Sony HB-10 Troubleshooting Guide

Power section of the HB-10P PU-37 board: CN11 inlet, S901 operate switch, C1 and the µPC7805 for +5 V, and the two half-wave supplies for ±12 V (Sony service manual, page 5-12)

Fault-finding on the Sony HB-10 starts with the AC adaptor and the three supply rails, then moves to the known design faults of the board: the /RESET line to the cartridge slots, the RAM mirror on the 64 KB versions and the PSG mixing resistors. Component references are from Sony's service manual for the HB-10P and HB-10B. Niga reports that the Japanese HB-10 board is the same design.

Quick reference

Symptoms, causes and actions

Symptom	Probable cause	Action
Nothing at all, power LED dark	No DC from the adaptor, adaptor fuse F1 or F2, operate switch S901, inlet CN11	Measure across C1 with the switch on; check the adaptor (No power)
Power LED on, no picture, +5 V correct	Failed video chip, video RAM, or the video output path	See No picture
Some cartridges will not start or crash, others work	/RESET held at about 3 V by the slot diode; dirty slot contacts	Add a pull-up or replace the diode (Cartridge problems)
Some games crash on an HB-10B, HB-10D or HB-10P	RAM mirrored into slot 0 (port A8h quirk)	Typed workaround or board fix (RAM mirror)
Chords from the PSG distorted, single notes clean	R22 and R15 at the wrong values	Change R22 to 1 kΩ and R15 to 4.7 kΩ (Sound)
Sound cartridges or disk interfaces dead, BASIC fine	Missing +12 V or −12 V	Check the adaptor's AC output and fuse F3, then D1, D2, IC19 and IC20

No power

The AC-HB1 adaptor supplies DC 6 V at 900 mA and AC 16 V at 160 mA through a proprietary connector. Only the DC output feeds the +5 V rail, so a completely dead machine has lost its DC input or its 5 V regulator, or has a short on the +5 V rail.

1. Unplug everything from the computer. Plug the adaptor into the wall and switch the computer on.
2. Open the case (General Maintenance) and measure across C1, the 10,000 µF 25 V reservoir next to the inlet. This is the adaptor's DC output after the operate switch.
3. If C1 has no voltage, check the adaptor on its own, then S901 and CN11. Inside the adaptor, F1 and F2 (T1.25A) are in the DC rectifier circuit and a 133 °C thermal fuse is in the transformer primary.
4. If C1 has voltage but +5 V is missing at C35, the fault is IC1 or a short on the +5 V rail. Disconnect the keyboard and measure the resistance of the +5 V rail to ground with the power off. A dead short points to a failed chip or capacitor on the rail.
5. If +5 V is present but low, compare the voltage across C1 under load with the reading at C35. The regulator needs its input to stay above its output; a sagging C1 points back to the adaptor or to C1 itself.

sd_snatcher recommends fresh thermal compound between IC1 and its heatsink.

No picture

Niga's Japanese HB-10 powered up but gave no picture, and a previous owner had already replaced the video RAM without success. The fault was the TMS9118NL video chip. A replacement chip brought the machine back. Along the way Niga found two traps:

• The MSM38256RS system ROM uses pin 1 as an active-low chip select. To read it in an EPROM programmer, lift pin 1 and connect it to ground; otherwise the ROM reads as faulty when it is not.
• The TMS9918 used in early MSX1 machines and the TMS9118 in the Japanese HB-10 are not interchangeable. The TMS9918 drives eight 16K × 1 DRAMs with separate address and data lines; the TMS9118 drives two 16K × 4 DRAMs with bidirectional data lines.

Niga found the video chip runs warm and fitted a heatsink to it, noting that TMS9118 failures are not unusual.

The European HB-10B, HB-10D and HB-10P use a Toshiba T6950 video chip at IC5, with its two MB81416 video RAMs at IC6 and IC7. Its clock is set to 22.168 MHz ±10 Hz at pin 38 by trimmer CV1. With no picture on one of these, check +5 V at pin 34 of IC5 and the clock at pin 38 before suspecting the chip. Try both the RF output and the composite output on the 6-pin DIN socket to rule out the modulator.

Cartridge problems

On the MSX Resource Center forum in 2017, ericb59 traced erratic cartridge behaviour on the HB-10 and HB-20 to the /RESET line. Sony fitted pull-up resistors to /WR, /RD, /IORQ and /MREQ but not to /RESET, which passes through a diode. With no cartridge fitted /RESET sits at 5 V, but with a cartridge it falls to about 3 V. Cartridges that need a clean reset suffer: MegaROMs, 64 KB ROMs such as Uridium, the Beer HD cartridge, SD card cartridges and floppy disk interfaces. Niga mentions a report of a MegaSCSI interface that would not start on an HB-10.

There are three fixes, all reported on working machines:

• Add a 4.7 kΩ pull-up resistor from /RESET to +5 V (ericb59).
• Replace the diode with a 100 Ω resistor, which still offers some protection against a cartridge that drives /RESET (sd_snatcher). On his Japanese HB-10 the diode was D22, and replacing it also removed a lot of noise from the picture.
• Link the diode out, as on Panasonic's FS-A1 (Niga).

Clean the slot contacts before blaming the /RESET line, and never fit or remove a cartridge with the power on.

RAM mirror

The 64 KB HB-10B, HB-10D and HB-10P, and some very early HB-20P, show the top 16 KB of RAM in slot 0 as well as in slot 3. Software that keeps its variables in the mirror and assumes all RAM is in slot 0 can crash. Workarounds from the MSX Resource Center:

• Type OUT &hA8,240 before loading the program.
• For a cartridge, boot with the cartridge's slot switched off on a switched slot expander, switch it on and type OUT&hA8,240: DEFUSR=&H7D75: ?USR(0).
• Load ROM images with Execrom, LoadROM or OPFX /X from MSX-DOS, which sets up the RAM slots correctly.

The permanent fix removes the S3527's CAS3 signal from the RAM. The MSX Resource Center shows it on an HB-10P board: cut the track from the S3527's CAS3 output (pin 45) to pin 9 of IC9, the 74LS08, and link pins 9 and 10 of IC9. Niga made the same cut and link on a Japanese HB-10 upgraded to 64 KB, after which RAM appeared only in slot 3. On the HB-20, ericb59 describes a variant that lifts pin 9 of the 74LS08 and adds a 4.7 kΩ resistor from pin 9 to pin 14.

Sound

sd_snatcher found that R22 in the PSG output is 10 kΩ, in Sony's schematic and on his Japanese HB-10, where 1 kΩ is needed. The PSG saturates on many tunes, which can sound like a failed capacitor. His HB-20P already had 1 kΩ there. The fix is R22 to 1 kΩ, then R15 to 4.7 kΩ to restore the level against sound cartridges such as SCC and FM. The MSX Resource Center lists the same change for the HB-10B, HB-10D and HB-10P.

Niga measured the effect on a Japanese HB-10 with PLAY "v15o4a" and a C-E-G chord. Before the change a single note gave a clean square wave of 671 mV peak to peak at the audio output, but the chord was distorted. R15 and R22 are near the left-hand cartridge slot. After the change the single note measured 238 mV peak to peak and the three channels added cleanly. Cut the old resistors out and remove the leads one pad at a time, because the tracks lift easily.

Missing ±12 V

The +12 V and −12 V rails come from the adaptor's AC output, rectified by D2 and D1 into C4 and C3 and regulated by IC19 and IC20. Niga notes that they go only to the cartridge slots, where sound cartridges with op-amps and some disk interfaces use them. Niga tested his machine with only a 5 V supply connected, and BASIC started normally. Check the adaptor's AC output and its fuse F3 (T200mA), then the voltages across C4 and C3, then the regulators.

Test points

Supply pins on the HB-10P/10B PU-37 board (Sony service manual, page 5-10)

IC	Type	+5 V pin	Ground pin
IC5	T6950 video controller	34	13
IC6, IC7	MB81416 video RAM	9	18
IC8	Z80A, LH0080A or µPD780C-1 CPU	11	29
IC10, IC11	MB81464-12 main RAM	9	18
IC12	MSM38256-78RS ROM	28	14
IC15	S-3527 MSX-Engine	40, 90	6, 31

The CPU clock is 3.579545 MHz. The rail check points are in General Maintenance.

Related pages

• Sony HB-10
• Sony HB-10 General Maintenance
• Sony HB-10 Capacitor Replacement Guide
• Sony HB-10P and HB-10B Service Manual
• Capacitor Failure Symptoms