JCS "Bakery of Noise" BN-1

4-Bit Linear PCM Synthesizer

Designed and built at JCS, the BN-1 is a rudimentary linear PCM synthesizer, with 4-bit sample width and with the ability to be controlled by an external frequency source. It contains storage for 512 samples, half of which are stored in programmable SRAM, and the other half of which are stored in ROM. It was designed with versatility as well as serviceability in mind. All components are through-hole, and all ICs are socketed for easy replacement. The design most closely resembles digital circuitry of the late 70s—early 80s.

In terms of ICs, the majority are standard TTL 74xx or 74LSxx series units, or equivalent. As for equivalence, in the schematics, these are also denoted with another part number starting with “К155” (74xx) or "К555" (74LSxx), and denoting the equivalent type of chip made in the former Soviet Union. In fact, one of the aims of this project was to make partial use of a large batch of Soviet ICs available in the JCS parts supply, most being clones of various digital logic chip types. There is one Soviet chip used that is there is not a pin-compatible “Western” equivalent for, and that is the КР140УД1Б (transposed: KR140UD1B), which is in fact a clone of the µA702 op-amp, the first type ever offered, but with a different pinout. Even so, all of the components are, as of early 2016, commonly available, and in many cases are able to be replaced with even more common equivalents with minor modification.

Features

  • 1/4" jacks for input & output
  • Input and output level potentiometers
  • 3 clock selections, selected by a 3-position toggle switch:
    • Internal Oscillator
    • External Input
    • Frequency Doubled External Input
  • Internal clock frequency potentiometer
  • 8 "Address Entry" switches for selecting the current programming address (in Program mode) or the address that the counter starts at (in Run mode)
  • 4 "Data Entry" switches for entering data to be written to RAM in Program mode
  • Count reset button
  • RAM Write button
  • ROM / RAM memory selection switch
  • Program / Run mode selection switch
  • Loop Enable counting switch
  • Power switch
  • Input voltage: 9 to 12 V DC, tip positive

Design and Construction

Block Diagram and Schematic Preparation

As is often the case with electronic devices, this one started out with a few basic ideas, with the overall idea being to design and build some type of linear PCM synthesizer. Mainly, these were:

  • A 4-bit sample width, for ease of programming, sufficient low-fidelity-ness, and to make use of some of the 4-bit SRAM and ROM chips in the parts supply
  • An 8-bit address bus, also for ease of programming, and since the memory chips chosen have 8-bit address inputs, allowing for 256 memory locations
  • Ability to be controlled by an external frequency source, as well as an internal oscillator
  • Ability for the external frequency to be multiplied (i.e. doubled, quadrupled), since one cycle of the external source will advance the sample counter only once, and a chain of samples can be very long

From these ideas, a block diagram was drawn on paper. As this block diagram progressed, so did a schematic that was started in the light edition of CadSoft EAGLE 7.2.0. Changes were made to the block diagram based on what was done in the schematic, as well as what was eventually done in the prototype. After many changes, a final block diagram was drawn on paper, and is shown below:

The schematic was prepared in four sheets using mostly standard parts from the EAGLE library. A few components had to be created, such as the 2112 SRAM cand КР140УД1Б op-amp chips, which were not part of the default library. As well, the solder pad size of some of the default components (i.e. resistors) was increased for ease of soldering as well as durability (in the event something has to be desoldered). The finished sheets are shown below:

Alongside the preparation of the schematics, a prototype was assembled on breadboards. This allowed for testing the circuits as they were being designed, with revisions being made if the circuit did not work properly in practice. Certainly a lot more fun than rewriting some code and reprogramming a microcontroller! Here's the prototype, minus a few wires in the lower left, which were removed before the decision was made to photograph it.

Note: I have decided to omit the detailed circuit description from this article. However, I have prepared a zip file containing this detailed description, as well as some other useful information for those who are interested in understanding the circuit and possibly building it themselves. This can be found in the "Dowloads" section at the bottom of the page.

Design and Fabrication of the Main Printed Circuit Board

The BN-1 uses a single main PCB that carries most of its electronics. It is single-sided, measures almost exactly 4" x 6", and was designed in the light edition of EAGLE. Since the light edition restricts the maximum board size to 4" x 3.2", it had to be designed in two sections of about 4" x 3" each, which were later edited together in Photoshop. This resulted in considerable wasted space and an excessive number of jumper wires, particularly around the middle of the board, since the traces had to be arranged in such a way that they would line up on both sections. Another cause of wasted space was that I decided to use a relatively wide default trace width of 0.024" to reduce the chances of etching defects. Thus, the board design, while functional, is not ideal. Though, it was my second attempt at designing a PCB, and for that, I think its flaws can be partially excused. Nonetheless, I do recommend to anyone interested in building this instrument to try designing their own PCB based on the schematics, instead of using my design. I will only distribute my design by special request.

Once the board was designed, it was fabricated using the toner transfer method with Pulsar's "Toner Transfer Paper". The design was printed onto a sheet of this, cut out, and applied to a cleaned 4" x 6", 1-ounce, copper-clad fiberglass (FR4) board of thickness 1/32" using a small pouch laminator. Some "GreenTRF" foil was also applied as a coating on top of the toner using the laminator, in order to make the traces as difficult for the etchant to penetrate as possible. It was etched in ferric chloride, and with the copper traces tinned using MG Chemicals' "Liquid Tin," for greater solderability and resistance to oxidization. Quite a few holes had to be drilled as well (using a somewhat flimsy Dremel drill press), as can be seen on the finished, but unpopulated board:

Components were mounted and soldered, with the finished board looking like this:

Enclosure & Wiring

For the enclosure, a Hammond Manufacturing 1411UU utility case was used, with dimensions 10.00" x 6.00" x 3.50". A design for the component layouts on the top and side were produced in Photoshop, printed onto standard letter paper, and glued to the enclosure with Sulky KK2000 temporary spray adhesive. Then the component holes were drilled with a handheld drill, the visible surfaces sanded and cleaned, and about four coats of silver "hammered metal" paint applied after a single coat of grey primer (which may not have been necessary). The sanding included rounding the sharp edges of the enclosure for personal safety reasons.

Lettering was applied to the top of the enclosure using Pulsar's "DecalPRO" system. It is truly a miracle that it turned out as well as it did, because there are so many opportunities for failure in the use of this system. Here are the general steps: First, the design, which is a binary image with large black borders around it, is printed as densely as possible using a laser printer onto the "Toner Transfer Paper", which is then cut to size, and dried using a hot air gun. The direction that the paper curls when drying is memorized. Then, a sheet of "Toner Reactive Foil" is chosen that will give the black toner a coating of color. (In this case, I chose iridescent gold.) The paper is laid toner-side-up on a fiberglass board with the "TRF" laid face-down on it, and with a clear mylar sheet on top of all. It is fed through a small pouch laminator, which bonds the TRF to the toner. Once the TRF is removed, any excess left behind is taken off with blue painter's tape, and both the toner transfer paper and a sheet of mylar cut to be the same size as the paper are cleaned with isopropyl alcohol. Then, this is run through the laminator on the fiberglass board with the mylar over all, and afterwards dipped in a bucket of water in the orientation such that the paper curls as it enters the water. After about a minute, the toner transfer paper separates from the toner image, and the image is left on the mylar sheet, partially by the melting of the laminator, and partially by electrostatic attraction. Then, the sheet is carefully removed and dried, then sprayed with an even and conservative coating of KK2000 adhesive. The borders are cut off, and then, finally, the design is applied with extreme care to the intended surface. There is only one chance once you touch the surface with the decal, and luckily, in my case, it worked out perfectly.

On the bottom of the enclosure, four rubber feet are attached with bolts, and the screws securing the board's posts are visible. The side has two components mounted: the fuse holder, and the LM323K 5-volt regulator. The area under the regulator was masked during painting so that the heat wouldn't be transferred through a layer of paint, and so that silicone heat transfer compound could be applied between bare metal surfaces. The enclosure itself partially acts as the regulator's heatsink, but I also put a heatsink on the inside (with heat transfer compound as well) to try to keep the enclosure from warming up too much, at least in such a small area.

Wiring was done in such a way that the unit could be opened with the top panel moved to the left side, for ease of service. The wires can also be easily removed from the PCB since each group of wires has its own removable connectors. These connectors are made with male and female pin headers; the male headers are on the main PCB, while the female headers are soldered to some very small PCB sections that make gripping the connectors easier, and to which the wires are also soldered. This arrangement is very inexpensive, since single-row 40-pin headers can be purchased very cheap and cut to whatever width is necessary. Of course, the male headers can be split without any loss of pins, while the female headers will lose one pin every time they are split.

Videos

Two videos have been made of the BN-1: one of the prototype, and the other of the finished product. These videos include some basic demos of the sounds of the unit, in unprocessed form. They can be seen below:

Downloads

  • Various Info - This includes the schematics, as well as a detailed circuit description and the top enclosure graphics.
If you notice any errors or have additional information that you would like to add, please contact me!


Last Updated: 1/4/2016