An example of stupid design

18^th September 2015

The history of engineering is littered with examples of poor design. Usually an otherwise excellent design is let down by one simple mistake in a critical component. Today I shall be looking at something that isn't life-threating and is more of an annoyance.

The Audio Alchemy DAC-in-the-box

In the early 90s this product was well regarded in hi-fi circles. It was a worthwhile upgrade for a budget CD player (assuming you had a digital output, which was far from standard back then). It gave good quality sound, and was well constructed - apart from one small detail.

The power supply.

The problem

The design fault was simple and breathtakingly obvious. The unit required a split-rail +/-12V DC power supply. Audio Alchemy had chosen to use a 3.5mm minijack plug (the same type as used on Walkman headphones) with +12V on the tip, -12V on the ring and 0V on the sleeve.

Of course, the problem - which should be obvious to all - is that when you connect or disconnect one of these jacks, the contact in the socket briefly shorts out at least two of the contacts on the plug as the plug moves in or out of the socket.

As a result of this, if you connect or disconnect the power supply without turning it off first you'll blow the transformer. Nice bit of bad design.

There are many many dead Audio Alchemy power units, no doubt in landfill somewhere.

The solution

The whole issue could have been avoided if Audio Alchemy had been more careful at the design stage. The most obvious answer would have been to have used an internal power supply such as used on Arcam's Black Box range of DACs. However it appears that a compact form factor was part of their design goal, and as such having an offboard PSU was a requirement. It also meant that the main unit could be identical for all parts of the world, and only needed to be supplied with a different PSU for 110V or 220-240V countries.

So, assuming we're stuck with an offboard PSU, what else could have been done differently?

Option 1 - Use a more appropriate connector

Clearly a 3.5mm minijack plug is completely inappropriate for this application. A more suitable connector would have been a 3-pin locking DIN plug. There is no danger of the pins touching each other or being shorted out.

For the ultimate reliability the "gender" of the connectors could be reversed. The "male" connector (the part with the pins) fits to the back of the equipment. The "female" connector is moulded onto the end of the power lead. That way there's no chance of shorting out the power supply, even by the free end of the DC cable coming into contact with something metal.

Of course it may be that this option was far too expensive. But this wasn't a budget product. The extra cost (particularly with volume discounts) would have been insignificant. And probably cheaper than all the warranty repairs that resulted - and then failed again when the power cable got pulled out when the unit was still powered up.

There is another option though, which would have worked around the problem.

Option 2 - the polyfuse

Let's assume that the main unit and power supply were in production when this issue was discovered. There would have been a simple workaround. It's called a polyfuse.

Basically this little device works like a normal fuse by interrupting the supply when an over-current condition is detected, eg when there's a short circuit on the output of the power supply. But unlike a regular fuse where the wire melts and the fuse needs replacing - or even like a circuit breaker which "trips" and needs resetting - this device will automatically restore the supply once the fault condition is removed.

When the PSU went bang on my DAC-in-the-box after someone yanked the DC power cable out when the unit was powered, I built a replacement PSU. For the sake of less than a couple of quid I ran the +12V and -12V outputs through a pair of polyfuses. Now if someone yanks the cable without switching off at the mains, the power supply won't go up in smoke.

But why wasn't it spotted earlier?

That's a very good question. Surely someone at Audio Alchemy would have looked at that design and realised there was a built-in self destruct on the PSU? Or is this like the HCF (Halt and Catch Fire) instruction in one of the early microcomputers (one of the Commodore PET series if memory serves) where the PIO pins of a VIA chip were all grounded. Which was fine until someone set them as outputs and turned them all on. Smoke would then be emitted from the computer. This of course was revised in later versions and the pins were left floating.

I mention this as a lesson in what not to do when designing anything. This isn't a mission-critical application, nor is it life-threatening. The worst you might get is a bit of smoke as the transformer windings melted. It's unlikely that anything would have caught fire. But in a higher powered application something may have caught fire. Worth thinking about.

Comments

No comments to show.