DEQX Premate 8 digital active crossover / DSP

[Reddoc]

Tri-amped Magneplanar MG 3.6 + 4 twelve-inch servo open baffle subs

WOW! Enjoy

 

[Rikpost]

It looks like the DEQX has officially launched.
Prices are listed at dealers, and the unit is available—though it seems to have happened rather quietly.

 

[Vds]

They’ve been so quiet. I thought there might be one more software update before a retail release.
Question/observation… when I finish calibration I often check by using Rew with a usb mic and the frequency response is always no where near flat, and my ears confirm this. I then use eq to adjust accordingly. Are others experiencing this, or if not, how are you ending up with a smooth Frequency response?
Thanks

 

[Keith_W]

They’ve been so quiet. I thought there might be one more software update before a retail release.
Question/observation… when I finish calibration I often check by using Rew with a usb mic and the frequency response is always no where near flat, and my ears confirm this. I then use eq to adjust accordingly. Are others experiencing this, or if not, how are you ending up with a smooth Frequency response?
Thanks

There are multiple points of uncertainty here.

1. You have to ask yourself what measurements you are feeding into the DEQX and asking it to correct. Your verification measurement(s) need to be taken the exact same way. For e.g. if you are correcting over a listening area and not a listening position, your verification measurement needs to cover the same area.
2. There is always a possibility that you have used the software incorrectly.
3. The DEQX uses an XLR mic, and you are doing your verification with a USB mic. Have you compared the responses of the two mics? Maybe one of your mics is off?
4. Are you pointing the mic at the speaker, or at the ceiling? This alone can create a high frequency drop-off if you don't have the correct cal file loaded. When taking measurements for DSP, the mic should ALWAYS be pointed at the speaker with the correct cal file loaded. The only exception is if you are correcting a surround system.

I recently had a very good look at a DEQX. Someone bought one and wanted me to help him take measurements. I read the manual, then I went to his place and watched him make a correction from start to finish. My biggest beef is that the software takes a lot of control away from you. The measurements go to the cloud and then get sent back to your unit. I don't want that, I want the unit to do exactly what I tell it to do, and not some mysterious algorithm sitting in a server somewhere. And I also don't like it's inability to export measurements or import finished filters. I find devices that blocks me from doing certain things to be very frustrating.

 

[Vds]

There are multiple points of uncertainty here.

1. You have to ask yourself what measurements you are feeding into the DEQX and asking it to correct. Your verification measurement(s) need to be taken the exact same way. For e.g. if you are correcting over a listening area and not a listening position, your verification measurement needs to cover the same area.
2. There is always a possibility that you have used the software incorrectly.
3. The DEQX uses an XLR mic, and you are doing your verification with a USB mic. Have you compared the responses of the two mics? Maybe one of your mics is off?
4. Are you pointing the mic at the speaker, or at the ceiling? This alone can create a high frequency drop-off if you don't have the correct cal file loaded. When taking measurements for DSP, the mic should ALWAYS be pointed at the speaker with the correct cal file loaded. The only exception is if you are correcting a surround system.

I recently had a very good look at a DEQX. Someone bought one and wanted me to help him take measurements. I read the manual, then I went to his place and watched him make a correction from start to finish. My biggest beef is that the software takes a lot of control away from you. The measurements go to the cloud and then get sent back to your unit. I don't want that, I want the unit to do exactly what I tell it to do, and not some mysterious algorithm sitting in a server somewhere. And I also don't like its inability to export measurements or import finished filters. I find devices that blocks me from doing certain things to be very frustrating.

Let me simply and slightly change my statement. The frequency response displayed by Deqx on the EQ tab is nowhere near flat or smoothly sloping, it’s very irregular, and that is what it sounds like. When I “verify” with REW, I’m getting a similar FR when mic is in same position as Deqx mic. ( For this test I never moved the mic, it stayed at 1 meter from speaker). I assume, if I’m doing it right, Deqx should be smoothing and flattening FR.
Do others get FR display that looks smooth?

I wish Deqx would give better directions, suggestions , explanations, of what is happening at each measurement, and how mic placement choices effect the outcome.
Ted

 

[Rikpost]

I’d like a bit more insight, too—though on the other hand, it does sound really good.
I’ve done separate measurements with REW as well, and indeed, the results aren't what you’d expect.
I do make some minor adjustments using the PEQ and levels, and that’s it!
I’m just going to leave the REW measurements be.
It is strange, though.
The clarity and the speed are from another world .
Still i do think they have Some work on the software but these are minor isues

 

[AudioJester]

I recently had a very good look at a DEQX. Someone bought one and wanted me to help him take measurements. I read the manual, then I went to his place and watched him make a correction from start to finish. My biggest beef is that the software takes a lot of control away from you. The measurements go to the cloud and then get sent back to your unit. I don't want that, I want the unit to do exactly what I tell it to do, and not some mysterious algorithm sitting in a server somewhere. And I also don't like it's inability to export measurements or import finished filters. I find devices that blocks me from doing certain things to be very frustrating.

To be fair, this is what DEQX is designed to do and what its buyers want.
What you want is your own custom device tailored to your specific needs. Its similar to your complaints about Roon.

The vast majority of end users prefer simple automation, we tinkerer's are really a minority, over-represented on ASR.

Excludong cost, DEQX is going to be a way better option for the average audio enthusiast than most other choices.

 

[Keith_W]

What you want is your own custom device tailored to your specific needs. Its similar to your complaints about Roon.

No. All I want is a device that does what it is told. Let me design my own filters and load them, and i'll be happy. There are devices on the market that do that - MiniDSP, Danville DSP Nexus, Xilica ... unfortunately, all are IIR. There's also the upcoming Hang Loose Processor, which will be FIR. And then there's "rolling my own" which is what I ended up doing.

My complaints about Roon are twofold: (1) any subscription based software can eff right off. (2) it's supposed to be able to find music, but it can't do that. A player that can't find music is useless. If I type "Beethoven Symphony", I want Beethoven Symphonies. Not Shostakovich, not Mozart, not Prokofiev. Roon contaminates the search with results I did not ask for, and then the exact album that I want does not show up, even though I know it's there. Tidal can find it, why not Roon? Once again: it is software that doesn't do what it's told.

I don't have "specific needs", unless you think that a device that simply does what you want it to do is a "specific need". I wouldn't buy a robot lawnmower that decided it wanted to mow my flower beds despite being told to stay away from the area. I wouldn't buy an oven that decided I don't really want 150C and gave me 250C. And I wouldn't buy one of those Tesla self driving monstrosities that thinks it know how to drive better than I do. These machines have one single and simple job: do what I say and don't try to second guess me.

For DEQX, it would be a pretty simple fix. Let users import filters that they designed themselves. They could go use Acourate, Audiolense, Focus Fidelity, REW, or any DSP filter design software that they like. No need to force all their customers to use their software.

 

[Keith_W]

Let me simply and slightly change my statement. The frequency response displayed by Deqx on the EQ tab is nowhere near flat or smoothly sloping, it’s very irregular, and that is what it sounds like.

Which part of the frequency response? The bass part should have the living daylights DSP'ed out of it. But the high frequencies should be mostly left alone. If you want a smooth top end, you will need quasi-anechoic measurements of your speaker. Only then can you DSP the living daylight out of it.

I wish Deqx would give better directions, suggestions , explanations, of what is happening at each measurement, and how mic placement choices effect the outcome.

I agree with you. I just had a look at the DEQX manual that was emailed to me about a year ago. Under the section on "Measure drivers", there are no instructions on how to take a driver measurement. I seem to recall seeing a more updated manual on his PC when I visited him, and from memory it said to place the microphone 10cm in front of the driver. This is wrong.

Taking a quasi-anechoic measurement of a speaker driver is not an easy task. These are some things you need to keep in mind:

1. The mic needs to be spaced far enough to capture the baffle step. Which means you need to calculate the baffle step. If the driver radiates into 4pi space (omnidirectional), the mic is spaced 0.11x diameter of the driver on axis (Keele 1974). If it's 2pi space the mic needs to be spaced about 1-2 baffle widths away.
2. The measurement needs to be examined for reflections from nearby surfaces and appropriately windowed.
3. If the signal-noise ratio is poor, multiple measurements need to be taken, examined, rejected, and merged.
4. Measurements must be taken within the linear SPL range of the speaker.

There is a lot to it. I can understand why DEQX chose to leave that out of the manual, it would make the manual very long indeed. And the manual is written to teach you how to use their software, it's not a book on how to take loudspeaker measurements.

Remember that if you do not know what you are doing, it may be better to leave it alone. From the dozens of people I have helped on ASR, I have a simple observation: people who are new to DSP usually make the mistake of making aggressive corrections on poorly taken measurements. Bad results inevitably follow. For now, I suggest you take a look at the link in my signature. There are two books in there written for Acourate. Although it's for Acourate, there is enough general DSP advice in there for you to follow. Skip all the Acourate specific sections.

 

[Rikpost]

The meassurement taken so close to the driver 18 cm is to not meassure the room inflence .
The baffle step is calculated when you meassure the whole speaker on a dinstace .
The software is calculate what the reflections are and whhat the speaker is .
What you get back are the basics , you can fine tune as mutch as you like with the peq and the levels .
I use to own the Groundsound dnc 28 ,al manuall settings but i must say the deqx sounds way better

Gretings Richard

 

[milosz]

DEQX' modus operandi seems to be obscurity

 

[john61ct]

"Most people" know nothing about measuring nor creating DSP filters, and want to keep it that way, if possible get things fixed at the press of a button.

If you could just tell an appliance "cook this roast" not knowing anything about temperatures

if self-driving already worked well, these would sell very well.

I personally agree that "openness" between tools is critical FOR ME, I prefer systems where I understand what's going on, and can override automation shortcuts.

But from a marketing POV there will always be products that are specifically designed to ignore those preferences

and I accept I should simply avoid them.

Or in some categories, as with laundry machines, just accept there is little choice.

 

[milosz]

"Most people" know nothing about measuring nor creating DSP filters, and want to keep it that way, if possible get things fixed at the press of a button.

If you could just tell an appliance "cook this roast" not knowing anything about temperatures

if self-driving already worked well, these would sell very well.

I personally agree that "openness" between tools is critical FOR ME, I prefer systems where I understand what's going on, and can override automation shortcuts.

But from a marketing POV there will always be products that are specifically designed to ignore those preferences

and I accept I should simply avoid them.

Or in some categories, as with laundry machines, just accept there is little choice.

There's also the issue of proprietary products - the DEQX algorithm and how they build filters is not open source. Trade secret and all that.

 

[Keith_W]

The meassurement taken so close to the driver 18 cm is to not meassure the room inflence .
The baffle step is calculated when you meassure the whole speaker on a dinstace .

I don't see how software is able to do that. Let me show you.

In red: actual measurement of my woofer from 10cm away. In blue: simulation of the baffle step.

Comparison of the actual measurement (red) with the predicted response at the MLP (brown). The predicted response was obtained by convolving the measurement with the baffle step simulation.

Left graph: comparison of the predicted response (dark brown) with an actual measurement from the MLP (green). What you see here: the general trend of rising response is present on both graphs, confirming that my simulation is likely a true representation of the anechoic performance taken from a distance.

Right graph: comparison of the 10cm measurement (red) vs. the MLP measurement (green). What you see here: the MLP measurement is corrupted by room reflections which can not be windowed out since they occur too early with respect to the direct sound.

Take away: an actual MLP measurement of the driver is too corrupted by room influence to be used for fine DSP correction. There is no software compensation that can fix that.

 

[Rikpost]

Hi Keith,
Thanks a lot for the effort.
But doesn't the DEQX compare the close-range measurement with the measurement taken at a distance?
You would think the baffle step could be derived from that.
Of course, you have room influences, but that is something analyzed in the cloud.
At least, that’s the idea; I don't know how well it actually works—after all, you don't have any visibility into that process.
I do understand why DEQX chose this method, given the limited size of the market.

 

[Keith_W]

Hi Keith,
Thanks a lot for the effort.
But doesn't the DEQX compare the close-range measurement with the measurement taken at a distance?
You would think the baffle step could be derived from that.

I would be very skeptical if they could infer a high resolution correction from a measurement taken from the MLP. The rules are: if you want to correct a loudspeaker, you need an anechoic measurement. If the driver is on a baffle, the measurement needs to account for the baffle step. End of story. Short of using a Klippel, there is only one other way to take a quasi-anechoic in-room measurement: Krol's beamforming method. Otherwise, there is no magic here.

This is what I mean when I said I was frustrated by the DEQX. Because of the way the software works, there is only one way to take that measurement. You can't do something like a ground plane measurement, or any type of measurement which requires mathematical manipulation prior to correcting the result. In my case, I convolved the actual measurement with a simulation, and used that as basis for correction. Sometimes you need a workaround to get past your limitations as a hobbyist (i.e. no anechoic chamber).

I think I said in my first post that it was a bad idea to develop their own room correction software. It would push their cost up for no appreciable benefit. There are only downsides - the software would be less full featured, they would have to hire coders, maintain the code, your expensive hardware would be bricked if DEQX goes under and the servers are turned off, etc. I said this to Alan Langford many years ago. Just make it a "dumb processor", let people use their own filter design software, and at a single stroke you reduce your cost and simplify your customer support. Look at them now, years of having their software in beta, ironing out all the bugs, and it's still less full featured than say, Acourate. There was never a need for any of that.

 

[Rikpost]

You’re right, of course, but it really depends on which target audience you want to reach.
I think they want to expand their market by offering a product that requires a bit less specialized knowledge.
And it has to perform well; for me, it’s the best-sounding DSP I’ve ever heard.

 

[Digital_Thor]

The meassurement taken so close to the driver 18 cm is to not meassure the room inflence .
The baffle step is calculated when you meassure the whole speaker on a dinstace .
The software is calculate what the reflections are and whhat the speaker is .
What you get back are the basics , you can fine tune as mutch as you like with the peq and the levels .
I use to own the Groundsound dnc 28 ,al manuall settings but i must say the deqx sounds way better

Gretings Richard

Good you found a better successor for you.
I still own the DCN28 and haven't tried neither a deqx or minidsp that does better . . . Further I haven't heard any improvement with FIR... Yet.
Groundsound made a fully discrete built version with way more power and everything optimised, and that one sounds awesome... But still under development
FIR is actually on the drawing table.... but let's see what happens

 

[Rikpost]

As far as I know, Groundsound ceased operations a long time ago.
Anyhow I have tried the mini dsp flex 8 but that didn’t do it for me .
I have sold the Groundsound after I had installed the DEQX .
But its Much newer and much more expensive .
The Groundsound is a verre good and flexible divice

 

[El Geffe]

it's supposed to be able to find music, but it can't do that. A player that can't find music is useless. If I type "Beethoven Symphony", I want Beethoven Symphonies. Not Shostakovich, not Mozart, not Prokofiev. Roon contaminates the search with results I did not ask for, and then the exact album that I want does not show up, even though I know it's there. Tidal can find it, why not Roon?

A bit off topic: After more than 10 years there is a Roon update today that finally fixes a lot of search issues. It finally is able to cope with mistyping and finally only gives you your Beethoven Symphonies instead of many things you did not ask for.