This article originally appeared in TidBITS on 2012-08-19 at 3:20 p.m.
The permanent URL for this article is:
Include images: Off

Aging Cable Modem Causes Performance Decline

by Michael E. Cohen

I got my current iMac a couple of weeks before the release of Mac OS X 10.7 Lion last year — in fact, Lion was the reason I decided to upgrade from my 2007 model iMac. As a registered Apple developer, I was eligible to download the golden master candidate of Lion a couple of weeks before release, and as someone who writes about technology I felt it was important to learn as much as I could about it as soon as I could. I also knew that my old iMac was reaching the end of its supported life, and that a number of Lion features (such as AirDrop) would not be implemented on it. So I migrated my Snow Leopard installation from my old iMac to the new one, and then downloaded Lion and installed it over Snow Leopard.

As with any migration and upgrade, some things felt snappier, and some didn’t. Overall, though, the performance of the new software on my new iMac was pretty good, but, at that time, I wasn’t so much interested in performance tests and specs as in learning the various new features and exploring the interesting idiosyncrasies of Lion. But gradually I began to notice a certain sluggishness. Web pages took just a little longer to load; videos seemed to buffer more often than before; Dropbox files seemed to take just a little longer to sync. It wasn’t so much as to be alarming, just a little annoying.

At the time, I chalked it up to the early version of Lion not being fully optimized, coupled with various digital detritus that had made its way onto my new iMac from my old one. I made a note that I would, when I had the time, take a look into Joe Kissell’s “Take Control of Speeding Up Your Mac [1]” and see if I could track down what was making my shiny new iMac behave like it had a mild case of arthritis.

Of course, when it comes to dealing with housekeeping chores like that, there is usually something more urgent, or more interesting, to do. So I just kept putting off any serious attempt to fix things, especially since things weren’t really broken, just sluggish.

The slowness problem got worse. But it did so gradually enough not to make me take action. When I did think about it at all, I thought maybe it was my ISP being overloaded: I had noticed a steady lengthening of my AirPort menu as more and more neighbors activated more and more wireless access points around me, and I knew that cable modem connections were apt to show slowdowns as local usage and congestion increased. Besides, I had articles to do, as well as books to write and edit and help produce.

Then came the first TidBITS Presents event (see “Watch Joe Kissell and Adam Engst in TidBITS Presents: Adieu MobileMe [2],” 16 June 2012). Adam had asked me to lurk in the background and monitor the presentation, notifying him via a separate chat window if I spotted any problems. I was unable to do so: my link to the presentation kept losing sync, or freezing, or just being dropped. Finally, it was time to do some real investigation into my slowdown issues, about the same time Adam was experiencing his own bandwidth-related problems (see “Are You Getting the Bandwidth You’re Paying For? [3],” 11 July 2012).

The first thing I did was fiddle with the Wi-Fi channel my base station was using: I figured that with so many base stations sprouting up around me, maybe I should move to a non-default Wi-Fi channel where I might expect less congestion or interference. Doing that seemed to make things snappier when I restarted everything, at first, but the slowness soon (and by soon, I mean within hours) recurred.

Then I powered everything off, disconnected my base station, and connected my cable modem directly to my old iMac: the Ethernet cable wouldn’t reach to my new iMac, and besides, I wanted to see if it was the Lion installation that was the issue (my old iMac was still running Snow Leopard). Again, the slowness was somewhat alleviated, but only temporarily, just like it had been when I fiddled with Wi-Fi channels. I reconnected my base station and went back to my Lion-equipped iMac.

On my iMac, I went to [4], ran the test, and saw that I was getting roughly 1.5 Mbps on downloads: about twice what I used to get with DSL years earlier, but well below what I was paying my ISP for. Finally, I made the phone call to my ISP to complain.

Over the phone, the technician ran a test or two and claimed that the cable modem seemed to be working fine. He instructed me to disconnect my base station, and to connect via Ethernet directly. I told him I had done that and didn’t want to crawl beneath my desk and fiddle with cables again. He said he really couldn’t help me, since the problem was probably at my end, but, after I complained more insistently, he reluctantly scheduled a service appointment for me for later in the week.

When the next technician arrived, he took one look at my cable modem and said, “I’ve seen this before.” According to him, the ISP had been delivering this particular model of cable modem to customers along with a power supply (a typical wall-wart-type transformer) that delivered just slightly more voltage than the modem required. This power supply apparently caused those modems to generate slightly more heat in operation than was optimal, and it caused the modems to fail gradually over time as internal components began to emit extra radio frequency noise — static, in short — that caused packets to drop and be resent. It all sounded somewhat specious to me, but it did seem to explain the gradual performance decay. What’s more, when he swapped out my old modem and power supply with a different modem, I suddenly had ten times the downstream bandwidth that I had been getting earlier.

To this day, I don’t know if it was indeed overheating or something else that caused my old cable modem’s performance to degrade slowly, but the cable modem, for whatever reason, really was the cause of my bandwidth woes.

I learned several things from this: