General Tech Info
So, this is the first of a planned series of analogue poly synths based on that “new platform” Behringer were talking about. Bjoerngiesler has posted detailed pics of the final PCBs here, and they also described/listed a couple of the ICs used. This sandwich structure is divided into a “brains” PCB, while while the lower board contains the complete sound section. I suspect this type of modularity is key here, as it makes adaptations and variations easier.
So, this is advertised as a remake of the Prophet 600 with GLIGLI software mods (faster envelopes) – Ironically, apart from adding two voices (cool), Behringer not only delivered this thing initially with quite lame envelopes and some filter frequency “blopp,” which they quickly addressed per firmware update. They also substituted those parts of the circuitry that are probably most characteristic of the Prophet 600. Instead of using a CEM3372 remake (as reissued by Alfa Par, for instance, see datasheet here), a combination of LM13700-clone OTAs and a CEM3320 clone of the Coolaudio lineup were used (V13700 as VCAs and V3320 as VCF). Now, turn of events: nothing yells Prophet 5 Rev3 like 3340, 3320 and OTAs. Yet, on closer inspection, the surrounding circuitry is set up very differently. (By the way – tech details and modifications for the Prophet 5 Rev 4 can be found here.)
References:
The Prophet 600 Technical Manual can be perused here. The Prophet 5 Rev 3 Technical Manual can be viewed here.
Loopop’s fun blind test between Prophet 5 and Prophet 6 can be viewed here.
Single Voice
Here’s a preliminary trace of a single voice of the Pro-800. A couple of things I left out (the 3340 ICs, which work the same as in the Sequentials) and a couple of things I might have gotten wrong. Yet better than no schemo at all, I guess.
Now, let’s inspect some elements.
VCO Mixing stage: The basic Waveform switching circuitry is the same between the P600 and the Pro-800. Some resistor values differ (and that’s what I have yet to trace better, to my embarrassment), but the general direction is the same (as it is different from a P5). Now, volume control looks like this on the P600:
Since there are no such level controls on the 3320, Behringer used two standard-type OTA VCAs here with U50. The signal and control signal inputs of these OTAs are more on the conservative side; I assume, to minimise distortion, and to bring things up to level again for the filter, we find an added op-amp stage (U42) per VCO mix right after the OTAs. The amplification factor is quite high here (almost times ten), and then we get attenuation again. Moreover, the AC coupling at the mix input stage of the 600 is replicated in the Pro-800 too, yet with 47uf capacitors instead of 1uf. This works, and having caps here is a boon since the 3320 filter is prone to sideband distortion when a lot of DC content is pushed through.
Filter and passband attenuation compensation (QCOMP): With some filters the passband level drops when resonance is increased. On the CEM3372, as used in the Prophet 600, your signal level stays high when resonance goes up, and so it does on the Prop-800. As the CEM/V3320 usually thins out, Behringer employed a simple yet neat trick in order to circumvent passband level loss entirely. I only know this method from Jeroen Allaert’s exquisit 2019 paper on filter design (see p.14 onwards here) and know it to work well from some circuits of my own, so who knows, maybe someone at Behringer read that paper too? Thanks, Mr. Allaert, thanks, SSI, for sharing your knowledge!
Filter Frequency Control Slew: on the P600 this is a cool/weird one. Since the original envelopes were so steppy and the CV output so low, Sequential introduced a neat way of making sure *small CV changes were softened* while *strong CV changes went fast*. A classic RC filter softens CV (think portamento) but when that CV exceeds a specific voltage change (think rapid decay) a diode circuit jumps to action and lets that CV pass without RC filtering (like your up/down switch at a portamento circuit – the principle is the same). Since the CPU of the Pro-800 is much faster (and since we now have faster envs anyways per firmware update), I’m not quite sure why Behringer kept this slew circuit in place at all. For testing purposes I removed the slew on my unit and don’t hear any steppy stuff going on…
The rest? Modo grosso like P600. Not sure how the LM13700 differs from the VCA stages in the CEM3372. If you think this is a P5 in disguise – nope! The CA3280 used in the latter are high precision stuff and the entire mixing (inverted pulse wave, for instance) and filter support circuitry (filter output buffer to keep resonance stable) is different.
Modifications
First of all, I am not interested in some so-called “panorama mod” as I find this modification musically much more limited than, say, single voice outs. Secondly, be reminded that if you mod one aspect of a voice, you have to do this eight times – this can take time (might be a good thing in winter though).
VCO SECTION
Increase Polymod Amount: I find the polymod amount too little on this thing, so I change the simply reduce the CV attenuation of the respective VCA. This means you only need to change one resistor on the VCA control input rather than two (VCO1, Filter cutoff) on the VCA output.
Simply bridge R715 (10k) with wire – this bumps up polymod and is reversible in case Behringer ever decide to increase the effect per firmware update. You can keep in or take out the 10pf capo that follows R715. Note, this only increases the effect of VCO2 to polymod since ENV1 to polymod is done per software alone.
VCO MIX LEVEL BOOST: If you want to pull up the VCO mix level for reasons of overdriving and such, you can do this most easily by reducing the value of R571 and R571 to, say half of the original 47k. There might be a bit of 3320 crunch (not my fave) in this case.
A bit more roundabout but nice is to bridge R258 and R282 for VCOB and R528 and R527 for VCO A respectively with a single 10k resistor. Like so:
This way the OTA pushes the signal more into the following op-amp. Sound like this:
Note that the PRO-800 handles this latter VCO mix boost *plus* additional VCA boost together well. Just saying… 😉
Complete nutbar VCO MIX Boost Deluxe: What I did on my unit is to bypass the entire op-amp stage after the level OTA and simply drive that one harder. This still leaves enough headroom for the 3320 filter to sound nice and clear but uses less fluff. Listen to this XMOD example, where volume at the VCO stage gets quite hot – this is after the “deluxe” mod, so note how it handles inharmonic signal content quite well.
There’s a fair bit to do here…. First, wire a 10k resistor across the VCA control pins as described in the “roundabout” VCO Mixer Boost mod (one 10k across R258 and R282 and one 10k across R528 and R527). Then Remove R571 and R572. Then wire the OTA output Pins (Pin 5 and Pin12 of U50 *or* Pins 3 and 5 of U42, which are connected to the OTA outputs) to U58, PIN1. (There might be through vias for all of these). Optionally, remove R33 and R34, but this can turn crunchy. Worth it? Best test for one voice and then decide!
Roland-style amplitude modulation (“X-MOD” on JX8P): I loved the XMOD sound of my JX8P and often peruse Tom Wiltshire’s excellent writeup of the principles behind. When doing the roundabout VCO Mix Boost mod above, it occurred to me that a similar thing could be done halfway easily by modding the level VCA of VCO A by means of the polymod circuit. So, on my unit I sacrificed filter FM and rewired that switch circuit to modulate VCO A volume. This can sound like so:
What you do: Remove R339 (left of the dual op-amp) first. This deactivates polymod to Filter cutoff. Then, if you have not yet done the roundabout VCO level mod, wire a 3.3k resistor across R528 and R527. (If you did the roundabout mod, replace additional 10k resistor with a 3.3k). Then connect the upper terminal of your former R339 with the emitter of Q26 (where one terminal of the 3.3k resistor is connected). Boom – switch on polymod “filter” and pull up OCB B amount to get that “X”-Mod going!
VCF SECTION
Increase external filter CV mount: Nt sure what Behringer were testing this with but even with a 3340 VCO I get only say, mild filter cutoff control. I wired some 3.3k resistors across each R340 (23.2k) and am fine with the max amount now. ((demo soon))
Change filter resonance to level loss style: Counterintuitive maybe, but hear me out. I got a p5 module that has two excellent filters with nifty QCOMP circuits. So I want the Pro-800 to deliver something else (andm to be honest, the max resonance on this thing is a bit too shrill for me). So I changed resonance to level loss and took the maximum amount down a notch as well.
If you want to go there, easy: best pull out C1 and wire its cathode back in (NB I reversed the polarity by mistake on my schem!), then simply wire a 47k resistor between the ANODE of C285 and the freely dangling ANODE of C1. Super easy to undo as well!
With 47K for your “R_MFB” (modded feedback) resistor, you get quite a deal of self oscillation with resonance. As I prefer a bit less, I took 56k as a value here. Note, however, how this also affects the lowest possible frequency at which your filter can oscillate.
In this clip you hear me simply changing cutoff, the level loss happens by itself. (Explains why Sequential used a filter buffer thing in the Prophet 5 as this stabilises things).
Change filter response to -12db: Easy as pie! Lift the ANODE of C285 and wire IC58, PIN6 to that anode (there is a through via at the back of the PCB that allows you to do this without needing to solder onto the IC itself! Simply check for continuity with a multimeter…).
By the way, you could also tap the -18db or the -6db out (both have through vias!) – that’s the beauty of the 3320 as it offers buffered output pins.
VCA SECTION
Increase final VCA impact/level boost: the final VCA has quite some headroom left, so here’s what I did. I first increase the control signal level (wire a 10k resistor across R733) and then also the audio input level (wire a 33k resistor across R757 and, optionally, remove R662). This gives you a point where the signal has more oomph but does not yet clip.
I am really embarrassed but I forgot to take some A/B recording of this… Well, you can test this with one voice fi you want to. A good setup is filter all open and all waveforms on. As soon as the tri wave starts to distort slightly, you take stuff back a notch.
BTW – just realised this is my 30th modding project (Eurorack excluded) posted here…
Apologies for not providing any further updates/ or pending mod descriptions for this project any longer. I describe many mods here (more than the classic mods for the original), including mistakes and spelling glitches, and feel I rather want to spend time on other synth modding/development projects.