I started Jim Gamble Associates in 1977 with the Gamble HC40 and SC32 console models, moving the company to Lake Tahoe in 1981. The Series EX56 FOH console came out in 1985 and in 1989 Crest Audio, Inc. bought the rights to manufacture the EX console line.
During the 1992 AES Convention Bob Lentini showed me a Windows mixing console on a PC computer that knocked my socks off. I had been dreaming about designing a fully recallable console for years and this made it all possible.
I started researching and trying out some of the latest fully digital audio technologies. But even the best devices had unacceptable noise levels and distortion when the signals were at lower levels, with phase shift starting at 2.4 kHz, and there were—and still are—no answers to these basic digital problems, so these problems were never going to be fixed in digitized audio. Designing a new product like the DCX is very life consuming so I wanted the result of all that hard work to have no compromises. Because of all the problems that full digital audio has I decided to keep the signal path analog with digital control.
Any capacitors in the signal path will cause frequency related smearing of the signal, which shows up as harshness in the 3 kHz range. The larger the capacitor’s value the more pronounced the harshness will be.
The biggest design compromise in analog circuits today is electrolytic capacitors used for DC voltage blocking, to isolate the analog circuitry from the offset voltages found on the inputs and output of any typical op-amp. In the Series EX Console design I used only small value polycarbonate caps in the signal path, which only reduced the smearing problem to a lesser degree.
Another problem with using capacitors in the signal path of a large mixing console design is that the signal will travel through dozens of capacitors. This progressively rolls off the signal’s low end, requiring larger capacitor values in the design to get the low-end 3dB down point throughout the console to the 20 Hz range (phase shift starts at 200 Hz and increases as the frequency down).
The DCX design has no capacitors in the signal path—it's all done with servos. The frequency response with all stages active—EQ, compressor, gate, etc.—through the input channel, selected to a Subgroup and to the Master outputs, is 3dB down at 0.5 Hz and 200 kHz.
The low frequency (LF) phase shift point is from 5 Hz down, and the high frequency phase shift point is from 20 kHz up, so the DCX has no phase shift within the audible range of 5 Hz to 20 KHz.
In 1994 Texas Instruments (TI) started sending me preliminary production samples of an 8-pin Dip package dual BIFET op-amp, the TLE2072, for my evaluation. I must say that this is the closest thing to a perfect op-amp for audio I have come across yet, with 100 MOhm inputs, symmetrical clipping and symmetrical slew rate.
This puppy can drive a +20dB signal into a 100 Ohm load without any increase in distortion—and it sounds good, too. So that's what is used in the DCX for all internal analog signal processing.
For the servo ICs we use TI’s LF412, a dual BIFET 8-pin Dip op-amp with a laser trimmed output offset voltage of less than 3mV. The LF412 Servo IC is paired with a TLE2072, and the combination results in 0mV offset voltage on the output of the TLE2072.
The Analog Devices AD2143 balanced line input and AD2142 balanced line driver IC's are the answer for connecting external signals to the DCX. I have tested both inputs and outputs with up to 1,000 feet of Belden cable connected and there is no degradation of the signal.
Both devices are laser trimmed for -90dBu or better common mode rejection, so even with complicated external wiring situations the system remains quiet.
The mic preamps and balanced summing in the DCX design are done with Analog Devices’ AD2017, which has a very low-noise Class A transistor front-end, coupled to an Instrumentation Amp three op-amp configuration, all in one package.
While searching for the best DACs to use for the digital control of analog I discovered that the method most commonly employed was to place the DAC in the feedback of an op-amp.
But those type of DACs require the signal to be reduced in level so as not to exceed the upper THLD of the MOSFETs used, typically around -10dBu.
The lost gain must eventually be made back up. And as is often the case with gaing, the noise floor increases to the point that there is little remaining headroom.
I discovered a DAC that used JFETs, not MOSFETs, had a THLD of +18dBu, and you didn't need to place it in the feedback loop of an op-amp. The analog circuitry could be connected directly to the three isolated pot terminals, and it came in a quad package for about the same price as four Alps rotary pots.
With typical +15VDC and -15VDC power supply rails the internal signal level clipping point is +21dBu. So +17dBu was only a 4dB loss in headroom, which I could live with, because the overall noise floor of the DCX is very low.
One of the major problems with digital is that you cannot exceed the THLD level or all hell will break loose, and the result is 100-percent distortion. In the DCX design I insured that the THLD would never exceed +18dBu by adding back-to-back Zener diode clamping in the feedback (it sounds like tube squishing) to all the necessary circuits.
With the exception of the mic preamps, which can be clipped, the signal cannot be clipped anywhere else in the console. Even if the user runs the console with very high overall levels (VU meters maxed out at +15dB) the DCX does not sound distorted.
Using the Audio Precision system to measure distortion and noise I connect the monitor output to a ‘scope to see what the distortion consists of. With a typical signal level of 0dBu through the DCX, from input to output, the distortion reads around 0.005-percent from 20 Hz to 20 kHz, with no increase in distortion as the frequency rises, unlike most digital products.
The majority of that distortion is prefect sign waves, so in my opinion the DCX design has no perceptible distortion.
One of the criteria in the DCX design was to have maximum control and flexibility. For example, rather then use switches to select Subgroup busses, the DCX has Level Controls, so you can use the eight stereo Subgroup busses just like you use the 16 stereo Aux busses, therefore all 24 stereo busses can be used for monitors.
On Subgroup 1-8 the Level Controls go up to 0dB and are always post fader and pan, while on Auxes 1-16 the Level Controls go up to +10dB and each Aux left and right has Pre/Post Fader and Mute buttons. Center Level, Mono Level and Master Level controls are also used on the inputs.
Level controls are used on the Stereo Matrix, for the Master to Matrix, with Pre/Post Master Fader buttons, Center to Matrix, Mono to Matrix, Subgroup 1-8 to Matrix, and Aux 1-16 to Matrix.
In normal operation the VU meters on the inputs read around the 0dB to +5dB range, giving 12dB to 17dB of headroom, including any EQ boost. Because all of the input Level Controls are always being summed together (they are always connected, unlike a bus selector switch), and because the signal level cannot exceed +17dBu (remember the ‘tube squishing’ circuits), the gain is increased 4dB on the ICs that feed the busses.
There are no clamping circuits on the ICs that feed the busses, so there is 4dB of headroom left before clipping (+21dBu). The summing resistor values are set for 4dB of gain reduction in the summing amps, so the signal is still at 0dB to +5dB at the Subgroup and Aux outputs.
Stereo Subgroups 1-8 have a 2.5dB gain reduction into the Stereo Master, to mix the eight Subgroup signals together and maintain some headroom in the Master pre fader summing. All the Level Controls on the Matrix go up to 0dB and are unity gain to the Matrix 1-8 Stereo outputs.
With 40 inputs summed into eight Subgroups summed into the Master the noise floor is -83dBu on the Master outputs, which is 100dB noise floor to clipping. That is the worst-case noise test for any mixing console design.
The most carefully designed digital console in the world is totally useless without great software to go with it. The software geniuses at CorTek Software Inc., Las Vegas, NV have made the DCX dream come true. They started writing the Virtual Audio Console Software (VACS ) for this project in July 1997 and have worked for endless hours on it ever since.
I have been working with CorTek Software on the software for three years and we have been able to get everything working correctly, but it has been an overwhelming project to do and was accomplished one step at a time. Simulating in software how controls work in real life is not as easy as one might think.
It was an almost impossible thing to pull off. The DCX is a true virtual console because every function is controllable on-screen. Without the hardware/firmware working, it was hard to get the software working correctly, and vice versa. It had to be done in little steps: first a little software, then a little hardware, and so on. We have been working on the software non-stop for the last three years.
The lookup tables that are displayed on the screen for each control took one year to perfect, and we had to write a special program that would compile the lookup tables into the DCX.bin file.
The firmware for the Altera chip (100 programmable gates on one chip, one on each DCX module), took one-and-a-half years to prefect, including the State Machine, (a set of rules that keeps the Data flowing on the module no matter what happens).
We even had to write our own Windows NT port driver, which required a 650 MB NT driver program to create the 8 KB DCX port driver. Windows has its own set of rules that must be observed, and it can take hours of research, and trial and error, just to get one little part working correctly in the software.
The same guys that designed the Sound Droid digital mixing system used on the first “Star Wars” movie designed the BIOS firmware and the Backplane Digital Network System, applying the basics of that technology to the DCX project over a four-year period.
The DCX ROM BIOS firmware, the operating system for the on-card computers, is programmed onto a ROM chip, one on each DCX Module.
The software that runs the DCX Rack is Server Software (GServer), which also has special functions for testing and programming the DACs in the rack.
The DCX Interface Module has an Ethernet 10/100base-T that connects to a 3Com hub in the Workstation Roadcase and a Dell Precision 420 that runs the VACS client/server software. The software is written in C and C++, (MS Compiler 6.0). VACS client software is compatible with Windows 95-98 and NT.
The Zoom View in VACS is like looking at a portion of a typical console surface through a piece of glass. With 2,560 x 1,024 pixels on two monitors, in 65,536 colors with 3-D shading, it looks very realistic and uncomplicated to the user.
The software is as easy to learn how to use as falling off a log. The VACS software is continually evolving, which is the greatest reason to have a virtual console surface because new features can be added as users ask for them.
All the computers on the modules are parallel-processing the data over the backplane, to the Interface Module/Server Computer Ethernet connection.
There are four DCX models: the DCX Cabaret, the DCX Showtime, the DCX Event 40 and the DCX Event 60. The DCX Showtime (24 channels, 48 mic inputs, 48 line inputs) does not have enough channels for FOH. The DCX Event 60 (60 channels, 120 mic inputs, 120 line inputs) is more console then most situations require. The DCX Event 40 (40 channels, 80 mic inputs and 80 line inputs) is just the right size for most users, so that's what we are building in 2001.
The DCX Event 40 comes packaged in a self-contained double-bay rack with pull-on cover. All of the electronics are on 60 plug-in module cards that are housed in NASA-approved cardcages, with separate power supplies for each cardcage.
When Kevin Korecky and I designed the DCX packaging we decided on a VME-type cardcage because it was a very reliable and time-proven Euro (worldwide) standard design, and not too expensive.
The size of the overall rack had to be no more than 30-inches (762 mm) deep to meet today's truck pack requirements. It needed to be a rack, shock-mounted within a rack, with big beefy wheels that can survive being used on tour. The DCX Event 40 is 56 inches (1.422 m) tall including the wheels, 45 inches (1.143 m) wide and 30 inches (0.762 m) deep.
We added some oak trim and two Littlites in the front with a DCX dimmer knob—the only knob on the DCX—and we included a pull-out, flip-up 14-inch (355.6 mm) LCD with keyboard/touchpad for the 233 MHz Dell server computer.
A DCX Module is 15.5 inches (393.7 mm) high and 0.8 inches (20.3 mm) wide. Each module has TT jacks on the front panel for all of the mic and line inputs, which are normalled through to the patchbays, and 160 XLRs with ground-lift switches. There are 48 output XLRs. There are also four sets of 48-pair tie lines in the patchbay and 10 Ramlatch multi-way connectors.
The 40 channels have two separate mic inputs and two separate line inputs. Each channel has a compressor and a gate with EQ on the gate key. The four band fully parametric EQ features peak/shelf selection on the HF and LF, a 24dB per octave lowcut filter and a visual EQ graph display. There are pre- and post-compressor/EQ inserts, compressor and gate side chains, standard or true left-center-right panning with Center bus, Mono bus and Direct Out.
The Event 40 has eight stereo Subgroups with inserts, 16 stereo Aux sends, eight stereo Matrix sends with four band stereo EQ and inserts, and Stereo Sum-Inputs (with 10dB of gain) on all of the busses on the console. Two stereo Cue (Solo) systems, with Cue outputs and headphone jack outputs, stereo talkback system with mic and line Inputs, and stereo Pink Noise and Tone are included.
Lawrence Rambo of Ramtech Industries, Inc., Gainesville, FL wires the DCX Racks then sends them to Tahoe where the power supplies, cardcages and DCX Modules get installed and tested. Ramtech supplies all the cabling, including custom cabling if necessary, to connect the DCX to the sound system. That includes a 100 meter CAT-5 wire as well as the Ramlatch snake cable so that the FOH DCX Event 40 Rack can stay up on the stage if necessary, and then only the DCX Workstation Roadcase needs to be out at the FOH mix location.
The DCX Workstation Roadcase comes with two 21-inch (533.4 mm) monitors, Dell computer, and 3Com hub. We decided to use Dell Computers, Sony monitors and 3Com hubs because they have proven the most reliable and have the best support. The operating system we are using on the Dell computers is Window NT 4.0 SP4, and in all the years I have used Windows NT it has never crashed on me.
All the ICs needed to be in high-quality sockets so that the modules could be repaired on location. A full set of replacement ICs is included with each DCX System.
The power supply chassis needed to be a welded steel 19-inch (482.6 mm) rack mount design and the PSU needed to work on worldwide AC voltages. We use four UL-approved linear supply module units from International Power DC Power Supplies rated at +5V@25A, +15V@15A,-15V@15A, +48V@1.5A. We decided against switching supplies because they introduce switching noise back into the AC distribution.
The rack ventilation system consists of two 4.5 inch (114 mm), 58 CFM fans in each power supply, sucking air into the front of the PSU through a cleanable fan filter. The PSUs are always positioned below the cardcages, and on the back of the DCX Rack, blowing the air up through the cardcage modules and sucking the hot air out of the top, are eight 3.0 inch (76 mm), 28 CFM fans in a low-noise fan enclosure.
Making VACS foolproof to use was one of the chief requirements. The first time you open VACS the Zoom View comes up with all the controls set in their default positions. When you turn on the power to the DCX Rack all the DACs on the Modules power up with all the controls set in their default positions, also. After making changes to the controls, and maybe adding some Sequence Scenes, you can Save Mix As. The next time VACS is opened, all the changes to the controls and the Sequence Scenes are all there, without having to load a mix file.
The user can load the mix file at that point, or make changes and Save Mix As with a different mix file name. The mix file is saved every three minutes into the Mix Folder. This is the file that is used when you open VACS, so if the AC power goes down in the middle of a show, when the power comes back on again and you open VACS you are within three minutes of where you left off.
All the DCX hard drives are formatted with NTFS, and you can add thousands of Sequence Scenes to a mix file. The mix files are saved as compressed files, and consequently load faster.
The computer comes with a CD burner, so you can copy mix files to a CD-RW or to a 3 ½-inch floppy disk. VACS has a Work Offline function that allows the user to use their laptop to edit and save mix files before or after the show.
Mixing a live show on a 40 channel console, using only a mouse for control, sounds like it would be difficult to have control over everything at the same time, but the DCX Event 40 System and VACS software make it very easy to do, for the following reasons:
CorTek Software, Inc.
Las Vegas, NV
(702) 263-0419
http://www.corteksoft.com/
Jim Gamble Associates
(530) 583-0138/0139 (Secret Mountain Laboratory)
(530) 583-5603 (FAX)
http://www.gambleboards.com
Ramtech Industries, Inc
(800) 817-2683
(352) 466.0906 (FAX)
http://www.ramtech.net/