How to Choose a Sunrise Alarm Clock: 5 Criteria

Five things actually matter when you’re choosing a sunrise alarm clock: whether the light is bright enough at your pillow, whether the ramp itself is any good, what the device demands from you (app, account, subscription), what happens when the power or internet goes out, and whether it’ll still be working in ten years. I judge these differently than a spec sheet does, because I’m coming at it from an odd angle: I own the category’s long-standing top pick, the Philips SmartSleep HF3520, and I liked it enough that I have spent months building my own sunrise engine (Horizn) to fix the one thing it doesn’t do. So consider this a buyer’s guide from someone who liked the top pick enough to build his own from scratch.

1. Brightness that reaches your eyes, lux at the pillow, not on the box

Light intensity falls off fast with distance, so the only number that matters is what arrives at your face, not what the LED emits at the lens. The frustrating part: most vendors won’t tell you. Philips is the honest exception, per Philips’ manual, the HF3520 delivers 1–300 lux at 45 cm, with 20 intensity levels. Hatch publishes no lux figure at all for the Restore 3; I checked their documentation looking for one. For the budget JALL, an independent light-meter test (OptimizeYourBiology, which tested 15 clocks) measured 545 lux at one foot and 275 lux at 18 inches: genuinely bright for a clock in the $35 range.

If a product page won’t state lux at a distance, treat the brightness claim as decoration. I go deeper on what the research says about how much light you actually need in how bright should a sunrise alarm be.

2. The quality of the ramp: the invisible spec

This is the criterion nobody prints on the box, and it’s the one I care most about as a builder. Two things separate a good ramp from a lamp on a dimmer:

Perceptual smoothness. Your eye doesn’t perceive brightness linearly, a light ramping linearly in power looks like it sits dark forever, then lurches to full. A well-engineered ramp is shaped to how humans perceive light, so the brightening feels continuous. You can’t read this off a spec sheet; you can only notice its absence as visible “steps” in the early minutes.

A pre-dawn lead. A real dawn doesn’t start at zero and sprint. The gentlest wakes I’ve had come from a long, barely-there warm-up before the main ramp. Per Philips’ manual, the HF3520’s sunrise runs 30 minutes by default, adjustable to 20, 25, 35, or 40: a sensible range. Hatch’s support docs put the Restore 3’s lead time at 5–60 minutes. Longer options are better if you wake hard.

This is the part of the problem Horizn is built around: a 60fps render engine on its own dedicated core, a gamma-corrected brightness curve with four tunable points, and a separate pre-dawn phase, because the ramp is the product.

3. What needs an app, an account, or a subscription, and what happens offline

Read this section before you buy anything, because it’s where the category quietly split.

The Philips is fully standalone: per its manual, no app, no account, no Wi-Fi, no subscription. The JALL is the same, per its manual, no app, no account, no subscription; everything is set on the device itself.

The Hatch Restore 3 is a different animal. Does it need a subscription? For core alarm use, no, but per Hatch’s own support docs, the Hatch Sleep app and an account are required for setup, and setup requires starting a Hatch+ free trial. That’s not me editorializing; Hatch has a support article literally titled around the question “I can’t set up my Restore 3 without a free trial, is this normal?”, and the answer is yes (it’s cancelable). Hatch+ runs $4.99/month or $49.99/year; the free tier keeps the core alarm and 40+ sounds, while the full light, sound, and routine library is subscription content. Wi-Fi is required, schedules run when connected, and there’s no documented offline mode.

None of that makes it a bad device. It makes it a device with dependencies, and dependencies are a reliability question, which is why my own device runs entirely on the local network with no account and no cloud, and keeps its schedule with the internet down.

4. What a power outage does to your alarm

An alarm clock has exactly one job at exactly one moment, so ask the ugly question: what happens if the power blips at 3 a.m.?

Per Philips’ manual, the HF3520’s backup preserves clock time only, for up to 8 hours, the alarm and light won’t fire during an outage, but the clock won’t be blinking 12:00 when power returns. Per the JALL manual, all functions stop in an outage; a button cell saves your settings only. The Hatch’s dependency is connectivity rather than just power: schedules run when connected.

Nothing in the Philips or JALL manuals, and nothing I could find in Hatch’s support docs, describes catching up an alarm that a power outage swallowed. That specific failure is why Horizn has a power-loss catch-up mode and an on-device log of every alarm decision it makes, so I can audit exactly why it did or didn’t fire.

5. Will it still work in ten years?

Per Wirecutter and the Philips manual, the HF3520’s LED is sealed, rated for roughly 7 years of regular use, and not replaceable. There are no user-serviceable parts except the adapter. It carries a 2-year warranty and has been on the market for about a decade, which is its own kind of endorsement. The JALL, per Tom’s Guide, offers a lifetime warranty via email registration: unusual at $35. For the Hatch, the long-term question isn’t the hardware, it’s the ecosystem: a device whose setup and full feature set route through an app, an account, and a subscription is only as durable as the company’s willingness to keep those servers on.

The three you’ll see recommended everywhere

Philips SmartSleep HF3520 (~$150 list)

This is the one I own, and it earned its reputation, about $150 list per Philips’ product page, and often discounted. It’s the long-standing Wirecutter/NYT top pick, it’s completely standalone, it publishes real lux numbers, and it reliably wakes me. Wirecutter’s flaws are fair, the small 3/4-inch display, small same-size buttons, tinny rear speaker, and awkward programming, but none of them touch the core function.

My honest first-hand take, as the person this device inspired: the “colored sunrise”, soft morning red through orange to bright yellow, roughly 1500–2800 K per Philips’ product page, reads in the room as essentially one warm color that gets brighter. It’s a glow that intensifies, not a dawn with a color story. Real dawns move through violet, magenta, ember, amber across the sky. That gap is literally why Horizn exists.

Hatch Restore 3 ($169.99 list)

Credit where due, and the $169.99 list price is per Hatch’s own page: Wired rated it 8/10, found it bright enough to wake, and praised the physical controls. Tom’s Guide, after 2–3 months of testing, judged it the best sunrise alarm you can buy online, praising the design and its content library. The free tier is genuinely usable, too. It keeps the core alarm and 40+ sounds, per Hatch’s support docs. Judged by my five criteria, though: no published lux figure, mandatory app and account (and trial-start) for setup, Wi-Fi required with no documented offline mode, and Tom’s Guide’s own cons, price plus subscription, the mandatory app, and rivals like the Philips having brighter light.

JALL Wake Up Light (~$35)

The budget pick, and a legitimately strong one: independently measured at 545 lux at one foot (OptimizeYourBiology’s 15-clock light-meter test), 20 brightness levels per its manual, fully offline with no app or account, and a lifetime warranty via email registration per Tom’s Guide. The trade-offs are ergonomic and structural: roughly a dozen side buttons that Tom’s Guide says will have you consulting “the instruction booklet a few times,” a tinny speaker per Vice, and, per the manual, every function stops in a power outage. Sunrise ramp options vary by revision, so check the listing for the unit you’re actually getting.

So which one should you get?

Today, honestly: if you’re budget-first, get the JALL: the light itself punches far above its price. If you want a content library, soundscapes, and don’t mind living in an app ecosystem, the Hatch Restore 3 is the best of that breed. If you want a standalone premium device that just works with no strings, the Philips HF3520 is still the safe pick, and it’s the one on my nightstand. (If you’re still deciding whether the category works at all, start with do sunrise alarms actually work.)

What none of them deliver is the dawn itself, the color trajectory, the perceptually smooth ramp, the pre-dawn lead, the reliability engineering. That’s the gap Horizn is being built to fill: local-first, no account, no subscription ever, and a sunrise that moves through violet, magenta, ember, and amber like the sky does. It’s pre-launch, I wake up to the prototype at 4:34 every morning, and there’s a waitlist if you want to follow along.