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

ALL CyberTruck discussion

This site may earn commission on affiliate links.
Based on this, I’m guessing no other vehicle does speaker and microphone based active noise cancellation since nothing else has as high a communications bandwidth.
The Model S&X currently have active noise cancellation. I think the microphones are only in the front headrests. But they likely don't use CAN, they may be wired directly to the head unit. (I think it was rumored that there are sensors on the struts that feed into the system as well.)
 
  • Like
Reactions: mongo
The Model S&X currently have active noise cancellation. I think the microphones are only in the front headrests. But they likely don't use CAN, they may be wired directly to the head unit. (I think it was rumored that there are sensors on the struts that feed into the system as well.)
I have it in a MS Plaid, and it is real junk (all FSDj kidding aside. Turned it off long ago. Perhaps @WilliamG can chime in on his experience?
 
At first I thought they were moving all the DACs/ADCs all the way out to where the actual speakers/microphones were, but it sounds like (pun not intended) that instead they moved from the ICE unit to the nearest etherloop stop, as for example the left front body unit that was shown, which seemed to have a bunch of left speakers, I am guessing the left front speakers, then right front on the right front body unit, similarly someplace one or more rear body units that handle ADC/DACs for that area of the vehicle.

So we have not totally gotten rid of legacy wiring in favor of "Etherloop" but realistically it is the sane balance. Running the network all the way to the speakers/microphones would actually require more wiring and complexity than this model where it runs to nearest module, then is broken out more traditionally, since from what they said about using a normal ethernet connection on the body module they were showing the PCB for, it seems likely they're using 8-conductor ethernet wiring to handle gigabit.

Better to have just two or three wires for the speakers or microphones than 8 wires plus power running to them, after all (or 8 wires and the added complexity of PoE). Similarly, rather than running power plus 8 wires to door motors and such, just run the few wires actually needed (2 or 3). And since we're operating at 48V (or 24V apparently for some things), wiring is smaller than it was for 12V for these motor loads, so you still have less wiring than before, even with the "last mile" being more traditional rather than "Etherloop".
 
  • Like
Reactions: MP3Mike
There’s a cost benefit analysis between wires, routing, assembly, multiple mini-amplifiers, etc. that justifies why Tesla mechanized the wiring and circuitry the way it did.

We’ll just never see it. Until we see it torn down and analyzed by Munro or explained by Tesla, we won’t know exactly how it’s done.
 
I wonder where this 48V battery is? I haven’t seen it in any of the videos.

Doesn’t seem like it would be good design necessarily since you’d only be able to do it for a limited time, but a 48V battery could provide plenty of 48V surge power.

Has anyone heard about implementation other than my rambling (likely incorrect) about it being potentially 22 4680 cells 11s2p (this was only to make the 1366 number make sense, could be a garbage number, and it only works out to 40V nominal (46V max) too, not to mention various other issues)? Anyway 2kWh of lithium-ion cells would have plenty of surge power.
As @verygreen reported, the part catalog is available. I looked for the 48V - it is under the dash.


It’s called the MV system (mid voltage) not LV.


IMG_9748.jpeg



Has an interior radar. I think that was sort of known.

Has a super horn!

Tires listed for 20” are 285/65R20 Goodyear Wrangler Territory R/T (Rugged Terrain)

Fascia Camera Nozzle for front camera as previously reported - nothing anywhere else.

PCS looks like 48A
 
Last edited:
At first I thought they were moving all the DACs/ADCs all the way out to where the actual speakers/microphones were, but it sounds like (pun not intended) that instead they moved from the ICE unit to the nearest etherloop stop, as for example the left front body unit that was shown, which seemed to have a bunch of left speakers, I am guessing the left front speakers, then right front on the right front body unit, similarly someplace one or more rear body units that handle ADC/DACs for that area of the vehicle.

So we have not totally gotten rid of legacy wiring in favor of "Etherloop" but realistically it is the sane balance. Running the network all the way to the speakers/microphones would actually require more wiring and complexity than this model where it runs to nearest module, then is broken out more traditionally, since from what they said about using a normal ethernet connection on the body module they were showing the PCB for, it seems likely they're using 8-conductor ethernet wiring to handle gigabit.

Better to have just two or three wires for the speakers or microphones than 8 wires plus power running to them, after all (or 8 wires and the added complexity of PoE). Similarly, rather than running power plus 8 wires to door motors and such, just run the few wires actually needed (2 or 3). And since we're operating at 48V (or 24V apparently for some things), wiring is smaller than it was for 12V for these motor loads, so you still have less wiring than before, even with the "last mile" being more traditional rather than "Etherloop".
Mostly.
The Etherloop isn't standard four pair, but rather automotive single pair per link. Based on color and shape it looks like it may be TE brand connectors.
 
Are you sure? I'm pretty sure in that Munro video they called out using a standard connector. Plus getting gigabit on single pair is not easy.
Yes, a standard automotive connector. There are around a dozen "standard" ethernet connectors depending on the application. And yes, gigabit on a single twisted pair is very much doable and is done.

Edit: BTW, the reason you can do 1 gbps on 1 twisted pair is that you don't have to run it 330' which is what "normal" Ethernet is specified at. Since vehicle lengths are so much shorter, they can up the data rate on a smaller number of pairs. Also, using shielded twisted pair helps too.


 
Last edited:
Are you sure? I'm pretty sure in that Munro video they called out using a standard connector. Plus getting gigabit on single pair is not easy.
Yeah, I rewatched that part a bunch to verify before posting.
Sandy said thst does not look like an ethernet cable and Pete said it's shielded twisted pair ethernet. 1000Base-T1 is an automotive standard that provides full duplex on a single pair.
 
Yeah, I rewatched that part a bunch to verify before posting.
Sandy said thst does not look like an ethernet cable and Pete said it's shielded twisted pair ethernet. 1000Base-T1 is an automotive standard that provides full duplex on a single pair.
Yeah, I wondered if Sandy too was thinking standard CAT 5/5e/6 Ethernet cable/connector, which Tesla is obviously not using.
 
  • Like
Reactions: JusRelax and mongo
Yeah, I rewatched that part a bunch to verify before posting.
Sandy said thst does not look like an ethernet cable and Pete said it's shielded twisted pair ethernet. 1000Base-T1 is an automotive standard that provides full duplex on a single pair.

Amazing really. Top execs at Tesla and at least come across as nice, down-to-earth and easy to talk to people.
 
  • Like
Reactions: JusRelax and jebinc
There is a central head media unit. The difference is that rather than having it direct wired to all speakers and/or amps and microphones, it links via Etherloop to local nodes with the speaker amps and DACs along with microphone amps/ ADCs. Toslink on steroids, if you will (copper vs optical, of course).
With deterministic jitter and latency, the nodes can include a fixed delay to resync the audio outputs, that's the easy part. The more performance requiring aspect is doing:
Microphone(s) -> ADC -> network -> headUnit -> processing -> network -> DAC -> speakers
Fast enough for noise canceling which is dependent on signal phasing of the live noise. Of course, the only new sections are the network transport ones. Signal processing and digital<->analog exist in both implementations.

For reference, on playback 1 uS at 20kHz is 7.2 degrees of phase shift. If noise cancelation is bounded to lower frequencies, that makes it easier, 1 uS at 10kHz is 3.6 degrees.
Super-informative, mongo! I hope the SQ is important. Regardless of what many say, car audio quality is rarely outstanding, when compared to a reasonable mid-fi or true hi-fi home system. And I don’t mean analog LPs and tubes, but a quality digital system and good speakers
 
  • Like
Reactions: jerry33 and jebinc
Lol, tubes. I came across a box of vacuum tubes and packages of ancient carbon resistors and caps in 50s to early 60s envelopes and showed a few tubes to some colleagues asking “what do you think this is?” and they all said “a light bulb?”.

Granted, I could have shown them an 8-track tape or a laserdisc and got the same right, but wrong response, so I won’t judge. I never learned of tubes in school, either. They were not even mentioned, but I think magnetic core memory was.
 
There is a central head media unit. The difference is that rather than having it direct wired to all speakers and/or amps and microphones, it links via Etherloop to local nodes with the speaker amps and DACs along with microphone amps/ ADCs. Toslink on steroids, if you will (copper vs optical, of course).
With deterministic jitter and latency, the nodes can include a fixed delay to resync the audio outputs, that's the easy part. The more performance requiring aspect is doing:
Microphone(s) -> ADC -> network -> headUnit -> processing -> network -> DAC -> speakers
Fast enough for noise canceling which is dependent on signal phasing of the live noise. Of course, the only new sections are the network transport ones. Signal processing and digital<->analog exist in both implementations.

For reference, on playback 1 uS at 20kHz is 7.2 degrees of phase shift. If noise cancelation is bounded to lower frequencies, that makes it easier, 1 uS at 10kHz is 3.6 degrees.
I didn't understand all that, but thank you nonetheless
 
  • Funny
Reactions: navguy12 and jebinc
I didn't understand all that, but thank you nonetheless
I understood most of it, but I didn't understand this part:

For reference, on playback 1 uS at 20kHz is 7.2 degrees of phase shift. If noise cancelation is bounded to lower frequencies, that makes it easier, 1 uS at 10kHz is 3.6 degrees.

@mongo ? What's the significance of this?
 
I understood most of it, but I didn't understand this part:



@mongo ? What's the significance of this?


Noise cancelation involves using the speakers to generate an out of phase (180 degree) version of the sound. If the phasing is closer to 90 degrees, it adds to it instead. So that sets a limit on processing latency.

However, most noise is repetitive at the cycle level so one can likely shift by multiples of a full cycle and still be effective as long as the latency is constant. Though lag does mean it will be less responsive to changes in noise spectrum. This would also requires a linear phase shift vs frequency to back out the latency.

At 10kHz, the waveform repeats every 100 uS. So 50uS of jitter would make the system a noise adder instead of a noise canceler.