Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[CraigNZ]

I am using PEQ to provide an active crossover for my front speakers.  So I setup PEQ like this:

Copy Left to CH 9 (12 channel DAC)
Copy Right to CH 10
HPF Left and Right
LPF CH9 and CH10

I then look at the signal levels at the DAC.  When I enable the HPF I see no change in the amplitude.  But when I enable the LPF I see a -20 dB (approxmate) drop in the signal level.  Loudness is disabled, there are no other DSP components in use.

This is confirmed with an SPL meter.  No change in SPL reading when enable HPF, but -20 dB drop in SPL when LPF enabled.

Shouldn't the LPF and HPF filters keep the same output levels going into the DAC?

[CraigNZ]

I just thought of something.  The calibration pink noise for the Left and Right front speakers used in JRiver is 'narrow band' .. I suspect that means there is very little low frequency content in it and thus why the level drops.  I'll jump on the system shortly and try the full band pink noise and see if that makes a difference.

[natehansen66]

Hopefully that's your problem. I've been running active xo's in MC for several years and I've never seen that. Output level never changes with h or lpf in the passband.

[CraigNZ]

Yes .. that was it.  When I used the full bandwidth pink noise calibration file everything was as expected.

I have a question though since you have been working with this quite a while.  When I add the LPF and the HPF, I set both to 24 dB slope, does this result in an out of phase result?  I read somewhere that a LR filter is used because it preserves the phase, but a IIR filter inverts the phase .. I think the JRiver LPF,HPF combination is an IIR .. is that right?

[mwillems]

Yes .. that was it.  When I used the full bandwidth pink noise calibration file everything was as expected.

I have a question though since you have been working with this quite a while.  When I add the LPF and the HPF, I set both to 24 dB slope, does this result in an out of phase result?  I read somewhere that a LR filter is used because it preserves the phase, but a IIR filter inverts the phase .. I think the JRiver LPF,HPF combination is an IIR .. is that right?

JRiver's filters are butterworth filters; to get correct summing at the crossover point you need to cascade two of them to create a linkwitz riley filter that will be phase coherent.  So you'd need four PEQ entries total: set two LPF with a 12dB slope at the xover frequency, and two HPF with a 12dB slope at the same frequency.  That will give you a 4th order (24dB) symmetrical Linkwitz Riley slope, which sums flat and is phase coherent.

Both types of filters (Butterworth and Linkwitz-Riley) are IIR, that's not the issue.  A fourth order butterworth is phase coherent but doesn't sum flat (there will be a 3dB lump).  A fourth order LR both sums flat and is phase coherent. Other orders work differently.

[natehansen66]

I'll add to mwillems' response that he's talking about the ideal response of the electrical filters. Once you add the acoustical response of the drivers things can change significantly unless you are crossing in a region where your drivers are flat for a couple octaves on either side of the xo.

[mwillems]

I'll add to mwillems' response that he's talking about the ideal response of the electrical filters. Once you add the acoustical response of the drivers things can change significantly unless you are crossing in a region where your drivers are flat for a couple octaves on either side of the xo.

Yes absolutely, you need to take your drivers response into account.  I was talking about how to create an "ideal" 4th order filter for "ideal" speakers  

The only caveat I'll offer to your observation is that the drivers need not be flat for a couple of octaves past the XO with a 4th order filter.  By the time you're an octave past the crossover point you have so much attenuation (24dB) that anything but truly horrifying resonances should be irrelevant.  Linkwitz himself says the goal is to find drivers that will be flat until they can reach -10dB of attenuation, which happens with an LR 4 about half an octave past the XO.  I think that's a little too optimistic based on my own experience, but he knows a lot more about crossovers than I do   

[CraigNZ]

Thanks guys .. working perfectly now.

My AVR recently died so rather than replace it I moved everything back into JRiver, hence all the recent questions.

Balance is good now, XO is good, final step is using REW to generate room correction filter settings.  I am currently using a Behringer ECM8000 microphone, coming in through my sound card (Audiofire 12).  I set the sound card in REW to JRiver because I need the XO settings.  But then noticed there are no inputs reported back to REW so looks like I can't use the mic .. will have to find a USB mic instead.  I think REW can use a sound card for the output and a separate USB mic for the input.

[mwillems]

Thanks guys .. working perfectly now.

My AVR recently died so rather than replace it I moved everything back into JRiver, hence all the recent questions.

Balance is good now, XO is good, final step is using REW to generate room correction filter settings.  I am currently using a Behringer ECM8000 microphone, coming in through my sound card (Audiofire 12).  I set the sound card in REW to JRiver because I need the XO settings.  But then noticed there are no inputs reported back to REW so looks like I can't use the mic .. will have to find a USB mic instead.  I think REW can use a sound card for the output and a separate USB mic for the input.

If you are using REW in ASIO mode, try using it in Java/Directsound mode instead.  When REW is set to ASIO mode it requires the same device be used for output and input, so a USB mic won't help if that's your problem.

[CraigNZ]

That would make sense.  The JRiver audio driver can present to the windows side any type of interface needed.  On the hardware side it would be limited to what the hardware requires.  In my case the hardware uses an ASIO interface, but the JRiver virtual sound card acts as a translator between the clients on the windows side and the hardware.  I'll give this a try later today, hopefully it works because I have everything in place to use the ECM8000 microphone.

Just had another thought.  Since the JRiver sound card sits in between the Windows side and hardware sound card, it should be able to control what it passes through.  For example, the default sound card could be set to the JRiver virtual sound card.  Then any windows client (like a web browser) which always uses the default sound card will now use the JRiver sound card.  But the problem is windows sounds now come through also.  Disabling windows sounds in windows helps somewhat, but some still get through and also it might be important to know something is wrong.  Ideally then what JRiver could do is have an option for Windows sound:

a) do nothing .. let it pass through
b) suppress .. no audio passes
c) visual alert -- no audio but a small indicator displays showing an alert from windows

As a minimum, option (b) would be ideal when listening to music or watching a video, but would revert back to (a) otherwise.

[natehansen66]

Good point about the LR xo mwillems.

Craig - I have Windows set to no sound in my setup (I use an AF12 as well) and I don't have anything coming through other than the audio I want when using WDM or Loopback.

I'm curious about your setup. You say you have the ECM8000 hooked up to the input of your AF12....is your mic hooked up to a preamp? That Behr mic needs phantom power which the AF12 does not provide.

I use an older M-Audio MobilePre USB for a mic pre. The WDM isn't stable enough in my setup and doesn't keep the stream open long enough to be useful so I end up using the Loopback or ASIO Bridge. Windows audio device is set for whatever audio "grab" method I'm using (not the MC device), Windows recording device is set to my mic preamp, and REW is set to Java. The output device in REW is set to match the device I have set in Windows (I think setting to default here would work), and the input device is set to default. MC obviously is set to my AF12. When using ASIO Bridge (which is similar to MC's WDM but much more stable for me) I set the windows default device to ASIO Bridge or my "dummy" soundcard when using the Loopback.

If you are in fact using an analog i/o preamp with the Behr mic then I would think it should work going into the AF input, and disregard all the above 

[CraigNZ]

All working perfectly.

Correct, the Behringer mic connects to a M-Audio DMP3 which in turn connects to CH 1 input on the AF12.  I used REW to calibrate everything and set levels.

Using Java as the sound card for playback in REW and outputs Default.  This resulted in LEFT and RIGHT channels active in JRiver.  With the XO setup, LEFT is CH1 which is the Horn, CH9 is the woofer.  So I saw signals on both.  Using the MUTE function on the AF12 I could selectively choose either LEFT or RIGHT, Horn or woofer or both.  This was useful when analyzing the crossover.

I then got a good sweep and generated the filters which I manually entered into the PEQ.  I put the filters before the crossover because it saved me having to enter them in twice, once for the horn channel and again for the woofer channel.

My options are:

a) Sweep the woofers and horns separately, generate a filter set for each and add to PEQ after the crossover filters, or
b) Sweep the woofer and horn combined, generate a single filter set for the 'speaker' and then add after the crossover which means adding for horn (CH1) and woofer (CH9) separately (ignoring those filters that are pass the xo, or
c) Add the filter set before the XO in the PEQ to the LEFT speaker

I decided to go with (c) first because it was simpler.  Listening test afterwards resulted in an excellent sound from the speakers.

It was interesting that on my core i7 PC while listening to music with all the JRiver processing that the cpu utilization registered 0% .. too low to show it.  I'm assuming from that the JRiver DSP algorithms are very efficient.

The next step is to do the Center speaker and surround speakers.  I guess what I need to do is insert temporarily a COPY filter into the PEQ to copy the LEFT channel signal to the other channels and then use MUTE feature on the AF12 control panel to measure each speaker separately.

So far so good ..

[natehansen66]

Glad you got it working.

I put the filters before the crossover because it saved me having to enter them in twice, once for the horn channel and again for the woofer channel.

I'm trying to figure out what you've got going on. You're using the same filters for the woofer and the tweeter? I'm sure I'm just not awake yet but what you've got going on isn't making any sense to me. Are you breaking out L&R using mix channels in the PEQ?

[mwillems]

Glad you got it working.

I'm trying to figure out what you've got going on. You're using the same filters for the woofer and the tweeter? I'm sure I'm just not awake yet but what you've got going on isn't making any sense to me. Are you breaking out L&R using mix channels in the PEQ?

Yeah, I'm with Nate, what's going on there? If you just attempt to correct the "whole speaker" response you're missing out on a huge benefit of bi and tri-amping which is that you can actually resolve frequency response ripples in the crossover region; if you try to make those adjustments "before the crossover" you'll have a hard time getting good correction because the "problem" is often with one of the speakers, but you're correcting both to try and fix it.

FWIW, the "correct way to do it" (insofar as there are any best practices in this discipline) is to EQ each stage (HF, MF, and LF in a 3-way) separately to get the best response.  After the channel copy, if you're doing it all in PEQ the order of EQ vs. crossovers is irrelevant. All minimum phase filters will add up the same regardless of the order of operations as long as you do it after the channel copying. 

So you should copy the channels so they're all in the right places; then apply the EQ and the crossovers (in whatever order).  To the extent you have the same EQ applied to multiple channels (i.e. to the extent that high frequency left and right have the same EQ), you don't need to enter them twice, just use the channel checkboxes in PEQ to select both channels. 

None of this applies if you have a convolution step as FIR filters don't necessarily "add up" the same way; if you're convolving you should put the filter wherever it was in the chain when you measured it and tweak with PEQ afterwards to the extent necessary.