So this almost never happens...

[Silas Pradetto]

I picked up four of EAW's vintage KF650es a few days ago and figured I'd better get about building DSP settings for my Itech HDs. Since EAW doesn't offer settings for anything other than their DSPs, and I didn't have any of EAW's processors around to measure, and I wasn't about to build settings from scratch, I posted in the Marketplace looking for DSPs to borrow and measure.

Fortunately, I didn't have to get an MX8750 to measure because Langston linked me to where EAW has the Smaart traces of the the MX8700/MX8750 online (thank you!).

This made matters quite easy - I simply had to measure the transfer function of the Itech HD and compare it to the traces EAW provided.

My test setup was as follows:

Pink noise originating from Smaart played through an ART USB DualPre USB audio interface, left channel out into analog input on the HD. Left channel link thru on the amp was then connected to the XLR input of a Radial Pro AV1 direct box, which was then connected via XLR to the DualPre's right channel input as my reference. The direct box was required here for two reasons: one, to pad down the reference signal since the interface does not have line inputs or pads, and also to put a transformer in line for both the reference and measured inputs, keeping the loopback transfer function flat.

For the measured input, I first connected the actual KF650e to the amp outputs (as a real-world load, since class I/D amplifiers have a low-pass filter on the output which uses the connected load as part of the circuit. Translation: the amplifier's transfer function changes depending on what's connected.) I also made a banana to 1/4" adapter and plugged that into whichever channel I was measuring at the time. The 1/4" end plugged into a Radial ProDI with the 15dB pad engaged (and was measured separately to ensure ProDI and ProAV1 had same transfer function, which they do. The ProDI doesn't connect the 1/4" ring to anything and I didn't want to leave anything floating, hence the use of the AV1 on the reference channel). The ProDI was then plugged into the left channel of the DualPre.

Before doing any measurements, I punched EAW's stock crossover and PEQ settings into the amp. First, I measured the MF/HF passband (I'm only running these in biamped mode). Smaart reported - amazingly - that the stock settings were pretty much interpreted perfectly by the HD processing. This never happens. The only slight difference was in the top end, which was slightly lower than EAW's transfer settings mainly due to the amp's own HF rolloff due to the class I design. A slight .6dB increase of one PEQ and the addition of another PEQ at 11.25k with a .7dB boost made it match about perfectly.

The LF was equally amazing - both PEQs and crossover filters matched perfectly. Absolutely no change required.

To finish up my custom HD presets, I adjusted the bandpass gains to provide a good gain structure through the amp, and calculated out limiters using the standard formula.

So, for a couple hours work, I now have excellent IIR presets for my KF650s!

Pictures: test setup and the transfer functions. Loopback test is the lowest, mostly flat setting. That is the HD amp with all DSP off and the speaker connected offering an 8 ohm nominal load. The highest traces are EAW's traces, and the two lower ones are my traces, lowered by 2dB so you can see the difference (or lack thereof).





Next: making these presets into flat-phase presets using all-pass filters, at least the best that I can.

Eventually: translating the Gunness-focused greybox settings to the HD platform.

 

[Stuart Hannon]

I would have just listened to them and used my EARS. Isnt that what us audio engineers are supposed to do?

Sent from my DROID RAZR 2

 

[Ivan Beaver]

Re: So this almost never happens...

I would have just listened to them and used my EARS. Isnt that what us audio engineers are supposed to do?

Sent from my DROID RAZR 2

I would challenge most "engineers" to attempt to dial in a loudspeaker as good as a measurement system can 9with a qualified operator).

Yes there are a VERY VERY FEW-who can get pretty close. But I would argue that most "sound men" can''t even get close.

Yeah a good person should be able to find the big issues-but it is often the little things that are harder.

Determining crossover points is a lot harder.

It is one thing to "voice" a system to taste. Quite another to arrive at proper DSP settings. At least accurate settings.

 

[Rob Timmerman]

Re: So this almost never happens...

I would have just listened to them and used my EARS. Isnt that what us audio engineers are supposed to do?

Sent from my DROID RAZR 2

Yep. Which is how you can spend several hours and still not get results as good as what the manufacturer with modern measurement tools got for recommended settings. And which is also why few systems with external non-blackbox processing are consistent from one user to the next.

 

[Evan Kirkendall]

Re: So this almost never happens...

It is one thing to "voice" a system to taste. Quite another to arrive at proper DSP settings. At least accurate settings.

Oh yes. I'll leave it up to the manufacturer to get me crossover, delay and basic EQ points. Then the voicing part is done by ear, and double checked with SMAART(just to make sure I'm not losing my mind).



Evan

 

[Helge A. Bentsen]

Re: So this almost never happens...

FWIW I tested the same basic eq and xo settings on a Itech HD and a PLM14K and got identical traces on my labsubs from both amps.

 

[Stuart Hannon]

Its terrible that the industry is coming to a point where engineers are becoming sound men. Reliance on smaart and measurements is killing the skills that actually make us worth paying, like being able to determine an appropriate crossover point in a simple box such as a 650.
This would have been a perfect opportunity for the op to spend an afternoon with a process, a mic and the speakers. This is an easy way to really get to know your speakers and their limits; much better than manufacturer's specs who are concerned about return rates. Its easy to find points that work well and may or may not sound better.

Try it sometime, its not hard.
Sent from my DROID RAZR 2

 

[Stuart Hannon]

For context, I am not talking about blackbox systems. I am talking about 2-3 way active systems.

Exercise your "ear muscles" and they will grow strong. Listen harder and learn the subtle difference between two types of xo slopes, there is a difference, I assure you.

Sent from my DROID RAZR 2

 

[Geri O'Neil]

Re: So this almost never happens...

Stuart, this is an old argument and the answers are too easy. It's as simple as two parts...

(A) Varsity guys like a consistent place to start. So just how fresh are those magic ears (I do NOT use that term in jest) after an 8-hour flight or a long bus ride (or truck ride/drive in Kombat Audio land)? And how are those sinuses on any given day?

(B) Varsity guys work out a system using BOTH ears and measurements to get the results that they are looking for, presumably the same results that keep them employed and sought after.

And an added footnote: Varsity guys are more interested in the results than the method. And (usually) don't tell other people that they are doing it wrong, especially if the results are acceptable. And we all know guys that claim to have golden ears, but would probably benefit from a measurement system, as well as training to use that system.

Interesting that you exclude the blackbox systems from the discussion, considering how the manufacturers derive those results.

And I hope that you understand that I'm not critical of your methods, just the argument you present regarding other people's methods.

Respectfully, as always,
Geri O

 

[Justice C. Bigler]

Re: So this almost never happens...

So just how fresh are those magic ears (I do NOT use that term in jest) after an 8-hour flight or a long bus ride (or truck ride/drive in Kombat Audio land)?

This is especially true. When I flew out to Las Vegas for NAB and back my ears were pretty thrashed after 8 compression and decompression cycles in three days. It took three or four days before I felt like my ears were back to normal. I was lucky that I came back to an opera and didn't have to do any real mixing.

 

[Jay Barracato]

In my mind, being able to measure and optimize based on measurements is what lifts the skillset of an "engineer" above a "technician".

 

[Ivan Beaver]

Re: So this almost never happens...

FWIW I tested the same basic eq and xo settings on a Itech HD and a PLM14K and got identical traces on my labsubs from both amps.

Here are the basic differences I have found between the Itech HD amps and the Lab gruppen lake processing and our own DSP. This is based on the same DSP for the same 4 way loudspeaker.

With the Lake/lab-the Qs were all off. Everything else was fine. On the HD amps-the freq were off.-specifically on the crossover freq-I don't remember about the eq filters.

So the results should be different between the two different models. Not a make or break difference-but enough to easily measure.

 

[John Roberts]

Re: So this almost never happens...

[satire] Ah you guys are wimps.. back in the day we set up crossovers with a dog, a cat, a stick and a candle... And it wasn't even a sharp stick,,, and the candle was one of those birthday cake candles, but already used so we had to work fast. [/satire]

JR

PS I'm a little disappointed that Silas was criticized for using too much science before I could lecture again about using factory presets, but he was going for the factory curves so kudos, and carry on.. While EQ and DSP settings can sometimes experience translation errors, occasionally you will end up on a platform that agrees with the original manufacturer's bench setup. Stranger things have happened. .

 

[Silas Pradetto]

Re: So this almost never happens...

Its terrible that the industry is coming to a point where engineers are becoming sound men. Reliance on smaart and measurements is killing the skills that actually make us worth paying, like being able to determine an appropriate crossover point in a simple box such as a 650.
This would have been a perfect opportunity for the op to spend an afternoon with a process, a mic and the speakers. This is an easy way to really get to know your speakers and their limits; much better than manufacturer's specs who are concerned about return rates. Its easy to find points that work well and may or may not sound better.

Try it sometime, its not hard.
Sent from my DROID RAZR 2

Stuart, unfortunately, it is hard. I would maintain that the KF650 is not at all a simple box - it has 3 passbands with the drivers being quite physically separated. I would challenge you to design the tuning for this box with just your ear and have it come anywhere close to what EAW provided.

Crossover points take into consideration many, many other things than sound. In fact, sound is probably the least concern of selecting proper crossover points of a box. The main concern of selecting crossover points, in my opinion, is polar response. The crossover points should be located where the patterns of adjacent drivers combine well, where the drivers are still operating in their comfortable range, and loading whatever horn/box they are placed in correctly.

Another huge component of building a DSP tuning is phase alignment between passbands. This is hard, because with the exception of a coaxial speaker, there is no one setting that is correct; they are all a compromise depending on listening position. Even if there was one setting that was correct, it's still not that easy. Different crossover types and different drivers will create instances of varying phase slope. I dare anyone to phase align drivers by ear when their phase slopes are not identical, if you could phase align anything by ear at all.

Without making complete polar measurements in an anechoic chamber, it's really quite doubtful that you'd come anywhere close to what EAW's engineering team could come up with. I have built DSP settings from scratch, using up entire days trying to do it. There is really no way to get it done without lifting a speaker far enough into the air that ground bounce is not a concern, or using ground plane, which is only semi-functional with the upper end of things, because it changes the loading from full space to half space. When doing this myself, I do multiple measurements off-axis and on-axis to confirm pattern and crossover compatibility, and still feel like I only get halfway there.

The EQ of the box is probably the easiest part to do by ear, but it's still hard. When doing my own tunings, I usually set the EQ such that each passband is flat beyond an octave past its expected bandpass. This ensures crossovers are consistent. Still, even simple things that exist in the EAW tunings, like 2dB cuts with a Q of 10, are so subtle that I bet most people couldn't hear them, even doing an A/B comparison with and without that filter. This is why a controlled measurement environment is so critical.

There's more to it than just getting the crossovers, phase alignment, and overall magnitude response correct. As I hinted above, phase is a big deal, and it directly corresponds to impulse response. If the tuning is done, the crossovers are set, the pattern is good, it's nice and flat, guess what? It could still be just OK from a time perspective. I did a quick measurement with an Audix TM1 when I was done with the KF650 - the magnitude response was pretty decent (given the terrible measuring conditions), but the phase trace showed a few thousand degrees of phase wrap. This not only shows the bad measuring conditions, but it shows that the KF650 impulse response is going to be terrible. As I mentioned in my OP, I'm planning to implement all-pass filters to flatten the phase of the box as soon as I get a chance. This will drastically improve the impulse response and subjective sound quality of the box without affecting magnitude response at all.

I've challenged you to design crossover settings by ear, which is one thing. Let's see someone design a flat-phase tuning with all-pass filters by ear alone.

And after all this basic IIR stuff is done, then the really hard stuff comes into play. Think about horn resonances or throat distortion or all those other things that cannot be corrected with any amount of EQ or phase adjustment. Most notably, Dave Gunness can build FIR tunings that combat as much of this as possible. The average magnitude/phase FFT measurement doesn't even show things like throat distortion, so where would you even start building an FIR (truely, even I don't know) if you aren't Dave Gunness?

EAW provides tunings as a baseline that takes the environment out of the equation. They are designed to be flat out of the box, and the engineer is expected to EQ to taste at each show.

PS- I hate the EP series speaker connectors.

PPS- consider also that these factory-provided tunings are as good as they may have gotten with the technology available at the time (and with the ears they had at the time :lol, but with the advent of Dave Gunness, it got even better. From everyone I have talked to, the processing that he created for EAW with Gunness Focusing makes a dramatic improvement to the sound quality of the box. Guess what? I bet they both measure flat!

PPPS- Black Box systems exist because 'engineers' almost always make the system sound worse when they are allowed to change things themselves. The Black Box locks the system down so it can't be ruined by incompetent wannabes (or anyone else)...

 

[Ivan Beaver]

Re: So this almost never happens...

[satire] Ah you guys are wimps.. back in the day we set up crossovers with a dog, a cat, a stick and a candle... And it wasn't even a sharp stick,,, and the candle was one of those birthday cake candles, but already used so we had to work fast. [/satire]

JR

But I am sure that if you had a sharper stick-you would have used it.

Back in the day there simply weren't the tools available (on s o many fronts) that we have today.

I remember working on a crossover for a loudspeaker a couple of decades ago and I was using a goo quality RTA (1/3rd oct) with fine resolution and I could hear the changes I was making-but the RTA wasn't showing any difference. I could not tell which part value was correct. Very frustrating.

It wasn't until I started using higher resolution measurement tools that I started to realize what i was doing.

The ears are a good final judge-but I would not want to start out using them.

But having the experience of using my ears over most of career doesn't hurt.

 

[Michael John]

Re: So this almost never happens...

Here are the basic differences I have found between the Itech HD amps and the Lab gruppen lake processing and our own DSP. This is based on the same DSP for the same 4 way loudspeaker.

With the Lake/lab-the Qs were all off. Everything else was fine. On the HD amps-the freq were off.-specifically on the crossover freq-I don't remember about the eq filters.

So the results should be different between the two different models. Not a make or break difference-but enough to easily measure.

Hi Ivan,

Before the transfer to Lab, our guy who made presets used to have a spreadsheet that could covert many other brands of "Q" into Lake "Q." He would first use this, then verify with SMAART and adjust to within a fraction of a dB and degree. I think there were occasionally other differences in the high frequencies, mostly due to sample rate differences and filters very close to Nyquist.

Interesting you mention the Itech differences. We recently had to convert some JBL cinema speaker presets into the DLP and the conversion wasn't straight forward. For the one 3-way speaker, JBL had significantly different preset tables for their different processor/amp products - iTech's, Driveracks etc. We ended up renting one of their higher end Driveracks and measuring it...

Cheers,
Michael

 

[Helge A. Bentsen]

Re: So this almost never happens...

Hi Ivan,

Before the transfer to Lab, our guy who made presets used to have a spreadsheet that could covert many other brands of "Q" into Lake "Q." He would first use this, then verify with SMAART and adjust to within a fraction of a dB and degree. I think there were occasionally other differences in the high frequencies, mostly due to sample rate differences and filters very close to Nyquist.

Interesting you mention the Itech differences. We recently had to convert some JBL cinema speaker presets into the DLP and the conversion wasn't straight forward. For the one 3-way speaker, JBL had significantly different preset tables for their different processor/amp products - iTech's, Driveracks etc. We ended up renting one of their higher end Driveracks and measuring it...

Cheers,
Michael

My measurements was done inside my warehouse, not an ideal place to measure subs. And it was a basic HP/LP and two bands of EQ, hence the "FWIW"
I'm not saying that they always measure the same, but on that occasion, they did

Do you have a method of recalculating presets between Itech HD and Lake?
I'm planning on measuring that difference for a couple of presets when I find the time. For now I run with some I made that sounds ok and I haven't blown any drivers. Yet.

 

[Christian Tepfer]

Re: So this almost never happens...

But I am sure that if you had a sharper stick-you would have used it.

Back in the day there simply weren't the tools available (on s o many fronts) that we have today.

I remember working on a crossover for a loudspeaker a couple of decades ago and I was using a goo quality RTA (1/3rd oct) with fine resolution and I could hear the changes I was making-but the RTA wasn't showing any difference. I could not tell which part value was correct. Very frustrating.

It wasn't until I started using higher resolution measurement tools that I started to realize what i was doing.

The ears are a good final judge-but I would not want to start out using them.

But having the experience of using my ears over most of career doesn't hurt.

Using ears is good, but ears can be fooled. They adapt and that can be counter-productive. On the other hand tools only give answers to the questions we give them. The hardest part in tuning systems is the questions imho.

 

[Ivan Beaver]

Re: So this almost never happens...

Using ears is good, but ears can be fooled. They adapt and that can be counter-productive. On the other hand tools only give answers to the questions we give them. The hardest part in tuning systems is the questions imho.

And to add a little bit more-you need to know the answer (or kinda close) to the question-BEFORE you measure. If you don't then how can you be sure that the measurement is correct?

We see all kinds of measurements that are invalid-but often people think that because something shows up on a computer screen it is "the truth".

How a person reads that data (and the "question" that was asked) determines what (and how valid) the shown data is.

I know of many cases in which my ears have been fooled.

How many times have you made an eq "adjustment"-only to realize that the eq was not engaged-or that you adjusted the channel next to the one you wanted to?

If you haven't-a person is either lying or hasn't done it long enough.

Or listen really hard to a musical passage-then mute the channel. Now unmute the channel and listen again. Do you hear a difference? Or change the level 0.5dB while the channel is muted and see how different the sound is.

It is quite amusing how many "things" that the audio fools hear when they think that various "mods" to the system are making a real difference.

That is where measurements (at least accurate ones) come in handy. At least they bring a reference to the situation.

OK rant off.

Yes the ears are VERY important. But proper measurements can show up all kinds of things that the ears can't here.

I see statements all the time that people can hear things that can't be measured. But that is only because they are either using the wrong tool or not measuring the right thing.

Sorry to rant.

 

[Adam Black]

Re: So this almost never happens...

[FONT=Helvetica, Arial, sans-serif]An egregious misconception about measurement applications is that they somehow nullify the required expertise of the user. I believe this is the impetus for antithetical sentiment regarding their use. Ironically, the effectiveness of a tool is directly related to the expertise of the user.

Measurement applications don't indicate what is wrong with the system, nor do they indicate what's right. They simply depict what is. This information is meaningless without a skilled user to interpret the data and act upon it.

Sorry for the rant, I realize I'm preaching to the choir.


-A[/FONT]