Skip to content
LiteNetCONNECTIVITY SOLUTIONS

Support

Settings, and
what to do when.

Everything on this page fits on an index card, which is deliberate: most of it has to be read down a telephone to somebody sitting in front of a machine with no working Internet connection.

The whole configuration

LiteNet connection settings
Telephone number+1 610 922 1996 — 16 lines, 2,400 bps
+1 857 342 5877 — 1 line, 14,400 bps
Modem dial stringATDT16109221996
Connection typePPP
AuthenticationPAP — see the note below
IP addressAssigned by the server. Do not set one.
DNSOffered over PPP. If your stack cannot ask, use 10.64.0.1.
ProxyNone. Leave every proxy field empty.
Modem initAT&F
Serial port speed38400 or 57600 — higher than the line rate on purpose
Flow controlHardware (RTS/CTS)
Data format8 data bits, no parity, 1 stop bit

Use PAP, not CHAP

LiteNet authenticates against a local PAP secrets file. MS-CHAPv2 is not configured and has never been tested against a real client, so a stack set to "require encrypted password" will negotiate down or fail outright depending on how strict it is. Set it to accept PAP, or to "accept any authentication".

PAP sends your password in the clear over the modem link. That is worth knowing and it is not worth panicking about: the link is a point-to-point telephone call to us, and the same paragraph in the acceptable use statement explains the much larger thing to be careful of, which is that this service terminates TLS.

There is no proxy to configure

Page transformation is transparent. Requests to port 80 are redirected into it by the router, so there is no host and no port to type into Trumpet Winsock, Netscape, or anything else — and nothing to explain down a telephone. Leave the proxy fields blank.

One consequence: https:// addresses do not work. Ask for http:// and the proxy fetches the encrypted version on your behalf. If your browser insists on a secure connection, it will fail — none of its ciphers have been accepted by a real server for a decade.

By machine

The settings above are the same everywhere. These are the places each system hides them.

Windows 95, 98, NT 4

My Computer → Dial-Up Networking → Make New Connection. Name it LiteNet, pick your modem, enter the area code and number.

Then right-click the new icon → Properties → Server Types. Set PPP: Internet. Clear Log on to network, NetBEUI and IPX/SPX — they cost negotiation time on a 2,400 bps line and do nothing here. Leave Require encrypted password unticked. Under TCP/IP Settings, leave both address options on Server assigned.

Windows 3.1 with Trumpet Winsock

File → Setup. Tick Internal PPP. Leave IP address as 0.0.0.0 so the server assigns one. Set the name server to 10.64.0.1. Baud rate 38400, hardware handshaking on.

Trumpet dials from a login script, so the username and password go in login.cmd rather than a dialogue box. If you would rather not write one, dial manually with Dialler → Manual login, type ATDT16109221996, wait for CONNECT, and press Escape.

Mac OS 7.5 to 9

Open Transport PPP or Remote Access: enter the number, your name and password, and in the TCP/IP control panel set Connect via: PPP with Configure: Using PPP Server.

On an older system running MacTCP with MacPPP, configure the server in ConfigPPP first, then set MacTCP to PPP and Obtain address: Server. MacTCP caches its settings aggressively — restart after changing them.

Amiga

Miami or AmiTCP with ppp.device. Interface type PPP, dynamic IP, PAP authentication, serial rate 38400 with 7-wire handshaking. Point the resolver at 10.64.0.1.

If you are using the internal serial port on a stock A500 or A600, drop the port speed to 19200. Its UART loses bytes above that under load, and the symptom is a PPP session that comes up and then stalls, which looks exactly like a line fault.

Anything else with a PPP stack

OS/2 Warp, NeXTSTEP, an early Linux with pppd, a DOS packet driver — all fine. LiteNet is a plain PPP server. Dial, authenticate with PAP, accept the address it gives you, and use the DNS it offers.

For pppd specifically: noauth, usepeerdns, defaultroute, and your credentials in pap-secrets. Do not pass require-mschap-v2.

Which browser

Pages arrive as HTML 3.2, so anything from Mosaic 2 upwards renders them. Netscape 3 and Internet Explorer 3 are the comfortable choice. Mosaic and MacWeb work and get a simpler layout, because tables are held back on the slower profiles rather than risked.

No period browser has yet rendered a page from this service. The output is built against a documented set of period rendering bugs and checked automatically, but until real hardware dials in, that is a claim about the markup and not about your screen.

When it does not work

Four symptoms, in the order you meet them. Each one rules out everything before it, so read from the top.

Symptom 1

The modem dials and you get NO CARRIER

The call is not reaching us, or it is reaching us and your modem cannot hear ours. Check the obvious first: an outside line prefix, a tone versus pulse setting (ATDP rather than ATDT), and whether the line has voicemail stutter tone on it, which some modems refuse to dial through.

If the number rings and rings, all sixteen lines are busy. There is no queue; hang up and redial.

Symptom 2

CONNECT, then nothing — PPP never comes up

The carrier trained but the two ends did not agree. Almost always authentication: a stack demanding an encrypted password against a server offering PAP. Turn off "require encrypted password".

The other cause is flow control. If your serial port is set to no handshaking, a 38400 port feeding a 14,400 line overruns the modem's buffer during LCP and the negotiation dies part-way. Hardware handshaking, always.

Symptom 3

PPP is up and you have an address, but nothing loads

Check first whether name resolution is the problem: ask for http://10.64.0.1/ by address. If that page appears, the connection is fine and your machine simply has no working DNS — set the resolver to 10.64.0.1 by hand.

If your address is not inside 10.64.0.0/16, tell us. Every subscriber address has to come out of that range, and one that does not gets every forwarded packet dropped by LiteNet's own anti-spoofing rule — while pings to the gateway keep working, because a local address never reaches the forward chain. It reads exactly like a modem fault and is not one.

Symptom 4

Pages load, but the pictures are missing

That is the service working as designed, not a fault. LiteNet reads the rate your modem actually negotiated and picks a budget from it:

Transformation profiles by measured goodput
ProfileApplies atWhat you get
textup to 400 B/sNo images at all. Pictures become numbered links you follow one at a time. Tables become lists.
lowgfx400–800 B/sUp to 3 images, 320 px, 32 colours. Still no tables.
gfxabove 800 B/sUp to 6 images, 480 px, 64 colours, and simple tables are kept.
turboyou asked for itThe text budget on a fast line, when you would rather have the page now.

http://10.64.0.1/line tells you, while you are dialled in, exactly which profile you are on and why. A 2,400 bps connection lands on text, and on that line a single 20 KB photograph is over a minute of waiting.

Still stuck

Write to [email protected] with the machine, the modem, what the modem said, and how far it got. If you saw a CONNECT message, the speed it reported is the single most useful line in the whole message.