[AdonisP91]

Enough with the chit chat, let’s Begin:

Step #1: Timing Alignment:

If you have an optical cable and a usb to optical adapter, you can then plug into the laptop and connect to the DSP. This would make this process slightly easier since you can just run the impulse sweeps straight from REW. However, I am going to assume most people don’t have that extra hardware. So what we need to do is generate sweep files that we copy onto a USB drive and then play them from the car while measuring with REW with our Umik1 or other calibrated microphone. Here are my default REW Preferences settings, I will not explain them all but will point to a few important selections. You don’t need to copy them all but it wouldn’t hurt either.



Note the circle. I selected the Exclusive driver for the Umik1. Also make sure to select the 90 degree calibration file. When running the sweeps we make sure the tip of the microphone is pointing to the sky.

Next we generate the measurement sweep files. For technical reason I can expand upon in another post, I chose the front left speaker as my reference speaker (UK drivers will want to make the front right the reference). Click on Generator in the main REW menu in the middle.



Next we click on Sweep at the right and the save to file on the bottom right, and make sure we are running 0-20,000 hz frequency range, 20-20,000 hz also works. Copy my settings here.



Select the left channel only, and click wav, save it as Left. Then select right channel only and save it as Right (you might as well save a Left and Right channel combo as well, we might use it later). Next we change the frequency range from 0-300 hz. We leave the channel to right only and we save it as Subwoofer.

Find a USB drive, format it to FAT32 or exFAT and copy the wav files on to it. Keep a copy of those files on the computer. REW will also need them.


We now want to run our sweeps. REW is ready but we need to get the car and ATF DSP PC tool ready. Connect to the DSP and load your sound profile from BimmerTech or the ones you downloaded from ATF directly. We go to IO tab and select Virtual to Output Routing. Take a screenshot and save this information, we will need to revert back to it.



For the purposes of the sweeps we need to set the Front Left Full to the Front Left Full, for everything else, we want Front Right Full to all those speakers.



Now click on Output Tab and right click on the Amp A out speaker icon, it turns green and disables all other channels. Only the front left speaker will play the sounds.



Back to REW, from the Main screen, we select Measure on the top left.



Copy these settings:



Now click on From File. We load the Left file.



Click Start, it will capture the background noise and then say waiting for timing reference.



Now you can play the Left sweep file from the USB you plugged into the car with the generated wav files. You end up with a result that looks something like this (don’t worry if your graph is different from mine this is just an example):



From my left circle we see there are some impulse measurements. We need to calibrate REW. We click on Actions, circled on top right. A new window pops up:



We select Estimate IR delay. A new windows pops up showing:



We select shift and update timing offset. We rerun the sweep on the left channel.



We see compared to the previous measurement the delay is very small, REW is calibrated to the Front Left speaker now. Delete the first Front Left measurement, keep only the latest calibrated one.

In ATF DSP Tool, we enable the next speaker to be measured, front right.



As you can see my Front Left reference is still in green, but I also enabled the next speaker to be measured in grey. We now load the Right sweep into the measurement tool in REW.



I didn’t take a new screenshot here, but it should say Right.wav. Now hit start in REW and then play the Right file from the USB when you have the waiting for timing reference in REW after it captured the background noise. Next we deselect the Front Right in ATF DSP and select another channel, say the rear right.



We now take a new measurement in REW, we will keep it on Right.wav for all the speakers until the Subwoofers. For that we need to load in Subwoofer.wav file and play the Subwoofer file from the USB.

We repeat measurements for all the speakers. I highly recommend doing them in order and immediately naming them properly in REW.

Now that all the impulse sweeps are done (which can we use to EQ the system with even more precision later on), we can read off the delays directly in REW and input that information straight into ATF DSP PC Tool. It should look like this



As you see from the measurement tabs, the delay is right there for every speaker. I circled 2 of them. Input those delays into ATF DSP tool under the Output tab or the Time tab.

If you like, rerun all the sweeps to confirm alignment and fine tune it more. I didn’t rerun them all, I just wanted to show quickly with 3 speakers



We see the delay is now within 0.01ms using the above technique. This is very good already and like I said above, we need to balance between timing and phase, so any slight discrepancies at this stage are ok. The next post will cover phase alignment, which will be much faster and easier. But before we do that let’s restore our original Virtual to Output Routing since we are finished with running the sweeps.

Attachment 1

[AdonisP91]

Troubleshooting:

If all the measurement delays were negative relative to the Front Left, setting the delays will be easy and fast in the ATF DSP tool. But what if the REW impulse measurements sometimes give negative and sometimes positive numbers? Why are some speaker measurements positive and some negative anyway?

REW is measuring the impulse when it is received at the microphone relative to the reference. If the values in REW are negative, that means the impulse from that speaker is arriving early relative to the reference. If the value is positive, that means the impulse is late. If it was exactly 0, it would be perfectly on time. If an impulse is arriving early, we need to add a delay that speaker in ATF DSP tool. If the impulse is arriving late, we need to decrease the delay relative to the reference.

Wouldn’t a positive delay in REW mean we need to input a negative delay in ATF DSP? How do we do that? It only takes positive values for delays!

Hypothetical: The FR, C, RR, SL, SR, SWR are all negative values in REW, but due to our seating position, the RL and SWL actually measure closer by giving us positive delays. What do we do?

We note that the absolute numbers don’t mean anything, what matters is the relative timing between all the channels. So we start by imputing all the delays that are negative in REW into ATF DSP PC tool. We then click on the Time tab and group all those speakers into a group, say group 1.

Next we look at the values for the RL and SWL. We start by adjusting for the one with the smallest delay. Suppose that is the SWL. We now delay the entire group 1 by the amount of delay REW measured for SWL. Adding a delay to the entire group 1 is like putting in a negative value for the SWL in ATF DSP tool. We now create a new group with all the previous speakers and include the SWL, selecting Group 2. We now delay the entire group 2 by the value from the RL REW measurement MINUS the value from the SWL. We already delayed by the SWL value, we only need to now add the difference to the RL.

Now all the impulses should be properly lined up. Take a screen shot or record all the values for the individual speakers as our baseline for the timing alignment. Rerun all the sweeps if you would like to double check. In the next sections when we adjust the phases small chances might happen to our baseline.

Screenshots to be added later if this is too abstract.

ETA: If the .wav files can’t be played directly from the USB drive, then we need to convert them to FLAC files using VLC. I was just reading the owner manual and it didn’t mention wav files only mp3, mp4, m4b, and aac.

[AdonisP91]

Step #2: Phase Aligning the Pairs

Now we phase align our pairs using the alignment tool built into REW. Select ALL SPL and then select Actions top right.



Then select Alignment tool.



Make sure the Front left and Front right are selected and the phase drop down is selected (Red Circles).



We now use the fine tuning adjustment slider (purple circle) to look for the best overall response curve (green circle). Start by moving it to left extreme and right extreme and watch how the curve changes. We want to try to reduce as many dips as possible, but we also don’t want to have massive changes to the timing (purple circle). Play around, you can even click on invert polarity (purple circle) to find the best overall response curve that still keeps the delay to a minimum. We want SMALL changes not big ones. Here are a few screenshots to show how things are moving. Pay attention to the differences in the Response Curve (green circle):



Edit: For some reason it is displaying the other 2 screenshots at the bottom of the post and my edits can’t fix it.

When you are happy you found the best balance between phase and timing, click on Align Sum on the bottom right of the alignment tool panel. That adds a new measurement and the details there tell you what changes to make to your timing in the ATF DSP tool to match the phases as you want. Rename it to something like Front Pair. Then make the changes in the ATF DSP software.



Repeat this step for each pair, rear, surrounds, and subwoofer. We now have our 4 pairs phase matched. It should look something like the following (remember if the delays recommended are very large something is off, we need an overall balance between timing and phase):



Step #3: Phase Aligning the Groups

Just like we did to phase align the pairs, we need to now phase align the groups. We start by selecting our Front Pair and the Center.



We now enter the Phase alignment tool like we did in the previous stage and match the front pair to the center channel. We align the sum and rename it to something like Group A, or Front Pair + Center Group A.

Now we match this Group A with the rear pair and name the result Group B.



We repeat the same process in the prior step to phase match the Group B (front pair, center, rears pairs) with the surrounds creating a new group C. Rinse and repeat Group C with the Subwoofers. You end up with something like this:



We have now phase and time aligned the entire system. All we need to do is to input that information into ATF DSP PC Tool.

Next we go to the Time tab and will use the group functions here to input the REW information. Save your work previously as we will reuse the group functions here if you previously used it before in the timing alignment step.



We enter the changes recommended for the Center channel. We then group Front Left, Right, and Center in group 1. We change the rears, as say group 2, relative to our group 1. We now make a new group 3, containing all the front, center, and rear speakers. We adjust the surrounds, as group 4, relative to group 3. Finally, we use group 5 to adjust the subwoofers.

The system is now time and phase aligned and is ready for EQ. Final touch ups to timing and phase alignment can be done after the EQ stage. I know was a long post, but once you do it once and understand the process, the tuning becomes very fast. This was more of a crash course how to use REW in conjunction with the ATF software.

The 2 screenshots below are supposed to be above where my red edit is, but the forum isn’t displaying things properly. I don’t know how to fix it.

[BMW DIY Guy]

Originally Posted by IanH

Do you have any tips for the sound deadening process?

I always start with the environment. I would've gone with resonix but decided to go with a newer entry in the 120mil silless.

Here's what I have so far. 120mil dead mat for the sheet metal. Foam to treat over and potentially use as vapor barrier. Speaker rings to isolate the drivers and stop panel buzz. I also have quite a bit of thinsulate 3in batting to fill voids.

Gonna start with the doors and trunk, eventually I'll have to tear out the rear sides because that speaker is buzzing on the driver side. I could ask bmw to fix it but frankly I hate trying to convince dealers to do warranty work on all the small noises I hear.

Longer term will add sub first then look at amp and speakers when you all have paved the way for me

Big fan of Soundskins and done several whole series of videos for installation into BMWs, including my M2. The E53 X5 was the most dramatic improvement (not surprising) but even the G87 showed a reduction plus white noise frequency shift. Absolutely worth it to me doing, even for daily driving let alone setting a better sound stage.

[///MoT]

Originally Posted by kenkamm

All I have to say Tom is, awesome progress!

But are all your EQs going to need to be re-done once you figure out the input EQ issues and get that set up? Certainly the input affects the output, right?

Maybe the input EQ settings don't matter that much - if the output EQ is where it needs to be... but I guess in an ideal world, the DSP would be starting with a "clean slate" (input EQs optimized as much as possible.)

Yes, if I figure out how to fix the input EQ, I would need to re-do the TuneEQ process. But, I was thinking like you alluded to.... as long as the output EQ is where it needs to be with regard to the reference curve, I think that'll be good enough. But, I'm still going to try to correct the input EQ if I can.

I've read thru the rest of your posts here, and just wanted to say you guys are quite a bit more advanced in this stuff than I am lol. I ultimately wanted more clean power (I'm old, and was a gigging musician, so my hearing is less than ideal), and some more tweakability with the sound than I was getting with the head unit, and I'm getting it with the Alpha 1 solution. I don't feel the urge to upgrade anything else in the system. I suppose if I ever get tired of using the long USB cable, I would purchase the WiFi expansion card to add to the DSP, but that would probably be it.

I enjoy learning from you guys!


-Tom

[J-RO]

All those charts and graphs are making my head spin.

[AdonisP91]

Originally Posted by J-RO

All those charts and graphs are making my head spin.

Yeah, I realize it is hard to present it in a write up. Maybe I should have made a video it would have been easier to follow. Most of the charts can be ignored, there are only a few bits of information that are actually needed which I circled in red, but I took screenshots of the entire screen.

[IanH]

Originally Posted by BMW DIY Guy

Big fan of Soundskins and done several whole series of videos for installation into BMWs, including my M2. The E53 X5 was the most dramatic improvement (not surprising) but even the G87 showed a reduction plus white noise frequency shift. Absolutely worth it to me doing, even for daily driving let alone setting a better sound stage.

I hadn't seen this second video. Very helpful thank you sir.

I'm probably going to do selective coverage but I am seeing there is a lot more area than I anticipated. I will post my install when I get around to it.

[kenkamm]

AdonisP91 Thank you SO much for all the work you put into explaining all that. I will be running through it step by step and will be very thankful for it when I get my DSP in the spring.

[AdonisP91]

Originally Posted by kenkamm

AdonisP91 Thank you SO much for all the work you put into explaining all that. I will be running through it step by step and will be very thankful for it when I get my DSP in the spring.

Sounds good, if you do end up giving it a try, let me know if you run into issues or spot mistakes so I can edit and fix the instructions or even improve upon them.

[BroDoze]

Microphone just arrived. It’s 10 degrees here, and the garage isn’t much warmer. Not going to be today!

[///MoT]

Originally Posted by BroDoze

Microphone just arrived. It’s 10 degrees here, and the garage isn’t much warmer. Not going to be today!

Lol, only 15 here and has been in single digits and teens for the last few days, so no further progress made here either

[kenkamm]

Originally Posted by BroDoze

Microphone just arrived. It’s 10 degrees here, and the garage isn’t much warmer. Not going to be today!

Originally Posted by ///MoT

Lol, only 15 here and has been in single digits and teens for the last few days, so no further progress made here either

Same kind of weather here. That's why I figured I would just purchase my DSP in the spring, when I can install it in some warmth. 😭

[BMW DIY Guy]

I hear a Go Fund Me account for a space heater coming up!

I can understand though.. been in the 20s here in Seattle which means the garage is like 35-40 degrees at best.

[BroDoze]

11 degrees right now!

Yes, I should really invest in some sort of garage heater. Or move a couple of my existing space heaters out there. Maybe I'll do that. I'm eager to get going on this stuff!

[///MoT]

Hey guys,

Was perusing ATF's website last night, and for shits and giggles decided to read the individual DSP PC-Tool release notes, and found this interesting info in the "Special Features of Version 5" section:

Advanced Input Signal Analyzer (AISA)

The new Advanced Input Signal Analyzer (AISA) allows to measure the incoming analog input signal in a very fast and precise way. Especially in modern Car Audio Systems the OEM Integration aspect becomes more important than ever. Therefore, our Advanced Input Signal Analyzer is equipped with a wide range of innovative features to make the compensation of OEM processing functions as easy as possible: Factory Equalization can be defeated by using our proprietary TuneEQ algorithm which automatically de-equalizes the incoming signal. Factory time alignment can be compensated with our Delay Finder which semi-automatically compensates the input time delays. Factory phase processing (Allpass Filters) can be compensated semi-automatically by using a peak frequency detector and a simple and easy to use workflow.
Each input can be measured individually as well as the summation of up to four individual input channels. Each correction (EQ, Time & Phase) is analyzed and displayed in real time to make the effectiveness of the adjustment immediately apparent.
To use the AISA, a specified RCA output of the device is converted into a measurement output, which is used to send a signal (max. 1 VRMS) to any universal USB audio interface / USB sound card.

Not sure about the last bit about using a specified RCA output, but I might brave the 10 degree cold this evening in the garage just to try this stuff out. This basically sounds like it will automate the manual process of fixing the input EQ, as well as any time alignment or phase processing coming from the head unit.

These features are probably new enough that there are no RAW-CAt or ResoNix videos to help us out.

Here's a link to the ATF page where I found this info: https://www.audiotec-fischer.de/en/t...ool/downloads/
Scroll down to the "What's new" section, the expand the "Special Feature of Version 5" section.


-Tom

[BroDoze]

Interesting ^^

I was literally looking that download page over last night but seemingly glossed over this section.

If we can get somebody to 100% completely (as best as possible) FIX the input signal, then that should really be shared as a starting place for all of us to work off of.

[AdonisP91]

Originally Posted by ///MoT

Hey guys,

Was perusing ATF's website last night, and for shits and giggles decided to read the individual DSP PC-Tool release notes, and found this interesting info in the "Special Features of Version 5" section:

Advanced Input Signal Analyzer (AISA)

The new Advanced Input Signal Analyzer (AISA) allows to measure the incoming analog input signal in a very fast and precise way. Especially in modern Car Audio Systems the OEM Integration aspect becomes more important than ever. Therefore, our Advanced Input Signal Analyzer is equipped with a wide range of innovative features to make the compensation of OEM processing functions as easy as possible: Factory Equalization can be defeated by using our proprietary TuneEQ algorithm which automatically de-equalizes the incoming signal. Factory time alignment can be compensated with our Delay Finder which semi-automatically compensates the input time delays. Factory phase processing (Allpass Filters) can be compensated semi-automatically by using a peak frequency detector and a simple and easy to use workflow.
Each input can be measured individually as well as the summation of up to four individual input channels. Each correction (EQ, Time & Phase) is analyzed and displayed in real time to make the effectiveness of the adjustment immediately apparent.
To use the AISA, a specified RCA output of the device is converted into a measurement output, which is used to send a signal (max. 1 VRMS) to any universal USB audio interface / USB sound card.

Not sure about the last bit about using a specified RCA output, but I might brave the 10 degree cold this evening in the garage just to try this stuff out. This basically sounds like it will automate the manual process of fixing the input EQ, as well as any time alignment or phase processing coming from the head unit.

These features are probably new enough that there are no RAW-CAt or ResoNix videos to help us out.

Here's a link to the ATF page where I found this info: https://www.audiotec-fischer.de/en/t...ool/downloads/
Scroll down to the "What's new" section, the expand the "Special Feature of Version 5" section.


-Tom

This is a great find Tom. The last part means we can use the Line Out RCA to output a line level signal that we send to our computers. We can then run signal analysis within REW to calibrate our input signals and compare it to an unprocessed signal to make sure the EQ’ing matches everything up. However it is even better than that as their TuneEQ algorithms might already fix everything for us automatically.

I can’t run any input eq tests from the Demo software, but I look forward to tinkering with this or seeing progress from others like you as I wait for my car and equipment.

Also I’ve never done input EQing. I only tuned outputs and time and phase aligned on my home theatre and learned how to do it for the car. I guess it will be fun to learn something new and maybe we can finally fix the problems with the input signal to repurpose our underseat subwoofers into mid-bass woofers.



ETA: I think I figured out how to do the signal analysis in REW to double check all the Input EQ’ing we do in the DSP PC tool. I’m going to order a few adapters and cables from Amazon and then run simulations and tests using my home theatre AVR. It also looks like we can use REW’s EQ wizard to get the perfect filters. I don’t know why I didn’t think of all this before but it makes sense and the task shouldn’t be as daunting as I originally said upthread.

[BroDoze]

Ugh. Got some time tonight with the car (bout froze to death.)

This is the first time I’ve attempted to auto time-align with the PC-Tool, as I just got the mic a few days ago.
Well, it was pretty goofy to get going and the results are not what I expected. It’s almost like my IO routes are messed up or something. When I’m done time-aligning, it appears some of the wrong channels were changed or time-aligned, and some channels didn’t get time-aligned at all, based on default delays still showing on them. Also, the auto time-alignment wants to align the subwoofers that it thinks are G&H when they’re actually I&J - once I re-routed that, it shows the correct channels. However, it never made any adjustment to one of the channels, despite playing/thumping the correct ones and passing the TA process successfully (if it can’t hear a certain channel/speaker, it will fail.) Sorry if this isn’t being explained well.

Does anybody know what the factory crossover points are for the midrange high-pass and the underseat sub lowpass? My PC-Tools shows no highpass filter for the midrange+tweeters - it shows full-range. How can we know what the crossover points are? We need to know these and set them in the PC-Tool, in order to properly EQ each channel.

Kind of a pain in the ass as something is definitely off in my channel assignments/routes. Some things are behaving as expected, some aren’t.

When I was done time-aligning, it sounded off compared to manual measurements. The stage/image was way to the right. Also, it really sounds like the speakers could use some phase alignment with each other.

Also, while still using the subwoofer inputs I completely disabled the low-pass filter and the speakers definitely did not go full range, so they’re getting crossed over (low-pass before the DSP) I did run them up to the maximum 300hz and was able to hear a difference. Again, I’d LOvE to know where these channels are getting crossed over at. I forgot to try running the underseat subs as full range inputs. I’ll try that tomorrow.
Surprisingly, the midranges seem to play down fairly low and have halfway decent mid-bass output. The mid-bass speaker concept may not be needed after-all. I found running the underset subs up to 140/150hz still sounded ok and brought a lot more mid-bass to the stage vs being low-passed at 120hz stock (or wherever the hell they’re crossed at before the DSP.)

I definitely need more time with the car and software. I gotta figure out what’s happening with the auto TA not adjusting certain outputs/channels and adjusted others I had disabled.

Another thing.. the center channel sounds like hell. It does bring more fuller sound, but it screws up the image and you can REALLY hear that sound reflecting off the windshield. I’ve got it disabled. Maybe when I get to the point where I’m able to properly EQ individual channels, it’ll get fixed/toned down, but for now it just adds a ton of mess and reflection to the sound.

[kenkamm]

Originally Posted by BroDoze

Does anybody know what the factory crossover points are for the midrange high-pass and the underseat sub lowpass? My PC-Tools shows no highpass filter for the midrange+tweeters - it shows full-range. How can we know what the crossover points are? We need to know these and set them in the PC-Tool, in order to properly EQ each channel.

Originally Posted by AdonisP91

Here is where it gets interesting. The other forum user reports that all the input signals for the full range speakers look similar like this:

You immediately note there is a built in high pass filter. Similarly, the sub input signals are reported to look like this:

Does this help? AdonisP91 posted this upthread and it shows a -3dB down point of maybe 100Hz for the sub and (ignoring the hump from 160hz to 1khz) somewhere between 100 & 125Hz for the midrange drivers.

[kenkamm]

Originally Posted by BroDoze

Another thing.. the center channel sounds like hell. It does bring more fuller sound, but it screws up the image and you can REALLY hear that sound reflecting off the windshield. I’ve got it disabled. Maybe when I get to the point where I’m able to properly EQ individual channels, it’ll get fixed/toned down, but for now it just adds a ton of mess and reflection to the sound.

Is your goal to get the best possible sound quality, imaging, and staging for yourself as the driver and not worry too much about what everyone else in the car hears? Then the center channel may not be for you.

I believe the goal of a center channel is to attempt to provide a center image for both passenger and driver at the same time which would otherwise be impossible due to time alignment issues inherent in any car due to the various seating positions. But if we then time align all the speakers to the driver, the driver will hear a naturally generated center image - the center channel becomes redundant, and I wonder if it even makes things worse.

[BroDoze]

Originally Posted by kenkamm

Does this help? AdonisP91 posted this upthread and it shows a -3dB down point of maybe 100Hz for the sub and (ignoring the hump from 160hz to 1khz) somewhere between 100 & 125Hz for the midrange drivers.

Perfect - I need to upread parts of the thread, again, apparently - thanks. And I think you’re right - given a 24db rolloff, loos to be a 100hz lo-pass & 120hz hi-pass. The hump masses with it. I will run my underseat lo-pass up to at least 120.

Originally Posted by kenkamm

Is your goal to get the best possible sound quality, imaging, and staging for yourself as the driver and not worry too much about what everyone else in the car hears?

… the center channel becomes redundant, and I wonder if it even makes things worse.

I could give a rip about any passengers. Looking to have best SQ for my ears, in my specific seating position. I’ve never needed / wanted a center in a vehicle and this one reinforced my reasons.