Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

MCU2 Upgrade Audio EQ Balancing - Coming From MCU1

This site may earn commission on affiliate links.
Background as to why I did this project:

I own an October 2016 model S 90D. This was one of the first autopilot two cars made (which I paid for), ordered with the premium sound upgrade, and came with MCU 1. When I first took delivery of the car, I thought the sound system was excellent and much better than most I had experienced in a factory car (all much lesser premium vehicles). Over the years, with ongoing software updates, that sound changed and became more limited, but I was always able to retune the system back to sounding very good. I have never been unhappy with the sound system.

Then came my decision to upgrade to MCU 2. With all the great benefits to the new MCU, the one detriment I noticed immediately was the sound from the system. It was drastically different than MCU 1. My equalizer settings from MCU 1 carried over to MCU 2, but they sounded like crap. The bass was overwhelming; the vocals were harsh; it was just not well balanced. I have always considered myself to be a "budget audiophile." I can appreciate when an audio system is well-balanced, but I am not sure I could tell you a $2,000 amp sounds that much worse than a $20,000 amp.

With MCU 2, I spent a lot of time messing with the EQ settings trying to get something close to what MCU 1 sounded like. I could tell things were still wrong, but I was struggling with manual adjustments. Even bringing out my trusty 25-year-old RadioShack SPL meter and recordings of my Sheffield/A2TB Test CD was only partially helpful. I needed better data, so I decided to make some objective measurements.

Equipment used:
  • Rode NT1A microphone mounted at head-height between the front seat headrests.
  • Focusrite audio interface connected to my laptop
  • Audio samples were recorded directly into Adobe Audition software, and then frequency analysis grafts were generated by Audition.
  • MP3 recording of track 57 on the Sheffield/A2TB Test CD (320 kbps) played from a USB drive. This track is a 20 Hz to 20 kHz continuous warble designed to evaluate the overall smoothness of the acoustic response in a listening area. The sweep is warbled to eliminate the standing waves that would normally occur with a simple swept tone. This is essential when evaluating the frequency response of a car stereo within a very enclosed listening environment where standing waves reflecting off glass are all over the place.

What I did:

I imported the test MP3 into Adobe Audition and used this as a baseline to compare recorded samples obtained in the car. This test sample maintains a constant volume across the frequency spectrum, and therefore the dB grafted at the low-end are much greater than those produced at the high end of the spectrum. Human hearing perceives high-frequency sounds to be much louder than lower frequency sounds produced at the same dB level. Nonetheless, I am using a perfectly recorded sample as the baseline audio file and directly comparing the playing of the sample recorded directly in the car from the sound system.

The sound samples recorded in the car are then overlaid on top of this baseline to show the relative dB difference throughout the frequency spectrum. A perfect sound system in a perfect listening environment with perfect recording equipment should produce an identical graft to the test MP3 sample.

There are several limitations to my set up.
  • First, I am not an audio engineer. This was a makeshift recording set up using equipment that I have available. I recorded with all the windows up in the sealed car in a quiet garage. I was seated in the driver seat of the car with my laptop. Climate control was off. The laptop fan was running, but the base recording audio level was at about -45 dB. Recorded samples peaked at about -5 dB.
  • The Rode NT1A microphone is known to be a particularly good basic studio recording mic, but it has its own frequency response curve that is well documented. This frequency response will taint recordings and must be taken into consideration when interpreting the recorded samples. It is also a directional mic with a cardioid polar pattern. I faced the mic pointing forward just behind and between the front seat headrests. The balance of the sound system was set at the center default point. Most of the sound in the Tesla system is produced by the front speakers, with rear speakers producing more fill and obviously the subwoofer base. Most of that sound reflects off of multiple surfaces before reaching the driver seat. Although the mic pointed away from the rear speakers, I believe the recordings reproduce a good sample of what is heard by front-seat passengers.

The Rode NT1A microphone response curve:

M0eRxwS.jpg


And now the data:

First, I recorded the sound system with all EQ settings at zero. The graph shows the baseline MP3 test file in orange versus the recorded sample in blue. The positioning of the baseline is somewhat arbitrary, and simply what I chose to be a good neutral target to allow subsequent adjustments to the equalizer bands. This baseline could easily be moved up or down and was just a choice I made looking at the initial frequency response graft. It is also important to note that the graft is a plot of the relative dB change on the X-axis versus the logarithmic scale of frequency on the Y-axis. The zero point on the dB axis is also placed arbitrarily and is only a reference point. Each horizontal line on the graph represents a 2 dB change in volume. Click on any image for full-size versions to read more carefully.

DAPM7Uj.jpg


The results in this graft show a huge dB boost in frequencies below about 1500 Hz. There is also a significant bump in the 3500 to 6500 Hz range. Somewhat surprising is a very steep drop off after about 14,000 Hz. However, the human ear in a moving car is going to have great difficulty in appreciating anything above that frequency.

Next, I needed to find out where each of the 5 EQ adjustments affects the overall frequency curves. I went through the process of recording five samples, each with individual EQ bands set at "-8" (slider fully down). I will talk about the EQ bands numbered from one through five, left to right on the settings page, from bass to treble, respectively. Each graph has the baseline target again shown in orange, with the recording of the default EQ setting in blue, and the "-8" EQ setting is in green. I am only altering one EQ slider at a time, with the rest kept at the zero default. The green line should, therefore, mimic the blue line (within sampling error) with the deviation only where the targeted EQ slider is affecting frequencies.

EQ slider one:

p36lYaO.jpg


The slider begins to affect frequencies below 600 Hz to frequencies off the graph to the left, well below 100 Hz. I suspect the peak influence is off the graft below 100 Hz.

EQ slider two:

kSLuKdc.jpg


The slider begins to affect frequencies at about 800 Hz with a maximal deflection around 450 Hz but continues to affect even well below 100 Hz. There is a crossover with slider one in the lower frequencies. This EQ band is well within the human singing voice.

EQ slider three:

kmP8Ye9.jpg


The slider has a frequency range beginning at 2000 Hz down to about 400 Hz with maximum effect at about 900 Hz. The band will again significantly affect human female singing voices, especially high soprano.

EQ slider four:

5NdCHOS.jpg


The slider has effects on frequencies between about 1.5 kHz and 6.5 kHz, with maximal effects centered around 3 kHz. This is well within the range of hearing the crack of consonants in speech and "S" sounds.

EQ slider five:

IzWHCc3.jpg


The slider has effects between 4 kHz all the way through the very higher frequencies with a maximal broad impact between about 6 kHz to 12 kHz. This is the range of "airiness" or "breathiness" in recordings and the high-pitched crack of symbols.

Now to try and make it sound better:

So now what to do with all this information? Unfortunately, Tesla has chosen only to give us five frequency bands to play with. These bands do have some crossover at certain points, but some of them are much broader frequency ranges than I would like. I miss the eight-band equalizer that was present on my model S when I first bought it. Why did they get rid of this? But we are left with what we have, and I set forth in trying to do the best I could with what is available. I ended up with the following changes to the five equalizer bands.

Vn3mn0d.jpg


And here is the final frequency curve that I was able to generate with the above settings.

G9AKMZG.jpg


It was the best compromise. I left the base frequencies slightly pushed since car stereos tend to need extra base to overcome road noise. I wish I could have pushed the 2-3 kHz band higher, but unfortunately, this lies at the crossover between slider three and slider four. Pushing slider four higher ends up producing undesirable increases at the 4 kHz to 6 kHz range. Pushing slider three higher significantly affects female singing voices, and is the reason I initially thought voices sounded so harsh. After listening to many different types of music, I am happy with the results, and the system sounds infinitely better than when I started.

I hope this might give some general starting points for those in a similar situation. Ultimately, EQ choices are based on your individual ear and how you like your music to sound. My settings are for my desire to have a very flat response out of the system. I listen to a vast variety of music, everything from classical, the 1930s and 40s standards, jazz, current and classic alternative rock, and on the rare occasion, even modern pop music (I do have teenage daughters). I am happy with this baseline EQ setting but may overtime tweak it slightly.

I have no idea if these settings crossover from my October 2016 model S to the current model S, X, or even 3 and Y. I am not sure if speaker manufacturers used by Tesla have changed over the years, and I suspect the 3 and Y are entirely different. The settings are probably utterly wrong if you do not have the premium sound option. I would love to hear what others have come up with within their own experimentation.

This is fantastic stuff! Thank you for all of this work. I currently have the MCU1 in my 2014 and am planning my upgrade. I have struggled with the EQ and came up with a profile that works for me currently. But it took forever to get to it. Your work will help me get a good starting point. And since I only know the basics of what you have explained, I expect it will sound better than my fumbling.

Much appreciated
 
Someone on the official Tesla forum said they had the problem and the service center resolved it by,
"disconnecting and reconnecting all the connections". They don't know why it happened or got fixed,
but it did for that customer. Anyone else having the issue please ask SC to do this and report back if
it worked.
 

I have this issue even with MCU1. What I realized is that when you slide the fader all the way to the rear, its only driving the two speakers that are in the tailgate, which have no bass. If you put the fader halfway between center and rear, you'll drive the rear speakers in the doors and the woofer, and you will hear bass from the rear. It's an odd setup.
 
  • Informative
Reactions: smilepak
Someone on the official Tesla forum said they had the problem and the service center resolved it by,
"disconnecting and reconnecting all the connections". They don't know why it happened or got fixed,
but it did for that customer. Anyone else having the issue please ask SC to do this and report back if
it worked.
JImHarris do you have a link to the forum you are referring to?
 
I think this is it:
2020.32.3 killed rear audio/bass

And @Joelgjr , that doesn't work if you have standard sound system. There is no sub in the rear. And I tried what you said and you just get a little more volume and bass out of the 2 rear door speakers, not much more. Put on a heavy bass song and put your hand on the front doors speakers and then the rear doors speakers, no matter where you slide it, you will never feel bass coming out of the rear doors and volume will always be way less. If you think you are feeling or hearing bass from the rear doors, you are just feeling the vibration from the front doors a few feet away.

P.S. I just got a software update an hour ago, checked the stereo system. Still no bass and 1/2 volume from rear door speakers.
 
I think this is it:
2020.32.3 killed rear audio/bass

And @Joelgjr , that doesn't work if you have standard sound system. There is no sub in the rear. And I tried what you said and you just get a little more volume and bass out of the 2 rear door speakers, not much more. Put on a heavy bass song and put your hand on the front doors speakers and then the rear doors speakers, no matter where you slide it, you will never feel bass coming out of the rear doors and volume will always be way less. If you think you are feeling or hearing bass from the rear doors, you are just feeling the vibration from the front doors a few feet away.

P.S. I just got a software update an hour ago, checked the stereo system. Still no bass and 1/2 volume from rear door speakers.

hmm - my car, with premium sound, has been like this since I got it from the factory in 2018. never occurred to me something was broken because when fader is centered it sounds good
 
  • Funny
Reactions: drew.c
With all these complaints, I guess I should consider myself fortunate with a nice sounding LH speakers and AMP running on an MCU1 setup. I think I'll replace just the daughtercard and stick with MCU1 when the time comes!

Yes, do that. DO NOT "upgrade" to MCU2 if you value the quality of your audio. I am kicking myself. I never gave it a second thought that the awesome sound would take a step backwards.

I HATE, HATE, HATE it.

I am not one of those audiopile types who can tell you difference in sound based on the color of your speaker wire, but I ABSOLUTELY notice a considerable step backwards in the imaging and overall quality. To me, it's almost as if they chose 8 bit DAC's to save a few pennies.

(did I mention I hate it?)

I do thank the OP for the time and analysis, and my EQ settings just happen to be very close to his based on trial an error on my part - nice to see some science backing my ear. But EQ settings alone will not fix this issue. I feel like a fool for complaining to Tesla - A broken headlight is one thing - but "my stereo sounds bad" is so subjective - yet I *know* it is different(worse).
 
So i found this thread a while ago, all of which describes my thoughts exactly on the sound quality with the new MCU 2. I've been playing with my tone settings for a long time but still couldn't get it to sound right. it always sounded... muffled and not crisp at all.

I one thing i learned in this thread is that the rear speakers are actually driven from the MCU and not through the amp! this got me thinking...
If i put the balance all the way forward, the muffled sound doesn't seem so bad anymore, though overall more flat being just the front speakers. Did the same with the back and got same results. So for the last few days I've been only using the front and it's overall better (but not ideal).
So, here's my guess as to what's going on. All amps will cause a very slight delay in its audio output. just how they work. thus is not a problem if all speakers have the same delay. With MCU1 there's probably a software delay in the rear output to make sure the audio in the back speakers line up with the slightly delayed front speakers. With MCU2, that delay was forgot/removed and thus there's now a desync in timing between front and back which produces a very slight echo of sorts which is muddling up the sound quality.
I just now set the car to about 80% rear and 20% front balance, and with the rear speakers louder than the front i can somewhat hear a sort of echo effect going on, but could be just my ears.
If anybody else wants to test this theory? Be interesting to get a laptop and try and measure the difference somehow.
 
  • Like
  • Informative
Reactions: Paul42 and Hank42
I am happy that I am not the only one noticing the difference. I have a 2017 S with premium audio. Since the upgrade to MCU2, indeed I experience less bass in the rear speakers. The car went in for service yesterday and Tesla claims it is perfectly ok. (Their test being that they got into another car and it sounded exactly the same like mine. That is about the worst possible way to diagnose a system. I know a lot of people that cannot tell a good red wine from a crap one. To them it all tastes the same. It is the same with sound. If you are not experienced or educated, it'll all sound the same. I would be very surprised -and impressed- if the technician who listened to my car is educated and experienced in recognising true quality sound...)
So there I am. 2770 €'s down and a crap sound system in my car, with Tesla claiming it is just fine.
Therefore I am indeed looking for people who can measure an MCU1, premium audio Model S and do the same with an MCU2 upgrade. The measuring should be done in the center of the rear seat with audio settings in neutral: flat equaliser and of course Dolby off. Set the volume at let's say 5 and fader fully set to rear speakers. Then test signals of different frequencies should be played and measured with a Decibel meter. The exact same procedure should be done to a car that is upgraded to MCU2. That is the only way to prove that a change occured. It can than be presented to Tesla with the demand to put things back the way they were. It is what we chose when we bought the car and they have no right taking it away from us. (Same applies for taking out the radio on MCU2.)
Finally a question: what does UHFS stand for?
 
  • Like
Reactions: Hank42
So there I am. 2770 €'s down and a crap sound system in my car, with Tesla claiming it is just fine.Therefore I am indeed looking for people who can measure an MCU1, premium audio Model S and do the same with an MCU2 upgrade. The measuring should be done in the center of the rear seat with audio settings in neutral: flat equaliser and of course Dolby off. Set the volume at let's say 5 and fader fully set to rear speakers. Then test signals of different frequencies should be played and measured with a Decibel meter. The exact same procedure should be done to a car that is upgraded to MCU2. That is the only way to prove that a change occured. It can than be presented to Tesla with the demand to put things back the way they were. It is what we chose when we bought the car and they have no right taking it away from us. (Same applies for taking out the radio on MCU2.)
Finally a question: what does UHFS stand for?

Ultra-High Fidelity Sound.
 
  • Informative
Reactions: Hank42
In follow up to my previous post on rear speakers being out of sync with the front, I grabbed a laptop with audicity to have it play a click track and record what the microphone hears. I recorded one with the front speakers only, and a second with the rear speakers only, Here's the results I got. Note, this was done in an isolated garage with all AC off.

Top line is the click. Next one down is the Front Speakers only. Bottom is the Rear Speakers only.
audiolag.PNG
view


Upon Zooming in:
audiolag3.PNG
view


Looking at the data, the rear speakers actually lag behind the front speakers by ~.012 seconds. Opposite of my first prediction, but they're still different. 1 / .012 = 83.3 Hz. I know very little bit about frequency analysis, but enough to guess that any harmonics of 83.3 will come through fine since the waves will align up, but any frequencies outside that will interfere with itself between the front and back and will disrupt the audio quality. Course, perhaps just the slight echo is annoying enough for our ears. I'll have to do some more tests on if a .012 delay makes a big difference on just how things sound. I.E. I think I'll take two tracks of the same song and shift one that much and see how it sounds, if I notice a difference. But that's for another day, too busy for now.

MCU2 Audio - Google Drive
Here's the link to the drive if the pictures don't show up here.
 
In follow up to my previous post on rear speakers being out of sync with the front, I grabbed a laptop with audicity to have it play a click track and record what the microphone hears. I recorded one with the front speakers only, and a second with the rear speakers only, Here's the results I got....
Looking at the data, the rear speakers actually lag behind the front speakers by ~.012 seconds.
Since the speed of sound is 1125 ft/s, 0.012 seconds is approximately the time is takes sound to travel 13 ft. That is not very significant. There is a greater difference in timing for different instruments in an orchestra to get to the listener's ear. If that 12 thousands of a second was significant, people would have been complaining about the problem long before you.
 
Since the speed of sound is 1125 ft/s, 0.012 seconds is approximately the time is takes sound to travel 13 ft. That is not very significant. There is a greater difference in timing for different instruments in an orchestra to get to the listener's ear. If that 12 thousands of a second was significant, people would have been complaining about the problem long before you.

That's a neat way to think about it, just 13 ft different.
And I see what you're saying, but each instrument in an orchestra plays something completely different from the others. There is a reason all the trumpets sit next to each other and not far apart. And yeah, there may be a slight delay, but the sound each instrument plays is very unique to the other, so they blend in very well. Most live theaters are designed to actually direct the sound off the roof towards the audience, not directly from the instruments to the ears.

Biggest thing with orchestras is there is no speech. No talking involved. Speech is where there is so much going on so fast that a .012 delay does make a difference. I do notice in my car that lyrics are much harder to hear and understand after the MCU upgrade.

Anyway, biggest difference between our cars and a live orchestra is the fact that the rear speakers and the front speakers in the car are playing the exact identical audio track. And have you ever sat down in front of a TV with one speaker sitting right next to you and the other 13 feet away? generally not, cuz it would sound terrible. you want them on either side of the TV equally spaced to keep them balanced, both volume but also for the delay. Most high end home theater systems have a balancing they do when you set it up, for that reason. Though again, people always try to sit right in the middle of the 4+ speakers; not have one 1 foot away and another 13 ft away.

I could say more examples, but better hear for yourself. I grabbed a vocal speech track off the internet and combined it with itself but with that same .012 delay. It sounds robotic and not even close to the same as the original. That's what I believe is going on with our cars with the upgrade. I could be wrong, but this thread shows that something is wrong and I'm hoping to just point tesla in the right direction to fixing our audio.

ORIGINAL track
DELAYED track

(I don't post much, so hopefully those links work)
 
I grabbed a vocal speech track off the internet and combined it with itself but with that same .012 delay. It sounds robotic and not even close to the same as the original. That's what I believe is going on with our cars with the upgrade. I could be wrong, but this thread shows that something is wrong and I'm hoping to just point tesla in the right direction to fixing our audio.

ORIGINAL track
DELAYED track
You have very effectively proved the point that the 12 millisecond delay does indeed degrade the sound experience, so I retract my argument to the contrary. The delay is effectively functioning as a reverb. I highly respect someone who actually presents actual evidence over opinion.

Now I should ask, what were the Immersive Sound settings inside of the Tesla when you measured the 12 millisecond delay? And how much does changing it affect the measured delay?
 
  • Like
Reactions: Guy V
What's frustrating is that speaker delay is such an easy thing to fix. This is basic DSP stuff in creating a proper soundstage.

And yes, the human ear is very capable of perceiving a 12 milliseconds difference in sound signal. The human ear/brain can localize sound direction with an interaural time difference down to about 10 microseconds depending on the sound's frequency. The most sensitive is at the frequencies of the human voice. It makes sense that a poorly balanced sound system is most evident with muddy vocals.
 
  • Informative
Reactions: Cheburashka
Now, the real question that I don't have the answer for... how do we Tesla's attention with this? The EQ balancing as well as proper DSP timing. Clearly the settings between what we had on MCU1 and what's being used on MCU2 are different and has affected the sound quality a lot. I might start a new post here to highlight it better and hope somebody at Tesla sees it, but that's a very low chance to begin with. Might bug my local service center too. Any suggestions appreciated.

Knowing the fact that since V9 it seems anything related to the Model S/X in terms of UI (and thus, Audio quality) are pretty much last priority with Tesla. My gut says we won't ever see a fix since we're a small amount of folks with older cars. I'm sure the original testing of the MCU upgrade went like this: "does it play audio? yes? ok check it off the list". Sucks the premium audio I paid for back in the day lost all it's value and lowered the enjoyment of my car significantly with the upgrade. Still torn if the faster UI is a good trade off.

Worse case if they never do, I'll probably look into some sort of pre-amp that I'll wire before the premium amp that will add a delay back into the front speakers to balance it back out.