Dev Investigation Notes
How to check hotspot connection limits before attaching multiple phones to a laptop server
This is a translated article. Read the original in Korean ↗
I was building a web app called formation for dance studios. Everyone in the room follows the same choreography video on their own phone, kept in sync. A laptop acts as the server and the phones connect to its address. The question was whether we could open a hotspot and connect about 10 people to it. I asked an AI coding assistant. It was a design question, not a bug, and on paper the bandwidth looked like plenty.
Conclusion
Check two things before you use a hotspot. (1) How many devices the phone hosting the hotspot can take at the same time. Ten viewers plus the server laptop all have to join the same hotspot, so the limit has to be 11 or more before the hotspot is even a candidate. This article could not confirm that number. Check it in the manufacturer's or carrier's documentation, or with a real connection test. The "5 devices" figure in the body has no source, so don't treat it as the answer (how to check is in step 1 of the Fix section below). (2) Whether one viewer phone on the hotspot can actually open the laptop server's address. If it can't, first check that the laptop's firewall allows Node.js (or the server port) on both private and public networks . If it's allowed, you've confirmed the address is right, and it still won't open, suspect client isolation (a setting that blocks connected devices from talking to each other). Bandwidth math can pass and you can still get stuck on either of these. In that conversation, only the math was done. Neither check was run, and the hotspot was dropped in favor of the studio Wi-Fi .
Bottom line: a hotspot only becomes a candidate after it passes both checks. Until then, the router (studio Wi-Fi) is the default. Nobody has measured yet how many devices that router can handle.
Situation
In this web app, a leader phone sends the playback position and the viewer phones follow it . The server runs on Node on the laptop. It reads video segments from disk and streams them out in the response . The question came up in a conversation with an AI coding assistant . The address viewers use to reach the server changes every time the network changes, so the project later moved to opening it with a QR code .
Symptom
All anyone saw on the surface was one question: "Is it OK to connect 10 people?" The math was simple. One video is about 4.5Mbps, so 10 devices come to about 45Mbps . Once the hotspot came up, the numbers were broken down by quality level: 10 people × 480p is about 11Mbps, and × 360p is about 6Mbps . Going by the math alone, the answer stops at "lower the quality and it'll fit."
Cause
The math looks only at bandwidth. On a hotspot, though, the connection itself can fail before bandwidth ever becomes the issue.
(a) Simultaneous connection count. The answer said "iPhone Personal Hotspot has long been capped at 5 devices." It added the caveat that this varies by model and carrier, and gave no source . With 10 people, that one number would leave more than half unable to connect. The number stayed unconfirmed in that conversation too.
(b) Client isolation. Some hotspots block traffic between the devices connected to them. The internet works, but the phone next to you can't reach the laptop. In this app the phones connect directly to the laptop, so if isolation is on, not a single viewer will load .
On top of that, the claims the bandwidth judgment rested on were calculations or assumptions, not measurements. That covers the line "the only variable is WiFi," the figure "measured combined 200–400Mbps," and "CPU/memory load is close to zero" .
Fix
Check these two things before you use a hotspot.
- Find the maximum number of devices the hotspot phone allows (it needs 11 or more, counting the server laptop). Look up that model's simultaneous connection limit in the manufacturer's or carrier's documentation, or first test whether 10 viewers and the laptop can actually connect together. If you can't confirm the limit, you don't know the number, so drop that hotspot as a candidate. Treat the "5 devices" from the session as an unsourced number .
- Connect the laptop and one viewer phone to the same hotspot. On the viewer phone, pick the hotspot name from the Wi-Fi list and enter the password. The laptop has to be on this hotspot too, or the phone can't reach the server. For the laptop's address and port, copy the "📱 뷰어 접속" ("📱 Viewer access") URL from the server page (
server.html) . If that URL looks wrong, it may be because the server automatically picks the first non-internal IPv4 address, so it can grab the wrong one when the laptop has several adapters . If the laptop is hosting the hotspot, the gateway is usually192.168.137.1. If a phone is the hotspot, the address is in the192.168.x.xrange that phone hands out . The port used in the session was 3000 . Open the resultinghttp://노트북주소:포트/viewer.html(laptop address : port) in the viewer phone's browser.
- If it opens, there's no isolation. Move on to the next step.
- If it doesn't open, suspect client isolation first. Also check whether the laptop's firewall is blocking the server port.
- If you can't find a setting to turn isolation off, don't use that hotspot.
Only put the hotspot back on the table once both checks pass. In the end, the decision was a judgment call: the user dropped the hotspot and went with the studio Wi-Fi . That wasn't based on a measurement, so how many devices the studio router can handle needs to be checked the same way.
Bonus — keep calculated numbers separate from measured ones
45Mbps for 10 devices is a calculation, and the "measured 200–400Mbps" was never actually measured . If you put both kinds of number in the same sentence in a write-up or a meeting, listeners will take the calculation for a measurement.
Reporter's Notes — 45Mbps checks out, but can they even connect?
The conversation from the day a hotspot was being considered instead of the studio Wi-Fi.