Apple versus the analog monster

Let me spoil it for you: Apple loses.

No, this is not a post about DRM or the analog hole. It’s about something much less important, but potentially just as infuriating.

In August of 2003, I ordered a Power Macintosh G5 with dual 2HGz CPUs. This was a “revision 1” product; it was the very first Mac with a G5 CPU, and it came in a brand new enclosure. I ordered it sight-unseen, knowing the risks.

Well, I thought I knew them, anyway. My previous machine was also a revision 1 product (a Power Macintosh G3), also top-of-the-line (400MHz baby!), and also in a brand-new enclosure. It had a few issues. The IDE controller had a bug that prevented it from supporting two drives. The internal drive sleds were a single-level only even though there was room for a stack. These are the kinds of things I expected from my new Power Mac G5: some errata in an i/o controller chip, or perhaps some buggy firmware. That sort of thing.

What I got was something very different. The silicon was pretty close to flawless. The “mechanicals” were exemplary: all nine(!) fans ran flawlessly1. All the ports, both internal and external, worked as advertised. I thought I was home free.

Then I noticed a strange noise…a chirping sound. I chased down the source, I searched the net for solutions, and I eventually wrote about it here at Ars in March of 2004. The summary: the power supply in the revision 1 Power Mac G5 made chirping noises, and there was no hardware-based fix in sight.

In fact, the only remotely useful work-around was to disable the “nap” feature of the G5 CPUs. This worked because the noise was caused by fluctuations in power draw. The nap feature allows the CPUs to briefly drop down to a lower speed, then “wake up” instantly when needed. This constant cycling from low to high power caused the power supply to emit a rhythmic “chirp” noise in time with the fluctuations.

Any change in power draw could cause the same effect. One common example was the use of the exposé feature, which caused a momentary surge in power requirements as the CPUs and the GPU quickly scaled and moved windows on the screen. Even scrolling would do it sometimes. Google for "g5 chirp expose" or just "g5 chirping" to see several thousand hits on the topic.

Before I continue, let me dwell a bit on the peculiar nature of this problem. A “little thing” like a quiet chirping sound probably won’t bother a lot of people. In fact, it isn’t audible at all above the din of the average office (or retail store…). So right away, it seems like we’re only talking about people like me: anal-retentive “perfectionists,” people who agonize over one dead sub-pixel on an LCD containing over six million perfectly good ones them. Call us "the crazy ones," if you will.

Let’s think a bit deeper about this problem. Sure, all computers make some sort of noise. Even without fans there’s still the hard drive spinning with its disk heads moving to and fro. Moving parts mean noise. But a power supply? The only things moving in a power supply are electrons, right? AC in, DC out. What’s to move?

If you owned one of those 150-in-one kits from RadioShack as a kid (or if you have a degree in electrical engineering…pretty much the same thing, surely) you know full well that analog electronic components can and do move—ever so slightly, mind you—and therefore can make noise. The transformer, in particular, was quite a noise-maker in my 150-in-one kit. Perhaps not surprisingly, one of the many things a transformer can do is convert AC to DC and vice versa. Hmm. [Update: Transformers may be part of a circuit that converts AC to DC, but they cannot do so alone.]

But even the more boring electronic components can make noise under the right circumstances: resistors, capacitors, you name it. Throw enough energy through these tiny bundles of conductors and insulators and yes, Virginia, they can start to vibrate. Vibration == motion == noise!

In short, analog electronics suck. In the context of digital computers, they’re simply a necessary evil. Analog electronic design is a fuzzy world with lots of black magic and few absolutes. What separates a good piece of electrical equipment from a bad one is how well the designers wrestle the analog demons into submission.

Let’s get back to the Power Mac G5’s power supply. Presumably Apple gave its supplier clear specifications. “Peak power draw will be X, minimum will be Y, source is AC at such-and-such volts, result should be blah volts DC.” But maybe they left out the part about the power the exact nature of the power fluctuations? “Power draw may fluctuate from min to max N times per second.”

Whatever happened, someone dropped the ball. A power supply that makes noise may not be an "F" (a grade reserved for catching fire, perhaps) but it’s certainly a “D-”. This is amateur hour.

Anyway, that’s all in the past. In my original lament about the chirping, I expressed my hope that the “revision 2” Power Mac G5 would come with a silent power supply. Since then, Apple has released three more revisions of the Power Mac G5. Although the enclosure is externally the same as it was in revision 1, the CPUs have changed at least twice, the internal details of the case have been tweaked several times, and the motherboard even went through a few different designs.

Yet through all of this, one thing has remained unchanged. The stupendously sucktacular Chirping Power Supply of Doom™ is still a fixture of the Power Macintosh product line to this day! There’s a seemingly eternal supply of interminable threads on this topic in the Apple support forums, with smaller flare-ups elsewhere on the net.

Everyone goes through the same dance in these 2006 threads as we rev-1 G5 owners did back in 2003. “What’s making the noise?” (The power supply.) “What can we do to fix it?” (Try to minimize power fluctuations by disabling CPU nap.) “Will Apple replace my power supply under warranty?” (Probably.) “Will that totally silence the chirping?” (Probably not.)

The whole process often takes weeks to cycle through. Usually, those that know the drill have long since stopped reading such threads and therefore can’t share their past experiences to expedite the process. In my case, once I disabled CPU napping and put my G5 under my desk (my first Mac ever to be placed there…a sad day) I just tried to push the whole chirping thing out of my mind. But I was still at least aware of the continuing reports, and I remained amazed that Apple was not doing anything about this problem.

Crappy analog electronics have plagued several Mac models over the years. In particular, PowerBooks have “whined” in varying amounts since at least the dawn of the AlBook. But I still held out hope that the transition to Intel would wipe the hardware slate clean and that the Intel-based Macs would have competent analog electronics inside them.

The Intel iMac was a good start: quiet, cool, and no audible noises from the analog circuits. Then came the Mac Book Pro. In case you haven’t heard, it hisses. (Not all of them, mind you, but enough to be a minor story on the Mac web.) And—surprise!—adjusting the power draw (by, say, disabling a CPU core or changing the brightness of the display) is the only way to quiet the hissing.

Now you can squawk all you want about how a little noise isn’t the end of the world, and tell me I’m being an unreasonable perfectionist, blah blah blah. I have two things to say to that. The first is that crappy analog electronics actually have very real consequences for a certain class of user: audio professionals. You know all of my whining about my G5’s periodic chirping? Well audio pros had it a lot worse. They had G5s that put out periodic noise pulses on their audio inputs and outputs!

(Oh, and there were also a few ground loop problems for audio professionals, for good measure. Again, more analog electronic gremlins. Apple tried to blame this problem on the users, but I’ve read enough reports from what seem like very competent audio professionals to seriously doubt this assessment.)

And lest you think this madness can’t enter the digital domain, go find a Power Mac G5 and hook up a few oscilloscope probes to the USB or FireWire port on the front of the machine. Then try scrolling a few windows or triggering exposé and watch the scope come alive with noise. (There used to be a nice video of this, but it’s offline now. The way-back machine only has the web page text, but not the video.)

So that’s the first thing: this issue is more than just cosmetic. The second thing is this. So what if it was “just” cosmetic? Do I really have to spell this out? Apple is shipping computers with crappy analog electronics! It has been doing so for years, and it continues to do so, right into the Intel transition.

This is a sad, sad state of affairs. In what industry is it acceptable to ship an electronic device with analog components that make noise under normal operation? Seriously, even Joe Engineer making kitchen blenders is held to this standard. Wake up, Apple! Analog electronic circuits should not make noise! You are failing to produce adequate analog circuits! Forget about “excellent” or even just “good.”

How in the hell is this happening? Surely it is someone’s job inside Apple to either design these circuits or work with outside vendors to produce them. In either case, someone at Apple has the job of either accepting or rejecting these analog bits. To this person or group, I have this tip to offer. When your analog circuits make audible noises, you need to try again. Change the design. Send it back to the vendor. Whatever you have to do. This is not acceptable behavior!

For the love of Woz, people! Someone is spending a week designing tiny details of hardware enclosures at Apple, and we can’t get, say, one competent electrical engineer (or engineering firm, if sub-contracting) to make a noise-free circuit or component? Yes, I know it’s a complex problem encompassing the whole system and not just one piece in isolation. But it can be done! Apple did it for years and years, and continues to do so in most cases (e.g., the Intel iMac). That just makes it all the more unfathomable when they screw it up, and absolutely infuriating when they let an entire product line languish in chirpy hell for years!

The worst thing about this situation is that there are so few people who care. Analog noise problems exist in a nether-world between big things that everyone will yell about, like logic board failures, and little things that will annoy enough “normal” customers to cause a stink, like iPod scratches. Most Mac users don’t have Power Mac G5s and Mac Book Pros. There’s also great variability in the volume of the noises even among affected models. (See? Analog electronics! Fuzzy! No absolutes!) And even if you have a noise-maker, if you’re in an office or you’re far away from your G5 tower or your high-frequency hearing isn’t very good, you may not notice at all.

My appeal is to the better angels of Apple’s hardware design group. Someone, somewhere should be embarrassed by this. I’m embarrassed for Apple. I also don’t want any noisy analog electronics in my future Macs. A noisy fan or hard drive I can replace or fix. But a noisy custom-designed power supply buried deep within my G5 tower enclosure, or an entire laptop? If a component or system swap under warranty doesn’t fix these problems because they’re actually inherent in the design itself, I’m doomed.

The fundamental premise of this rant is a simple one. I will spell it out once again. Analog circuits in consumer electronics should not make audible noise under normal operating conditions. Period, end of story. No negotiation, no qualifications. I find this premise so obvious that it seems absurd to even write it. Either Apple does not agree with this premise, or it is willingly shipping incompetently designed analog circuits. I find both possibilities troubling, but the first is terrifying.

I’m hurtling towards a date with destiny in the form of a quad-core 64-bit Intel-based Macintosh. It will likely be yet another “revision 1” machine. I don’t expect perfection, and I certainly don’t expect total silence. All I ask is that all the noises it emits come from large-scale moving parts. Pride, sanity, and the eternal spirit of electrical engineering demands it.

  1. …flawlessly, that is, until two months later when the $0.02 fan on the $399 ATI video card began to make a noise like an electric toothbrush. I replaced it. I had to use thermal paste. It made me feel dirty…like a PC user.

This article originally appeared at Ars Technica. It is reproduced here with permission.