Bogen MXM Modifications

Please use your Browser's "back" button to return from all image links.

<<== Back: Considerations for the Direct Output Buffers

Implementation of Direct Output Buffers

Buffer specifications. My design specifications for the MXM's direct output buffers were: (1) A total of six buffer circuits (using three dual op-amp packages), five for the input channels and one for the master output channel. (2) High input impedance (particularly for the five buffers tapping channel faders). (3) Low output impedance (<600 ohm), and stability when driving shielded cables of moderate length. (4) Use of op amps with low noise and distortion and wide bandwidth; use of low-noise metal film resistors throughout. (5) Nominal audio output levels compatible with today's standard for ground-referenced audio lines, i.e., -10 dBV (as typified by the Alesis ADAT recorder's "unbalanced" inputs). (6) Unity voltage gain, and pads (voltage dividers) on inputs where appropriate. (7) Immunity of solid-state devices from damage caused by excessive input voltages. (8) Maintenance of consistent signal polarity from inputs to outputs. (9) All buffers must fit on a perf-board 2.3 x 3.7 inches in size, to be located above the chassis plane in a space formerly occupied by two filter capacitor cans. Figure 10 shows schematics of the four different buffer circuits that I settled on, and photos of their perf-board assembly installed in the MXM are shown in Figure 4.

Appropriate polarities. For polarity consistency (specification #8 above), all mic inputs must use today's standard of XLR pin 2 as "hot" or "+" (where positive voltage swings are analogous to the compression part of a sound wave), with XLR pin 3 as the inverse; on the direct outputs, the signal polarity on the "tip" contact of the 1/4-inch jack should match that of pin 2 on the input XLR. With two stages of triode amplification, the Channel 1 preamp is a non-inverting amplifier, while each of the other input channels are inverting stages with their single pentodes. Thus, I chose a non-inverting output buffer design for Channel 1 and an inverting design for Channels 2 through 5 (Figures 10). Relative to the mix bus, the master channel is an inverting amplifier, already compensating for inversion by the majority of the input channels (2 through 5), so I selected a non-inverting buffer to drive the mix output jack (Figure 10). Therefore, signal polarities are consistent throughout, with one exception: the mix output is inverted with respect to Channel 1's input (as in the original MXM).

Output coupling network. All output buffers employ TL072 op amps and share a common capacitor-coupled output design (Figure 10), which uses a 10-microfarad electrolytic capacitor bypassed with a 0.1-microfarad stacked polypropylene film capacitor. In audiophile circles it's believed that such bypass capacitors help electrolytics pass high-frequencies and transients, preserving "airiness" in the sound. From past experience with TL072s used as drivers, I've found it important to include the series 100-ohm resistors to insure stable operation into shielded cables under all conditions (especially feeding high-impedance inputs, such as a guitar amp). The 100K resistors further add stability by insuring that all outputs have some pathway to ground regardless of the world external to the MXM.

Overvoltage protection. I bridged the inputs of all buffers with opposing 12-V zener diodes to protect the op amps from instantaneous signal voltages exceeding +/- 12 V (Figure 10). This is a generous margin of safety given +/- 15V operation of the chips. Such overvoltage protection fixes the maximum undistorted output of all six buffers at 18.6 dBV (= 24 Vp-p); see the gain diagram in Figure 11).

The two non-inverting buffers. For the master ("mix") and Channel 1 output buffers, op-amps are configured as simple unity-gain non-inverting voltage followers (Figure 10). In the master (mix) channel, a voltage divider containing the 100K and 39K resistors forms an 11-dB pad (see section entitled "Pad Decision for the Master Channel Output Buffer"). This sets the input impedance at 139 K-ohm, which is low compared to the other buffers (with their 1 M-ohm input impedances). However, since this particular buffer is driven by a cathode follower (the second triode of the 6CG7 in Figure 8), this impedance does not overload the MXM's master output stage. Channel 1's direct output buffer (Figure 10) has essentially no input pad. The 10 K-ohm series resistor helps to isolate the buffer and a 1.0 M-ohm shunt resistor determines its input impedance.

The one ill-conceived feature. In the case of Channel 1's buffer only (Figure 10), the zener diode overvoltage protection bridge includes a series 1.0-M-ohm resistor to prevent the circuit's input impedance from dropping to just 10 K-ohm during protection events (instead it drops to 500 K-ohms). This was my attempt at op amp protection without zener breakdown causing distortion of Channel 1's signal on the mix bus. This was unnecessary because the summing amp begins to distort at a much lower input level (+2.3 dBV; see above) than is required for zener breakdown (+18.6 dBV). Protection would be enhanced without loss of performance if this 1.0-M-ohm resistor is shorted out the next time this MXM is further modified.

The four inverting buffers. The direct output buffers for Channels 2 through 5 use the remaining TL072 op-amps in a unity-gain, inverting configuration (Figure 10). Input impedances are set at 1 M-ohm. In parallel with the feedback resistors are 6- or 7-pF capacitors to promote stability. The pentode preamps of Channels 2 through 5 have slightly more gain than Channel 1's dual triode circuit (67 dB versus 61.5 dB; see Figure 9 and Figure 11). While not a big difference, I decided nevertheless to include a pad ahead of the output buffers for some of the pentode channels. Arbitrarily, I did this for Channels 2 and 3, where 9.5 dB pads are designed into the op-amps' input/feedback networks without affecting input impedance (Figure 10). No pad is used in the direct output buffers for Channels 4 and 5 (Figure 10). Effectively this means that, for a given input level, the faders on Channels 2 and 3 have to be set slightly higher than those on 4 and 5 to match direct output levels.

A calibrated gain diagram of the modified MXM. In Figure 11, I have drawn an overall gain diagram for the modified MXM and aligned it to absolute signal levels in dBV. It is based on gain data already presented in Figure 9. A condition has to be chosen for such an alignment, so I arbitrarily picked the following imaginary one: the master fader is fully clockwise; the fader for one input channel at a time is fully clockwise while the rest are fully counter-clockwise; the output buffer on the master (mix) channel is delivering its maximum signal before distortion, which is +18.6 dBV (as fixed by the zener diode bridge). For perspective, the dBV scale is annotated with maximum and nominal Alesis ADAT ground-referenced ("unbalanced") input levels. This provides real-world landmarks since the modified MXM is likely to be used with such a recorder. Notice that all solid-state output buffers have 13.6 dB of headroom above the ADAT's maximum input level. Track levels set properly according to the ADAT's meters thus preclude any possibility of solid-state distortion. The MXM's own V.U. meter is dedicated to displaying the post-fader master output level in relative "Volume Units." In the figure, I have aligned its scale with dBV as measured at the buffered master output; for example, the V.U. meter reads about -5 relative units when the buffered master output is +5 dBV.

Pad decision for the master channel output buffer. The modified MXM's unity-gain solid-state buffer on the master channel has an input pad (-11 dB; Figure 10). Without a pad, only very low master fader settings could be used with this buffered output. Referring to Figure 9, the total gain of the tube stages between the mixing bus and the output driver is 11 + 21.5 = 32.5 dB. Consider a (very hypothetical) case of the Channel 5 fader and the master fader both set at maximum. Overall gain from microphone input to master output would be (67 + 32.5 =) 99.5 dB! No known practical audio situation requires that much voltage gain (nearly 100,000-fold). Without a pad ahead of the master output's solid-state buffer, the master fader itself would need to handle all of the necessary attenuation, confining useful fader settings to its most counterclockwise region. My choice of -11 dB for this pad was pretty arbitrary, and one could advocate even greater attenuation. The choice would be critical only in the special case of simultaneously needing both the (original) transformer-balanced master output and the (added) buffered master output. An increase in pad value, or a variable pad, might be desirable if this situation ever arises.

Next: Suggested Further MXM Modifications ==>>

Back to Bogen MXM Main Page

Back to Clark Huckaby Main Page


Copyright 2005 Clark Huckaby. All rights reserved. Commercial use of this content without first obtaining the expressed permission of Clark Huckaby is prohibited by law. Distribution of copies beyond personal use and/or re-posting of this page is prohibited by law.