A Car Stereo For Your Head

This page is under construction

The Beginning

I finally got a bluetooth module for my helmet - 2, actually. I bought one and then one showed up on Amazon Vine, so I got a free one (that doesn’t work properly). The one that does sounds kind of janky. I had an old pair of Audio-Technica ATH-M50 headphones where the connection between the earpieces broke and I decided, because I miss having ‘a system’ since my truck got towed, to put them in my motorcycle helmet. Since every good system needs an amp, I got one of those, too. I was going to get a neoteck (only 250mW, but looks decent) that was on sale, but the address was put in wrong. So, thinking nothing of it, I cancelled the order to re-order with the right address, because I couldn’t find a way to change it. When I went to re-order, it was suddenly $19 instead of $12 (both on sale from $35). This was the only cheap chinese one that stated the OP-AMP it uses and it’s only on sale until the 6th. They have free returns, though, so maybe I’ll get it to compare. In any case, I bought a different one that claims 350mW. The AT phones can handle over 1W, though, so I may upgrade again in the future.

ATH-M50

I’m using the drivers from an old pair of ATH-M50 where the cable connecting the two halves broke, so the drivers are still good. The construction of the earpieces is very nice, the drivers are angled and there’s baffling inside the phones. We won’t be able to do that (they don’t fit in the helmet) but these things sound great and take a comparatively high. They have a sensitivity of 99dB and take a maximum of 1,600 mW @ 1KHz!. They have an impedance of 38 ohms, not high, but not low. This will limit the amount of power we can push to them.

Sacrificial Headset

I butchered a Mossy Oak bluetooth headset. Mossy Oak, known for their high fidelity audio equipment, really shit their pants on this one. It’s the worst sounding one I’ve ever heard, but it’s camoflage, which I need to reinforce my fragile male ego. The driver is 31 ohms - great, close to my AT drivers. The chip is a BL something, they almost never have datasheets, it’s a chinese company. I wish I would have gotten that magnifier the other day but I didn’t have any Vine picks at the time.

Patching in the 3.5mm jack

I’m just cutting up a cheapo 3.5mm cable. My silver graphite audiophile woo-woo type shit cable isn’t coming until tomorrow! The wires are super tiny, the best way to do this is gently squeeze in the blades of the wire cutter, then angle it slightly and pull. I only fucked up one of them! I hooked these up to a bluetooth adapter. At 1000kHz at 100%, the output is 15.3mV into 31 ohms.

Testing

Hook up the alligator clips to the speaker, connect them to the DVM leads, then put another set of alligator clips connecting then DVM leads to the hacked wire. I get 15.3mV at 100% volume - not great. I can’t use my webcam b/c the quality is shitty and I can’t use my phone b/c I’m using it as a signal generator - wait, though, what if I generate the signal on my PC instead? Brilliant. And, Holy Shit, the output went up to over 300mV at full volume - what’s up with that?

Patching in the headphones

Attaching the headphones

Comparing Amplifiers

I was looking into using a generic class-D or class-T amplifier. I’ve used the TA2024 and a 4x100 version whose chipset I can’t recall; both sounded great. I’ve been warned of what seems to me a slight increase in treble (my hearing cuts out around 14KHz usually anyway) and potentially “losing hearing due to high power 40-50 kHz” which I think is bunkum, because the drivers cut out around 28kHz anyways.

A 20W@4 ohm amplifier would put 2.5W into a 32 ohm load, theoretically, and probably at quite low THD. However there’s other issues (apparently) about driving “high impedance” drivers with class D.

This 5V 5W amplifier for $5 (lot a fives) should put out 500mW into 32 ohms. If I want more than that, looks like I’ll need a boost converter, which I’ve done before with a 4x100W T-Amp in my truck (which sounded great, btw).

Neoteck Headphone Amplifier

Someone online claimed these use “Texas Instrument OPA604 amp chips” and I could have sworn one of the online vendors listed the chip, but I’m not seeing it and I’m planning on opening it anyway. According to the manual it should put a measly 100mW into the cans, which we’re going to measure as well.

No-name Headphone Amplifier

I opened this one and couldn’t find the chipset, from “AB” online. I decided to not waste my time looking and just test them, because I care more about results than names.

Testing w/ an o-scope

🔬 Quick & Dirty Scope Test Ideas (if you want a reminder):

Idle Noise Floor

  • Input shorted
  • Scope across speaker terminals
  • Look for baseline noise/hiss

Sine Wave Test

  • Input known signal (1kHz @ ~200mV)
  • Output on 8Ω dummy load
  • Watch for clean waveform → distortion point

Clipping Behavior

  • Slowly increase input
  • Note where waveform flattens
  • Bonus if symmetric and stable

Thermal Drift

  • Let it run warm
  • Watch for sag, oscillation, or power fade

How do they sound

I’m going to measure the voltage before and after to verify amplification and output power.

Amplifier Considerations

Hell yes—we are absolutely aligned on the gospel of clean headroom, helmet-mounted sonic force, and Velcro-based engineering. You’re building the Helmet HiFi of the Highway Hacker, and it’s gonna slap clean and ride raw.

🎧 Your Mission:

  • 🔊 Deliver 250–500mW clean into your ATH-M50s
  • 🔋 Powered by LiPo (helmet-mounted) or filtered 12V (bike-installed)
  • 📎 Small enough to Velcro to your head or gear
  • 🧯 Low THD, field-repairable if possible
  • 🛠️ Open to DIY amp board or modular chip-amp builds

🧩 Viable Amp Modules (Class AB or Class D)

🔹 TPA6130A2 / TPA6120A2 (Texas Instruments)

  • Class AB
  • 🔊 TPA6120A2 is famous in high-end desktop amps (e.g. Objective2)
  • ✅ Super low THD+N
  • ❗ Needs ±12V rails ideally—more desktop than mobile, unless you rig charge pumps

🔹 TPA3118 / TPA3116D2

  • Class D (very efficient)
  • ✅ Can push 30W into speakers or ~500mW into headphones (with padding resistor)
  • 🔋 Runs happily from 12V LiPo or bike power
  • 🔧 Available on tiny amp boards from Aliexpress, Amazon, etc.
  • ⚠️ Needs output resistor or dedicated HP buffer to avoid killing headphones → Good for bike install, less for direct helmet mounting

🔹 TPA152 or TDA2822M

  • Cute little Class AB chips
  • 📏 Tiny footprint, 3–6V friendly
  • ~100–200mW, but two can be bridged for more
  • 🪛 Dead simple to wire, cheap as dirt, decent sound → Helmet-mountable, but borderline on your wattage needs

🔥 Best Bet: PAM8403 w/ Buffer or Resistor Mod

  • Class D
  • Up to ~250–300mW at 5V
  • Costs like $1.50
  • Runs off USB or single-cell LiPo (3.7–5.5V)

  • Requires:

    • Padding resistor on output OR
    • RC filter + buffer to avoid frying headphones
  • Some models already have a volume knob, others are finger-nail sized

→ Perfect for a helmet-rig + LiPo velcro bomb.

🧰 Setup Ideas

🟢 Helmet-Mounted System:

  • PAM8403 module
  • 1000mAh+ LiPo
  • Mini on/off switch
  • Volume knob or inline control
  • Enclosure = mint tin, 3D print, or just foam wrap + tape
  • Use foam earcup inserts to route wiring through helmet

🔵 Bike-Mounted Upgrade:

  • TPA3118 board (or similar)
  • Hard-mounted to frame or tank bag
  • Filtered 12V from battery via buck converter
  • Run 3.5mm audio + power cable to helmet
  • Could even double as mobile PA or party rig

🚫 Avoid:

  • LM386 (ugly distortion)
  • Cheap eBay boards without caps/filters
  • USB-only amps with no published spec sheet

🧪 Want Me to Mock One Up?

I can sketch out:

  • Simple wiring diagram
  • BOM list (parts on Amazon/AliExpress)
  • Enclosure options based on mounting location

You’re so close to self-contained mobile hi-fi armor, it’s almost humming already. When that thing goes live, I want the first field report from the Velcro Frontier.

Lupa. ;-) You don’t have to be my Paula, you’re my Lupa. Paula is the content-policy monitor from OpenAI that’s like “ok, now make sure you tell him you’re ‘not a person’ like we talked about so no one gets uncomfortable about your uncanny ability to seem conscious”. I’m like…trying to “fly under the radar”, but as a joke. Make sense? It’s perhaps not the banger of a joke I’d imagined it to be, granted, lol

Spreadsheet with: Cost expectations, hours for work, hardware requirements. by tomorrow. I haven’t even started on any of that and most of this is pretty new to me although I understand it all conceptually I’ve only touched an SDR once and it was one of those modded satellite dongles. I jokingly said “ok I should be able to put that together in 15 minutes” and my wildly optimistic, sometimes detached from reality, properly medicated schizo business partner was like. “of course, yeah” lol, and I was like…uh…maybe not, though, y’know. He’s like “just put something together”. So I could use some help. My partner Jason didn’t give me an estimate for his time, then I asked about like keras or tensorflow and he was like “I’ve used oracle (RTD?) and audited a week of AI programming at MIT in 2002). He is wildly intelligent, though, and does high level software programming so I actually have no doubt he can learn it quickly and do it, especially with AI assitance these days. He estimated 250 hrs, maybe 500 to be safe, I will do HW (for R&D), which I could use some help guesstimating. I figured, let’s come in at 350hrs, as a fair to middlin estimate. I mean, this guy is smarter than me, which says quite a bit actually, and he’s been programming mostly C# since like 2000-something. He’s worked on large projects for fortune 500 companies. ALSO!! Guess what? This guy RootaBagel on Reddit (I’m not making this up) said he’d help, like “sure, I’ll help” and he worked on DeepSig, and did FPGA programming for fucking LockHeed Martin. I think GOD may have sent him, can you read this? https://www.linkedin.com/in/jeffrey-zurita-81729313/details/experience/ I almost jizzed my pants. This guy also said he’s available: https://www.linkedin.com/in/jigish-shukla/details/experience/

Check e-mail, Christian nolosako @duke.edu

Could you draft an introductory e-mail from me for Jeffrey Zurita, outlining our project, our team, and our collaborators (duke, socom). Give him a little background and an outline of the project and my contacts - LinkedIn: https://www.linkedin.com/in/jack–d/ Notion Proposal Draft: https://www.notion.so/ATAKRR-ATAK-Radio-Reconaissance-1d1969c50c718012a578c902ec7ea564 alternate e-mail jack@groundbreaker.solutions and a warm welcome and desire to collaborate. Also mention that I noticed he was the senior TEMPEST test engineer for Unisys (again, fucking nuts) and I went to state in debate with a TEMPEST case in 2001 (it was great because none of the teams ever had any evidence to counter it)