Synth pages
   Synth story
DIY modules
   A110 sine shape
A117 clock in
A127 LP/BP switch
A135 indiv outs
A140 self-cycle
A150 switch voltage
A151 2/3 steps
A155 16-steps
A161 clock in
A163 DC couple
IEC power inlet
Plan B summary
M9 bleed-through
M10 envelope
M11 filter
M14 crossfade
M15 sine adjust
M15 others
M17 timer
M28 upgrade
M32 joystick
M38A Expander
Diode Ladder Filters
Korg MS10/20 Filters
Moog paper refs
Fooled by SPICE
Filter pole animations
Serge VCS
EDP Gnat schematics
Old brochures
PE Synth/Minisonic
About me
Photo links
Purple Mountain
Contact details
What's new
Cheltenham Circular Footpath
The Trouble with Christmas
this night wounds time
Behold the Computer
Solartron scans

Site map

I have a Facebook page for occasional blog-like articles:

Tim Stinchcombe's Synth Stuff


Plan B: Summary of design issues and modifications

The following table summarises:
     - all those Plan B modules I have had a detailed look at
- any design issues impacting on a module's functional operation
- links to suggested modifications to overcome some of these issues

[I didn't deliberately set out to critique Plan B designs, but simply fell into it when no one else stepped in to help after reading of problems other people were experiencing with their modules.

Since I started publishing these mods in Sept 2009, there has been virtually no comment or input from Plan B on these issues—the first, and as far as I know only, proper technical interaction was in the M9 thread at ModWiggler. Even at the now defunct Plan B Yahoo group ('PLAN B analog blog'), there was only a small (and somewhat flawed) attempt at addressing the incorrect clock speed exhibited by M28 modules fitted with third-overtone crystals, and then later a mention of the M28 upgrade, of which Peter Grenader had little-to-no technical input to—both of these were within the last 6 months of activity in that group, which lay dormant from Oct 2009 until Yahoo groups closed at the end of 2020. Thus to all intents and purposes it appeared to me that Plan B had abandoned all support of their modules.

During the intervening years Peter and I have had some interaction, and if these mods had a mention (technical or otherwise), any exchange would likely have ended abruptly and at an impasse. More recently, toward the end of 2017, Peter outright refused to accept that the cause of the M9 bleed-through is due to parasitic capacitance in the PCB layout, despite the very credible evidence presented here, and also said that the problems with the M38 were due to the use of the wrong sort of pots. Having awoken more fully to the implications of these denials, which includes the utter fruitlessness of any technical discourse with Peter, I shall no longer attempt any such interaction.]

ModuleSummary of issuesLinks to fixes
Unwanted capacitive coupling between tracks on the PCB allows signals to bleed-through to the outputs even when they are switched off, and may also cause undesirable high-frequency oscillations in the internal amplifier sections. Some simple resistor value changes largely mitigates the bleed-through problem and quashes the oscillations.M9 bleed-through
Envelope Generator
There are three diodes in the circuit that are prone to being blown given the right circumstances, causing loss of associated functionality—the addition of a few more diodes, plus some track cuts, overcomes these problems. An inconsistent choice of resistor values between the normal and inverting parts of the envelope output circuit causes the envelope slope to change near the extremes of its excursion, giving little 'peaks' in the envelope output—a single resistor change cures this. Also, some component fitments for the MkII version may cause double-triggering of the envelope.M10 envelope
Bandpass Filter
Though there are a few strange things in the M11 circuit, none are sufficiently bad to have enough impact on the module's functionality that they are crying-out to be corrected; however the module's defining feature is somewhat misnamed, so I have produced an M11 page to help clarify this.
State Variable Filter
The relationship between CV and cut-off frequency in the M12 is very linear, and when compared to a usual '1V/octave' control law, not surprisingly this makes it hard to control the cut-off, especially at lower frequencies, as was noted many years ago in this thread:Plan B M12 Frequency and Bandwidth response problems. I spent quite of lot of effort working out a modification to make it more exponential (there are some charts and details here: Plan B M12 - mod to improve the frequency control law), but because it so drastically alters the way the cut-off changes with the CV, the sound of the filter is drastically changed too (i.e. equals 'not so good'!). Consequently I've not published the full details of the mod, but may do so one day, perhaps after I have studied it some more and better understood the effects it has on the sonics of the filter.(perhaps one day when I get back around to looking at it again...)
Lowpass Gate
The M13 seems to have undergone quite a number of revisions: initially the wiring from the pots and jacks on the faceplate were wired point-to-point to the PCB; then the pots/jacks sprouted a small PCB, and wiring was via a 12-way ribbon connector; a little later the ribbon became 14-way. Some have reported (Plan B Model 13 bleed?) problems with bleeding between one channel and the other—it seems likely that this is caused by capacitive coupling in the ribbon, and replacing the ribbon with point-to-point wiring alleviates the problem.
The M13 is essentially a copy of the original Buchla 292 Lowpass Gate circuit, in which there is a 47kΩ resistor in the output buffer: in my example this resistor (in both channels) was actually fitted with a 4.7kΩ, despite the silkscreen showing '47k'. This increases the loading on the buffer and causes an asymmetry in the output waveform, with increased harmonics which are quite audible, making me think that perhaps the change in value is actually deliberate—in any case changing it to 47kΩ reduces the harmonics and re-establishes the symmetry in the waveform.
Post with photos showing the 47kΩ locations (actually in the channel bleed thread)
Voltage Processor
Inconsistent resistor choices within the circuitry of the M14 make the crossfade awfully asymmetric—a fairly extensive re-scaling (involving swapping 7 resistors and adding 3 more) improves this, and also reduces the voltage sweep necessary to execute the crossfade down to around 5 volts.M14 crossfade
If we include the Subcon version, there may have been upwards of 12 or so revisions to the M15, which were either enhancements, corrections, or form factor changes, etc. The M15 others page attempts to summarize the revisions: some issues corrected in later revisions (e.g. a reversed electrolytic capacitor and a stability cap on the pulse wave) can easily be applied to earlier versions; a change to the morph circuit for early versions is given (this area was still being played around with in the latest 'rev 4.0' Subcon VCO); another 'undecided' area is clearly the resistors in the 'sync in' circuit, but not having had time to examine this in greater detail I've just listed some component fitments I have actually seen or heard of; replacing the poor choice of 'W' law pot for the PWM pot (if fitted) is an easy way to gain greater control over the PWM range. The sine wave adjustment mod is probably not really worth the effort unless you get desperately annoyed with the original method (which I found counter-intuitive, in that you cannot iteratively move both pots towards 'optimum', but have to take it further away from 'best' with one, then bring it back with the other).M15 others
M15 sine adjust
Event Timer
Due to a defect in the design, spikes are output on the 'sum' output at the falling edge of each gate, and they shouldn't be—a couple of track cuts and the addition of a resistor removes these. The widths of the pulses output at the 'sum' out are not all the same, due to an inconsistent choice of the capacitors involved—simply changing the capacitor concerned removes this. An additional simple change (adding a resistor and removing a capacitor) lowers the maximum voltage threshold selectable, giving a greater range of usable pot travel for setting the gate levels.M17 timer
Analog Shift Register
If two or more M23s are chained together the clock signal passed down the chain is the wrong one, and so the 8-stage (or longer) combined shift register won't work as one would expect it to—a fairly simple mod can rectify this. There is also some doubt about the advisability of running the main S&H chip off 12V split supplies—a simple mod can change it to run off a single-sided 12V supply, and which shows a pronounced improvement in the S&H droop figures.M23 ASR
Heisenberg Generator
There are some very strange constructs in the design and implementation of this module, some of which look like desperate attempts to overcome some (presumably) perceived shortcomings in the circuit (for example turning two transistors around so that they are operating in the reverse-active mode of operation). In the 'Smooth' section there seems to be some sort of attempt to have an RC time-constant 'adapt' with the frequency of the clock, but it doesn't seem very tractable, with the end result that the 'Smooth' output doesn't really know what it is. So far it has defied my attempts to improve it, as to do so properly really requires access to the software in the microprocessor, which I don't have.
I did however modify a single module to allow the external trigger input to be switched through to the micro instead of the internal 'Stepped' VCO, which means that the random divided-down pulses it outputs are in sync with the external trigger. This is relatively simple to do, requiring a couple of track cuts and the addition of a switch and a diode—I might write up how to do it one day when I have the time.
(currently none)
Audio Processor
(no known issues having a functional impact)
Control Processor
There is such a strangely-placed resistor in the slew limiter circuit of the M26 I have (a 'rev 1.1'), that I inquired if others had it in theirs, or at a different value: Plan B Model 26 Slew Limiter component check, and a response there suggests it wasn't there in the original release, and so was placed there later for some purpose. Due to diodes and buffering in the slew circuit however, it cannot impact on the slew operation directly, and can only have an indirect effect, in that it severely loads the output of the module driving the slew, so much so that it is likely to distort the driving module's waveshape as it appears at the input of the slew circuit (and so that in fact may be its purpose..?). The loading effect may be so severe that it actually prevents the M26 from working with some modules, one instance being the unipolar square output from a Doepfer A-146 LFO (it more or less just outputs DC). (Included in that thread are a few scope shots illustrating the sort of waveshapes that may be seen.) Simply removing the resistor or cutting it off (it is 'R28') 'restores' the slew to a more standard arrangement, though I haven't spent much time checking to see if what results is then 'sensible' in terms of the slew response (and what is more in the short term I am unlikely to do so, as it doesn't seem that there are that many people out there with this module who are bothered by this strange behaviour).
Update Apr 24: There are op amp sections taking inputs of a 4011 chip below ground (in the 'Boolean Gates' section) and seemingly this can destroy the CMOS chip, so increasing the resistors there may help; also both the slew input and output sockets aren't grounded properly, which may lead to strange behaviour (if e.g. used with a Doepfer mini case/rack which doesn't have its own ground...). Get in touch if you need exact details.
(de-solder or cut the resistor off)
Tap Clock
The main thrust of the work on the M28 was to upgrade the 'phase lock' function, which in its original state was badly named as it simply didn't do what it said—the upgrade requires a new/re-programmed microprocessor chip and a small modification to invert the incoming clock signal. During this work several other issues came to light, the most notable being: the 8th note accent pulse amplitude levels are dissimilar; and the resistors defining the voltage levels for which the 'select' input determines which pulses appear at the 'serial' output are rather inconsistent (and indeed several different sets of values have been observed in different modules)—both these issues are easily rectified by simple resistor changes.M28 upgrade
Buffered Mult
(no known issues having a functional impact)
Vector Plotter
In terms of the number of design issues and mistakes in the modules I've looked at, this module is the clear winner, with 10 items noted. Fortunately most of these will have little impact on the functionality of the module (for example 5 components are useless, being either unconnected or rendered ineffective by surrounding components/wiring), though there are some issues which may have implications for the long-term reliability of the module (for example, an electrolytic capacitor inserted with the wrong polarity, or CMOS inputs taken outside their supply range by the op amps driving them, causing the op amps to run hot, and probably not doing the protection diodes in the CMOS chips any good either).
One feature that may have room for improvement is the kinetic gates: my unit exhibits inconsistent behaviour between the X and Y axes (reasonably easily demonstrated with a few minutes of waggling the joystick), and also between positive and negative going signals (really requires a scope to see). The circuit topology is the same for the X and Y axes, but there are four component value differences between both sides, when one would expect them to be identical (it's possible they are build errors, but to have four of them is perhaps a little unlikely). With a little concerted effort it could well be possible to find a decent set of values which would make the kinetic gates perform in a more consistent manner—however this is potentially a rather fruitless task, as there probably aren't that many M32s in existence, and even fewer owners who care that this function doesn't perform very well.
Update May 19: I have done this now!
(no known issues having a functional impact)
(no known issues having a functional impact)
In its original state the response of the attack, decay and release pots is so bad the module is almost unusable—fortunately the simple addition of a few components to the PCB changes this state of affairs.M38 ADSR
ADSR Expander
For reasons not understood, only 0V and +12V are passed to the M38A from the M38, meaning that the CV summing there is made particularly awkward due to the use of the single-sided supply, and necessitating the use of some clumsy DC offsets; in addition, on the M38, the siting of a couple of resistors with respect to the jumpers used for configuring use with/without an M38A are poorly chosen, resulting in some scaling changes when running an M38 with an M38A, and one of these is so bad (the sustain voltage level) that it renders the whole assemblage inoperative (at least it was for the example I saw), though as it looks to be device-dependent, others' units may actually work (though have poor control of the sustain level). A track cut, and adding/changing a few resistors improves things considerably.M38A ADSR
Headphone Amp
(no known issues having a functional impact)

[Page last updated: 17 Apr 2024]