A thread on our message boards about how we test loudspeakers in production inspired a new video and plant walk-about with Andrew Welker. ¬†Watch as he explains how we go about testing speakers and speaker parts in production.
Andrew Welker:¬†Axiom’s testing is very involved because there are a number of steps to testing the components: ¬†the crossovers, the drive units, and then the entire system. ¬†It is not simply a sweep to see if the speaker makes sound: we actually do very detailed measurements every step of the way.
The Crossover Network
One of the most important components of a loudspeaker is something called the crossover network (sometimes referred to as a filter network or a dividing network). ¬†It’s basically a circuit board that contains electronic components such as resistors, capacitors and inductors. It’s responsible for dividing the entire frequency band into the individual ‘chunks’ that the individual drivers will carry. ¬†For instance, you want to send the woofers in a loudspeaker only the low frequency sound. ¬†So the filter will actually prevent high-frequency sound from getting to the woofers. ¬†Similarly, for the tweeter, we don’t want low-frequency sounds getting to the tweeter, just high-frequency sounds, so the filter will make sure only those high frequency sounds get to the tweeter. ¬†The crossover network design, along with the driver design, is also what allows us to manipulate the family of curves which is the most important design element of a great-sounding loudspeaker.
Because we manufacture pretty much all of our components in-house, including all of our crossover networks, we have the benefit of being able to keep very tight controls on the quality of these parts. ¬†Because they are such a critical component of the speaker, every single one of them is tested and measured.
Here we’re sitting at what we’ve nicknamed the ‘hotdog cart’ where we have a measurement system which allows us to measure the electrical characteristics of each section of the crossover network individually: woofer, midrange and tweeter.
We go to the lengths of manufacturing our own drive units or woofers, tweeters and midranges at Axiom, even though it would be very easy to buy them off the shelf because there are lots of manufacturers making these types of drivers. ¬†One of the reasons – and really the most important reason to make them in-house, other than having full design control over the part – is that you then have full control over the quality of each part that goes into your loudspeaker.
There are a number of steps along the way while we are building transducers where we actually perform measurements on them. ¬†As a final test, just like the crossover networks, and just like the entire system is tested before it gets boxed up and shipped to a customer, each individual drive unit is boxed up and tested.
We made the investment a couple of years ago into a state-of-the-art ¬†measurement system from a company called ListenInc. ¬†It allows us with a very very quick signal sweep to perform an entire suite of measurements. ¬†In just three seconds, we are able to measure not only the frequency response or the amplitude¬†response¬†of¬†the¬†drive unit, but we’re also measuring for impedance, for buzzes or distortion. ¬†All of those items are measured individually in our mini anechoic box that has a microphone mounted in it. ¬†Each one of the items will get a green light or a ‘pass’ if they are within specification.
We determine our own tolerances and they’re very very tight. ¬†On something like an Axiom tweeter, a dB and a half is our measurement window, which is very hard to achieve if you’re buying a batch of tweeters from another company rather than making them yourself.
This is really one of the most critical components – along with the crossover network – that make up the speaker. ¬†Cabinets are important, but it’s also those components that go into a loudspeaker that are so important, and that’s why we go to such great lengths to make sure they are tested and they meet our stringent quality requirements every step of the way.
The Mini-Chamber “Booth” Test
After we’ve gone through and¬†individually¬†tested the crossover network and all of the drive units individually, we assemble them into a speaker cabinet. ¬†Now we test the complete speaker as a system, to make sure everything has gone together properly and everything is wired properly, and that there are no issues with a loose wire, etc.
We actually have a small anechoic chamber on the production line. ¬†We stop the conveyor belt and mount the speaker pointing inwards into the chamber, where there is a microphone mounted. ¬†We’ll do a frequency response curve using the same measurement and same system that we do in our large anechoic chamber. ¬†We’ll also do a polarity test. ¬†The reason we do a polarity test is that there is a possibility (although it happens very infrequently) for the polarity of the input terminal to be reversed from what it should be. ¬†What would happen in that case is that you would get one speaker that was wired in-phase and one speaker that was wired out-of-phase, the sound would be all wrong and you wouldn’t get any bass. ¬†So it’s a very important aspect to check.
Finally we’ll do a sweep at a high level where the calibrated measurement microphone is actually the operator’s ears,¬†because¬†in that case we are trying to detect things that are very difficult to pick up with a microphone: things like wires buzzing inside the cabinet, a loose screw, etc.
So all of those aspects are tested essentially by ear after all of the other steps are done.
The Ultimate Test: ¬†The Anechoic Chamber
There is one speaker in our line up – the LFR1100 Omnidirectional speaker – which can’t be measured at the same end-of-line test as our other products. ¬†There are a couple of reasons for this: one is that it’s a fairly complex system involving ¬†multiple amplifier channels and a DSP box.
The other is because it’s an omnidirectional loudspeaker, if we measured it in a non-totally anechoic environment, what goes on between the front and rear sections would mess up the signal. ¬†So we have to actually measure that speaker in the full size anechoic chamber.
When this Chamber Lab was built, we made sure to build it literally 20 steps from¬†the¬†production line. ¬†The reason we did that was so that when we encountered models like this that we had to test in this environment, it was near by.
The other reason is that it lets us do regular checks of quality by pulling product randomly off the line into the full-sized anechoic chamber, and then we can measure them against our references.
The measurements that take place on an LFR speaker are numerous. ¬†We measure the front and rear independently; we also measure them together. ¬†We do a full family of curves to measure the sound power.
The reason we do that on every model is because there are different interactions¬†occurring, and those¬†interactions¬†between the front and back drivers can change those individual response curves, so we have to do a very detailed measurement here whereas the other models just need a single on-axis frequency response.
Hopefully that’s answered the question of what we do for our production testing of our loudspeaker models. ¬†You can see that it’s very involved, and the luxury of manufacturing almost all of the components in-house is that we can do this sort of testing along the way and guarantee a high-quality product that matches very closely to our references.