AVR Frequency Response Graphs

[MACCA350]

This started from this thread.

I Thought we'd start a thread to compare the Frequency Response of different receivers/pre-amps.

Method: Basically its taking the Mic out of the loop and connecting the preamp outputs of your receiver to the inputs of your soundcard.

There are 2 ways we need to connect so we can get the measurements we are after.
1) RED line = this is for the full spectrum sweep and Mains when crossover is engaged
2) GREEN line = this is for testing what is actually getting to the sub when crossover is engaged

Note: do not connect both at the same time

Thanks to Brucek for the graph


There are 3 measurements that we can look at
1) Full spectrum sweep 2Hz -->20kHz]
2) Mains with the crossover engaged
3) Subwoofer with crossover engaged

A couple of things I did to keep things simple.
On the receiver
1) Set volume trims to 0.0
2) Set main volume to 0.0
3) Do not adjust volumes between readings(this will keep a reference for comparing)

On REW
1) Under 'Settings' Make sure No Mic calibration file is loaded
2) Under 'Settings' create a soundcard calibration file (include all cables you will use in the loop) load the file and run a sweep and make sure you get a dead flat line.
3) Under 'Measure', set measuring level(dB FS) to -10
4) Under 'Measure', set the End Freq(Hz) to 24,000
5) After the full spectrum sweep, set the 'Trace Offset'(under 'Trace Adjustments) so that the line around 1kHz is as close as possable to 0db on the graph(I had to use -84.8) This value will need to be input to the other measurements, do not change this number even if the other graphs don't line up with 0db(remember we measured at the same level so these differences are due to the crossover being in chain of processing) You can click the 'Add offset to Data' button if you like.
6) Under 'Graph Limits', set to:
Top = 12
Bottom = -12
Left = 2
Right = 20,000
7) After measurements and adjustments are taken click on the 'All Measured' tab and your ready to post

When posting add the Receiver/Pre-amp model information.
You can zoom into the graph to find -1db and -3db points.
You should end up with something like my next post.

Any suggestions are welcome,
Cheers

[MACCA350]

Denon AVR-3805

Full (Red)
-1db at 4.5Hz
-3db at 2.2Hz

Sub channel(blue)
-1db at 6.9Hz
-3db at 4.4Hz

cheers

[JohnM]

The slight ripple you see at the low end is caused by a narrower impulse response window than is needed. If you look at the Impulse Response graph you'll see a section of the response that is not being included, so extend the window duration to cover all the area above the noise floor and the low end should show a smoother curve. If you post a plot of the impulse response (with the Y axis set to dB) I can indicate exactly what should be included.

[Otto]

Hi there,

I would expect the sub and main signals to intersect at -6 dB from their nominal levels, but it appears that they intersect more around 4 or 5 dB down (yes, I see that their nominal levels are above 0 dB slightly).

If I look at where each signal actually does cross at -6 dB, it appears that the sub's -6 dB point is about 80 Hz, while the main signal's -6 dB point is about 60 Hz.

Is it possible that there are two different crossovers set up? Perhaps something else going on?

[brucek]

Is it possible that there are two different crossovers set up? Perhaps something else going on?

Of course. Most systems use a 2nd order HPF (x5) and a 4th order LPF. This then accounts for the 2nd order roll off of the mains to result in an effective 4th order.

Any suggestions are welcome

My feeling is that most receivers/processors will be perfect in their crossover execution and will have a flat line out to 20KHz without fail (your green line). The differences we're most interested in would be the low end response effect on our subwoofers. This is where you would find differences between products. I would rather see the graphs from 2Hz to 200Hz (or at least that in addition to what you're requesting), to better examine the drop below 20Hz. Certainly all the crazies aficionados of bass down to 5Hz would be extremely interested in the constraint their equipment is placing on the final sound.

sp2 sub with adc dac.jpg

brucek

[MACCA350]

brucek wrote:

My feeling is that most receivers/processors will be perfect in their crossover execution and will have a flat line out to 20KHz without fail (your green line). The differences we're most interested in would be the low end response effect on our subwoofers. This is where you would find differences between products. I would rather see the graphs from 2Hz to 200Hz (or at least that in addition to what you're requesting), to better examine the drop below 20Hz. Certainly all the crazies aficionados of bass down to 5Hz would be extremely interested in the constraint their equipment is placing on the final sound.

brucek

I agree that the majority of differences will be below 20Hz, but I would like to keep the full sweep also because you never know, we may just find some manufactures altering their response to have a "signature sound". That would be interesting

The other question is, do we adjust the second sub only graph so that the highest point is at the 0db line, even though it read higher under identical conditions(I'm leaning towards yes). I'm really wondering why the inclusion of the crossover increased the line by about 0.5db

The other thing that was interesting was the noisefloor as shown in the impulse response graphs. Both Full and Mains was -105db, but Sub was -60. I think this was due to signal being passed through above the crossover, I've zoomed out the graph, see here:


cheers

[brucek]

we may just find some manufactures altering their response to have a "signature sound".

doubtful........

The other question is, do we adjust the second sub only graph so that the highest point is at the 0db line

I would say no. You'll see in my graph that I am below 0dB since the crossover at 60Hz never hits 0dB before the response starts to drag the line back down. The proof is in the fact that at 60Hz it is exactly at its proper -6dB target.... if I had used 80hz cross, then it would have hit zero....

brucek

[MACCA350]

JohnM wrote:

The slight ripple you see at the low end is caused by a narrower impulse response window than is needed. If you look at the Impulse Response graph you'll see a section of the response that is not being included, so extend the window duration to cover all the area above the noise floor and the low end should show a smoother curve. If you post a plot of the impulse response (with the Y axis set to dB) I can indicate exactly what should be included.

I'm not sure I understand how to make those changes. Here are the Impulse Response graphs.

Full


Main


Sub


cheers

[MACCA350]

brucek wrote:

doubtful........


I would say no. You'll see in my graph that I am below 0dB since the crossover at 60Hz never hits 0dB before the response starts to drag the line back down. The proof is in the fact that at 60Hz it is exactly at its proper -6dB target.... if I had used 80hz cross, then it would have hit zero....

brucek

This doesn't explain my graph, since mine reached over 0db and the -6db line crosses at about 84Hz instead of the 80Hz crossover as set in the AVR. adjusting to 0db brings the -6db point to 80.9Hz. Any ideas why there was an overall increase when the crossover was engaged?

I'll hookup my Denon 1603 in the morning and see what that does.

cheers

[JohnM]

Macca,

Click the IR Windows button in the REW toolbar then adjust the Right Window value until the dotted line is out where the impulse response dispappears into the noise level. For the blue trace that is about 650ms for the response you captured (but see the noise floor comment that follows), for the red trace about 1200ms. After you adjust the value, hit the Apply Windows button to apply it. You will then find the responses roll off more smoothly at low frequencies.

The noise floor is high for the sub (blue) trace because you did a full range sweep but there is no useful output above a few hundred Hz, so the captured signal just has noise for much of the swept range. Do a sweep to 500Hz or so for the sub and you will see a much lower noise floor.

[MACCA350]

JohnM wrote:

Macca,

Click the IR Windows button in the REW toolbar then adjust the Right Window value until the dotted line is out where the impulse response dispappears into the noise level. For the blue trace that is about 650ms for the response you captured (but see the noise floor comment that follows), for the red trace about 1200ms. After you adjust the value, hit the Apply Windows button to apply it. You will then find the responses roll off more smoothly at low frequencies.

The noise floor is high for the sub (blue) trace because you did a full range sweep but there is no useful output above a few hundred Hz, so the captured signal just has noise for much of the swept range. Do a sweep to 500Hz or so for the sub and you will see a much lower noise floor.

Understood, and adjusted, thanks. I also noticed that increasing the window lowered the -1db and -3db points on the graph for the Full range plot. I'll remember to cut the sweep for the Sub plot next time, I may redo the above Sub plot in this way and see if it makes any difference to the -1db and -3db points.

Do you opinion on the issue of the different trace levels as stated above. I'll add a second graph for the Sub plot, but I'm still not sure whether to set the highest point to 0db for reference purposes, because this will affect the -1db and -3db points(I think these points should be compared to the highest point on the plot, but then again )

I'm about to run some tests for the Denon AVR-1603, so those graphs should be up soon.

cheers

[MACCA350]

Ok I've got an interesting one here.
This is the Denon AVR-1603 with 80Hz crossover

Red = V.AUX input and CDR output
Purple = EXT. input and SUB output
Green = V.AUX input and SUB output with speakers set to LARGE(LFE+MAIN on/pff didn't make any difference) set to both STEREO/DIRECT(this trace is also the same with speakers set to SMALL but with prologic cinema or DIRECT enguaged)
Blue = V.AUX input and SUB output with speakers set to SMALL set to STEREO

Something really screwy going on there I even reset the microprocessor but it didn't make any difference.

Any ideas?

cheers

[MakeFlat]

MACCA350 wrote:

Blue = V.AUX input and SUB output with speakers set to SMALL set to both STEREO/DIRECT

Something really screwy going on there I even reset the microprocessor but it didn't make any difference.

Any ideas?

cheers

I'll comment on the blue curve. Your result is consistent with my REW runs for the Stereo Mode on the Denon 1604. There is one difference: In the Direct Mode, the SUB response appears to roll off at 8Hz on my 1604's SUB output. So I guess Denon changed the design. Why they decided to roll off Stereo at 40Hz on the Denon 160x, it stumps me

Also, on my REW runs, in Stereo Mode, the mains and sub appear to be quite a bit lower in level when compared to all other modes: PL2, 5 Ch, Direct. The level is consistent with what you find on the sub's level.

[MACCA350]

MakeFlat wrote:

I'll comment on the blue curve. Your result is consistent with my REW runs for the Stereo Mode on the Denon 1604. There is one difference: In the Direct Mode, the SUB response appears to roll off at 8Hz on my 1604's SUB output. So I guess Denon changed the design. Why they decided to roll off Stereo at 40Hz on the Denon 160x, it stumps me

Also, on my REW runs, in Stereo Mode, the mains and sub appear to be quite a bit lower in level when compared to all other modes: PL2, 5 Ch, Direct. The level is consistent with what you find on the sub's level.

You're right. I just redid the DIRECT and STEREO measurements with speakers set to SMALL, and the DIRECT line is identical to the green plot(I've changed the previous post to add these changes).

So it seems that with speakers set to SMALL and in STEREO mode the SUB channel is seriously attenuated, yet setting the speakers to LARGE gives you full output on the SUB channel(and the LFE+MAIN feature makes no difference), to me this is backwards.

It seems to be only the STEREO mode when speakers are set to SMALL that cuts the SUB channel output. I ran a sweep with the crossover set to 120Hz and comparing both it looks like there is a subsonic filter engaged in this mode around 40Hz or so(as you suggested)

But the question is why, If you tell the receiver your mains are small and you have a sub, why is it cutting off the sub channel like this ONLY in stereo mode, really doesn't make sense

cheers

[Otto]

Hi MACCA,

I will agree that that seems strange in both cases. The way I think it should work is that there wouldn't be a difference between the LPF crossover between small and large -- the small and large settings should only affect the mains. I guess I should that that this is true as long as you have the "LFE+Main" feature set. But you already know that, which is why you're complaining about the seriously attenuated sub output in the "Small & Stereo" case above.

Unfortunately, I don't have any reasoning for its behavior, but I can offer a few other comments.

While I tend to agree with brucek that we are frequently concerned with the subwoofer band, I do believe there's some value in keeping the mains sweep in there. I would imagine that most of the time it will be correct. However, as you may have seen in my thread here, I was able to detect a problem with the HPF of my Outlaw 990 (I notified Outlaw, and after a bunch of discussion, they have indeed acknowledged that this is likely a real problem). I only measured the listening position response, but you can clearly see a HPF engaged on the mains when it should not be (in DIRECT mode, of all things!?!?!). So, I'd like to see the mains signal in there in addition to the others. I think that the upper frequency of 200 Hz is good.

Here's another thing. I don't know how many receivers you have access to, or how much trouble it might be, but the Outlaw discrepancy taught me something: the behavior of the different "modes" may be different, even though one would expect it to be the same. I don't think you started this thread/project as an effort to find bugs in these products, but as long as you have one "on the bench" you might consider scrolling through as many modes as possible. I didn't go through all the modes on the 990, but I did hit all the 2-channel modes. Anyway, just a suggestion.

Wow! All this talk of modes reminds me of "Direct Mode" a.k.a. "Bypass Mode". I tend to think of this mode as applying zero change to the original signal, so you might see some different behavior here. It was "Bypass Mode" in which I found the Outlaw was applying a rolloff to the mains when set to Small (though it had nothing to do with the actual crossover frequency selected). Do you know how the Denon is supposed to behave? I would somewhat expect that there would be no LPF on the sub out in Direct/Bypass mode. I do believe the implementation of "Direct Mode" could vary from manufacturer to manufacturer.

So, MACCA, I just want to say that this is an awesome thread -- very educational. Keep up the good work! I will be working to post this information for my Outlaw 990.

Thanks!!!