level directly. Put another way, gain structure is about making room for your volume control to operate.
 
Now let's say your sub amp has an input sensitivity of 1.5Vrms. If you actually need that to drive your sub to the output you want then you have a problem with respect to delivering sufficient clean signal. OTOH say you had an input sensitivity of 0.7Vrms then you need to trim a bit more somewhere to avoid clipping.
CV5000 Input Sensitivity 8 ohms 1.42V (+5.3dB) (Cerwin Vega CV5000 Power Amplifier)

This post on AVS suggests up to 4.6Vrms from the 818 pre-outs:
Onkyo 818 Pre-out Connection Rating - AVS Forum | Home Theater Discussions And Reviews
but I'm not sure if that applies to the SW output.

Either way, mine seems fine so far... lol
 
This post on AVS suggests up to 4.6Vrms from the 818 pre-outs:
Onkyo 818 Pre-out Connection Rating - AVS Forum | Home Theater Discussions And Reviews
but I'm not sure if that applies to the SW output

Either way, mine seems fine so far... lol
fwiw I don't think there is any point relying on random interweb data for this, you can either measure your own system and prove to yourself it is clean or just go with a sensible setup that works and sounds good to your ears.
 
You are correct, and i usually don't use smoothing, but to be honest i forgot ...

Please download the mdat file here.

I would be very grateful if you had the time to advise me on the best way to correct the response, so please accept my thanks in advance :)
some of the data looks a bit odd, what is your measurement setup?

for example, LSUR indicates the signal starts and then there is a v loud spike 8ms later.

upload_2016-1-20_22-9-25.png


LH is even funkier with 20ms of signal before the peak level recorded

the front 3 look more normal in that the initial peak is at 0ms rather than later in the signal

the sub measurement is also seems rather noisy as judged by the resulting FR, it's v jagged

upload_2016-1-20_22-16-13.png


I would look into your measurement signal chain before going further.
 
one other question/comment, I'm not sure what the data is. The sub measurement is much louder than the main channel measurement but the sub xo 140 is at a similar level as the main, not sure what is going on there.

actually make that 2, how are you intending to set this up? are you manually applying a crossover and then purely manual PEQ? what is dealing with time alignment?
 
some of the data looks a bit odd, what is your measurement setup?

I'm using a Laptop, connected by HDMI to the UMC-200, and a miniDsp Umik-1.
Subwoofer is powered by an American Audio VLP 2500 (CV 3200 clone), and the measure was done with a -6db gain. The subwoofer is connected to the UMC-200 by a XLR cable.

Before addressing each questions, just let me mention that i had several problems when measuring. Sometimes when changing the speaker on REW, without changing anything else, from one measure to another the levels were 15db higher.

for example, LSUR indicates the signal starts and then there is a v loud spike 8ms later.

This speakers is placed in the roof, here:
ueuRclR.png



LH is even funkier with 20ms of signal before the peak level recorded

LH is place here:
lIiAE4R.png


the front 3 look more normal in that the initial peak is at 0ms rather than later in the signal

the sub measurement is also seems rather noisy as judged by the resulting FR, it's v jagged

What do you mean by noisy?
 
The sub measurement is much louder than the main channel measurement but the sub xo 140 is at a similar level as the main, not sure what is going on there.
Yes, although the Sub is aligned with the speakers (using the method described above), the sub measure was much louder. I really don't know how to explain this, but this is exactly what i mentioned above, for some reason, sometimes, there was an enormous level different between two measures without changing anything.

For example, the Sub+Left and the Sub+Right measures show that the Sub is aligned with both speakers.

are you manually applying a crossover and then purely manual PEQ? what is dealing with time alignment?

The individual measures were taken with the speakers on full range, and the sub at 250Hz.
Levels and delays were also at 0.

After the measures, the filters were calculated, and applied using the UMC-200 PEQ.
Then, i set the levels, delays, speakers were set to Small and applied a 140Hz XO.

Then, i took the Sub+Left and Sub+right measures, tune the PEQ, re-measure, tune again, until i reached a flatter response.
 
Before addressing each questions, just let me mention that i had several problems when measuring. Sometimes when changing the speaker on REW, without changing anything else, from one measure to another the levels were 15db higher.
I think you need to get to a position where your measurements are completely repeatable. To that end....

I'm using a Laptop, connected by HDMI to the UMC-200, and a miniDsp Umik-1.
I assume this means you're using Asio4All? Do you have the ability to make an analogue connection from your laptop to some aux input on the UMC-200? I think you probably need to rule out that signal chain as the source of your problems

other things to verify;

- no enhancements or anything applied to each windows audio device
- confirm all windows audio devices, in each direction, are configured at the same sample rate
- check levels for each windows audio device to ensure they're not really low or really high
- confirm that measurements in REW are within the expected bounds (i.e. input levels in the -15 to -30 range should be fine)

you could also try a different piece of software (like holm) to check whether it is a rew specific issue or not.

This speakers is placed in the roof, here:
the only plausible reason I can think of (for the measured output before 0) is that your signal chain is resulting in output on multiple channels, have you verified this isn't the case? e.g. play a sine wave using the generator and do a walk by of each speaker to confirm whether there is output or not.

What do you mean by noisy?
the resulting FR is v jittery and the IR is v spiky, this could be indicative of some sort of noise in the output that your mic is picking up. It could also be a problem with the mic I suppose.
 
I assume this means you're using Asio4All? Do you have the ability to make an analogue connection from your laptop to some aux input on the UMC-200? I think you probably need to rule out that signal chain as the source of your problems

I have an older laptop that has an optic output, could try it ...

- no enhancements or anything applied to each windows audio device
- confirm all windows audio devices, in each direction, are configured at the same sample rate
- check levels for each windows audio device to ensure they're not really low or really high
- confirm that measurements in REW are within the expected bounds (i.e. input levels in the -15 to -30 range should be fine)

The only thing that occurs to me is that when measuring the subwoofer, this measure produced clipping. I tried to lower the levels on both the sub amplifier and in the UMC volume, but it always clipped.


you could also try a different piece of software (like holm) to check whether it is a rew specific issue or not.
I doubt i will have the time for learning it, but if i cannot figure what is the problem with REW, i might have to try it.

the only plausible reason I can think of (for the measured output before 0) is that your signal chain is resulting in output on multiple channels, have you verified this isn't the case? e.g. play a sine wave using the generator and do a walk by of each speaker to confirm whether there is output or not.
It makes sense, and it could have occurred.
I will check it.

the resulting FR is v jittery and the IR is v spiky, this could be indicative of some sort of noise in the output that your mic is picking up. It could also be a problem with the mic I suppose.
Probably its because it was clipping, although i dont understand why it was clipping :(

And thank you for your help.:thumbsup:
 
The only thing that occurs to me is that when measuring the subwoofer, this measure produced clipping. I tried to lower the levels on both the sub amplifier and in the UMC volume, but it always clipped.
what was clipping exactly?
 
The REW measurement.
After measuring the Sub, REW produced an alert message saying something like the levels were too high and this could produce an incorrect reading or something like that (cant remember exactly).
 
After measuring the Sub, REW produced an alert message saying something like the levels were too high and this could produce an incorrect reading or something like that (cant remember exactly).
REW has 3 meters, one for output (which is driven by the sweep level) and one for each input. You need to confirm which one is clipping, I would think it's the input one. This means you need to reduce the mic input level til it is not clipping. Once you have this sorted out, you should get better measurements for sure.
 

The latest video from AVForums

TV Buying Guide - Which TV Is Best For You?
Subscribe to our YouTube channel
Back
Top Bottom