Self-made analog delay pedal using BBD (BOSS DM-3 MOD) Self-made analog delay pedal using BBD (BOSS DM-3 MOD)


Self-made analog delay pedal using BBD (BOSS DM-3 MOD)

Self-made Guitar Effect Pedals

My self-made effect pedalsThis is a list page of the effect pedals that I made. You can see articles on how I made them for each board type (universal board, through hole / surface mount printed board) and enclosure size (normal, mini).

Find an article about your favorite effect pedal.


Make my own full analog delay pedal using BBD
Self-made based on the circuit of BOSS DM-3
Design a 2-layer PCB and order from "Fusion PCB"
Component mounting on the board
About the main parts
Operation check
Built into the enclosure
Sound samples
What I felt after making an analog delay

[Additional MOD] Added delay time indicator
[Additional MOD Part 2] Add BBD to get longer delay time of 600ms
[Additional MOD Part 3] MOD that raises the delay sound quality

The second analog delay self-made on the surface mount PCB

Make my own full analog delay pedal using BBD

BOSS Analog DelayAmong the effect pedals for guitar, analog delay using BBD is very popular even now. BOSS continues to sell the cool analog delay model as one of the "WAZA" series which includes three BBD devices to get the longer delay time 800ms.

Regarding the available self-made delay effect pedal, a production article using the easy-to-produce one-chip IC for delay "PT2399" has been published by TONEPAD as "REBOTE Delay". However, I wanted to try making my own delay using a primitive 4096-stage BBD device.

The BOSS DM-2 analog delay uses the discontinued MN3005

MN3005 BBD deviceThe schematic for the BOSS DM-2 analog delay pedal can be found in the repair service manual. However, the DM-2 uses the Matsushita BBD device "MN3005". This device is nolonger manufactured and hard to obtain. The devices on the market are very expensive or suspicious used chips taken from some abandoned gear.

BBD "MN3205" with different specifications is produced.

Currently, the BBD device has been changed to another specification device "MN3205". This device can be purchased, but it is a little expensive. So I decided to use the cheaper Coolaudio "V3205" (the second source product of Matsushita's "MN3205").

You can buy V3205 for around $4. The clock generator V3102 used together is less than $1.

Difference between BBD devices "MN3005" and "MN3205 (V3205)"

Both of these are functionally the same as the BBD with a built-in 4096-stage bucket element, but the new MN3205 (V3205) is designed to operate with a negative power supply. Simply to say, the power line +9V and GND are the opposite of the MN3005. In addition, the maximum power supply voltage is +9V, which is lower than the + 15V of the MN3005. The recommended voltage of MN3205 (V3205) is +7.5V. Therefore, it seems that the power-supply circuit of BOSS DM-2 using MN3005 cannot be used as it is.

Self-made based on the circuit of BOSS DM-3

Tried to make my own based on DM-3 that uses "MN3205"

BOSS's analog delay DM-2 (released in 1982) is considered to be a famous pedal. On the other hand, there is a model called DM-3, which was sold only for a short period of time after the DM-2. The function is the same as DM-2, but the BBD device is changed to "MN3205" and it supports stereo output. So I searched for the DM-3 schematic. A service manual counld be found which contains the schematic using MN3205.

The BOSS effect pedals have an ON / OFF foot switch that is not a mechanical switch but FET electronic switches. I want to make my own delay pedal the True-Bypass specification, so I removed all the FET switches from the schematic, and stereo output as well.

Circuit diagram (DM-3 modify, PDF)

DM-3 has a double noise reduction circuit, which gives very good S / N characteristics!

The DM-3 has a double noise reduction circuit. First, it has pre-emphasis and de-emphasis circuits that use operational amplifiers for input and output. In order to improve the S / N of the BBD delay part, the treble range is raised at the input, and after passing through the delay circuit, the treble range is lowered at the output. This can reduce the high-frequency noise generated in the BBD circuit. Similar technology is used in FM radio broadcasting to improve S / N.
The DM-3 consists of a high shelving filter and a low shelving filter that use operational amplifiers. It raises 1kHz or more by a maximum of 14dB (amplitude 5 times), passes through a delay circuit, and then lowers it by 14dB.

Moreover, about the processing of a delay sound, the compressor and the expander are constructed by using half each of V571 before and after the BBD circuit. Thereby, BBD noise is reduced. This is a circuit similar to the noise reduction of DBX and Dolby used in analog tape recorders. This expander raises the volume only when there is a delay sound, and turns down the volume when there is no delay sound, so BBD noise when there is no sound is almost inaudible. The DM-3 is a delay effect pedal with low noise and very good S / N.

Design a 2-layer PCB and order from "Fusion PCB"

This time, I ordered a printed circuit board that I designed from "Fusion PCB" in Shenzhen, China, which produces 10 PCBs for $25. All board design processes were performed using the freeware "KiCAD". I input the circuit diagram, designed the double-sided circuit board, output the standardized board design data called "Gerber data", and sent it to “Fusion PCB”.

The manufacturing fee is only $5. However, DHL shipping costs $20, so the total is $25. Still, it's cheap enough to make your own board for $25, which used to cost hundreds of dollars a decade ago. It will arrive in about a week.

10 original printed circuit boards that arrived from Fusion PCB
10 original printed circuit boards that arrived from Fusion PCB

About board design

I designed the PCB using the freeware "KiCAD". First of all, I input the schematic with the schematic editor, assign the part shape data called 'footprint' to the used parts, and I can design the trace with the board editor. I wanted to put this analog delay in an MXR size enclosure, so I made it with a size of 60mm x 100mm. The actual area is smaller because the foot switch area is hollowed out. The trace design is fairly easy since the double-sided board can draw trace on the front and back. I connected all the lines except GND, and I filled all the remaining areas with solid ground.

Board design requires some experience (I worked for an electronics design company about 25 years ago). There are many books on CAD and board design methods, so you can refer to them and make your own.

Board design free software "KiCAD"
Board design free software "KiCAD"

KiCad PCB design data

The PCB data created using "KiCad (Windows 64-bit) Version: (6.0.4), release build" is provided. The ZIP file contains the schematic, board design data, and original component data. Please read the terms of use and disclaimer in the attached "readme.pdf" before using this data.

Component mounting on the board

I put the purchased parts on the printed circuit board that I ordered for design. For me as a self-making nerd, soldering is the most enjoyable task. Since there are many parts, it is challenging to make, but the fun parts mounting is completed in about 3 hours.

Analog Delay Completed mounting of components on the board
Completed mounting of components on the board

About the main parts

For BBD devices, I used Coodaudio's second source products "V3205, V3102". The DM-3 uses a noise reduction device (NEC uPC1571C) called a compander. This device was replaced with Coodaudio's second source product "V571".

As for the transistor for small signal, 2SC1685 and 2SC732 are used for the original DM-3. Since these were difficult to obtain, I used the alternatives "2SC945, 2SC1815". The FET "2SK30A" is also difficult to obtain, but I found it by chance and was able to purchase it. If you cannot get it, "2SK2880" can be used instead.

Coolaudio's second source ICs
Coolaudio's second source ICs

Operation check

"Oops, I made a mistake!"

I checked the operation with the board on the desk. I connected a guitar to the input and connected the output to an amplifier, but I heard a oscillating sound and no guitar sound. I connected a small digital oscilloscope that I recently bought for $ 50 and followed the signal from the input. The signal is coming up to the input operational amplifier (NJM4558DD), but there is no signal coming out from the output pin 7.

I looked closely at the schematic and found a mistake that pins 5 and 6 of the op amp were reversed. I reluctantly cut the trace and made the wiring to reconnect the pin numbers in reverse.

Oops, I made a mistake!
I cut the trace and reconnected it to fix the pin number of the operational amplifier that I made a mistake.

A delay sound came out!

The board has three semi-fixed potentiometers. I adjusted them as follows.

CLOCK adjustment: This adjustment determines the maximum delay time. Set the RATE pot in advance so that the delay time is the longest. Turning the CLOCK trimmer all the way to the right maximizes the delay time, but increases the aliasing distortion (the delay sound sounds jerky and muddy). Gradually move it back to the left and set it where you can't hear the aliasing distortion (where you can tolerate it) while listening to the delay sound. By design, the maximum delay time is about 300ms.

BIAS adjustment: This is a distortion adjustment. Input a 200Hz sine wave and monitor the output pin 10 of the compander V571. Gradually raise the input level and stop where either the top or bottom of the sine wave clips. Turn the BIAS trimmer to adjust to the point where the sine wave distortion is the smallest (where the sine wave shape is clean).
If you don't have an oscilloscope, set the effect level pot to MAX, connect it to an amp, listen to the guitar, and turn BIAS to find the place where the delay distortion is minimized.

[Reference] BIAS confirmation with an oscilloscope:
In the lower waveform, the upper (yellow) is the input to the BBD, and the lower (blue) is the output of the V571 (pin 10). It gives a level of input where the output begins to distort. The waveform on the left is distorted at the bottom, but not at the top. This is a condition where the BIAS (center level of the waveform) is too low. The waveform on the right is evenly distorted at the top and bottom. This is the state in which BIAS is appropriate.

I don't have a signal generator, so in order to generate a 200Hz sine wave, I use the Windows free software "WaveGene" to generate the signal. The signal from the audio interface "Steinberg UR22mkII" was used as the input for this delay.

CANCEL adjustment: Shorts the input to GND. CANCEL Monitor the semi-fixed pin 2. Turn the semi-fixed to adjust to minimize clock leakage.
If you don't have an oscilloscope, set the effect level pot to MAX, connect it to an amplifier, listen to the delay sound, and turn CANCEL to find a place where the noise on the delay sound (It sounds jerky) is minimized. If the input is silent, the V571 expander is closed and there is no delay sound itself. You need to add some input.

[Reference] CANCEL check with an oscilloscope:
The upper waveform (yellow) is the input to the BBD, and the lower waveform (blue) is the pin 2 of the CANCEL semi-fixed resistor.

As shown in the circuit diagram above, 4098 bucket elements (capacitors) are arranged in order inside the BBD. The bucket elements are grouped in even and odd numbers, and each has a switch driven by a clock (CP1, CP2) of opposite phase. According to the clock, the electric charge stored in the bucket is transferred in the order of odd-numbered stage -> even-numbered stage -> odd-numbered stage -> even-numbered stage.

In the part that outputs the charge from the last bucket, the even-numbered and odd-numbered clocks are driven by different clocks (opposite phase), so the 4097th and 4098th charges are taken out from different pins (OUT1, OUT2). increase. These are mixed with a CANCEL semi-fixed resistor to obtain the final output of the BBD.

In the waveform on the left side above, the mixing balance of CANCEL adjustment is bad, and the outputs of odd-numbered stages and even-numbered stages are not connected well. In this case, clock noise will be output to the final output. CANCEL adjustment is good on the right side, and the outputs of odd-numbered stages and even-numbered stages are well connected.

When using an oscilloscope, you can adjust CANCEL with the input turned off. In this case, adjust so that the output of pin 2 of the CANCEL semi-fixed resistor is aligned as shown in the waveform below.

With this adjustment, the repeated delay sound can be heard clearly.

Built into the enclosure

Overall structure

I designed the component placement so that the phone jack and foot switch do not interfere with the PCB. The small Switchcraft box-type phone jack (# 112BX) can be used. In order to take a 20mm space between the PCB and the other parts, I taped 2mm thick paper spacers inside the enclosure. Unfortunately there is no space to insert a battery because the PCB length is too long.

I made a 3mm hole and pushed the LED from the inside and fixed it with super glue. I don't use the uncool LED holder. If you want to make it in the same way, please refer to the figure below.

Enclosure processing

I housed this analog delay in a blue-painted enclosure of HAMONND's "1590N1". The enclosure was processed using a hand drill. After positioning, the holes were expanded in the order of 1.5mm, 3mm, and 6mm, and then expanded to the hole diameter of each part using a taper reamer.

My own analog delay enclosure processing

My own analog delay enclosure completed

Housing PCB into the enclosure

The board width is made according to this enclosure, so it fits perfectly. The board is inserted with the component side facing outward so that the board looks beautiful. The input / output jack and DC jack are located near the top of the enclosure so that the PCB can be stored. When I open the back cover, it looks beautiful and looks like a ready-made product, and I am very satisfied!

My own analog delay PCB mounting

My own analog delay beautiful looking

Sound samples

The maximum delay time is 300ms. When the FEEDBACK knob is raised above 2 o'clock, it will start oscillating. If I turn the delay time knob while oscillating, I can create an oscillating sound effect peculiar to an analog delay box. The EFFECT knob works exactly as I imagined and is perfect.

Below is a sample delay sound when the delay time is set to MAX 300ms. You can hear the thick and warm sound that is unique to analog delay and does not interfere with the original sound. The feedback repeat sound is gradually distorted and blurred. It has a nice analog feel.

Sample #1: Short chord phrase, 300ms delay (MP3)

Sample #2: Short melody, 300ms delay (MP3)

My own analog delay completed perfectly!

What I felt after making an analog delay

Music making has changed from a tape recorder to a personal computer Pro Tools. I think that digital technology is used for guitar pedals and amplifiers without being aware of it. However, analog delay is probably an effect pedal that goes against the times.

In the first place, electric guitars are still connected with old-fashioned thick phone plugs, and they are instruments that make a lot of noise from primitive pickups. Therefore, it may not match digital devices such as multi-effects.

In the sound of this analog delay, the high range is cut, and the feedback repeat sound is gradually distorted. This is it! I feel that this is the sound of an electric guitar that started in the 1950s, shortly before I was born. So is spring reverb.

I'm almost 60 years old and have been playing guitar for 45 years. I'm not in a band anymore, but I still find it most fun to play along with the songs of my favorite artists.

For those who want to make their own effect pedal from now on, I made a page explaining how to select a tool. Please refer to it.

[Additional MOD] Added delay time indicator

I added an LED indicator to show the delay repeat interval. This analog delay doesn't have a TAP function, and the delay time is as short as 300ms in the first place, so it may not be useful as a repeat indicator. This is a self-satisfying mod.

The method is to divide the clock for driving the BBD (V3205) generated by the clock generator (V3012). Since V3205 is a 4096-stage BBD, the clock is divided by 2048 (a half of the clock) to create a clock with a repeat interval of the delay sound.

Since 2048 is 2 to the 11th power, the clock of V3205 was divided using the binary counter "TC4040" with 12 stages. The schematic is below. (Click here to see the entire schematic.)

Since the TC4040 is a CMOS-IC, the current that drives the LED directly cannot be obtained, so it is driven via a transistor (2SC1815). Furthermore, in order to prevent noise at the edge of the blinking clock of the LED, a time constant is added by CR (R2 and C1) so that the LED lights are turned on gently.

I made it with a universal board, put it in a shield box made of aluminum foil tape and vinyl sheet, and stored it in the battery space.

It flashes as below. It seems to be a guide for the tempo (about quarter notes).

(This is a GIF animation created as an image. It is not an actual video.)

[Additional MOD Part 2] Add BBD to get longer delay time of 600ms

I added another BBD (V3205) and modified it to extend the delay time. As a method, the clock generator (V3101) can drive two BBDs even if they are connected in parallel, so simply connect another BBD in series to double the delay time.

BBD output requires a BBD odd-numbered and even-numbered mixing pot. The circuit diagram of the BBD part is as follows.

BBD addition modification for DM-3

The output of BBD1 is output at a level about +3.5dB larger than the input. A trimmer (RT101) for this gain adjustment is added because it will clip if it is input to BBD2 as it is. However, gain adjustment may be difficult without an oscilloscope. If you don't have any, you can adjust the semi-fixed BBD2 GAIN to about -3.5dB (x0.64 times). It is OK to replace RT101 with 56kΩ and 100kΩ resistors.

And the input to BBD2 is given the same bias as BBD1 via R125 by cutting DC with C102. Models that use multiple BBDs, such as MAXON, allow the BBD bias to be adjusted separately, but I think there is no problem with using one bias adjustment.

Yellow:BBD1 input (pin.13) / Blue: BBD1 output (pin.2 of RT2 trimmer)
The gain from input to output is about +3.5dB.

The BBD to be added is taped directly above the original BBD. The pins that connect in parallel to the original BBD are soldered directly, the other pins are flipped up and the wires are soldered to the pins. Additional parts such as semi-fixed capacitors, resistors, etc. are made on a universal board cut into small pieces and attached to the empty part of the printed circuit board with double-sided tape.

Sound sample with longer delay time

These are sound samples with a delay time set to about 560ms. With the original 300ms, the time is too short and it often interferes with the rhythm, but if you can set it up to nearly 600ms, the usage will expand.

Also, if two BBDs are used, the number of stages (the number of BBD buckets) will be doubled. Therefore, if you want to get the same delay time, you can double the clock frequency feeding to BBD. As a result, the folding noise (digital distortion caused by sampling and discretizing) can be reduced and the delay sound becomes clearer.

Sample #1: Short chord, 500ms delay(MP3)

Sample #2: Solo phrase, 500ms delay(MP3)

[Additional MOD Part 3] MOD that raises the delay sound quality

Sampling frequency that determines the sound quality of BBD

The BBD, like the AD converter in a digital circuit, samples an analog signal and converts it into an electric charge, and uses the bucket element of the BBD to create a delay. The AD converter converts it to a digital value, while the BBD converts it to an electric charge (voltage value), but both are sampling in the same way.

Technically speaking, when sampling is performed, it is converted from a "continuous system" to a "discrete system", but when it is changed to a "discrete system", the maximum frequency that can be handled becomes "1/2 of the sampling frequency". It is a theory called "Nyquist frequency".

For those who "do not understand at all", when using BBD to sample input sound at "8kHz" (8000 times per second), understand that the maximum frequency that can be input to BBD is "4kHz", which is 1/2 of 8kHz. If a signal higher than this frequency is input to the BBD, annoying noise called "foldback distortion" will occur.

This sampling frequency (clock given to BBD) is changed by the RATE knob of DM-3. BBD is fixed at 4096 stages, so the delay time can be changed by changing the sampling frequency.

DM-3 BBD sampling frequency

The original DM-3 circuit uses a 4096-stage BBD chip to create a delay of up to approximately 350ms. In order to make a delay time of 350ms with 4096 stages, the time to drive one stage is:

(0.35 / 4096) seconds

This corresponds to the sampling cycle, so to convert it to frequency, the cycle time should be reciprocal:

1 / (0.35 / 4096)

The calculation result is 11702 Hz (= 11.702 kHz). So if you sample at 11.7kHz with a 4096-stage BBD, you get a delay of 350ms.

Low-pass filter that reduces folding distortion

The maximum frequency that can be input to the BBD is 5.8kHz, which is 1/2 of 11.7kHz, due to the above "Nyquist frequency". If it exceeds this, folding distortion will occur and it will not be possible to use it as a delay circuit.

Therefore, the DM-3's original circuit has a 2.5kHz low-pass filter that uses a transistor in front of the BBD. This filter is called a "third-order Shallen-key filter" and has an attenuation factor of -18db / oct above 2.5kHz (when the frequency doubles, the volume drops to 1/8).

You might wonder, "Isn't the 2.5kHz lowpass filter too low in frequency, while the Nyquist frequency can tolerate up to 5.8kHz?" This filter cannot limit the frequency "steeply", but has an attenuation factor of -18db / oct (gradually decreasing). Therefore, in order to limit 5.8kHz or more (lower the level sufficiently), there is a margin of frequency.

A 5.8kHz signal that causes fold distortion drops to -20dB (1/10) after passing through this filter. I think you can still hear the fold distortion, but it seems to be a practical level for input to BBD.

After passing through the BBD, two more low-pass filters (3rd order 1.9kHz and 2nd order 4.4kHz) are included to reduce the clock noise mixed in the BBD (see the waveform of CANCEL adjustment mentioned above). .. After passing through all the filters, it seems that it will eventually be limited to 1.5kHz or less. Even the sound of AM radio extends to 7.5kHz, so you can understand how bad the frequency response is.

For this reason, the delay sound of the DM-3 is a round, fluffy sound, as if the tone knob was lowered. (This may be the warm sound peculiar to analog delay.)

MOD that raises the frequency of the low-pass filter

The only way to get high frequencies with BBD delayed sound is to increase the sampling frequency. Since the number of BBD stages is fixed, increasing the sampling frequency results in a shorter delay time. There is a trade-off between frequency characteristics and delay time.

The modification to add another BBD to the circuit of DM-3 was done above. As a result, the maximum delay time has increased to about 700ms. Therefore, I limited the delay time to 500ms so that it can be used with the sampling frequency increased, and modified it to increase the frequency of the low-pass filter before and after BBD within the range where folding distortion does not occur.

The sampling frequency for creating a delay of 500ms with two 4096-stage BBDs, a total of 8192 stages:

1 / (0.5 / 8192) = 16384 Hz (16.4 kHz)

Since this frequency is about 1.4 times the frequency of the original MD-3 (11.7kHz), the frequency of the low-pass filter can also be increased to 1.4 times.

Therefore, I modified the cutoff frequency of the low-pass filter before and after BBD to be 1.4 times.

NOTE: The BBD chip divides the internal bucket element into odd-numbered stages and even-numbered stages and drives them with clocks of opposite phases. Therefore, the clock input to the BBD (V3205 pins 2 and 9) is 1/2 of the sampling frequency.

Change the resistance constant of the circuit

The low-pass filter is a "Shallen-key filter". To increase this cutoff frequency by 1.4 times, change the 10 kΩ resistors in series to about 7.1 kΩ, which is 1 / 1.4 times (reciprocal).

Since the printed circuit board I made has a through-hole structure, it is troublesome to remove the solder and replace the resistors. Therefore, I soldered a 27kΩ resistor in parallel to the original 10kΩ resistor and changed it to 7.3kΩ.

Frequency characteristics of the delay line after MOD

I measured the frequency characteristics of the delay sound that passes through the BBD. It is a characteristic from the input jack to the output jack. It was measured by removing R47 (mixing resistance of direct sound) and preventing the direct sound from being mixed.

The band up to 2kHz is output almost as it is, but it is attenuated from 2kHz and dropped to -28dB (1/25 times) at 4kHz. With the original constant before MOD, I think that 1.4kHz or more is attenuated, so I think that the muffled delay sound has been improved.

The second analog delay self-made on the surface mount PCB

I made a second self-made unit using surface mount components, including modifications. The adjustment pot and phone jack are also mounted on the PCB, imitating the structure of the manufacturer's product. Please see the page below.