Over the last 18 months or so I have been designing, building, then redesigning a modular DJ mixer. I’ll admit that one of the biggest motivations was my repeated frustration at having to use DJ mixers with pots that had undergone immense abuse (both from use and from dirt), leading to noise, crackle and volume jumps, or DJ mixers simply sounded bad, or were too complicated to use. Generally I hate flashing lights and USB ports – so you won’t find many of those on things I’m building.
So how does one solve this problem? Well, other than regular maintenance which didn’t seem like a big priority for venue owners, you could use control voltages, VCAs and VCFs, much akin to an analogue synthesizer. This isn’t a brand new idea but I was yet to find a product that caught my interest. So I began down this path with a few ideas in mind:
-High quality pots are very expensive and cheaper versions that are generally used will wear out quickly.
-Cheap audio pots track poorly according to a logarithmic scale.
-Cheap pots track poorly between channels.
-Shorter signal path as all audio stays on the main PCB rather than being brought to the front panel.
I like to work back from the front panel, as I find it focuses the actual electronic design, so I began with a CAD file that wound up in a 3U rack format looking like this:
A high quality VCA was obviously essential, so the THAT Corp range of THAT2180 chips were the obvious choice. In addition to being low distortion and noise, the THAT2180 range is very well documented with both a long datasheet and numberous practical application notes. The design notes on the THAT Corp website were/are a great help and everyone I spoke to that THAT was quick the respond with good advice. Worth reading for this project were DN103/DN104/DN106/DN110/DN116. I would also like the acknowledge Rod Elliot’s project 141 as great reading during the design process. One of the major benefits with this VCA is that it has a built in exponential converter that is quite accurate, meaning the left and right channels with track very well, both each other and according to a logarithmic scale.
In order to manipulate the control voltage from the potentiometer I used a TL072 to provide me with accurately trimmable gain and offset. I was aiming for and overall range of -540mV to 0V (equivalent gain being -∞dB to 0dB) and this got me right on target.
The VCA card (on its third revision) wound up looking like this: (note, this particular circuit has been breadboarded and worked great considering the limitations of the breadboard. It is missing some trim potentiometers for the moment)
I had now replaced the classic but flawed dual gang logarithimic potentiometer with a single linear unit emitting a predictable control voltage to a pair of accurate amplifiers. Great. Seemed like a good jumping off point for the equalization too. I’m not a big fan of baxandall equalization on DJ mixers and opted for a pair of state variable filters on each channel to provide High Pass and Low Pass response – using my beloved THAT2180. Using a jumper, the same filter can be used for HPF or LPF so this minimized my design time and production costs. A great example of this is detailed in THAT Corp DN130 (are you sensing a pattern here?). This design requires a very low bias current opamp, I opted for the LT1169 as in addition to performing very well, it comes in a dual opamp package making my pcb layouts much easier than a single design.
The EQ card also performs the switching between the Phono and Line inputs for each card, so the PCB needed to have room for this. For this kind of application, nothing will ever beat a relay. I’ve used the NEC EA2 range in the past and decided to use them for this project. Annoyingly, after I had ordered the PCBs I found out that they had EOL’ed the relays used and moved this range to the Kemet corporation with a change of footprint. Damn.
I designed a small PCB to handle the switching of this relay, which houses all of the switching, led indication, relay driving as well as the power distribution for the VU meter illumination. The management of ground currents are essential in any low noise audio design, so the decision was made to keep audio and power grounds very separate. In regards to the relay switching, with a 5V/140mA required to actuate the coils held the possibility to disturb sensitive audio circuitry. I made sure the +V and GND terminals from the coils returned to the front panel logic PCB.
As an aside, the VU meter illumination looked great with 5V so I was able to use a single 5V rail to power this, the relays and the phono/line LEDs.
No good DJ mixer is complete without a few Phono preamplifiers, and I opted for a tried and true design I’ve used many times before. Based around 3 opamps it includes a high pass ‘rumble’ filter, something I’ve found useful in the past due to the common DJ practice of pitching down records. Records already have substantial low frequency content and this practice only contributes to that. Superfluous low frequency content is a great way to use up headroom quickly, not something I was after. I’m fairly traditional with my choice of capacitors, so the RIAA equalization was all performed with high quality polystyrene caps. They also happen to look great!
The summing amplifier was something I also didn’t need to give a huge amount of thought to, as I’ve used a design outlined by Douglas Self in his book of biblical weight ‘Small Signal Audio Design’ a few times and loved it. The one curveball, however, is the use of the classic 2SA1316 transistor in the original schematic. This is an ultra, ultra, ultra low noise device, with a spec’ed Rbb (base spread resistance) and is completely unobtainable (I don’t buy off ebay) and so I obviously needed to find a suitable replacement. On the first summing amplifier PCB I designed I had used another THAT Corp product, the THAT320P transistor array for the four transistors in the long tailed pairs. I had fooled around with a few other transistors and eventually found that I was getting great performance from a basic BD140. Cheap and will be available forever, two things I like.
The headphone amplifier is another design that I’ve used a number of times, it is similar to the popular Gyraf design, in that it is buffered by a BD139/BD140 pair and uses LEDs for the diode drop. This has been used by quite a few other designs too but the Gyraf design is nice and succinct. Something missing from his original schematic, in my opinion, are current limiting resistors on the output. Particularly important in a situation where headphones will be plugged in and out repeatedly over the course of an evening, day after day. The ground current for the headphones also has the potential to distrupt audio ground and has to be managed carefully. The ground line from the headphones returns to the star point.
And that leaves one very important card. A number of functions were relegated to this one which I dubbed the ‘master card’. On board are two discrete opamps for the left and right master channels, a buffer and trim for the VU meter send and two balancing amplifiers for the cue output.
The discrete opamp is a very old design, one I’ve used in projects before in the API2520 format, the Melcor 1731. The transistor substitutions I used are based on the work of Gary Barnett who brought this opamp back from the dead with his GAR1731 opamp, a great addition to any design. Ordinarily I wouldn’t go to the trouble of using a discrete opamp for the output of a device, but it will be driving a transformer for balanced output. Balanced outputs are the only way to interface audio professionally and I’ve always been a fan of how transformers colour the sound. At home I use a tried but true pair of Genelec 1031a monitors which are happiest with balanced signals.
The VU meter could be driven by the discrete opamp or the transformer but it is considered best practice to buffer the signal beforehand to avoid any ill effects on the integrity of the output signal. I use a NE5532 buffer with trimpots to allow for accurate adjustment of the signal running to the meters. I used Sifam clones made by Taiwan’s Nissei meter company.
The cue bus is balanced by (again) THAT Corp chips, the 1646 series. These don’t need much explanation.
So, those are the 6 core function cards explained. How are they all connected together? I decided to use a large motherboard design for signal routing and power distribution between the cards. I had this PCB fabricated in 2mm fibreglass instead of the usual 1.6mm I use for other boards so it was sturdy enough to hold everything. Honestly, I’m quite unhappy with the 20 pin sockets I used to socket the cards. They simply don’t have the support to hold the cards in place and the Y/tuning fork shaped contacts are pathetic. I believe they are only rated for 25 insertions or so which isn’t good in the long term. I’m keen to get my hands on some Molex 22-18-2201 connectors but they are so damned expensive that I haven’t gotten around to spending the money just yet.
Overall there are two channels each consisting of a phono preamplifier, input switch, high pass filter and low pass filter. The signal is then split at this point for either the program or cue bus where it is run through a VCA then the appropriate output circuitry.
Nothing about this mixer is revolutionary (though it is a rotary mixer -haha-) but I was motivated to build it because I couldn’t find anything that met my needs on the market. There are some great mixers being built at the moment, and I have to say the designs coming from Condesa, E&S and Isonoe were very inspirational.
For now I’ll enjoy the process of finishing this prototype off.