EMV2 Flanger – Revision 8

MN3207 based clone of the Original EHX Electric Mistress (V2)

Alex Lawrow (28-April-2021)

My aim in producing this circuit was to make the best possible clone of the original
18V Electric Mistress (V2) using a BBD that is still available, but without the noise
and volume drop issues of the original.

I found that in this particular circuit the difference in BBD gain between an SAD1024
and a suitably buffered MN3207 is typically less than 1 dB (and often close to 0 dB)
for all audio frequencies across most of the sweep range. So the overall frequency
response of the MN3207 circuit differs by no more than 0.5 dB compared to the
SAD1024 circuit (i.e. using an MN3207 instead of an SAD1024 makes no audible
difference to the sound). Claims that the SAD1024 is the key to the sound of the
original Electric Mistress and that no other BBD will sound the same are simply
incorrect, and that can be demonstrated by measurement.

So the main challenge was actually to redesign the noisy LFO and VCO of the
original circuit so that all the control pots work in the exact same way as the original.
I worked out a way of doing that for the SAD1024 circuit in 2018 (see 18V Electric
Mistress with reworked LFO & VCO). For the EMV2, I took my reworked LFO &
VCO design and then modified component values to produce the doubled clock rate
required by the MN3207.

There are two halves to the circuit running on different supply voltages. The LFO
and VCO that produce the sweep for the BBD run from a 12V regulator, while the
BBD, clock buffer and audio ICs run from a 9V regulator.
The circuit needs at least a 15V supply (18V maximum) to get the correct flanger
sweep range and sweep ratio. It is possible to use a cheap MT3608 DC-DC boost
board to increase the supply voltage from a standard 9V supply to the larger supply
voltage required by the EMV2.

1
2
 Name Type Value Quantity
R35 0.25W 10R 1
R17 0.25W 100R 1
R32 0.25W 200R 1
R5, R13, R14 0.25W 470R 3
R9 0.25W 1k 1
R19a 0.25W 1k2 1
R33 0.25W 2k2 1
R30 0.25W 2k7 1
R28 0.25W 3k3 1
R21,R31 0.25W 3k9 2
R4, R34 0.25W 4k7 2
R2 0.25W 5k6 1
R6 0.25W 6k8 1
R11, R12, R15, R20, R25 0.25W 10k 5
R8 0.25W 13k 1
R10 0.25W 14k (14k3 also works) 1
R19b 0.25W 15k 1
R16, R29 0.25W 22k 2
R23 0.25W 27k 1
R24 0.25W 33k 1
R22, R26 0.25W 39k 2
R3, R7, R18 0.25W 100k 3
R27 0.25W 180k 1
R1 0.25W 2M2 1
D1 1N4001 1
D2 1N4148 1
D3 1N5819 1
D4 LED 1
C12 Ceramic 100p 1
C17 Ceramic 470p 1
C16, C22 Ceramic 100n 2
C3 Film 820p 1
C7 Film 3n3 1
C4 Film 4n7 1
C2, C5, C8, C9, C10 Film 47n 5
C1 Film 100n 1
C11, C13 Film 220n 2
C6 Electrolytic (16V) 4u7 (10u also works) 1
C14, C15, C21 Electrolytic (16V) 10u 3
C19, C20 Electrolytic (16V) 100u 2
C18 Electrolytic (25V) 220u 1
Q1 PNP 2N5087 1
U1 12V Regulator LM78L12 1
U2 9V Regulator LM78L9 1
U4 Bucket Brigade Delay MN3207/BL3207 1
U7 D Flip Flop CD4013 1
U8 Hex Buffer CD4050 1
U3 Op Amp JRC4558 (or TL072) 1
U6 Comparator LM311 1
U5 Op Amp TL062 1
U9 Op Amp TL072 1
RT2, RT3 Trimmer 10k 2
RT1 Trimmer 20k 1
Color Control Pot 10k-B (linear) 1
Range Control Pot 100k-B (linear) 1
Rate Control Pot 1M-C (reverse-log) 1
SW Switch SPDT 1
3
Circuit Notes

The audio path is the same as an original V2 Electric Mistress apart from the BBD,
the optional capacitor C3, and the addition of an output gain stage that fixes the
volume drop of the original circuit. C3 can be used to reduce heterodyne noise. This
noise (which sounds like “whistles” or “tweets”) occurs if the input to the circuit
contains a high frequency signal, for example clock noise from another circuit further
up the signal chain. Adding C3 does not alter the sound of the flanger, so although it
is marked as optional I recommend including it anyway.

Note that 14k3 can also be used instead of 14k for R10, and 10uF can be used instead
of 4u7 for C6. Neither of those changes will affect the sound.

LFO/VCO: Use metal film resistors (1% tolerance) in the LFO and VCO sections.
That means all resistors with identifiers between R19 and R33 inclusive.

Clock Buffer: Try to use a CD4050 for U8 if you can get it. If not then use a
CD4049B, or CD4049UBE. I tested all 3 of these chips to see how well they work as
buffers for the MN3207 across a range of audio and clock frequencies. I found no
difference between the two 4049 chips, but the 4050 works better as a buffer at higher
clock frequencies, and gives the MN3207 the closest match to the BBD gain of an
SAD1024.

Op-Amps: I used a JRC4558 for the input op-amp U3 simply because that is what is
in the original Electric Mistress. You may want to use a lower noise op-amp such as
a TL072.

Using a DC-DC boost board: The EMV2 can be powered using a cheap MT3608
DC-DC boost board to increase the voltage of a standard 9V pedal supply to the
higher voltage required by the EMV2. These boards typically contain a trimmer pot
to set the output voltage. If a boost board is used, I recommend increasing the value
of R35 from 10R to 47R and setting the output voltage of the boost board to be at
least 16V. The boost board output voltage should be set to be sufficiently large for
the 12V regulator in the EMV2 to be giving an output of 12V. When housing the
boost board in the enclosure, try to keep it away from the audio circuitry of the
EMV2 and the input wire. It is a good idea to shield the boost board with aluminium
foil to stop it producing RF interference. Firstly wrap the boost board in insulating
tape to prevent any connections to the foil. Then wrap it in the aluminium foil, and
finally wrap the foil in more insulating tape to prevent short circuits from the foil to
any parts in the enclosure.

4
Trim Procedure

Bias trimmer: Start with a bias voltage (measured at pin 3 of the BBD) of 5.2 V.
Turn the bias trimmer to find the two voltage levels either side of this where
distortion first occurs and then set the bias voltage half-way between those levels.
For example, if lowering the bias voltage to below 4.6V or increasing it to above
6.2V gives distortion, then set the bias voltage to be 5.4V.

Color trimmer: In Filter Matrix mode, put the Rate pot at minimum and the Range
pot at maximum. Adjust the color trimmer so that the Color pot can be set to
maximum without the circuit going into oscillation. On my own build I have the
color trimmer resistance set to 5.4 kΩ.

Clock trimmer: In Filter Matrix mode, put the Color and Rate pots at minimum, and
the Range pot at maximum. Adjust the clock trimmer so that the clock frequency at
pin 2 of the BBD is 40 kHz. If you do not have a scope, you can set the clock
trimmer using the following trick. All you need is an in-tune guitar and a guitar tuner
(or a good ear for pitch). In Filter Matrix mode, with the Rate pot at minimum, and
the Color and Range pots at maximum, repeatedly slap the guitar strings over the
pickups, or repeatedly rake the strings while muting them with your fret hand. Use
the guitar tuner to measure the “pitch” of the resulting filter matrix sound (the pitch
should change if you vary the Range pot). With the Range pot at maximum, the
pitch of the filter matrix sound should be 78.125 Hz. That is lower than the bottom E
of the guitar (82.41 Hz) but higher than the E-flat that is below it (77.78 Hz). So
adjust the clock trimmer until you get a pitch that is slightly above that E-flat.

PCB Notes

Points marked 0V on the PCB: All these points are connected together on the PCB.

Trimmer Pots: The PCB provides multiple holes so that different types of trimmer
pot can be used. Bourns ¼ inch pots will all fit (3362P, 3362F, 3362R, 3362U), as
will Bourns ¼ inch precision multi-turn pots (3266W, 3266Y, 3266X, 3266Z).

Control Pots: The PCB provides multiple holes so that different sized board-
mounted pots can be used (16mm Alpha pots or 9mm Alpha pots). The pots will be
under the PCB so take measures to avoid accidental short circuits to the board.

Test Output: The test point on the board (marked “Tst”) bypasses the final output
gain stage. So the output from this point will have the same volume drop as an
original V2 Electric Mistress.

5
Comparing the sound of one build to another (or an original EM)

In an Electric Mistress the possible sweep rates are affected by the value of the Rate
pot and the rate capacitors (in this case C14 and C15). Electrolytic capacitors can
have tolerances of 20%, and the Rate pot can have a tolerance of 10%. This means
that when comparing two builds, there can be a significant difference between the
fastest possible sweep rates that the builds can achieve, and the slowest possible
sweep rates that the builds can achieve. It also means that if you give two builds the
exact same visual “clock” position for the Rate pot, then the sweep rates (and
therefore the flanger sounds) will not necessarily match.

The same applies to the Range pot. On the lower Range pot settings, even small
turns of the Range pot can produce a significant change on the range of BBD delays
covered by the sweep. So you cannot rely on using visual “clock” positions to set the
Range pots either.

Therefore the only way to properly compare sounds from two builds is to set the Rate
pot positions so that the sweep rates match on the two builds (with sweep period
measured by a scope ideally), and to set the Range pot positions so that the delays
match at the bottom (or top) of the sweep. This can be done by ear while the Rate pot
is set to give the slowest sweep, and the Color pot is set to give maximum feedback
(because the pitch of the feedback sound can then be used to estimate the BBD
delay).

Taking the above points into account, I performed a blind test against a SAD1024
version of the circuit. I repeatedly switched between the SAD1024 and MN3207
circuits while playing and the two circuits sounded identical.

6
7
Revision History

Rev2: Fixed supply section: C18: 220uF→33uF, C19, C20: 100uF→0.1uF.

Removed separate RC smoothing for LFO supply.
Rev3: Adjusted sweep at maximum range: R21: 5k6→5k1, R32: 220R→180R
Rev4: Fixed sweep at minimum range: R19: 18k→16k2, R21: 5k1→3k9
Rev5: Slightly increase sweep range to better match original: R32: 180R→200R
Rev6: Added R33 = 2k2 to limit lowest Range pot setting.
Rev7: Removed RT4 (balance trimmer). Added D3 protection diode at supply.
Added R34 to feed an off-board LED that flashes at the sweep rate.
Rev8: Changed C18: 33uF→220uF, C19, C20: 0.1uF→100uF and added R35 in
series with the protection diode D3. These changes allow a cheap MT3608
DC-DC boost board to be used, so a standard 9V supply can be boosted to
give the higher voltage required by the EMV2.
Added C22 close to the supply pins of the CD4050 to reduce some of the RF
ripple it produces on the 9V supply. Not essential but there was space on the
PCB.

The values of R19, R20, R21, R32, R33 affect several things including

• The ratio of upward and downward sweep times. Ideally this is 1 : 1.

• The swept range of BBD delays and how that range varies with Range pot
setting.

There is a trade-off in getting these two things correct while still using commonly
available resistor values. The resistor values in the latest revisions (Revisions 6, 7
and 8) give upward and downward sweep times in the ratio 1.03 : 1, and a range of
swept BBD delays that differs from an original EM sweep by less than one semitone.

Licensing

You may use a purchased EMV2 PCB for DIY or small commercial operations. You
may not use the PCB artwork in this document to sell your own version of the PCB
design or as part of a “kit” or similar commercial product.

8
9
Veroboard (Stripboard) Layout

If you cannot obtain a PCB then you can use this Veroboard layout.

10