GREG MACKIE & CARL PERKINS: Mackie Designs

Within just a few years, the Mackie brand has come from almost nowhere to a prominent position in the project studio mixer market. Now the company are looking to the future and deciding where they can next apply the Mackie philosophy. Paul White takes a busman's holiday to Seattle and looks behind the scenes in R&D...

To the majority of SOS readers, Mackie need little introduction — they're the company which reinvented the mini‑mixer, transforming it from a niche product to a mainstream audio commodity. Today, Seattle‑based Mackie are one of the manufacturing success stories of Washington state, and judging by the massive new premises already under construction at the time of my visit, they have no plans to slow down. Indeed, the next year will see massive diversification for Mackie. Playing a significant part in the company's future will be designer Cal Perkins, who might best be described as Mackie's 'fifth Beatle'.

Cal Tech

Cal Perkins has made a major contribution to Mackie's designs since the company started. However, he's only recently joined the team full time, so I asked him to bring me up to speed on some of his background before getting down to discussing new products.

"I started out doing sound contracting and live sound recording for Swanson Sound Service in the late '50s and early '60s. Cal Poly was foolish enough to give me a BSc and let me do a year of post‑grad work in analogue design. During that time, I got a gig designing a low‑frequency enclosure to go with JBL horns, which later became JBL's Pro Series of loudspeaker enclosures. After college I joined their Pro Sound division. Marantz eventually lured me away with some high‑end design work during the '70s. Greg Mackie had interviewed me for a job with his company Tapco, but I wanted to live in Portland. I helped start a manufacturing division for North West, building the Anchor power monitors, and developed the first low‑distortion, constant‑directivity horn in 1977, which the Eagles used on tour.

"After four or five years, I moved to Biamp, managing their engineering department. I'd also done design work for Yamaha, and because of my transducer experience, ended up doing a lot of work on their PA speakers. Yamaha's management migrated to Fender, so I went to head up their Pro Sound division. We looked for a manufacturing partner, which turned out to be Ramsa — part of Panasonic/Matshusita. CBS then sold Fender, so I stayed on for a while, then rejoined JBL and became technical director of UREI. I eventually did the electronic design for the JBL Eon system.

"All that time, Greg Mackie and I have stayed pretty close, and about six or seven years ago, he asked me to help design a mic preamp and an EQ for his new mixers which could be built inexpensively without sacrificing performance. Now I'm with Mackie full‑time, and my first design job was to get the CR1604‑VLZ mixer into production."

Mixer Matters

Perhaps this would be a good time to explain the VLZ concept.

Cal Perkins: "Basically, VLZ is shorthand for design techniques we discovered which dramatically reduce noise in a mixer circuit. Those techniques had eluded the entire industry up to that point — including our own team! What we found is that the thermal circuit noise is a function of the equivalent input noise voltage and current of the individual amplifier stages and the termination resistors. Noise is a function of absolute temperature; if you have molecular motion in a conductor, you have a certain amount of energy present. If you want low‑noise voltage, then the impedances have to be fairly low. The industry was using high values for summing resistors and high values in the post‑fader buffer, and using bi‑FETs, which have very high‑noise voltages. The noise level presented to the mixing buss was typically 20dB higher than you could achieve by using bi‑polar parts coupled with good design techniques. VLZ exploits that differential. The impedances are way down, both in the build‑out resistance in the mix buss and around the pan pot."

Greg Mackie: "We probably spend most of our time making our products intuitive."

I would imagine that a limit is set by how much output current your mix amp can develop. Is this the case?

Cal Perkins: "Exactly, and most ICs are limited to around 30mA peak. You have to walk that fine line so the resistors are as low as possible before the op‑amp runs out of current capability and has a hissy fit! This radically reduces both mix‑buss noise and crosstalk, which is important when you're working with closely‑spaced circuit‑board tracks. The other big problem with this approach is that circuit‑board design becomes non‑trivial. Even with an eighth of an inch wider ground track, if you're pumping a lot of current into it, you can get a lot of common‑mode impedance coupling between stages, so the ground impedance has to be made as low as possible, by using heavier copper. You also have to know where your current is coming from, because you don't want current from channel A flowing through channel B. The board layout is really important."

I understand you've changed the mic preamp design on your new 40‑channel live sound console.

Cal Perkins: "Originally, we could handle up to about a +12dBu level, with about 8dB of gain, but what Greg found out in his own home studio was that a close‑miked kick drum would cause him to pull the faders down below unity because of the very hot signal level. Traditionally, you'd switch in a pad, but a pad degrades the signal‑to‑noise ratio, so we changed a few things around, and the preamp will now accept +22dBu at unity gain. We're doing some things that I've never seen done before: essentially, the whole preamp is DC coupled — everything on a master current source, so everything moves together. More importantly, the design also holds the equivalent input noise [EIN] over about a 30dB trim range. With less stout mic amps, as soon as you go away from maximum gain, your equivalent input noise degenerates.

"Rupert Neve pointed this out a few years back. He said that a lot of mic preamps have an impressive spec at their full gain setting, but perform poorly at real‑life gain settings. So the mic preamp we have now holds an EIN of ‑126dB with only 20dB of throughput gain. That means the noise floor is ‑106dB, which is 10dB better than even our previous design. Over a 40dB range, the EIN degenerates only from ‑129dB to ‑126dB. At this point, you have to look at which other parts of the circuitry are adding noise, such as the phase splitter. In the VLZ, the phase splitter is fully differential; in a cheaper design, where you just run the signal through an inverter, the delay is different depending on whether the phase invert is switched in or out.

"The other thing we've done is to use a differential mix buss, which yields a substantial increase in performance. If you have enough headroom in the first place in the channel strip, where your noise is extremely low, there are two ways to do it. You can come off the pan pot with a fully differential drive, but we still come off with a single‑ended drive and the mix buss itself is differential. Essentially, each pan pot has two build‑out resistors, where one is for the ground and one is for the pot wiper. The practical advantage is that if you have a nearby bundle of wires from the lighting console, it is far less susceptible to interference than a conventional design. You can make the desk quiet using a single‑ended summing buss, but you don't get the protection against external interference. We believe customers really notice these 'designing for the real world' touches, and that may explain why we're able to compete so well against big companies who believe bench measurements are the supreme test."

Listening Without Prejudice

The Acoustic Team is also responsible for the new studio monitors, which are a radical department for Mackie, and I imagine this is a bit different for you too, with your live sound background.

Cal Perkins: "I haven't spent much time in studios, though I've done quite a lot of live classical recording. But being a loudspeaker freak, I have strong beliefs about what makes an accurate monitor. That being said, Greg and I were able to hire Mats Jarlstrom, a young guy who we thought was years ahead of the pack from his work on active speakers. We've gotten a big kick from watching really high‑level mix engineers listen to prototypes utilising his ideas — truly incredulous looks on their faces, with some even asking where we were hiding the subwoofer!

Greg Mackie: "It's not inconceivable that we'll bring out a stand‑alone reverb/effects processor at some time."

"We decided that active speakers were the way to go because we can eliminate a bunch of uncontrollables that can mess things up — even with the best components. You can do things like use active feedback to cancel out the voice coil resistance, to change the Q and improve the damping, include a limiter to prevent the customer from breaking the speaker, and optimising the amp size.

"One thing we wanted was an extended bass that was very tight and well‑controlled. With a reasonable box size, you can go a couple of ways. You can port the box, but the vent size is usually much smaller than the diaphragm area (because of the space constraints) so the vent velocity is extremely high, you get power compression, and it blows air in your face when you're mixing! You hear the vent noise and the frequency still doesn't go very low.

"If you want to go low and tune the box down, you're going to need a reasonable‑sized vent, and the way to get that is to use a drone cone, or passive radiator. Mackie uses two 6.5‑inch passive radiators, which is a little more area than the driven cone and with the same peak displacement. Once we had that, we put it in a well‑braced three‑quarter‑inch MDF box, totally filled with acoustic foam to make it acoustically inert. This converts most of the higher frequencies from the back of the cone to heat; in fact, if you measure the output from the passive radiators, it rolls off at 6dB per octave above 80Hz. This breaks up a lot of the resonances in the box that would be in the male vocal region for a box of this size.

"The amplifiers are bi‑polar designs with no current limiting or fancy protection to compromise the performance. We call it a Fast Recovery design, and it uses the same symmetrical, differential topology as our stand‑alone power amplifiers. There are stacked power supplies where the pre‑drivers are on higher power rails along with the front end; you can end up with much less than one‑tenth of a percent open‑loop distortion on a design like this, so you don't need much negative feedback to correct it. This allows you to set the amplifier roll‑off outside the audio spectrum and the feedback is always in the correct phase.

"One thing that's rarely understood in the context of Transient Intermodulation Distortion, or TIM, is that if the large signal bandwidth is smaller than the small signal bandwidth you get lower TIM. The open‑loop output impedance stays relatively constant with frequency, so if there is any transient overloading inside the amplifier loop, the speaker doesn't let go instantly. A radical change in output impedance equates to rate‑of‑change of damping factor, which is something that the industry doesn't want to talk about.

Greg Mackie: "In general, people who come from a digital engineering background don't understand why analogue mixers are laid out the way they are, so a lot of the systems tend to be pretty hard to use."

"The crossover is analogue and uses two separate filters; the woofer uses current feedback, where we synthesize a negative output impedance, and we also do some band shaping. The box itself is a 6th‑order system, which is flat to 40Hz, 3dB down at 37Hz, then it rolls off really fast. If you listen to impulse responses, the system gives a good account of itself — it's critically damped, which makes it sound very tight."

How did you arrive at the choice of drivers?

Cal Perkins: "If you're going to get a lot of bass, you need to move a lot of air, so you have to specify a transducer with a lot of peak swing. We went to an OEM [Original Equipment Manufacturer] vendor knowing what we wanted in terms of the magnetic structure, moving mass, compliance and so on. There's also a little art in choosing the cone material because of the damping and the higher‑frequency breakup modes of the cone. This determines a lot of what you hear in the upper‑mid range, and comes down to personal taste. Our woofer works up past 4kHz and the crossover is set at around 2.5kHz. If you get off‑axis, the power response is still good. Again, it's the slope of the crossover and the frequencies at which the two filters are sitting that determines how well the thing comes together.

"The top end is handled by an aluminium‑domed tweeter which works as a pure piston up to 22kHz, and there's a gentle roll‑off above that built into the amp. We mounted the tweeter in a very shallow exponential flare which acts as a waveguide or constant‑directivity horn, and because this increases the dispersion at higher frequencies, you'll find that at the top end, the on‑axis response is down about 4dB, measured with a B&K test mic. The reason is that in a flat baffle, tweeters will tend to beam at higher frequencies, which concentrates all the energy into a narrower angle, whereas in a constant‑directivity system, the dispersion is maintained up to a much higher frequency. This means that the top‑end response is headed off at 6dB per octave above the tweeter's mass break point, so you have to compensate for this in the active filtering. Again, this is something you don't have control of with a passive system, but by appropriate filter design, we've managed to achieve a wide, flat response, both on and off axis. A benefit of this is that you get a very wide sweet spot and, at the same time, very precise stereo imaging. And, ironically, getting all these things to happen actually reduces the cost of manufacturing. Frankly, we expect a lot of initial scepticism from consumers who have been trained to equate high price with performance. Hopefully, whatever credibility Mackie gained by bursting that assumption in small mixers will transfer to some degree."

What's the maximum SPL you can expect from this system?

Cal Perkins: "Realistically, we can achieve 105dB with programme material, if you put a sound level meter out there. We've decided to publish two sets of figures: one is calculated the same way as everybody else, for comparison purposes; the other is the way I honestly think it should be done. For example, if you take a speaker capable of 90dB/W@1m and you have a 100W amplifier, you'd get 110dB if there was no power compression, but if you're really honest and include, say, 2dB of power compression, the power supply sag, the voltage drop in the main feed, the real figure is likely to be 4 or 5dB less than claimed."

Digital Directions

So far, the only digital product you've formally announced is the Pro Tools interface. Can you give us an overview of where this product line is going?

Greg Mackie: "Basically, the interface you've seen is an analogue control surface designed to work with our upcoming digital mixers. A digital mixer can be controlled by a mouse, by a small control surface like the Pro Tools interface, or by a large‑scale control surface. If you need to stay in the middle of two speakers, you really need a small controller. The mixer itself will be a separate box to the control surface, and the good thing about that is that it keeps the analogue and digital circuitry separate.

"On the bigger digital console, we had one design proposal that had all the audio jacks on the control surface, but one of the problems there is that you end up with a huge console. We're trying to keep the size down, and by putting the jacks on the external audio hardware, you don't have that issue about whether the jacks should face upwards or backwards. It also allows you to have shorter cable runs to the patchbay. We've taken a semi‑modular approach to provide the greatest flexibility."

Cal Perkins: "Our goal is to be able to do an A/B comparison with an analogue desk and not be able to tell the difference."

Assuming good audio quality and the right mix of features, I imagine that the most important aspect of a full‑scale digital mixer design is the user interface. Most designers have settled on providing a full set of faders, plus a single set of channel controls that can be assigned to any channel as required. What can you add to this to make life easier for the user?

Greg Mackie: "Our background is in making analogue consoles, so we tend to think that way. People who use analogue consoles have got used to a certain control format, and that format has been developed over many years for a reason — by now it's nearly perfected. In general, people who come from a digital engineering background don't understand why analogue mixers are laid out the way they are or how a person works at the control surface, so what happens is that a lot of the systems tend to be pretty hard to use. We probably spend most of our time making our products intuitive. If a product is hard to use, the salesman doesn't want to sell it, because he can't use it himself! If the person who walks up to your product can't understand what's going on, there's going to be more sales resistance. The more money and time we put into making the product more understandable, the better off everybody is."

What I feel is missing from most of the current digital mixers is the global overview you get from an analogue mixer's control surface: you have to interrogate every channel to see where the controls are set, or you can only see one control at a time — like all the Aux 1 sends. Do you have any ideas for improving on this?

Greg Mackie: "On our smaller console, we have a system where you can query the mixer, so that if you check Aux 1, all the channels using it will light up. The level is less important at this stage. We also have a row of what we call 'V‑pots' (one per channel), which change their role depending on what you're doing. These have LEDs around them so you can see what your Aux 1 level is, or whatever. It's still not as easy as having a discrete knob for each function, but it's a lot easier than what we see out on the market now. There's also a case for reducing the knob count on a large console, because a human being can't scan a large number of knobs very easily. What we're trying to develop is a system where if you're not using something, it's not seen — it goes away. For example, with the V‑pots, if you're not using that control, the LEDs associated with it are turned off, so that you're not looking at a panel full of lights. On a conventional analogue console, you always see everything, even if the control is turned down, or if the pan is set central. If we work it right, I think there's an advantage in simplifying things to a certain extent."

Do you think there's a place for external VDU monitors for when you need more information?

Greg Mackie: "Absolutely, and that's part of the modular idea, where you can have software upgrades so that you can see, for example, your channel assigns all the way across. If there are certain things you want to look at all the time, all you need is a bit of extra software and you can do it. We will have a smaller LCD screen on the console itself, but obviously you can get more on a large colour monitor. Depending on what you want to do with the console, you can choose your work surface. For a little workstation, perhaps you don't need very much, but for an album project, you'd probably need the full‑scale console."

Market Pressure

It's no secret that a lot of mixer manufacturers are working on affordable digital consoles, and the Yamaha 02R has put more pressure on everybody to get something to market. What's a realistic estimate of when we'll see something from you?

Greg Mackie: "Very fast — we hope to have something by the end of the year. Mackie is going to be noticeably diversified by January NAMM. But it's my belief that analogue has still got a long time to go, and there's something about having an analogue desk in front of you, where you can reach out and just grab a knob. It's too convenient to dismiss, and it'll remain the preferred way to work for very many people. It's like the watches we're all wearing now — these are all analogue watches. Digital displays came and went. There will be some people who want analogue mixers with groups and faders, and others who want to go the digital way. The younger generation are used to computers, used to working with a mouse, and I think they'll be much more inclined to use digital desks than the older people, who might tend to cling to the analogue approach."

With Yamaha's 02R being so successful and so affordable, what do you feel your ideas bring to the party?

Greg Mackie: "The 02R is a real interesting product. They did a nice job and a lot of people like it. I think the usability is the weakest point, and tracking with it is difficult. You also have to buy optional I/O modules to make it a tracking console. It's fine for post‑pro, but we're not seeing albums getting cut on it. We feel that by making our console easier to use and by providing more controls than the competition, we'll get fewer phone calls. We've also had the benefit of being able to look at what other companies have done — Yamaha got some things right and some things less right. In other words, we had the advantage of not being first.

"We're going to try to build the digital multitrack interfaces in there as standard, because the ADAT and DA88 are the standards. It depends on cost, and if we can't build it in as standard, we'll make sure that it isn't too expensive to add as an option."

From what I saw in your digital division, you're also including dynamics processing for compression and gating. Will you be including effects too?

Greg Mackie: "We have effects, and the company has already developed a very nice‑sounding reverb which we're being told matches the best out there. We have the basic algorithms, and it's not inconceivable that we'll bring out a stand‑alone reverb/effects processor at some time. It's also planned to make the mixer expandable by adding cards, and we'd like to work with other companies with a view to them providing third‑party cards. That way we could offer their algorithms bearing their brand name."

A concern with all‑digital consoles is that there's a finite processing delay before an input signal appears at the output. What have you managed to get this down to?

Cal Perkins: "I think it's less than 3ms or so, it's not very much at all. A lot depends on the filter implementation, and we're not using up processing time trying to eliminate the phase shifts that occur in analogue EQs, as some digital designers do. Essentially, we have a digital emulation of an analogue EQ. We developed an FIR [Finite Impulse Response] crossover filter at JBL which was perfect and had no phase shift, but the delay was so long that you couldn't use it for live sound applications! Our goal is to be able to do an A/B comparison with an analogue desk and not be able to tell the difference. Is it as quiet as the analogue desk; does it do any strange things at low signal levels? From a corporate viewpoint, we're not going to do it before we succeed at that. We find existing digital consoles to be unacceptably noisy. For example, when you allow the same amount of headroom as you would on an analogue mixer, you find they're around 15dB noisier. If we came out with an analogue console that noisy, we'd be laughed at! You can make existing digital mixers work, but you can't afford to leave yourself much headroom. Our aim is to allow you to operate at zero level with 20dB or so of headroom above that. We don't want you to get to the middle of a song and get something clipping because the level has come up a bit."

Greg Mackie: "Digital has given us a superior storage medium in the form of the Compact Disc, but these use up all the available headroom. In real life, sound systems aren't run at maximum level all the time — 16‑bit digital's Achilles heel. At lower levels, there are artifacts which some people get very emotional about, while other people can't hear them. A 96dB dynamic range is wonderful, but during mixdown, most of your faders will be set a lot lower than that, which means that you need a huge dynamic range to keep out of trouble."

A Bit Secret

That implies using a high bit resolution. What resolution will your system be? 20‑bit?

Greg Mackie: "We don't want to say right now; we have to have some secrets. Let's just say that it will be sufficient."

You've already announced your intention to diversify your product range. What else can we expect from you in the near future?

Greg Mackie: "We won't get involved in a market area unless we believe we have something substantially better to bring to it. We feel that way about nearfield speakers, and about a number of digital products. I can assure you that we won't ever build guitars, for example, or anything that is a 'me too' product.

"As the company grows, we discover more technologies that we can utilise in our products. The areas that we can get into are probably obvious to our competitors; there aren't really any secrets out there. There are opportunities in broadcast, and by getting into power amps and speakers, we're better positioned for the conference market. It makes us a more valuable brand name for a contractor to specify and promote than if we just made mixers."

With so many tapeless recording systems coming onto the market, I would imagine that, ultimately, it would make sense to integrate the recording hardware with the mixer, so that propagation delays could be cancelled out automatically, by reading data off the disk slightly early. Is this an area you see yourself getting into?

Greg Mackie: "We're working on that right now!"