How Fast Is Your Modem?

by
David Rockwell, Columbia, MD,
Long Reach High School
Columbia, MD.

Synopsis

The purpose of my experiment was to test the actual download speed of conventional modems. Acting on the hypothesis that the actual download time was less than that advertised. To do this I designed a crude test. What I did was:
  1. Upload a test file onto America On-line(*test.txt- 100,000 bytes).
  2. Download the file from America On-line at 10:00AM with a 33.6k modem.
  3. Use the formula (file size)/download time(seconds)= bytes per second x 8= Bits per second
  4. Record the data for three different times of the day- morning(10AM), afternoon(3PM), evening(10PM).
  5. Repeat steps 2 , 3, and 4 with the 28.8k modem and the 14.4k modem.
  6. Take the averages of the three tests and enter all data onto a comparative chart.

With the limited supplies I had at my disposal I had to use a Pentium 133 for both the 33,600bps and the 28,800bps modems, using a Pentium 100 for the 14,400. I realize this may have altered results slightly. In fact, I couldn't eliminate all of the factors that acted on the experiments, such as phone line "noise" or interference, or the time it takes to transfer information from digital(for a modem to read) to analog(for a conventional phone line, (or analog, to digital).

The results of my experiment proved my hypothesis correct. There was a differential of 7200bps in the 33,600bps modem, a differential of 6400bps in the 28,800 modem and a differential of 4000bps in the 14,400bps modem. I explain this drop off in speed due to the factors that I mentioned previously and that a stop bit is needed to distinguish between bits, and I had no way to predict how many times stop bits were needed.

Walt Houser CPCUG Coordinator.

CPCUG Science Fair Home Page