Rotel X-Over Test

[maddog]

I have a Rotel RSX-1055 receiver with the latest firmware. I wanted to find out for certain how the crossover worked on the digital connections and on the multi-channel inputs. So I got the Stereophile Test CD 2 which has a bass sweep.

The test was done in 2-ch stereo using just my floorstanding mains. The subwoofer was set to Yes in Speaker Setup. However, my subwoofer's on/off switch was turned off. The volume value was set the same for each test. I did not try to match up dB's at a given frequency (e.g., 200 Hz) between tests. I used C weighting and Fast response on my SPL meter. The multi-channel inputs are supposed to pass the L & R signals "as-is" to the mains. The Large/Small front-speaker settings and Sub X-Over setting are not supposed to have any effect on multi-channel inputs. The bass < 100 Hz is also supposed to be copied to the subwoofer, which yields bass doubling on that frequency range with the L & R mains. As I'm sure everyone expected, my tests verified this. I got the same frequency response from the mains regardless of the Large/Small and X-Over settings when using the multi-channel input. Here are the full results:

Freq (Hz) Multi-Ch_In (dB) Dig_In_80_X-Over (dB) Dig_In_120_X-Over (dB)
---------------------------------------------------------------------
200_______68_____________67_________________67
160_______71_____________70_________________70
125_______74_____________73_________________72

100_______75_____________74_________________71

80________74_____________71_________________67
63________71_____________66_________________61
50________65_____________60_________________55
40________57_____________52_________________51
31.5_______60_____________52_________________50

Once again, although the subwoofer was set to Yes in Speaker Setup, it's switch was off. So, it is not contributing to the output. The output is due solely to the two mains.

The multi-channel inputs are the full-range response. I guess the roll-off there is due to my speakers and room acoustics.

But, what I was surprise at is how much of the signal the mains are playing below the crossover on the digital input. Comparing the 80 Hz crossover to full-range, there is just about a 5 dB difference at each frequency from 80 Hz to 40 Hz. I know that it takes something like double the power to get a 3 dB gain. So, maybe 5 dB is significant. But, it's still just 5 dB and just a very small turn on the volume knob.

I know this is a long and complex post (especially since you will not be getting any college credit for it), but here is what I would like to know:

1. Has anyone done a test like this on their Rotel? If so, did you get results like this?
2. Is this type of slow roll-off from the crossover the "norm" with A/V receivers and pre/pros?
3. Has anyone done a test like this using another brand of A/V receiver or pre/pro? If so, did you get a quicker roll-off or about the same as I got with the Rotel?

Thanks in advance for your responses.

- maddog

[Kevin D]

Yes, pretty much all crossovers have roll-off. They are not brickwalls. They usually come in 12, 18, 24, or 36 db per octave slopes. I'm pretty sure Rotel uses 24db or 12db slopes on theirs. That's why crossover frequencies aren't as easy as it first seems. The newer models have separate crossovers for the speakes and the subs and sometimes the 'correct' setting has a gap, ie: sub set to 40-, mains set to 80+

If you are using the Radio shack meter, keep in mind it is not very accurate at all at lower frequencies.. So using the same meter for your tests will give you trends, but not accurate values.

An octave difference is either half the frequeny or double the frequency. So at 80hz, you were 71db at 80hz, and 52db at 40hz. A difference of 19db. On 100hz, you were 71db at 100hz. At 50hz you were at 55db.. 16db difference..

So either room accustics, speaker design, and/or the meter is having an effect, or maybe Rotel using 18db per octave slopes..

Kevin D.

[maddog]

Thanks for the response, Kevin.

When you look at the full-range response (Multi-Ch_In) vs. the 80 Hz crossover response, it's really the difference between setting the speakers to Large vs. Small if you also use a sub with the Large setting. In both cases, there is bass doubling, but just to a lower degree with the Small setting. I guess that's just how crossovers work, as you explained in your post. But, in this particular case, there is only about a 4 or 5 dB difference from the mains at each frequency <= 80 Hz (e.g., 71 dB vs. 66 dB @ 63 Hz and 65 dB vs. 60 dB @ 50 Hz). Do you consider that to be a big difference with respect to listening and with respect to the demands on the amplifier driving the mains? From these results, it almost looks like there really is not a huge difference between Large vs. Small when using a sub with an 80 Hz crossover in both setups in my particular system.

[Kevin D]

Yeah, it's really about the amplifiers being more efficient. Bass is the hardest for amplifiers to drive and when you knock 5-6db off from what it has to do, you will get more power available to the rest of the frequencies.

In additional, lower frequencies have to be louder to sound as loud as everything else. Normally called a 'house curve' when setting up EQ's, you want a downward slope from 20hz to 20khz. So with that, louder at lower isn't bad. You just don't want middle bass frequencies to have a spike. Unless your mains are really big, they will slope down inheritantly. So let's say you have a high-pass 80hz on the mains and a low pass 80hz on the sub. The sub plays at 20hz-80hz and then slopes down, but say it really peaks at 40hz. Your mains normally slope down at 40hz, but now start sloping at 80hz. The sub naturally slopes it self from 40-80, but the reverse slope from the mains counteracts this. The mains are usually flat 80-120, but the slope from the sub boosts this (a lot at 80, then sloping down).

The bass doubling won't be much, but will boost the 40-80hz range more than normal. An abnormal boost in the range can be very noticeable and make the system sound boomy. On the flip side, if your position in your room has a natural dip in this area, the bass doubling might improve your situation.

So, with ALL the variables and the cost of the equipment needed to properly set everything, best bet is to play around and see what sounds best to your ears. I used to have 7 Rane 31 band EQ's, a behringer 8024 EQ/DSP, and a calibrated mic. Pretty much just have the 8024 for bass EQ, but dropped everything else. I picked my B&W's for their sound, and the other EQ's killed them.

Kevin D.

[Pitou5]

Hello,

I noticed the same as maddog on my rsx-1055, it seems that the large or small doesn't make any difference. My crossover is set at 80hz.

Now I'm very concerned about that.

Is it normal?

Pitou!

[Kevin D]

Are you talking about on the mutli-channel inputs? If so, that's correct, they don't apply to the mutli-channel ins.

[Pitou5]

No I'm not talking about the multi-channel inputs.

I'm talking about the crossover that is not very sharp. I understand that it's not a brickwall and that there is a slope, but speaking of a "digital" crossover, I would have expected better. I start hearing sound at about 40hz, which is way below 80hz (which is my crossover setting)

Pitou!

[Pwoida]

That's only an octave below, you would probably expect it to still be audible. It could however be harmonic distortion of your speaker(producing multiples of the input frequency).

Someone sould do a test with a scope and measure the output at the speaker outputs to eliminate the effects of the speaker and room.

I would have done it ages sgo but unfortunately I don't have a scope.

[Shawn Parr]

No I'm not talking about the multi-channel inputs.

I'm talking about the crossover that is not very sharp. I understand that it's not a brickwall and that there is a slope, but speaking of a "digital" crossover, I would have expected better. I start hearing sound at about 40hz, which is way below 80hz (which is my crossover setting)

Pitou!

As Pwoida already stated, that is only 1 octave below. From what I have seen, most HiFi equipment uses 6dB, 12dB, or in some extreme cases 12dB per octave.

6dB change equates to twice as loud or half as loud to your ear. So if you have your corner frequency for your filter set at 80hz, that means the curve should be at -3dB at 80hz, and as such with a 6dB per octave filter you will be at -9dB at 40hz, or at -15dB with a 12dB per octave filter.

One of many reasons to use gentle slopes, even with digital filtering, is that the sharper the slope is, the more phase distortion is created by the filter. The only way to avoid this is to use digital FIR filters which require a lot more processing power, and requires a non-insignificant delay into the system. This delay could cause problems with lipsync if you don't have a system that normally requires lipsync compensation. Because of this, when recording one usually will avoid any processing that utilizes FIR filters, and instead applies those processes during mixing or mastering when musicians don't require hearing things in real time.