don’t know what made me think of this but i went back and dug it out of my email from early 2000..
this is a series of emails i got from a customer who was having trouble getting connected through our dialup service back then..
it’s really terribly long but pretty funny..
i debated editing it down but i think it works better in its entirety..
———————————————————————————————-
This is Bill Covington with new account changedtoprotect@nwark.com having problems connecting. Using the exact same equipment, I connect to arkansasusa.com consistently at 49,333. If I connect to interface at all, I get 28,800.
The attached wavefile (8-bit 8kHz PCM) of the handshake contains a curious sequence of repeatable clicks. Analyzing the click spectrum shows each one to concentrate energy consistently around 2165 to 2175 Hz. I don’t hear these clicks when connecting to arkansasusa. I have disconnected all telephone equipment from the line with no improvement. I haven’t tried my other computer on another line, but I will.
My reach number is 555-8409.
——————-
I hooked a scope, dc coupled, to the phone line and watched the clicks. The clicks appear on the line as a series of square 1 volt positive pulses lasting about 2 milliseconds. Their magnitude is at least as great as the modem signals. The 2170 Hz ringing in the WAV file must just be some resonance in the line or microphone arrangement. I didn’t see any clicks after the first few seconds of training, but it may be that the line equalization procedure is already derailed.
Bill Covington, changed@nwark.com, 555-8409, 555-2605.
——————-
Here is a much better recording of the clicks I am experiencing. This was recorded without a microphone. I have transformer coupled the phone line to the line input of my sound card. I put a 10 microfarad capacitor between the transformer and the phone line for dc isolation. If I call arkansasusa with the same equipment, there are no such pulses anywhere.
——————–
The only thing I didn’t do is to put a resistor in the line to prove that the pulses are coming into my equipment from an external source. It’s possible my modem is creating the pulses, but it hasn’t done it before.
Examining the pulse pattern shows it to be very consistent. Each pulse is 2ms long (16 PCM samples). They frequently come in pairs with a 12ms center-to-center spacing. Near the end of the training sequence, they space at 20, 60 and 50 ms pretty much exactly.
———————
I guess I couldn’t stand the suspense. I put a resistor in the line and proved that the pulses are coming into my house from an external source (telco or IFWORLD).
———————
This morning, I tried a different experiement. I clicked connect on the PC normally, but as soon as the digits were dialed, I picked up a regular telephone on the same line and unplugged the (internal) modem for 1 to 2 seconds. I then plugged it back in and dropped the regular telephone. This appeared to connect to your modem but failed to negotiate with the server.
I tried it again with the same results. I tried it a third time and it connected successfully at 49,333. I then proved the connection by browsing the internet for a few minutes, confirming that I had a good, fast connection.
This experiment proves that the line is good, but something is hosing the handshake and equalization. The purpose of temporarily pulling the plug is to delay the handshake to throw off the timing of the clicks, hopefully with postive results - indeed this is occasionally the case.
———————-
At your suggestion, I called in a trouble report on my residential line and was fairly pleased with the response. There is a pair-gain multiplexer serving the Brookbury neighborhood. At a minimum, this system creates an additional A/D/A conversion in the path between me and the Central Office (CO). The best theory on clicks so far is that the pair gain system creates the clicking phenomena. This is consistent with my observations here in that the clicks are very sharp, square and clean, implying that they are generated physically close and not at the other end of 30,000 feet of twisted copper, which would cause them to appear smeared and rounded due to the natural filtering of the line.
The remaining question is then why is arkansasusa fine and nwark a disaster?
My guess is that your equipment is transmitting a signal that is too strong and overloading the pair gain system. The experiment to perform is to call both providers and use the same setup to record the signals and critically compare the received signals at my house, particularly regarding signal level. This will confirm any differences in the signals from the two ISP’s, and there is obviously a very significant difference that will negatively impact your business in this neighborhood.
———————–
cracker: “uh.. sure dude.. um.. ok.. great then.. nice work with the 10 microfarad deal and all that.. i like what you’ve done there.. i’ll check the flux capacitor and the hemi-friction modulator.. er.. ok then.. thanks..”