Interfacing the Line (Part 3)
Ben Duncan continues the saga.
In this, the penultimate part, Ben Duncan sorts out the detail on balanced pads — and line level impedances.
The Unbalanced attenuator we looked at in March HSR is often called an 'L' pad, because of the shape of the 2-resistor network in the drawing. For a balanced line, we could apply an L-pad to each input (hot and cold), the LAR on each going to ground. But the common ground connection isn't necessary, and may in fact upset the balancing if the grounding in the system is at all complicated. So, Figure 1 depicts the translation of a pair of L pads into the n or π pad for feeding balanced inputs. The source can of course be either unbalanced or balanced.
A kit of DIY register values are given in Figure 2.
Points to note are:
1. The input impedances have been set to lie to within the 'bridging load' standard. If the source equipment is budget and isn't capable of driving a 10K to 20K load impedance, you can safely double up all the resistor values pro-rata, to halve the loading.
2. The source impedance, looking up at the output, is quite high for the -5 to -20dB attenuation value, so it's a good idea to place the pad upstream, close to the gear you're feeding into, to cut out HF losses. For XLRs, this means wiring the pad inside a male line connector, A3M. Achtung! With budget source equipment, the output impedance could well be higher than the pad. That means a reversal - the pad being placed downstream, in order to keep the high impedance cable run as short as possible. If in doubt, test out the pad at both ends of the line. Some "ssssstt-sssssst" cymbal sounds will help you identify the position that offers the best top-end response.
3. For balanced inputs, small differences in the hot and cold line impedances will upset the common-mode rejection. In other words, if you're careless about choosing resistors, the anti-interference properties of the balanced input can go for a Burton. The moral is to pay out 15p extra for quarter watt metal film resistors, with one per cent tolerance. Failing this, you're at liberty to work your way through a bag of five per cent resistors, using a digital multimeter to select a pair (or pairs) with identical values. Remember - matching between UAR and LAR is what counts in a balanced pad, not absolute obedience to any specified value.
The picture shows a triad of metal film resistors mounted up inside an 'A' guage jack plug. Mounting pads like this saves money, but can be a frustration when the cable with the desired attenuation can't be dug out of the spaghetti fast enough - or the leads with on-board pads aren't identifiable. The remedy is to clearly label whichever plug contains the attenuation resistors. For those of a mathematical bent, the attenuation ratio of a balanced pad can be calculated as follows:
Attn Ratio = (LAR + (MAR/2)) / LAR
The exact result will depend on load source impedances as usual, and the ratio is converted into minus decibels in the normal manner.
In studios or on stage, the purpose of DI (Direct Injection) boxes is to provide protective interface between two alien sound systems - the musos' and the sound engineers' respectively patched. At least that's a cynic's viewpoint...
Other than attenuation and buffering (that's boosting up the load impedance while preserving a low source impedance, i.e. maintaining "high-in, low-out" relations), this implies a degree of isolation. In practice this means preventing a direct connection between the two system grounds. One technique relies on the high 'trans-grounds' impedance of an active, balanced DI, the sort that runs on two PP3 batteries (Fig 3a). Another more traditional technique breaks the ground loop with a transformer (Fig 3b). Note that despite its simplicity and lack of powering, the basic transformer model does provide a balanced output, ideal for driving the longer cable back to the console, given that DIs are customarily sited next door to their respective synths, drum machines or whatever. The trade off, as in any passive (unpowered) unit is attenuation. The 10:1 impedance conversion (gearing) also drops the level (speed) to one tenth, which represents a 20dB loss between input and output.
An active DI box plus output transformer can make up this loss or avoid it altogether and make the transformer less sensitive to the backline gear's source impedance. A classic example of this breed is Chas Brooke's BSS AR116 model. Otherwise we can dispatch the transformer altogether. The more sophisticated balanced input circuit topologies can cope with one big common mode error voltages that can arise when grounds 1 and 2 are on widely separated mains ring circuits and therefore differ from one another by a volt or more. Just to remind you, common mode voltages (CMV) are sneaky, not being measurable at the output, because a decent balanced circuit should cancel it all out. But CMVs of more than a few volts are capable, nevertheless, of driving a simple circuit into distortion. If you suspect this, test for DC and AC voltages between the grounds. Any reading below 1 volt should be satisfactory. A universal DI box's input impedance needs to be very high - ideally one million ohms or more if it's to cope with the direct output from a guitar, bass and some electric piano pickups. If the input impedance is too low - a mere 100k say, the harmonics will be wiped out or the sound may be drastically altered or both. The remedy is to pick up the DI feed from a low impedance source, namely the dedicated DI output socket found at the back of some instrument amps or failing this, come off the 'line' or tape output sockets, taking note of how this arrangement affects the controls. For example, if the line output socket is pre-eq, the adjustment of the amps eq settings won't affect the recording feed.
Feature by Ben Duncan