Print

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

[Langston]

I have more measurement toys than you can throw a stick at and it's still an exercise in frustration to use MC for the Mac to implement a multi-amped system with MC acting as the active crossover.

Adding the loopback mode you've had in the Windows version for several years is necessary for crossover alignment and a huge time saver for standard tasks such as EQ.

I'd be glad to beta test and provide detailed feedback in reference to this feature addition.

Thanks again!

[blgentry]

Can you point me to a reference on why loopback mode is helpful or necessary?  I know that loopback normally means to connect the output back to the input.  This let's you see two (or more?) things normally:  The delay inherent in the system.  The frequency aberrations of the system.

I would expect the delay would be constant.  I'd expect the frequency response of MC to be flat from output to input, but maybe that's just me being incredibly naive. 

Brian.

[Langston]

Hi Brian:

Good question. I think I understand it, but if not maybe this will at least be entertaining. :) I’m leaving out important lesser points - but here’s my pitch...

My goal is to eliminate the expense, etc., involved with using a hardware loudspeaker processor between MC and the amplifier(s). The huge advantage of an external processor with crossover and EQ filters is that I can send the output of my measurement device instead of music through the processor - amplifier - loudspeaker to see the real affect its electrical filter adjustments have on the raw driver passband.

It is the combination (you could say convolution) of these electrical filters and the raw driver response that yield the actual acoustic response at the measurement microphone or your ear. It’s that combined acoustic response that we’re ultimately trying to massage into as close a copy as possible with the electrical output of our prerecorded music playback system, generally speaking. A classic way of doing this that was more interesting than useful was to playback low level 100Hz and 1kHz square waves and see how badly the loudspeaker system mangled it in the acoustic domain.

If loudspeaker drivers had perfectly flat frequency responses with instantaneous time domain tracking of the electrical signal, then we could ignore the raw driver’s contribution to the mix and just implement filtering in MC and know that’s what we’d end up with getting to our ears. Not gonna happen in this universe, unfortunately. :)

Bottom Line

Very few MC users will ever have the knowledge, money and time to properly tune a loudspeaker system, especially a multi-way system, without closed-loop measurements. Open looped measurements leave out the critical phase information, without which it is impossible to align a crossover (among other things).

Definitions

Closed-loop means connecting the output of your measurement system to the system under test (MC - amplifier - loudspeaker) with the measurement microphone feeding the input of the measurement system.

Open-looped means you generate a test stimulus file, import it into MC, then play it back and record the acoustic result with your measurement system. The measurement system loses track of time, thus phase, with this method.

Conclusion

There’s a reason why the windows version of MC has the ability to accept closed loop measurement connections and the Mac version needs it for the same reason. There is no practical way at this time to use MC for Mac to replace an external loudspeaker processor for bi-amped, tri-amped, etc., setups. :(

Background

High Frequency Crossover Alignment

Subwoofer Crossover Alignment

[blgentry]

You know, I realized while driving to lunch today that when you asked for "loopback" what you really meant was "audio input".  For the exact reasons you describe.

I've actually done measurements like you are outlining:  Running the test equipment output through the entire signal chain, including EQs, crossovers, etc, and then measuring the output.  Then you can make measurements, change the EQ, and measure again to see the effect.  It's the most logical and straight forward way to do it.

I'm a little surprised that modern gear can't do a time correct "open loop" measurement though.  Back when I was using a TEF, the first thing we would do is an ETC curve (Energy Time Curve), which visually showed us the combined delay from input to output.  Using the cursor, we'd find the first arrival, and then set that as the delay time.  So the rest of the measurements used this same delay.

With open loop, the delay would presumably change every time, as a human being is pressing "play" and "record" on two different systems.  But, I've seen that some modern systems seem to be able to derive an ETC from the same swept sine source that we would normally use for EFC (Energy Frequency Curve, AKA frequency response).  This implies that it can easily figure out the first arrival by itself, or simply be told the first arrival time.  Which means time is "calibrated" and thus time and phase should be correct.

Besides, relative phase is all that's important when measuring a loudspeaker system.  In other words, I don't care if the signal arrived at the microphone 50 mS after it was sent, or 5000 mS.  What I really care about is the relative arrival time of the tweeter versus the midrange.  ...and of course I want to convert that to phase at some point, once the arrival time difference is very small (less than one full wavelength near crossover).  Then I can try to align or adjust based on what the relative phase does to the combined frequency response of both drivers.  I hope you follow my logical path.  Perhaps there's something technically wrong with my idea.

In any case, it certainly would be MUCH easier if MC for Mac had an audio input, like the WDM driver under windows, so you could just feed it the output of your test gear and get the whole audio chain into the measurement.

Thanks for the dialog. 

Brian.

[Langston]

Good point again - I have no idea what to call this feature. I've been trying to figure out the JRiver terminology for "audio input", "live mode", "loopback mode" or whatever. I just want to measure through MC with the ability to engage or disengage the DSP Studio components. :)

Here's a related tread from the OP that seems to code all the stuff in MC that I want the most...

---

BTW, you're a rare bird to know about postprocessed deconvolution (deriving the IR from open loop recording using a known sweep with something like GratisVolver). The problem with this approach, as you imply, is the loss of arrival time information. There's still no way to know exact time offsets (relative phase) when making separate measurements of a low passband and a high passband without closed loop measurement. The measurement system has to know when the stimulus was generated. Resolution within a one sample is standard (1/Sample Rate).

[blgentry]

Good point again - I have no idea what to call this feature. I've been trying to figure out the JRiver terminology for "audio input", "live mode", "loopback mode" or whatever. I just want to measure through MC with the ability to engage or disengage the DSP Studio components.

As I understand it, this is done through the WDM driver on Windows.  But thinking about it, WDM is a way of routing other windows sound sources through MC.  I'm not sure how you'd use a hardware input and route it to WDM.  I don't use the Windows version of MC, so I'm not sure how that works

BTW, you're a rare bird to know about postprocessed deconvolution (deriving the IR from open loop recording using a known sweep with something like GratisVolver). The problem with this approach, as you imply, is the loss of arrival time information. There's still no way to know exact time offsets (relative phase) when making separate measurements of a low passband and a high passband without closed loop measurement.

I only know about this because I saw REW do it.  Made a sweep, and it had an ETC curve as an option.  I was quite surprised.

Thinking some more about my experiences with testing multi-way systems, I see your point about absolute arrival times.  I always had a closed loop system, so I never considered how open loop would work.  Thanks for pointing out the error in my thinking.  Even if you had a real nice looking set of two spikes in the ETC curve for woofer and tweeter, it would be really hard to figure out where one stopped and the other started.  We always did several ETCs, one for each driver, and then time aligned them based on the numbers, by physically moving each driver. 

It's been too long since I messed with this stuff.  Fun. 

Brian.

[mwillems]

As I understand it, this is done through the WDM driver on Windows.  But thinking about it, WDM is a way of routing other windows sound sources through MC.  I'm not sure how you'd use a hardware input and route it to WDM.  I don't use the Windows version of MC, so I'm not sure how that works

There are three ways to get sound input on windows: the WDM driver, the ASIO driver, and the "live loopback" option.  All three effectively create a software loopback.  The ASIO input allows for routing a hardware input.