General points and a bit of backstory
It’s no secret that I’ve modded the one or other synthesiser for my dear friend Pedro Eustache. Accordingly, I was blown out of my socks when Pedro messaged that a certain Mr. Behringer had dropped by for a gig and gifting him a 2-XM. A couple of days later my own XM arrived (I got lucky with an early order), which is to complement Pedro’s, although in radically modified form. Before describing some modding options though, I want to describe/appreciate the circuitry of this synthesiser module first.
The Behringer 2-XM is oriented towards the Oberheim Two Voice, or, maybe more so, a dual SEM module with voice panning control and level settings. The case is nice, and I don’t mind the tall trimmers at all (at least they have a stabilised plastic shaft), and the circuitry looks neat. All components are placed on the top side of the PCB, all multi turn trimmers are at a convenient location on the panel sides.
Most importantly, the circuit design is en route de la monde belle vintage in crucial places while getting inspiration from the newer SMD-based SEM Pro as for midi control and utility VCAs. The implemented portamento is analogue and can go from snappy to long. Although I wish it were implemented *post* V/Oct inputs and dedicated utility VCA I/O, I get that there are limitsto the patch bay. Overall, as critical as I’ve been of some aspects of previous Behringer designs, this one I find awesome! Plus, it comes in recycled cardboard packaging (still styrofoam stuff inside, but I hope they find substitutes for that soon too).
Circuit description
Oscillators
The VCOs use some substitution parts for obsolete components but stick to the old SEM oscillator design, and that‘s where even the “NEWSEM9” VCOs of the Marion Systems 2010 SMD reissue went a different route. This may have to do with availabilities of “on spec” SMD packages at bulk costs, yet whatever the reason, I really appreciate that the people responsible for the 2XM went this way.
Here’s the VCO section of the second voice mapped onto the old SEM schematic. If you ever need better resolution/the rest, just be in touch per email.
Details: V3046 transistor arrays are used instead of obsolete CA3080s, and the fets switching the integrator cap (PPT) are 2SK209 (marking reads “XY”) selected for that application (the VCF section spots different Fet types). OPAS are LM4580.
Toying around with that circuit a bit, I swapped OPAs for 14580 and the V3046 for an NOS LM3046D, but the sonic difference is minimal and not quite worth the effort:
Filters
The filters are built around an V13700 OTA, two ppt caps (nice) and MMFBJ113 fets as buffers. Since the CA3080 is identical with the core of the LM/V13700, this substitution is a no brainer. The differences you hear in shootouts with the “NEWSEM9” Marion Systems/Tom Oberheim SEM from the 2010s may be on account of that NEWSEM9 not using FET buffers but keeping the filter integrators in an OPA loop. So, as far as fidelity to the 1970s design (sentimental or not) is concerned, I guess Behringer out-obeheimed Oberheim not only in the VCO design but in the filters as well.
As for OPA choice, we get the standard offering of LM4850, yet swapping them for 14580s (truer to the original and with much worse specs) I couldn’t spot significant differences. Luckily, the designers got the OPA buffering right this time, unlike on the Model D or the 2600, that is, they left the original signal path intact but added buffered patch bay I/Os – which was in both cases the right choice, in my opinion. and I could not hear any significant differences to my Doepfer SEM filter module that sports AS3080 clones.
If you think the 2XM sound too tame: What made a lot of difference though was driving the filter a little hotter and pulling up the resonance. See the modding section below for details.
The VCA(s)
The main VCAs are built around V13700’s, a PNP and 4580 OPA buffers, so nothing exciting but absolutely the right way of doing it unless you use CA3080s.
Like the old SEM, you get an offset trimmer and a relatively conservative gain staging that can be pumped up very nicely (pro hint: don’t overdrive the actual OTA signal input but reduce the value of the 47k resistor feeding the common base transistor i/v stage (that’s R17b on the‚ ole SEM schematics).
VCAs in plural also means that there are a bunch of utility VCAs (13700s and OPA buffers) for routing envelopes ver midi velocity to the VCOs, VCF (not labeled on the machine but described in the manual), and main VCA. We get this on the NEWSEM9 board of the Marion Systems version. There seems to be a bit of an Oberheim history of VCA-ing an VCA amount (as F. Anwander found out when repairing their OBXA recently); at least the service notes of the OBXA foresaw an option of velocity sensitivity…
Envelopes and LFO
The envelopes are as per original, transistor based and all that. The LFO is also as per original, a simple relaxation oscillator followed by a TRI to SINE converter per 13700 OTA, but the core circuit is tweaked for higher speeds (one 390k resistor replaced with a 100k value). Accordingly, the OPA choice is one with a faster slew rate, and SGM873-2 (not top notch but almost twice as fast as the 4580 and good enough anyways).
The envelopes are transistor based only, just like the old Moog Contours, to which they are not unrelated.
On account of Behringer using a lot of tall trimmers here instead of Dshaft pots and all, some folks are trying to find some nice knobs to put on top. TBH I found the tall trimmers horrible on the Roland Clone Modules but don’t mind them at all here. The only pot I wish were different is the LFO speed dial but not on account of the physical properties of the shaft but rather the response curve (which should be antilog IMO).
More pictures: https://imgur.com/a/behringer-2-xm-pcb-eilravr
“Hot rodding” – hotter input, lower output
On a forum I saw someone describing the sound of the 2-XM less wild and more production ready. However that’s transferrable to the reality of your studio/stage, I generally find that driving some circuits hotter than specified by old schematics can bring out some nice sound qualities. Especially state variable filters are susceptible for input level variations and can go relatively tame to aggressive. Equally, OTA-based VCAs can add nice overdrive when tweaked right. The downside to this is, however, that some synths then also can start distorting in places you don’t want them to at the same time. In case of the 2-XM that is the final voice mixer section, and even before that, the VCA output buffer.
So, here’s what I did for tweaks: decrease the filter input attenutating resistors from 220k to 150k, decrease the VCA control signal attenuation from 47k to 33k, and then dial back the VCA output buffer by reducing the 100k feedback loop resistor to 68k. Add a bit of resonance increase, and you’re good to go! If that’s all too much, dial back the 100k resistor of teh VCA output even more. Easy setup: pull both VCO 100% to square wave and pull resonance up, then tweak the final attenuation post-VCA.
Increase MIXER input levels: the easiest way can be accomplished from the backside of the pcb. Put a 470k resistor each between the wiper of each VCO/noise mixer potentiometer and the upper terminal of C254 (voice 1)/C28(voice2). These film box caps are part of the feedback path (C22 on the original schematics) and your easiest entry way into the filter. You find them right between the VCO and EXT/NOISE potentiometers.
Increase VCA control level: you need to wire a 100k resistor across R621(V1)/R180(V2). Attention: tiny resistors at work here!
Decrease VCA output: wire 100k or lower across R614(V1)/R173(V2). Same tiny resistors at work here…
If your direct filter out clips: increase R610(V1)/R169(V2) from 100ohm to 1k. Thankfully, these resistors are 0603 size.
Resonance boost: pull the top (terminal 3) of D40(V1)/D11(V2) via a 2k7 resistor to ground.
Modifications
Details coming through soon! Here are some previews: filter morph, sub osc, resonance cv