Car Audio Explained: Andy Wehmeyer's Lessons for Beginners, Start Here

4 days ago

36 min read

This is the introduction in the series Andy Wehmeyer on Car Audio, Explained for Beginners. Start here · Download the full guide as a PDF

Unofficial guide. Andy Wehmeyer's points are paraphrased from his public forum posts, blog, tuning guides, videos and podcasts, with a link to each source; the plain-English explanations were written for this guide. Not written, reviewed or endorsed by Andy Wehmeyer or Audiofrog.

What one of car audio's most experienced designers teaches about speakers, installation and tuning, with every point explained in plain English.

Please read first: what this guide is, and what it is not

This is an unofficial guide. It was compiled from Andy Wehmeyer's public forum posts, blog, tuning guides, videos and podcasts, dated 2008 to 2023 (plus one livestream from January 2024). Andy Wehmeyer is a co-founder of Audiofrog and was the Harman product manager behind the JBL MS-8 processor.

His points are paraphrased, not quoted. Each point is a summary in the compiler's words of what he said. Only words inside quotation marks are his own, and those quotes are kept short.
He did not write, review or endorse this guide. Neither did Audiofrog, Harman or JBL. Any mistakes are the compiler's, not his.
Check the sources for his exact words. Every point ends with a "Sources:" line linking to the forum post, blog article, guide or video it came from. Where it matters, follow the link and read what he actually said, in context.
The "In plain English" explanations are the compiler's. They are written to explain his meaning to a beginner, not to add to it. When an explanation gives general background (what a tweeter or a decibel is, for example), it is worded as general background ("A crossover is…"), not as his opinion ("He says…").
Product names belong to their owners. Audiofrog, JBL, Harman, MS-8, Logic 7, Bose, REW and every other product or company name mentioned belong to their respective owners. Naming a product is not a recommendation by the compiler.
He sells much of what he recommends. Weigh his comments on his own products accordingly.
Car electrical and installation work can damage a car or injure you. This guide explains ideas; it is not installation instructions. When in doubt, use a qualified installer.

How to use this guide

How each point is laid out

Every point appears in two parts:

His point. A bold one-line summary, then a short paraphrase of his explanation, any tags (see below), and a "Sources:" line with links. This part is copied word for word from the compiled list of his points, so it stays faithful to him.
In plain English. A short explanation of the same point for someone who has never installed a speaker: what the words mean, what he is saying, and, where he gave one, what it means in practice.

If the two ever seem to disagree, trust the first part and its sources.

What the reference numbers mean

Points are numbered by chapter, subsection and position. "3.9 #4" means chapter 3, subsection 3.9, the 4th point in that subsection. A few references point to a named list inside a subsection: "3.1 evolution" is the "How his procedure evolved" list inside subsection 3.1.

The chapters are:

Sound and speaker placement: how stereo works in a car, reflections, how speakers spread sound, where to put them, center channels, rear speakers and upmixing.
Installation and hardware: doors and midbass, sound deadening, pods, tweeter mounting, choosing drivers, tweeter protection and wiring.
Tuning and measurement: his step-by-step process, measuring, target curves, EQ, crossovers, phase, delay and the JBL MS-8.
Bass, power and factory systems: how the car boosts bass, blending the subwoofer, boxes, amplifier power, setting gains, noise and upgrading factory systems.
Myths, buying and industry: common myths, reading specifications, buying, choosing a shop, the business, competitions, listening, and his background.

What the tags mean

[RARE]: the point goes against, or well beyond, common car-audio advice. The explanation often says what the common belief is.
[Changed view: …]: he changed this advice over the years. The tag gives the earlier and later positions with dates. The guide explains both and does not pick a side.
[Conflict: …]: his sources say different things, for example a tuning guide and a forum post, or his own words and his company's sales copy. Again, both sides are explained without choosing.
[Era: …]: the advice belongs to a particular period or product, such as his years at Harman or the JBL MS-8 processor. Products and software change, so check whether it still applies.

What the source labels mean

Each "Sources:" line uses short labels. The most common are:

DIYMA 2021-09-14 (a date): a post on the DIYMobileAudio forum (diymobileaudio.com) on that date. The forum renamed his two accounts to the numbers "14642" (posts from 2008 to 2016) and "156546" (2017 to 2022). Posts under both numbers are his.
DIYMA MS-8 FAQ: his MS-8 question-and-answer thread on the same forum, from his Harman years.
DIYMA gain guide (attributed): a gain-setting guide on the forum that is attributed to him; it is not confirmed as his own writing.
Audiofrog blog (with a month and year): his signed Tech Blog articles on the Audiofrog website (December 2020 to October 2021).
Tuning PDF v1 / v2: his free Audiofrog tuning guide, first version about 2018, second about 2023.
COMPANY: an unsigned Audiofrog web page. It may reflect his views, but because it is not signed, it is not certain that he wrote it.
A-Series manual, MS-8 manual: product manuals (an Audiofrog amplifier manual; the JBL MS-8 manual).
OFCAP Ep. 19 @1:45:06: the Old Fashioned Car Audio Podcast, episode 19, at 1 hour 45 minutes 6 seconds. The link jumps to that moment.
SQOLOGY: a car-audio podcast from 2017–18, with the part number.
CMA or CMA Networks (with a year and timestamp): CMA Networks videos from 2021 to 2023. CMA Connected is its October 2021 livestream.
B1 and B2: two livestreams: B1 is the Absolute Electronix livestream (March 2023) and B2 is the CMA Connected livestream (October 2021).
D&F live: Dean and Fernando's livestream (January 2024).
EDUCAR: a technical tip published in EDUCAR magazine.
Other labels name the publication or site directly, such as a magazine, a retailer page, a product listing or a third-party measurement review.

Video and podcast points were taken from automatic captions, so timestamps are approximate and wording may be slightly off. Follow the link and listen if the exact words matter.

A suggested reading order for beginners

"Car audio in 10 minutes" (next in this guide). It gives you the basic words and ideas the rest assumes.
"Start here": 50 of his most distinctive points, each with a plain-English note. It is the quickest way to see how he thinks.
Chapter 5 (Myths, buying and industry), if you are about to buy a system or choose a shop.
Chapter 1 (Sound and speaker placement), which explains why a car sounds the way it does. Most of the later advice builds on it.
Chapter 2 (Installation and hardware), then Chapter 4 (Bass, power and factory systems).
Chapter 3 (Tuning and measurement) last. It is the most technical and makes most sense once the rest is familiar.

Keep the glossary at the back open as you read. Any term in bold or in quotation marks that you don't recognise is probably in it.

Car audio in 10 minutes

This primer is general background, written by the compiler for complete beginners. It is not a summary of Andy Wehmeyer's views; his views start in the next section.

The parts of a system

Think of a car audio system as a chain. The music passes along it from one link to the next, and the weakest link limits the result.

The source (head unit). This is the radio or screen in the dash, or a phone feeding it. It produces a small electrical signal, measured in volts, that carries the music. Its volume knob controls how strong that signal is. In many modern cars the factory head unit is tied into the car's other systems and is hard to replace.

The processor (DSP). A digital signal processor, or DSP, is a small computer for sound. It takes the music signal, splits it into separate channels for each speaker, and adjusts each one. It can change tone (equalization, or EQ), decide which speaker plays which notes (crossovers), hold back one speaker's sound by a few thousandths of a second (delay), and set each channel's level. Many factory systems already contain one, often hidden inside the factory amplifier.

The amplifiers. The signal from the source or DSP is far too weak to move a speaker. An amplifier ("amp") makes it powerful enough. Amplifier power is measured in watts (W). An amp has a "gain" control: it does not set power, it matches the amp to the strength of the incoming signal. If any part of the chain is pushed beyond its limit, it "clips": the tops of the sound waves are cut off flat, which sounds harsh and can damage speakers.

The speakers (drivers). A speaker turns electrical signals back into sound by moving a cone or dome back and forth. No single speaker can play every note well, so systems divide the work:

Tweeter: a small dome, often about 1 inch (25 mm) across, that plays the highest notes, such as cymbals and the "s" sounds in voices.
Midrange: a small cone, roughly 2 to 4 inches (50–100 mm), that plays the middle notes where most of a voice lives.
Midbass (or woofer): a larger cone, often 6 to 8 inches (150–200 mm) and usually in the door, that plays upper bass and lower middle notes, such as the body of a kick drum or a male voice.
Subwoofer: a large cone, often 8 to 15 inches (200–380 mm), in its own box, that plays the deepest bass.

A system with a tweeter and a midbass on each side is a "2-way". Add a midrange and it is a "3-way". The individual speakers are often called "drivers".

Crossovers. A crossover splits the music so each speaker gets only the notes it handles well. A "high-pass" filter lets high notes through and blocks low ones; a "low-pass" filter does the opposite. The point where one speaker hands over to the next is the "crossover frequency". A passive crossover is a small box of electrical parts (capacitors, coils and resistors) between the amp and the speakers; it is simple, but fixed. An active crossover sits before the amplifiers, usually inside a DSP, and gives each speaker its own amp channel, so every setting can be adjusted and fine-tuned.

Wiring. Power cables bring current from the car's battery to the amps. Signal cables (often "RCA" cables, named after their plug) carry the music from the source or DSP to the amps. Speaker wires carry the amplified signal to the speakers. Good connections, and correct polarity (plus to plus, minus to minus), matter: swap the two wires on one speaker and it moves backwards relative to the others.

The basic units

Hertz (Hz) measures frequency, the pitch of a sound: how many times per second the air vibrates. Low numbers are low notes. Human hearing runs from about 20 Hz to 20,000 Hz. Kilohertz (kHz) is a thousand hertz, so 20 kHz is 20,000 Hz. As a rough map: deep bass sits below about 80 Hz, midbass roughly 80–300 Hz, the midrange from a few hundred hertz to a few kilohertz, and treble above that.

An octave is a doubling of frequency: 100 Hz to 200 Hz is one octave, and so is 1,000 Hz to 2,000 Hz. Hearing works in octaves, so audio graphs usually space them evenly.

Wavelength is the physical length of one cycle of a sound wave. Low notes have long waves (about 4.3 m, or 14 ft, at 80 Hz); high notes have short ones (about 3.4 cm, or 1.3 inches, at 10 kHz). This is why the size of a car, and the distances between speakers and surfaces, matter so much.

The decibel (dB) measures differences in level (loudness). It works on ratios, not plain steps. An increase of about 1 dB is barely noticeable, 3 dB is clearly noticeable, 6 dB is a big change, and about 10 dB sounds roughly "twice as loud". Each extra 3 dB of output needs about twice the amplifier power, so 10 dB more needs about ten times the power. "dB SPL" is sound pressure level, the actual loudness measured in the air.

The millisecond (ms) is a thousandth of a second. Sound travels about 343 m (roughly 1,125 ft) per second in air, so in 1 ms it covers about 34 cm, or about 13.5 inches. In a car, where speakers sit at different distances from you, differences of a millisecond or two decide which speaker's sound reaches you first.

Phase describes where a wave is in its cycle, measured in degrees (°). When two speakers play the same note "in phase", their waves rise and fall together and add up. When they are 180° apart, one pushes while the other pulls, and they cancel.

What listeners mean by imaging, staging and tonal balance

Imaging is the illusion that a singer or instrument sits at a precise spot in front of you, even between the speakers, where there is no speaker at all. Good imaging puts a lead vocal in the middle of the dashboard, solid and small, as if someone were standing there.

Staging (the "soundstage") is the overall picture: how wide the band seems, how high (at dash level or down at your knees), and how far away. A good stage sounds as if the music is spread across the front of the car, in front of you, rather than coming from the door by your knee.

Tonal balance is how the low, middle and high notes sit against each other. Too much bass sounds boomy, too much treble sounds harsh, and a weak midrange makes voices thin. Tuners often aim for a "target curve", a chosen shape for the balance from low to high notes, measured with a microphone.

What tuning is

Tuning is adjusting the processor so the system sounds right from the listening seat. It usually involves four kinds of setting:

Crossovers: which notes each speaker plays, and how sharply it hands over to the next.
EQ (equalization): turning particular frequency bands up or down to correct the balance. A "parametric" EQ lets you choose the center frequency, the amount, and the width (often called "Q": high Q is narrow, low Q is broad).
Delay, or time alignment: holding back the nearer speakers by a millisecond or so (a few thousandths of a second at most) so their sound arrives with the farther ones.
Levels: how loud each speaker or side is relative to the others.

Tuners measure with a microphone and software. A common tool is an RTA (real-time analyzer), which shows the level at each frequency while a test signal plays, often pink noise, a hiss with equal energy in every octave. Another common approach is a sweep: a tone that glides from low to high, analyzed by software such as REW (Room EQ Wizard). The result is a "frequency response" graph: frequency from low to high along the bottom, level in decibels up the side.

Why a car is a hard place for sound

Home speakers are usually placed in front of you, at equal distances, in a room several metres across. A car breaks almost every one of those conditions.

You sit off-center. The driver is much closer to the left speakers than the right (the reverse in right-hand-drive cars). The near side's sound arrives first and louder, which pulls the image toward it.
There are two (or more) seats. A fix that suits the driver's seat can make the passenger's seat worse.
Hard surfaces are very close. Glass, the dash and plastic panels sit centimetres from the speakers. Sound bounces off them and arrives a fraction of a millisecond after the direct sound. Where the bounce and the original are out of step, they cancel at some frequencies and add at others, a pattern called comb filtering, because on a graph it looks like the teeth of a comb.
The space is small. At low frequencies the cabin is smaller than the sound waves, which strongly boosts deep bass. This is called cabin gain. In the bass range, the cabin's dimensions also create peaks and dips.
Speaker locations are limited. Speakers go where the car has room: low in the doors, in the dash, in the pillars beside the windshield, in the kick panels by your feet. Few of these are ideal.
It is noisy and it moves. Road, tire and wind noise mask the music, especially the bass, and panels rattle and buzz.
Factory systems are complicated. Many new cars process the sound before it reaches the speakers, with built-in EQ, delay and crossovers, which an upgrade has to work with or around.

That is why car audio leans so heavily on careful speaker placement, good installation and processing. The rest of this guide is about what Andy Wehmeyer has said on each of those, in his terms, with plain-English help along the way.

Start here: 50 things Andy says that almost nobody else does

Ten per topic, each with its key number or action and a reference to the full point and its sources.

In plain English: these are 50 of his points that go against, or well beyond, common car-audio advice. Each has his point in bold, a short note with his key number or action, and references (such as 1.1 #3) to the full points in the topic chapters, where the sources are listed.

Sound & placement (1 · Sound & speaker placement)

Door speakers put left and right 180° out of phase near 250 Hz, in both front seats. A ~28" path difference to the door mids causes it, right in the vocal range; people mistake those bands for "width". (1.1 #3, 1.9 #4)
In plain English: Your left and right door speakers sit at different distances from your head (about 28 inches, roughly 70 cm, of difference), so around 250 Hz, a pitch in the middle of the human voice, the waves from the two sides arrive exactly opposite and cancel, in both front seats. He says the spread-out sound people praise as "width" is really these cancelling bands, which make you hear where the speakers are instead of a centered image.
A car has no reverberant field, so you hear each speaker's power response: its output averaged over all angles. Reflections arrive within ~20 ms and fuse with the direct sound, and no EQ or delay can undo a reflection's comb. (1.3 #1, 1.3 #4)
In plain English: Every sound that bounces around a car reaches you within about 20 ms (twenty thousandths of a second) and blends with the direct sound, so what you hear is roughly the speaker's average output in all directions (its "power response"). Because the processor feeds the speaker before any bouncing happens, he says no EQ or delay setting can undo the dips a reflection causes.
Predict reflection dips with a tape measure: a dip lands where the reflected path is half a wavelength longer. A mid 6" from a surface dips near 500 Hz, so tuck mids into corners and keep tweeters away from surfaces. (1.3 #5)
In plain English: A bounced copy of a sound travels farther than the direct sound; where that extra distance equals half a wavelength, the copy arrives upside down and cancels, leaving a dip. He says you can predict this with a tape measure (a mid 6 inches, about 150 mm, from a surface dips near 500 Hz), so push midranges into corners and keep tweeters away from surfaces, so the dip falls outside the range each speaker plays.
A 6" + 1" two-way leaves an off-axis hole at its crossover, which is why cars need 3-ways. Aiming only moves the hole; studio monitors get away with 2-ways because their walls are ~10 ft away. (1.4 #4, 1.4 #6)
In plain English: A 6-inch woofer and a 1-inch tweeter spread their sound very differently near the crossover (the frequency where one hands over to the other), so off to the sides there is a "hole", and in a car that sideways sound reaches you by bouncing off surfaces right next to the speakers. He says aiming only moves the hole around, so cars need a 3-way (an added midrange); studio monitors get away with 2-ways because their room's walls are about 10 ft (3 m) away.
Aim only the tweeter. By his circumference ÷ wavelength rule, a 6" is omnidirectional below ~300–500 Hz and a 3" below ~3 kHz. (1.4 #2, 1.5 #1)
In plain English: Low notes spread out evenly in all directions; higher notes start to "beam" forward. By his rule (compare the cone's circumference with the sound's wavelength) a 6-inch speaker spreads evenly below about 300–500 Hz and a 3-inch below about 3 kHz (3,000 Hz), so aiming those changes little in their working range, and only the tweeter, turned in a little, is worth aiming.
Put a small midrange high in the door; an A-pillar is "a HORRIBLE place for a midrange". A pod a foot off the pillar corner digs a hole near 250 Hz, and correct crossover phase gives height anyway. (1.6 #2, 1.6 #6, 1.7 #7)
In plain English: He recommends a small midrange mounted high in the door and calls the A-pillar (the post beside the windshield) "a HORRIBLE place for a midrange": a pod hanging about a foot (30 cm) from the windshield, dash and side glass picks up reflections that dig a hole near 250 Hz. He says the height pillars promise comes anyway from high tweeters plus correctly phased crossovers, which pull the whole sound up to tweeter height.
The stage can't be wider than your speakers. Pan the balance fully left, then right: those are your real edges, and "width" from delay games is comb filtering. (1.7 #1, 1.7 #2)
In plain English: The "stage" is the imaginary band of sound you hear spread across the dash. He says it can't be wider than the speakers themselves: pan the balance control fully left, then fully right, and where the sound lands are your true edges; extra "width" made by playing with delay is really comb filtering (a pattern of cancellations).
Delay fixes one seat and doubles the error in the other; an all-pass near 250 Hz helps both. His 2016 starting point: one 2nd-order all-pass near 250 Hz, Q about 1, on one side only. (1.9 #2, 1.9 #4, 1.9 #7)
In plain English: The driver sits nearer the left speakers and the passenger nearer the right, so delaying one side to fix the driver's seat doubles the timing error at the passenger's seat. An all-pass filter shifts timing (phase) in one band without changing loudness, and near 250 Hz it helps both seats, not perfectly in either but better; his 2016 starting point was one such filter (2nd-order, Q about 1, a fairly broad setting) on one side only.
"L+R is not a center." A real center plays only what left and right share, and that content stays in L and R about 6 dB down. (1.10 #1, 1.10 #2)
In plain English: "L+R" means simply adding left and right together into one center speaker, which he says is not a real center. A proper center plays only the sound that left and right have in common, and even then that shared sound stays in the left and right speakers as well, about 6 dB quieter.
For rear fill, use plain in-phase stereo, then delay it until the stage returns to the front. High-pass at 100 Hz or more, raise the rears until the stage drifts back, then add 10–20 ms (35 ms becomes an echo). (1.11 #4)
In plain English: Rear fill means playing rear speakers quietly to add a sense of space without pulling the sound backwards. His recipe, after the front is tuned: feed the rears normal stereo with bass below at least 100 Hz filtered out, raise them until the stage starts drifting back, then add 10–20 ms of delay until it sits in front again; about 35 ms is long enough to be heard as an echo.

Installation & hardware (2 · Installation & hardware)

A door midbass needs a sealed baffle, not a box; tiny boxes are "bass prevention". With a high-pass, just isolate front from back: seal it to the trim panel or cover the sheet-metal holes with deadener. (2.1 #1, 2.1 #2)
In plain English: A baffle is the panel a speaker is mounted on; a box encloses its back. He says a door midbass whose deepest bass is already blocked by a high-pass filter needs no box, only its front sound kept apart from the sound off its back (seal it to the door's trim panel, or cover the holes in the door metal with deadening sheets), and that tiny boxes behind it actually reduce bass ("bass prevention").
Pull out foam cup "baffles", and stick foam rings to the bracket, never the speaker. Foam stops rattles; it doesn't "make midbass", so keep at most a cup's top as a rain shield. (2.1 #6, 2.1 #7)
In plain English: Foam cups are sold to go behind door speakers; he says pull them out, keeping at most the top part as a shield against water running down inside the door. Foam rings are only for stopping the door panel rattling, so stick them to the speaker's mounting bracket, never to the speaker, and don't expect them to add bass.
Choose a flat, low-Q midbass, because the door and its corner already add the bass. Each halving of radiating space adds ~6 dB below ~500 Hz, so a "full", high-Q driver turns into a peak. (2.1 #8, 2.1 #9)
In plain English: "Q" describes how peaky a speaker's own bass response is; low-Q means flatter, with no built-in bass bump. He says the door and the corner it sits in already boost bass (about 6 dB below about 500 Hz each time the space the sound can spread into is halved), so a speaker designed to sound "full" ends up with an unwanted peak in a car.
Midbass "punch" lives at 1–4 kHz, not in a bigger woofer. Prove it on headphones: low-pass a punchy track at 100 Hz, then 200 Hz, and the punch is gone. (2.1 #13)
In plain English: "Punch" is the snap you feel from kick drums and bass. His point is that it lives in the 1–4 kHz range (1,000–4,000 Hz: the thumb slap on a bass, the snap of toms), not in a bigger woofer; to prove it, play a punchy song on headphones with everything above 100 Hz, then above 200 Hz, filtered out, and the punch is gone.
Leave room beyond Xmax: keep grilles about 1.25" off a woofer's front. The cone travels past Xmax through the top-plate thickness, so a surround hitting a grille or panel is an install error. (2.1 #19)
In plain English: Xmax is how far a speaker cone can move in and out while still behaving cleanly. He says the cone actually travels further than that, so keep grilles about 1.25 inches (about 32 mm) off a woofer's front; if its rubber edge hits a grille or panel, that's an installation mistake, not a faulty speaker.

Deadening is rattle control: do the trim panel first, then the inner skin. Play a 250 Hz tone: if nothing buzzes, a 250 Hz peak is a cabin reflection to cut with EQ. (2.2 #1, 2.2 #3)
In plain English: Sound deadening means sticking heavy damping sheets and foam onto car panels; he treats it as rattle control, starting with the door's inner trim panel (the worst rattler) and then the inner metal skin. To check a loud spot at 250 Hz, play a steady 250 Hz tone: if nothing buzzes, the peak comes from the cabin's shape reflecting sound, and you turn that frequency down with EQ instead of adding deadener.
Quieter tires were "the best midbass upgrade" he ever made. For road noise, deaden first, then fit quieter tires, then save per-road DSP presets; never EQ for noise above 300 Hz. (2.2 #4)
In plain English: He says swapping noisy tires for quieter ones was the best midbass improvement he ever made, because road noise drowns out bass. His order for road noise: deaden the car, then fit quieter tires, then save separate DSP settings (presets) for roads you drive often, and never use EQ to fight noise above 300 Hz.
Map locations before cutting anything: a speaker at the driver's head, a mic at each candidate spot. Sound paths work both ways, so this finds good sub, midbass and mid spots, but not tweeter spots. (2.3 #13)
In plain English: Before cutting any holes, he "maps" the car: a small speaker goes where the driver's head would be, and a measuring microphone goes at each spot you might mount a speaker. Because sound paths work the same in both directions, this shows which spots suit subwoofers, midbass and midranges, though not tweeters, which spread their sound too narrowly for the trick to work.
A pod is an enclosure, and a leaky one "rattles". Size the pod to the driver, gasket it, and run an impedance sweep: a leak shows up as an extra, port-like resonance. (2.4 #1, 2.4 #3, 2.4 #4)
In plain English: A pod is a custom housing for a speaker (on an A-pillar, for example), and he says it is really a small speaker box: it must be sized to suit the speaker, sealed with a gasket (a soft seal) and airtight, because a leaky pod buzzes. To check it, run an impedance sweep (a test of the speaker's electrical load across frequencies): a leak shows up as an extra bump like the one a ported box makes.
Keep a cheap series capacitor on tweeters even in active systems, against amp failures and DSP typos. Use about 68 µF for a 4 Ω tweeter as a "parking brake"; a 10 Ω resistor across it cures turn-on pops. (2.9 #1, 2.9 #3)
In plain English: Even when a DSP and separate amps do the crossover (an "active" system), he keeps a cheap capacitor wired in line with each tweeter as a "parking brake" that blocks bass if an amp fails or someone types a wrong DSP setting, since bass quickly destroys a tweeter. His example is about 68 µF (microfarads) for a 4-ohm tweeter, and a 10-ohm resistor across the capacitor stops pops when the system turns on.

Tuning & measurement (3 · Tuning & measurement)

Measure and fix everything first; listen once, at the end. If hours of by-ear touch-up follow, the measurement, your reading of it or the process is wrong. (3.2 #1)
In plain English: He tunes by measuring with a microphone and fixing everything the measurements show first, and listens once, at the end. If hours of tweaking by ear follow, he says something went wrong: the measurement, how you read it, or the method itself.
One mic on the headrest, measured from outside the car with the door shut, is enough. With 1/6-octave smoothing it gets about 90% of a spatial average; waving it around your head from the seat eats the treble. (3.4 #1, 3.4 #2, 3.4 #5)
In plain English: He straps one measuring microphone to the headrest at the listening position, runs the cable outside and measures with the door shut. With 1/6-octave smoothing (averaging out tiny wiggles on the graph), that single mic gets about 90% of the accuracy of averaging many mic positions, whereas sitting in the seat waving the mic around your head lets your head block the treble and skews the graph.
Leave narrow peaks and dips above ~1 kHz alone, and move the mic before fixing one. If a feature vanishes when the mic moves a few inches, fixing it probably makes things worse. (3.4 #4, 3.4 #10)
In plain English: Narrow peaks and dips above about 1 kHz come from sounds arriving at slightly different times, and he says we don't hear them because tiny head movements average them out, so leave them. Before "fixing" one, move the mic a few inches: if the feature vanishes, correcting it will probably make things worse.
Put the target on the dips and cut the peaks down to it. A 64 dB peak over a 57.5 dB dip gets a ~7 dB cut; then restore overall level with amp gain. (3.6 #3)
In plain English: A "target curve" is the shape you want the frequency graph to follow. He says line the target up with the dips (the low points), not the peaks, and turn each peak down to meet it (with a dip at 57.5 dB, a 64 dB peak gets about a 7 dB cut), then raise the amplifier gain to get the overall loudness back.
Matching left to right beats hitting the target. Match within a couple of dB, and if one side has a dip you can't fix, copy it to the other side. (3.6 #2)
In plain English: He says making the left and right sides measure alike, within a couple of dB, matters more than hitting the target exactly. If one side has a dip you can't fix, add the same dip to the other side, which protects the centered sound.
Never significantly boost a null; first ask whether the dip responds to EQ. Driver and box problems do; reflection dips don't, and a dip only with both sides playing is a phase or delay problem. (3.6 #9, 3.6 #10)
In plain English: A "null" is a deep dip where sounds cancel; boosting it with EQ wastes amp power and strains the speaker, so he says never boost one by much. First ask whether the dip responds to EQ: dips from the speaker or its box do, dips from reflections don't, and a dip that appears only when both sides play is a timing (phase or delay) problem, not an EQ one.
Target-curve wars are pointless: popular car curves differ by about 2 dB, and the "Harman curve" began as his. His target today: about +10 dB below ~60 Hz, flat from ~160 Hz, and −6 dB at 20 kHz. (3.5 #1, 3.5 #2, 3.5 #3)
In plain English: Hobbyists argue about which target curve is best; he says the popular car curves differ by only about 2 dB, and the so-called "Harman curve" began as his own installer curve, which he took to Harman. His target today: bass about 10 dB louder below about 60 Hz, flat from about 160 Hz up, easing down to 6 dB quieter at 20 kHz.
Delay doesn't move speakers: center an off-center image with level, never delay. Turn the whole side the image leans toward down a little; about 6 dB moves an image halfway. (3.9 #1, 3.10 #2, 3.10 #4)
In plain English: Delay (holding back one speaker's sound by a few thousandths of a second) changes how speakers' sounds combine, but it can't change where the speakers seem to be. If a centered voice sits off to one side, he turns the whole side it leans toward down a little rather than using delay; about 6 dB of difference moves an image roughly halfway.
Set delays with a tape measure, keyed to the midrange, then leave them alone. Impulse peaks aren't arrival times for subs and midbass, and auto-alignment measures dash speakers via the windshield. (3.9 #3, 3.9 #5, 3.9 #6)
In plain English: He sets delays by measuring with a tape measure from the headrest to each speaker, using the midrange's distance for the front speakers, and then leaves them alone. He says graph-based and automatic methods mislead: impulse peaks (spikes on a timing graph) don't show true arrival times for subs and midbass, and auto-alignment measures dash speakers by the sound bouncing off the windshield.
Read phase from the left-plus-right sum of mono pink noise. About +6 dB over one side is in phase, +3 dB ~90°, equal to one side ~120°; leave anything within ±90° alone. (3.8 #6, 3.8 #7)
In plain English: Phase describes whether two speakers' waves move in step. Play the same pink noise (an even hiss containing all frequencies) through both sides and measure: about 6 dB louder than one side alone means in step, about 3 dB louder means roughly 90° apart, the same level as one side means about 120° apart, and he says anything within ±90° can be left alone.

Bass, power & factory systems (4 · Bass, power & factory systems)

Cabin gain belongs to the car: build the most efficient box you can fit, then EQ. Cut a boomy in-car peak by less than the graph says; he cuts 6–10 dB where it shows 15. (4.1 #1, 4.1 #6, 4.1 #8)
In plain English: A closed car naturally boosts deep bass ("cabin gain"), and he says that boost belongs to the car, not to the subwoofer. So build the most efficient sub box that fits, then correct it with EQ; where the graph shows a boomy peak, cut by less than the graph suggests (he cuts 6–10 dB where it shows 15).
Bass gives away its location through distortion and rattles, not arrival time. Bass up front needs just two things: sub and midbass in phase at the crossover, and no rattles or distortion. (4.2 #1, 4.2 #2)
In plain English: He says what gives away where bass is coming from is distortion and rattles, not when its sound arrives. For bass that seems to come from the front, you need just two things: the sub and midbass in phase where they hand over, and no rattles or distortion.
Cross the sub at 60 Hz and door midbass at 80 Hz, both LR4; never run door midbass down to 40 Hz. Lower the sub low-pass half an octave for every 12 dB you turn the sub up. (4.2 #6, 4.6 #1)
In plain English: The crossover is where the sub hands over to the door midbass. His default is 60 Hz on the sub and 80 Hz on the midbass, both LR4 (steep, 24 dB-per-octave filters), because you will turn the sub up; never make door speakers play down to 40 Hz, and lower the sub's crossover by half an octave for every 12 dB you turn it up.
He prefers vented sub boxes tuned for minimum excursion, with the in-car peak EQ'd out. Short on space but not on power, go sealed plus EQ; with no EQ at all, sealed is probably better. (4.3 #1, 4.3 #4, 4.3 #6)
In plain English: A vented (ported) box has an opening tuned to reinforce bass; a sealed box doesn't. He prefers vented boxes tuned so the cone moves as little as possible, with the boomy in-car peak removed by EQ; with little space but plenty of amp power he says go sealed plus EQ, and with no EQ at all sealed is probably better.
Music averages about 1/32 of its peak power; give mids and highs 50–100 W and the sub at least ~800 W. Only the sub amp ever reaches rated power; a tweeter crossed at 2.5 kHz gets ~0.16 W of a "20 W" calibration. (4.7 #1, 4.7 #2, 4.7 #3)
In plain English: Music's average power is only about 1/32 of its loudest peaks, and most of it is bass, so his rule of thumb is 50–100 watts for mids and highs and at least about 800 W for the sub. Only the sub amp ever reaches its rated power: in his example, with the amp calibrated to 20 W at 20 Hz (deep bass), a tweeter crossed at 2.5 kHz gets about 0.16 W.
Set the DSP input so it can never clip at full volume, and give only the sub amp a sine wave. About 70% of GB25 warranty claims were DSP input clipping; a 4 V radio needs a 4–5 V input setting. (4.7 #6, 4.8 #1, 4.8 #4)
In plain English: A DSP (digital sound processor) turns the incoming signal into numbers, and too strong an input "clips" (squares off the waves), which sounds like broken speakers and can damage them; he says about 70% of warranty claims on Audiofrog's GB25 speaker came from this. So set the DSP's input so it can't clip even at full volume (a radio putting out 4 volts needs a 4–5 V setting), and use a sine-wave test tone only for setting the sub amp.
Amps with built-in DSP can't win back output lost to EQ cuts. After tuning cuts, a "1,000 W" DSP amp may make ~10 W; for high-end systems use a separate DSP and amps with analog gain. (4.8 #9)
In plain English: Some amplifiers have a DSP built in, and he says they cannot turn the signal back up after tuning has cut the EQ, so a "1,000 W" DSP amp may end up making only about 10 W. For high-end systems he recommends a separate DSP feeding amps that have their own analog gain controls.
Hiss is not a ground loop; loops whine. Ohm the RCA shield to the amp's power ground: near 0 Ω means single-ended (use coax), hundreds of ohms means differential (use twisted pair). (4.10 #1, 4.10 #5)
In plain English: He says steady hiss is not a ground-loop problem; ground loops (faults in how equipment shares its ground connection) cause whine. To pick the right signal cable, measure the resistance from the RCA plug's outer shield to the amp's power ground: near 0 ohms means a "single-ended" input (use ordinary coax), hundreds of ohms means a "differential" input (use twisted-pair cable).
Tell factory all-pass filters from factory delay before you add any delay. On mono pink noise, one big L+R dip at 200–500 Hz means all-pass (add no delays); an even comb means delay. (4.14 #2, 4.14 #4)
In plain English: Many factory systems already adjust timing, using either delay or all-pass filters (which shift timing in one band). He says find out which before adding any delay: play mono pink noise and look at left plus right together; one big dip between 200 and 500 Hz means all-pass filters (add no delay), and a row of evenly spaced dips (a "comb") means delay.
In factory systems that upmix, keep the factory processing and wire each output one-to-one to new amps. Then just re-EQ and set levels, and "DO NOT TRY TO SUM" a Logic 7-type car back together. (4.14 #11, 4.15 #4)
In plain English: Some factory systems "upmix" stereo into many channels for different speakers (Harman's Logic 7 is one). He says keep that factory processing, connect each factory output one-to-one to a channel of a new amp, then just redo the EQ and levels, and in Logic 7-type cars "DO NOT TRY TO SUM" (recombine) the channels back into plain stereo.

Myths & buying (5 · Myths, buying & industry)

"You cannot blow a speaker with too little power." Clipping pushes a wave toward a square, roughly doubling its average power, and all the extra is treble that cooks tweeters. (5.1 #26)
In plain English: Many people believe an underpowered amp blows speakers; he says it cannot. The damage comes from clipping (driving an amp, radio or DSP past its limit), which squares off the waves, roughly doubles their average power and puts all the extra into treble that overheats tweeters.
A correct tune matters more than $200 versus $2,000 speakers. His priorities are processing, then installation, then brand; EQ is the only thing that overcomes the car. (5.3 #2)
In plain English: He says setting up (tuning) a system correctly matters more than whether the speakers cost $200 or $2,000. His order of priority is processing (DSP and EQ), then installation, then brand, because only EQ can overcome what the car does to the sound.
Upgrade a factory system in stages: speakers on the factory amp and EQ first, then power, then sub, rears and DSP. Only about 10–20% of a factory EQ corrects the stock speakers, so with similar new speakers it keeps doing useful work. (5.3 #6, 4.16 #1)
In plain English: For a car with a factory system, he suggests upgrading in steps: new speakers on the factory amp and its built-in EQ first, then more amplifier power, then a sub, rear speakers and a DSP. Only about 10–20% of a factory EQ is there to correct the original speakers, so with similar new speakers the rest of it keeps doing useful work.
Go to the best shop in town and ask it for a mid-level system. Paying ~30% more for the shop that can "build the space shuttle" beats pushing a cheap shop past its level. (5.4 #1)
In plain English: He advises going to the best shop in town and asking it for a mid-level system, telling them your budget, rather than pushing a cheap shop to work beyond its skill. In his view paying about 30% more for the shop that could "build the space shuttle" is worth it.
"Correct" is the beginning of expertise; backing off it for the customer is the rest. His disco fan boosted bass ~12 dB at the head unit and kept blowing speakers; a higher, pro-style crossover fixed it. (5.4 #15)
In plain English: His point is that knowing the technically correct setup is only the start; expertise is also knowing when to back off it to suit the customer. His example: a 1970s-disco fan kept boosting bass about 12 dB at the radio and blowing speakers, and the fix was a higher crossover like those in pro sound systems, giving up up-front bass the customer didn't care about.
Learn what imaging is with a matched $5 thrift-store pair of speakers. Sit centered in an equilateral triangle, then flip one's polarity; that, not a concert, is a tuner's first reference. (5.7 #5)
In plain English: "Imaging" is hearing sounds appear at exact spots between the speakers. He says learn it with any matched pair of cheap speakers (about $5 at a thrift store): sit centered so you and the two speakers form an equal-sided triangle, then swap the wires on one speaker (flip its polarity) and hear the difference; to him, that, not a live concert, is the first reference a tuner needs.

Distortion on only some tracks is usually in the source. Many streaming masters are clipped and phone "volume levelers" add more; if it stays with the phone turned well down, it's the track or codec. (5.7 #12)
In plain English: If only some songs sound distorted, he says the problem is usually in the recording: many streaming masters are already clipped, and phone "volume leveler" settings add more. If the distortion stays even with the phone's volume turned well down, it's in the track or the codec (the audio compression format).
Damping factor above ~10 and channel separation above ~60 dB buy nothing in a car. A damping factor of 1000 moves Qts only from 0.707 to 0.708, and each ear already hears the opposite speaker about 6 dB down. (5.1 #19, 5.1 #24)
In plain English: Damping factor (how firmly an amp controls a speaker) and channel separation (how little of one channel leaks into the other) are often advertised. He says beyond about 10 and about 60 dB they make no difference in a car: a damping factor of 1000 changes a speaker's Qts (a measure of its bass behavior) only from 0.707 to 0.708, and each ear already hears the opposite speaker only about 6 dB quieter.
Break-in matters mainly for subwoofers. It's the spider's resin cracking, which matters at and below resonance; mids and tweeters crossed above it barely change. (5.1 #30)
In plain English: "Break-in" is the idea that new speakers change as they are played. He says the real change is resin in the spider (the corrugated fabric that keeps the cone centered) cracking and loosening, which matters at and below the speaker's resonance (its natural bass frequency), so subwoofers change but midranges and tweeters crossed above that barely do.
Know when to stop: "imagine how much better…" is the biggest threat to satisfaction. Log tracks that annoy you, check that band with pink noise and fix only real peaks, or you end with a "smile" curve. (5.7 #13)
In plain English: He warns that "imagine how much better…" is the biggest threat to being happy with your system. Instead, note songs that annoy you, check that frequency band with pink noise and fix only real peaks; otherwise you keep cutting until you end with a "smile" curve (bass and treble up, middle scooped out) that sounds bad.

Where Andy changed his mind, and where sources conflict

In plain English: over the years he changed his advice on some topics, and a few of his sources say different things. The first list shows each topic's earlier and later view with years, and the second list shows where sources disagree; neither is ranked here, so read the full points under "See" for context.

His advice shifted on 26 topics over the years; the first list below gives them newest change first, and source disagreements follow it.

What he teaches. Earlier view (year): Tuning (about 20 years, before 2023); Later view (year): System design: map the car with a mic and speaker before choosing locations (2023); See: 3.2 #15, 5.8
Rear fill. Earlier view (year): Differential (L−R) fill, delayed 12–20 ms (2017); Later view (year): Plain in-phase stereo, high-passed and delayed 10–20 ms; L−R sounds "phasy" (2021); See: 1.11 #5, 3.9 #18
REW settings. Earlier view (year): FFT 16384, rectangular window, "dB not dBFS" (Tuning PDF, ~2018); Later view (year): 32k FFT; window type and dB vs dBFS make no difference (2021); See: 3.3 #5
Speaker break-in. Earlier view (year): Not needed (2015); Later view (year): Real, but it is the spider (2017); matters only for subs (2021); See: 2.8 #6
Efficiency vs power. Earlier view (year): "Efficiency is overrated. Power is cheap." (2008–13, arguing for active crossovers); Later view (year): Efficiency first in very loud systems (2021); a change of context more than a reversal; See: 4.7 #11, 5.2 #10
All-pass on sub and midbass. Earlier view (year): Never made one work (2010–13); used them in his own car (2015–16); Later view (year): Fix the crossover instead; if you must, put it on the high-pass side (2021); See: 4.2 #15
Dash-top center. Earlier view (year): A response "nightmare" that still raised the stage (2010), yet already his preferred center in the MS-8 FAQ (2011); a refinement, not a reversal; Later view (year): Mount it close to the windshield–dash junction; its ~2 kHz dip helps (2021); See: 2.3 #11
Midbass Q (partial). Earlier view (year): Car midbass should be low-Q (2015; Jun 2021 blog); Later view (year): Crossed above resonance, its Q "is not really important" (Sep 2021); See: 4.6 #3
Two-seat imaging. Earlier view (year): Phase EQ inaudible next to delay (2013–15); passenger seat a trade-off while delay was the only tool; Later view (year): Drop L/R delay for phase EQ (2016); it needs fiddling, so he won't teach it (2021); "forget the passenger" is a cop-out (2023); See: 1.9 #4, 1.9 #10, 3.8 #15, 3.9 #14
Factory integration. Earlier view (year): Sum, de-EQ and remove factory delay with two processors (2013–16); Later view (year): Keep the factory all-pass filters and processing; wire outputs one-to-one (2021–23); See: 4.14 #11
Gain setting. Earlier view (year): Scope and −10 dB sine, with ~9–10 dB of deliberate clipping overlap (2008–15); by ear (2014–16); Later view (year): Sine on the sub amp only, the rest by pink noise and RTA; overlap "never appropriate" at the DSP input (2018–22); See: 4.8 #4, 4.8 #5, 5.1 #27
Uncorrelated pink noise. Earlier view (year): Suggested for 31-band adjustments (2014); Later view (year): Useless except for upmixers, L−R rears or gaming an RTA (2020–22); See: 3.3 #4
8" door woofers. Earlier view (year): Audiofrog would make no 8" and offered a 6×9 (2016); Later view (year): The shallow GS8ND2 (2020), framed as a woofer crossed near 300 Hz; See: 2.1 #15
Stage height. Earlier view (year): Comes from physical location (2016); Later view (year): Crossover phase also holds everything at tweeter height (2020–21); See: 1.7 #7
Sub/midbass crossover. Earlier view (year): Both near 80 Hz LR4 (2010–15, MS-8 era); Later view (year): Sub 60 Hz, midbass 80 Hz, because the sub gets boosted (2018–22); See: 3.7 #9, 4.2 #6
Sub delay. Earlier view (year): "Don't time-align subs" (2009–10); delay everything to the sub (2015); Later view (year): Tape-measure the sub like any other speaker (2018–23 PDFs); See: 4.2 #12
Tweeterless widebanders. Earlier view (year): Must be on-axis (2012–14); Later view (year): Aim them at the windshield; on-axis sounds "clinical" (2018–21); See: 1.8 #4
Main measuring tool. Earlier view (year): REW sweeps work "much better" (2012); Later view (year): A live RTA on mono pink noise for production tuning (2018 on); See: 3.3 #1
Target bass lift. Earlier view (year): +6 to +9 dB (2008); +9 dB (MS-8, 2011); Later view (year): +10 dB (2018 PDF); the curve's shape never changed; See: 3.5 #4
Tuning order. Earlier view (year): EQ before delay (2011); Later view (year): Delay first (2013 on); with one EQ per driver, EQ each driver to its crossover target first (2018); See: 3.1 evolution #2–#6
Single mic vs average. Earlier view (year): One mic is worse than an MS-8-style average (2013); Later view (year): A smoothed single mic is ~90% as good, fine for production (2017–21); See: 3.4 #5
Matrix upmixers for music. Earlier view (year): "The ticket" for 2-channel music (2008–10); Later view (year): Poor for music; use upmixers that extract common content (2017 on); See: 1.12 #6
How much speakers matter. Earlier view (year): Little, unless badly designed (2008–11); Later view (year): No DSP removes a bad speaker's distortion (2016 on); See: 5.3 #3
Windshield corner. Earlier view (year): Keep speakers, mids especially, away from the junction (2012, still in Mar 2013); Later view (year): Tuck mids tight into the corner; a pod hanging near it is the problem (Aug 2013, firmly 2021); See: 1.6 #8
Firing mids at the glass. Earlier view (year): Did it in his own car (2009); Later view (year): Narrows the stage (2011); "the most dense comb filter possible" (2021); See: 1.6 #23
Tweeter aiming. Earlier view (year): Co-invented an aimable swivel tweeter at Infinity (filed 1997); Later view (year): Sail panels; aim only the tweeter, and only a little (2011–21); See: 2.5 #7

Where his sources disagree:

Audiofrog's founding year: sources give 2011, 2012 or 2014. Products shipped by mid-2015, and a 10th anniversary was marked in May 2023 (5.8).
Tweeter protection cap: Tuning PDF v1 puts its −3 dB point about an octave below the tweeter's Fs; the amp manual says 50 µF plus a resistor. PDF v2 (2023) drops the step, though his Dec 2020 blog recommends a cap (2.9 #2, 3.1 #1, 3.7 #18).
Tweeter crossover: documents give 1.5–2× Fs, about 2× Fs, 2.2 kHz, 2.5 kHz or 3–3.5 kHz, all well above Fs (2.9 #7, 3.7 #2, 3.7 #7). His claim that LR4 protects better than even 48 dB/oct rests on power near the filter's knee. The research notes could not square it with attenuation an octave lower (2.9 #8, 5.1 #57).
Sub delay: some 2013–21 posts and a 2020 blog say not to time-align the fronts to the sub. The Tuning PDFs and other 2021 posts tape-measure the sub like any speaker (3.9 #12, 4.2 #12).
Tuning PDFs vs his other material: the PDFs set the midrange at 0 dB and show a Q 10 cut at 1.6 kHz. His 2021 video and posts put the target on the dips, and his blog avoids Q above ~6 (3.1 #7, 3.6 #3, 3.6 #7). The video and PDFs also number the test-CD noise tracks differently (3.11 #3).
The attributed DIYMA gain guide vs his own posts: the guide starts DSP channels at −12 dB and uses "7 drum beats" for centering. His 2022 post starts at −6 dB, and he says drum tracks can't check midbass (3.1 #2, 4.8 #4, 3.11 #4).
Mids aimed at the glass: in Sep 2021 he condemned firing mids up at the windshield. From Oct 2021 he praised windshield-aimed factory dash "middlers", a factory-location strategy (1.6 #9, 1.6 #23, 2.3 #9).
A-pillars and crossfiring: he calls pillars "a HORRIBLE place for a midrange", yet Audiofrog copy sells "the ultimate active system for A-pillars" (1.6 #6). He calls crossfired pillar tweeters a "band-aid"; a 2015 Harman/JBL/Infinity tip sheet of unconfirmed authorship recommends it (1.5 #6).
Where bass "impact" comes from: the midrange near 3 kHz in his 2009 and 2021 posts, but correct midbass-to-sub phase on an Oct 2021 dealer livestream (4.2 #3).
Smaller discrepancies:
- CES 1996 class-D blind demo: three listeners after 700–800 demos (told in 2022) vs four after ~1,500 (2023) (4.9 #3).
- Harman patent US6876748: he recalls 250 Hz; its text cites about 200 Hz (1.9 #5).
- MS-8 auto-EQ: 8 biquad filters per channel (2011) vs an FIR-like correction convolved with the audio (2017, 2022) (3.12 #2).
- 1st-order all-pass: flips polarity below its corner (2015) vs pulse checkers show no reversal (2021) (4.14 #3).
- Xmax: one-way overhang "with no +10% fudge" at JBL (2008) vs (coil − gap)/2 + 15% at Audiofrog (2020) (4.3 #21).
- Lexus tap point: a secondhand ClubLexus report has MS-8 working only when tapped before the factory amp, against his usual advice (4.13 #3).

The full series

Download the complete guide as one PDF (351 pages, every point with its explanation).