# Newtech Model 6 Music Board — 6-bit DAC, amplifier and speaker Sources: Newtech Music Board.pdf and Computer Music part 1 (0976).pdf Newtech Computer Systems, Inc. (131 Joralemon Street, Brooklyn, N.Y.), *Model 6 Music Board Users Manual*, © 1977, Rev. A, June 1977 — 22 scanned pages. The Model 6 is an S-100 board with one write-only output port: a 6-bit latch, an R/2R digital-to-analog converter, an LM380 audio amplifier, a 2" speaker, a volume control and an RCA phono jack. It was sold assembled and tested for $59.95. The manual gives two programs, MICROSCORE (BASIC) and MICROPLAY (8080), and two test routines. The second source is the article the manual names as a source of more software: Hal Chamberlin, "Computer Bits — Computer Music", *Popular Electronics*, September 1976, pp. 116–119 (4 scanned pages). It is part 1 of 2; the October part is not in the scan. The manual has no printed page numbers, so the page numbers here are **PDF pages**. The article is cited by its printed pages. This file omits the marketing sheet, the warranty and the speaker-safety advice. --- ## 1. Quick reference | Item | Value | |------|-------| | Ports | **One output port.** No input port, no status, no interrupt. | | Default address | **`24H`** (044 octal). Newtech's software uses it. | | Ports answered | **Four adjacent ports**: A1 and A0 are not decoded. Default `24H`–`27H`. | | Decode | SOUT high · pWR low · A7–A4 as jumpered · **A3 = 0** · **A2 = 1** | | Address choices | `x4H`–`x7H` for x = 0…F (16 blocks, A7–A4 by jumper) | | Data latched | **DO7–DO2** (6 bits) in a 74LS174. **DO1 and DO0 are ignored.** | | DAC | 6-bit R/2R ladder, DO7 = MSB. Output from 0 V to almost +5 V, unsigned. | | Step | 5 V / 64 ≈ 78 mV at the ladder (derived) | | Reset | **None.** The latch clear pin is tied to +5 V through R13. | | Audio | AC-coupled to an LM380 (gain about 50), then to the speaker and to J1 | | Bus power | +18 V (pin 2) → +12 V; +8 V (pin 51) → +5 V; ground pins 50 and 100 | The board holds the last value written. Sound comes only from the program changing the value at an audio rate; the board has no oscillator and no timer. --- ## 2. Address decode (pp. 4–6) IC1 (74LS30, 8-input NAND) goes low only when all eight inputs are high. The latch clock (IC3 pin 9) is the IC1 output (pin 8). | IC1 pin | Signal | S-100 pin | Path | |---------|--------|-----------|------| | 1 | A7 | 83 | direct or through inverter 2A, by jumper | | 12 | A6 | 82 | direct or through inverter 2B, by jumper | | 11 | A5 | 29 | direct or through inverter 2C, by jumper | | 2 | A4 | 30 | direct or through inverter 2D, by jumper | | 3 | A3 | 31 | through inverter 2F — must be **0** | | 6 | A2 | 81 | direct — must be **1** | | 4 | pWR (active low) | 77 | through inverter 2E | | 5 | SOUT | 45 | direct | A1 and A0 do not go to the board. ``` A7 A6 A5 A4 A3 A2 A1 A0 0 0 1 0 0 1 X X = 24H (the address as supplied) \--selectable--/ \-fixed-/ \don't care/ ``` ### Table 1 — output port address selection (p. 5) The jumpers go in a 16-pin position: J1 is pins 1–16, J2 is 2–15, J3 is 3–14, J4 is 4–13, J5 is 5–12, J6 is 6–11, J7 is 7–10, J8 is 8–9. One jumper of each pair is fitted: | Bit | Jumper for 1 | Jumper for 0 | |-----|--------------|--------------| | A7 | J1 | J2 | | A6 | J3 | J4 | | A5 | J5 | J6 | | A4 | J7 | J8 | | High digit | Ports selected (hex) | Jumpers fitted | |------------|----------------------|----------------| | 0 | 04, 05, 06, 07 | J2 J4 J6 J8 | | 1 | 14, 15, 16, 17 | J2 J4 J6 J7 | | **2** | **24, 25, 26, 27** (standard) | **J2 J4 J5 J8** | | 3 | 34, 35, 36, 37 | J2 J4 J5 J7 | | 4 | 44, 45, 46, 47 | J2 J3 J6 J8 | | 5 | 54, 55, 56, 57 | J2 J3 J6 J7 | | 6 | 64, 65, 66, 67 | J2 J3 J5 J8 | | 7 | 74, 75, 76, 77 | J2 J3 J5 J7 | | 8 | 84, 85, 86, 87 | J1 J4 J6 J8 | | 9 | 94, 95, 96, 97 | J1 J4 J6 J7 | | A | A4, A5, A6, A7 | J1 J4 J5 J8 | | B | B4, B5, B6, B7 | J1 J4 J5 J7 | | C | C4, C5, C6, C7 | J1 J3 J6 J8 | | D | D4, D5, D6, D7 | J1 J3 J6 J7 | | E | E4, E5, E6, E7 | J1 J3 J5 J8 | | F | F4, F5, F6, F7 | J1 J3 J5 J7 | Fitting both jumpers of a pair shorts an inverter output to its input and can destroy IC2. --- ## 3. Latch and DAC (pp. 6, 8) | Data bit | S-100 pin | IC3 (74LS174) in → out | IC4 (CD4050) buffer | Series resistor | |----------|-----------|------------------------|---------------------|-----------------| | DO7 (MSB) | 90 | 14 → 15 | 4F | R1 | | DO6 | 40 | 6 → 7 | 4A | R2 | | DO5 | 39 | 4 → 5 | 4E | R3 | | DO4 | 38 | 3 → 2 | 4D | R4 | | DO3 | 89 | 13 → 12 | 4B | R5 | | DO2 (LSB) | 88 | 11 → 10 | 4C | R6 | - R1–R6 are 20 kΩ 1% (the "2R" legs). R7–R12 are 10 kΩ 1% (the "R" chain, R12 to ground). - The CMOS buffers swing very close to 0 V and +5 V, so the ladder output goes from 0 to 5 V. - The ladder output is taken at the R1 end and goes to C7. - IC3 pin 1 (clear) goes to +5 V through R13 (4.7 kΩ). No bus signal clears the latch. --- ## 4. Audio path and power (pp. 7–9) | Stage | Parts | Value | |-------|-------|-------| | Coupling and divider | C7 (0.1 µF), R14 (2 MΩ), R15 (25 kΩ volume pot) | 5 V × 25K / (2M + 25K) ≈ **0.062 V** p-p at IC5 | | Amplifier | IC5, LM380 | gain about **50** → about **3.08 V** p-p | | Maximum square wave | | 3.08 / 2 ≈ 1.54 V RMS | | Maximum power into 8 Ω | | (1.54)² / 8 ≈ **0.3 W**; peak current 1.54 / 8 = 0.19 A | | Output filter | R16 (1 Ω), C10 (0.1 µF) | for amplifier stability | | Output coupling | C6 (220 µF) | | | On-board speaker | SPKR, 8 Ω 0.2 W, through **JPR1** | remove or cut JPR1 to use an external speaker | | Jack | J1 (RCA phono), through **JPR2** | external 8 Ω speaker, or the auxiliary input of an audio system | Power: IC6 (7812 or 340T-12) makes +12 V for IC5 from +18 V (pin 2). IC7 (78L05) makes +5 V for the logic and the DAC from +8 V (pin 51). The manual warns that a program can make frequencies below and above the audible range, which a speaker system still receives. --- ## 5. Parts list (p. 19; p. 20 is the same page again) | Ref | Part | |-----|------| | IC1 | 74LS30 | | IC2 | 74LS04 | | IC3 | 74LS174 | | IC4 | CD4050 | | IC5 | LM380 | | IC6 | 7812 or 340T-12 | | IC7 | 78L05 | | R1–R6 | 20 kΩ 1% | | R7–R12 | 10 kΩ 1% | | R13 | 4.7 kΩ 10% | | R14 | 2 MΩ 10% | | R15 | potentiometer, 25 kΩ | | R16 | 1 Ω 10% ½ W | | C1, C2, C3 | 1 µF 35 V tantalum | | C4, C5, C7, C9, C10 | 0.1 µF 12 V disc | | C6 | 220 µF 25 V | | C8 | 4.7 µF 10 V tantalum | | SPKR | 8 Ω 0.2 W | | J1 | phono jack | --- ## 6. Shipped software Both programs assume an **8080 at 2 MHz with no wait states**. ### 6.1 Test routines (p. 18) Ramp. The manual calls it a triangular ramp of 312.5 Hz: 25 clock cycles a step, 256 steps. ``` 0000 3C START INR A 0001 D3 24 OUT 24H 0003 C3 00 00 JMP START ``` Maximum-amplitude square wave, the reference for setting the volume controls. The manual gives 1005 Hz. ``` 0000 97 START SUB A 0001 06 40 LOOP1 MVI B,64 0003 05 DCR B 0004 C2 03 00 JNZ $-4 0007 2F CMA 0008 D3 24 OUT 24H 000A C3 01 00 JMP LOOP1 ``` ### 6.2 MICROPLAY Rev. A (pp. 13, 16–17) Origin `0000H`. The score starts at `SCORE` = `0100H`. The port is `MODL6` = `24H`. **Score format — 3 bytes a note:** | Byte | Content | |------|---------| | 0 | pitch constant (a delay count). **0 = end of score.** | | 1 | duration, low byte | | 2 | duration, high byte **plus 1** | The main loop copies the three bytes into the operands of four instructions in `PLAY` (`XFER1`…`XFER4`). The program changes its own code, so it must run in RAM. At the end of the score it loops at `HERE` (`JZ HERE`). ``` 0000 31 7F 00 BEGIN LXI SP,STACK 0003 21 00 01 INIT LXI H,SCORE 0006 22 6D 00 SHLD PLACE 0009 2A 6D 00 NEXT LHLD PLACE 000C 3E 00 MVI A,0 000E BE CMP M 000F CA 0F 00 HERE JZ HERE ;end of score: loop here 0012 7E MOV A,M ;pitch 0013 32 41 00 STA XFER2+1 0016 32 60 00 STA XFER4+1 0019 23 INX H 001A 7E MOV A,M ;duration LSB 001B 32 3A 00 STA XFER1+1 001E 32 5D 00 STA XFER3+1 0021 23 INX H 0022 7E MOV A,M ;duration MSB 0023 32 3B 00 STA XFER1+2 0026 32 5E 00 STA XFER3+2 0029 23 INX H 002A 22 6D 00 SHLD PLACE 002D CD 33 00 CALL PLAY 0030 C3 09 00 JMP NEXT 0033 21 65 00 PLAY LXI H,TBL1 ;envelope pointer 0036 1E 08 MVI E,8 ;segment count 0038 7E MOV A,M ;starting amplitude 0039 01 00 00 XFER1 LXI B,LNGTH ;duration count (patched) 003C 40 LOOP2 MOV B,B ;waste time 003D C3 40 00 JMP XFER2 0040 16 00 XFER2 MVI D,PITCH ;pitch constant (patched) 0042 D3 24 OUT MODL6 ;output half wave 0044 34 INR M ;waste much time 0045 35 DCR M 0046 34 INR M 0047 35 DCR M 0048 34 INR M 0049 35 DCR M 004A 15 LOOP3 DCR D ;delay according to pitch 004B C2 4A 00 JNZ LOOP3 004E AE XRA M ;"COMPLEMENT A." — see Quirks 004F 0D DCR C ;count down duration 0050 C2 3C 00 JNZ LOOP2 0053 05 DCR B 0054 C2 40 00 JNZ XFER2 0057 D3 24 OUT MODL6 0059 23 INX H ;next segment 005A 1D DCR E 005B C8 RZ ;all segments done 005C 01 00 00 XFER3 LXI B,LNGTH ;(patched) 005F 16 00 XFER4 MVI D,PITCH ;(patched) 0061 7E MOV A,M ;new amplitude 0062 C3 4A 00 JMP LOOP3 0065 95 30 D0 FE FF FF 30 85 TBL1 ;envelope, 8 segments 006D 00 00 PLACE DW 0 ;score pointer SCORE EQU 0100H MODL6 EQU 24H STACK EQU $+10H ;= 007FH ``` - A note is **8 envelope segments**. Each segment plays the duration count in half waves. - The six `INR M` / `DCR M` write to the envelope table and leave it unchanged. - The envelope comment (lines 1140–1147): maximum amplitude is output when the accumulator is complemented from `0FFH` to `000H` and back; minimum when it is complemented from `080H` to `07FH`. The 8 values in `TBL1` set the envelope for each note. Timing of one half wave on the usual path (derived from the listing, standard 8080 cycle counts): `MVI D` 7 + `OUT` 10 + six `INR/DCR M` 60 + N × (`DCR D` 5 + `JNZ` 10) + `XRA M` 7 + `DCR C` 5 + `JNZ` 10 + `MOV B,B` 5 + `JMP` 10 = **114 + 15N** clock cycles = **57.0 + 7.5N µs**, where N is the pitch constant. ### 6.3 MICROSCORE Rev. A (pp. 11–12, 14–15) North Star BASIC (Rev. 6). It reads note strings from `DATA` statements and writes the MICROPLAY score into memory with `FILL`, starting at `U` = 256 (`100H`). The sample score is "The Entertainer" by Scott Joplin: 120 notes and the end mark (the printed run counts 1 to 120, and the `DATA` lines hold 120 note strings before the `X`). **Note string — 4 or 5 characters:** | Character | Values | Meaning | |-----------|--------|---------| | 1 | `A` `B` `C` `D` `E` `F` `G` | pitch in the octave (N = 1, 3, 4, 6, 8, 9, 11) | | 2 | `#`, `!`, space | sharp (N+1), flat (N−1), natural | | 3 | `1` `2` `3` | octave: `1` starts at A = 220 Hz, `2` at A = 440 Hz, `3` at A = 880 Hz | | 4 | `S` `E` `Q` `H` `W` | sixteenth, eighth, quarter, half, whole (T = 16, 8, 4, 2, 1) | | 5 (optional) | `.` | dotted: half as long again (T = 2T/3) | The one-character string `X` ends the score (it writes a zero pitch byte). An illegal character stops the run with `ERROR IN NOTE #n`, `DATA STRING`, `CHARACTER #c`. **The calculation (lines 160–170, 400–420, 560–680):** ``` K1 = 2^(1/12) K6 = 1.2 tempo; speed is proportional to K6 P = M, M+12 or M+24 for octave 1, 2 or 3 (M = N, N-1 or N+1) F1 = 220 * K1^(P-1) frequency, Hz T1 = 10^6 / (2*F1) half period, microseconds K3 = (T1 - 56.5) / 7.5 pitch constant K4 = F1 / (K6*T) half waves in one envelope segment D3 = INT(K4) D4 = 2*D3 - 2*INT(D3/2) "make duration even" D5 = INT(D4/256) D6 = D5 + 1 MSB as stored D7 = D4 - D5*256 LSB score byte 0 = INT(K3 + .5) score byte 1 = D7 score byte 2 = D6 ``` An octave runs from A up to G#, so `C 1` is above `A 1`. Derived from these formulas: a note lasts 8 × K4 half waves = 4 / (K6 × T) seconds, whatever its pitch. At K6 = 1.2 a sixteenth is about 0.21 s and a whole note about 3.3 s. The manual says each note takes roughly 8 to 10 bytes in the BASIC program and 3 bytes in the score. --- ## 7. The Popular Electronics article (September 1976, pp. 116–119) The manual (p. 3) says the programs in Hal Chamberlin's "Computer Bits" columns of September and October 1976 "can easily be implemented on the Model 6". This scan is the September column. The frequency table and the touch-tone tables that the manual mentions for the October column are not in it. The listings are in split octal, at `000:100` and `000:000`, with the port address left as `XXX`. ### 7.1 Fig. 1 — tone subroutine (p. 116) Enter with the frequency parameter in C and the duration parameter in D and E. The speaker goes on any bit of the output port. ``` 000:100 076 000 TONE MVI A,0 (7) 000:102 323 XXX OUT port (10) printed "232" — see Quirks 000:104 101 MOV B,C (5) 000:105 005 DELAY1 DCR B (5) 000:106 302 105 000 JNZ DELAY1 (10) 000:111 033 DCX D (5) 000:112 172 MOV A,D (5) 000:113 263 ORA E (4) 000:114 312 136 000 JZ RETURN (10) 000:117 076 377 MVI A,377Q (7) 000:121 323 XXX OUT port (10) 000:123 101 MOV B,C (5) 000:124 005 DELAY2 DCR B (5) 000:125 302 124 000 JNZ DELAY2 (10) 000:130 033 DCX D (5) 000:131 172 MOV A,D (5) 000:132 263 ORA E (4) 000:133 302 100 000 JNZ TONE (10) 000:136 311 RETURN RET ``` | Fact | Value (p. 117–118) | |------|-------| | Overhead a half cycle | 46 clock cycles | | Wait loop | 15 clock cycles × N | | Half cycle at 0.5 µs a clock cycle | **23 + 7.5N µs** | | N | 1…255; 0 counts as 256 | | Lowest frequency | 500,000 / (23 + 7.5 × 256) = **257.33 Hz** | | Duration parameter | M = 2TF half cycles (T seconds, F Hz), 16 bits in D and E | | Step between notes | 7.5 µs | ### 7.2 Fig. 2 — table of musical notes (p. 117) | Note | Hz | Period µs | Note | Hz | Period µs | |------|--------|--------|------|--------|--------| | C (middle) | 261.62 | 3822.3 | C | 523.25 | 1911.2 | | C# | 277.18 | 3607.8 | C# | 554.37 | 1803.9 | | D | 293.66 | 3405.3 | D | 587.33 | 1702.7 | | D# | 311.13 | 3214.2 | D# | 622.25 | 1607.1 | | E | 329.63 | 3033.8 | E | 659.26 | 1516.9 | | F | 349.23 | 2863.5 | F | 698.46 | 1431.8 | | F# | 369.99 | 2702.8 | F# | 739.99 | 1351.4 | | G | 391.99 | 2551.1 | G | 783.99 | 1275.6 | | G# | 415.30 | 2407.9 | G# | 830.61 | 1204.0 | | A | 440.00 | 2272.8 | A | 880.00 | 1136.4 | | A# | 466.16 | 2145.2 | A# | 932.33 | 1072.6 | | B | 493.88 | 2024.8 | B | 987.77 | 1012.4 | | | | | C | 1046.5 | 955.58 | ### 7.3 Fig. 4 — the article's 8-bit DAC (pp. 118–119) This is a circuit to build on any output port. It is **not** the Model 6 circuit. Two CD4050 packages buffer the eight bits; a 4.5 V battery powers them to keep system noise out. | Bit | Resistance to the summing point | |-----|---------------------------------| | 7 | four buffers, each with 47 kΩ, in parallel | | 6 | two buffers, each with 47 kΩ, in parallel | | 5 | 47 kΩ | | 4 | 47 kΩ + 47 kΩ in series | | 3 | four 47 kΩ in series | | 2 | 390 kΩ | | 1 | 820 kΩ | | 0 | 1.5 MΩ | Output is **0.0176 × N volts**, N being the last byte written. Sending 0 and 255 alternately gives 0 to +4.5 V; sending 0 and 128 gives 0 to about 2.26 V, about half the amplitude. The article notes that an amplitude envelope during a tone is hard to do without upsetting the loop timing. ### 7.4 Fig. 5 — white noise (p. 119) A 16-stage shift register in H and L with feedback from bits 15, 14, 12 and 3. The upper 8 bits go to the port. ``` 000:000 021 000 000 WHITEN LXI D,0 000:003 041 001 000 LXI H,1 ;shift register, non-zero 000:006 174 LOOP MOV A,H 000:007 323 XXX OUT port 000:011 017 RRC 000:012 254 XRA H 000:013 017 RRC 000:014 017 RRC 000:015 254 XRA H 000:016 017 RRC 000:017 255 XRA L 000:020 017 RRC 000:021 017 RRC 000:022 017 RRC 000:023 346 001 ANI 1 000:025 051 DAD H ;shift left 1 000:026 137 MOV E,A 000:027 031 DAD D ;new bit 0 000:030 303 006 000 JMP LOOP ``` ### 7.5 Fig. 6 — "weird sound" routine (p. 119) ``` 000:000 021 001 000 WEIRD LXI D,1 000:003 041 000 000 LXI H,0 000:006 175 LOOP MOV A,L ;MOV A,H for 8 more sounds 000:007 323 XXX OUT port 000:011 031 DAD D 000:012 322 006 000 JNC LOOP 000:015 023 INX D 000:016 172 MOV A,D 000:017 263 ORA E 000:020 302 006 000 JNZ LOOP 000:023 023 INX D 000:024 303 006 000 JMP LOOP ``` A different sound is heard on each bit of the port. ### 7.6 Fig. 3 — NOTRAN (p. 118) A sketch of a music language, with no program given: `TEMPO 1/4=500` (a quarter note is 500 ms), a note statement such as `1C#4,1/8` (voice, note name, `#` sharp or `@` flat, octave with C4 = middle C, length), and a rest statement such as `R,1/2`. ### 7.7 On a Model 6 - Fig. 1 writes `000` and `377`. On the Model 6 that is the full 6-bit swing. - Figs. 5 and 6 write a whole byte. The Model 6 converts the upper 6 bits. - The article's amplitude rule holds: 0 and 128 give about half the amplitude of 0 and 255. --- ## Quirks worth carrying - ⚠ **MICROPLAY: the comments say "complement", the code is `XRA M`.** The flow chart (p. 13: "COMPLEMENT OUTPUT VALUE IN ACCUMULATOR"), the comment at `004E` and the envelope comment all describe a wave that goes between a value and its complement. The listing at `004E` is byte `AE`, mnemonic `XRA M`, and the two agree; HL points at the envelope byte that A was loaded from, so A goes between **that byte and `00`**. The two readings give different swings for `TBL1` = 95 30 D0 FE FF FF 30 85: | Reading | Swing (byte − other level) for the 8 segments | |---------|-----------------------------------------------| | value ↔ complement (the comments) | 2B 9F A1 FD FF FF 9F 0B — rises, holds, falls | | value ↔ 00 (the code) | 95 30 D0 FE FF FF 30 85 | **The listing wins**: it is the shipped program, and that is what the board is sent. The table looks as if it was made for the complement reading. Not resolved; there is no second copy of the program to compare. - ⚠ **Four ports, not one.** A program that also uses `25H`, `26H` or `27H` for another board collides with the Model 6. - ⚠ **The low two data bits do nothing.** A program that puts a speaker "on any bit of the output port" (article Figs. 1, 5, 6) is silent on bits 0 and 1. A ramp of 256 steps is 64 levels, each held for four writes. - ⚠ **No reset.** The latch keeps its value through a bus reset, and its power-on value is not defined. The output is AC-coupled, so a held value is silence. - ⚠ **Write-only.** A read of the port address is not decoded by this board (SOUT and pWR are both in the decode). - ⚠ **MICROSCORE's lowest notes do not fit in the pitch byte** (derived, not in the manual). K3 must be 255 or less, so T1 ≤ 56.5 + 7.5 × 255 = 1969 µs, about 254 Hz. `A 1` (220 Hz, K3 = 296), `A#1` and `B 1` are over; `C 1` (261.6 Hz, K3 = 247) is the lowest note that fits. "The Entertainer" uses no octave-1 note below C. - ⚠ **56.5 or 57.0 µs.** MICROSCORE uses 56.5 µs for the fixed part of a half wave. The MICROPLAY listing counts to 114 clock cycles, 57.0 µs (derived). The manual's number is transcribed as printed. - ⚠ **1005 Hz or about 1009 Hz.** The square-wave test routine is quoted as 1005 Hz. Its listing counts to 991 clock cycles a half wave (64 × 15 + 4 + 10 + 10 + 7), about 1009 Hz at 2 MHz (derived). - **`JNZ $-4` is `C2 03 00`.** In the test routine the target is `0003` (the `DCR B`), so this assembler's `$` is the address of the next instruction. - **Article Fig. 1 prints the first `OUT` as `232 XXX`.** The second `OUT`, and Figs. 5 and 6, print `323`. `323` octal (`D3H`) is `OUT`; `232` is a misprint. - **The duration high byte is stored plus 1**, because MICROPLAY tests it with `DCR B` / `JNZ` after the low byte reaches zero. MICROSCORE does this at line 630. - **MICROPLAY must run in RAM**, and its envelope table too: it patches four operands and writes to `TBL1` as a delay. - **The author's name.** The article byline and the manual both print "Hal Chamberlain". The usual spelling of his name is Chamberlin. - **The sample run's last line reads `STOP IN LINE 760`** after `SCORE COMPILATION COMPLETE!`; the `STOP` that follows that message is at line 750. Printed so in the scan.