Idiot's Guide to breaking & Fixing the Sage 930A

The Sage 930A is a handy piece of test equipment, which can be had for a fairly reasonable price on eBay. Being an 80's vintage piece of electronics, it is also very repairable, so even units being sold as non-working can be quite easily brought back to life. From a design standpoint, it was quite leading edge, containing two break-through devices: the first DSP (the TI TMS320) and the first FPGA (the Xilinx XC2064).

Sage 930A Hardware Breakdown

CPU Board

The supervisor brains of the system; a 4 MHz Z80 CPU and supporting logic.

Daughter Board

• Intersil ICM7170 Real-Time Clock
• AMD Zilog Z8530 SCC
• Lithium Iodide Battery

Main Board

• Z80 CPU, 4 MHz
• Z80 CTC
• 7 ROM
• 1 6264 RAM (64 kbit, 8 kbyte)
• Lithium Iodide Battery
• clock (mhz?)

PCM ESF Board (two in system)

There are two identical PCM cards in the system, the only difference between them being a jumper setting and their position in the chassis. The ROMs on these cards are part of the system address bus, and the set will fail its boot-up ROM tests if either of the PCM cards are removed.

• Z180 CPU, 6 MHz - integrated clock generator, 16-bit counters/timers, interrupt controller, wait-state generators, serial ports and a DMA controller.
• Xilink XC2064 - 64 cell FPGA, Xilink's first FPGA
• ROM
• 6264 64 kbit ram (8 kbyte)
• 1 rom
• SCC
• Xilink XC2018 - 100 cell FPGA, confusingly numbered sibling to XC0264
• Analog Devices ADC0808 - 8 bit ADC with 8 channel MUX
• LF147/LF347 Wide Bandwidth Quad JFET Input Operational Amplifiers
• National LM319 High Speed Dual Comparator
• Pulse transformers - T1 in and out
• Crystal Semi CS61534 - PCM Transceiver
• Crystal Semi CS2180A - T1 Transceiver

DSP Board

Handles signal processing and tone generation functions.

• Burr Brown PCM53 - 16 bit audio DAC
• DG212CJ - Quad SPST CMOS Analog Switches
• DG211 - Quad SPST CMOS Analog Switches
• LF356N - JFET Input Operational Amplifiers
• LF353N - Dual JFETInput Operational Amplifier
• LF347N - Wide Bandwidth Quad JFET Input Operational Amplifiers
• National LM161N - High Speed Differential Comparators
• DAC08C - 8-Bit high-speed multiplying D/A converter
• 6264 RAM (64 kbit, 8 kbyte)
• 74HC193 Presettable synchronous 4-bit binary up/down counter
• 6116 - 16kbit, 2kbyte SRAM
• TI TMS32010 - DSP

DC Board

• Analog Devices ADC0808 - 8 bit ADC with 8 channel MUX
• Motorola MC14569 - Programmable Divide-By-N Dual 4-Bit Binary/BCD Down Counter
• LM324 - Low-Power, Quad-Operational Amplifiers
• DP8311 - Octal Latched Peripheral Drivers (100 mA driver)
• Motorola MC145436P - Low Power Dual Tone Multiple Frequency Receiver
• TL074CN - Low-Noise JFET-Input Operational Amplifiers
• CD4052B - 2-Channel, 4:1 Analog Switch with Low ON-Leakage Current
• TL072CN - Low-Noise JFET-Input Operational Amplifiers
• NE571 - compandor
• DG212CJ - Quad SPST CMOS Analog Switches

AC Board

Not documented yet.

Display Board

Not documented yet. Largely dominated by VFD driver circuitry.

Backplane Board

Not documented yet. Very little active electronics.

Power Supply

In both of my Sage 930As, the Power-One SP584 power supply is a custom Power-One unit that appears to have been special ordered from Power-One. Later models carried a SP594, which is a very similar design. I'm guessing that the 930A power supply is based on the Power-One SPL53 design, an 80 watt power supply platform that was available in a number of output configurations, with up to 4 outputs. Here's an SPL53-1005:

Compared to a Sage SP594:

The Sage 930A power supply is based on a 4 output design, +5, +15, -15 and -52V, which is regulated down to -48V by the circuitry hanging off the left side of the board. Also, an opto-isolated 120 Hz powerline-derived timing reference signal is generated, the diodes constituting the bridge used for this are tucked in wherever there is free space on the PCB.

At first glance, this power supply has a number of what appear to be custom semiconductors. In one of my units, the main regulator IC carries a proprietary part number. Also, the main chopper transistor (marked SGS/ST) and the transistor connected to the feedback toroid (marked National) are undocumented. On the newer of my two supplies, the regulator IC is revealed to be a boring old LM723.

The Idiot's Guide...

Why this obsession over the power supply? My first unit worked when I first received it, then it refused to turn on when I tried it out a couple days later when I received my second 930A. They were supposed to play together. I checked the obvious things, and discovered that the power supply was not producing any output, but the high voltage DC supply side was working. The input filter capacitors weren't bleeding down, one 5 watt power resistor was missing, and a 1 watt power resistor was open-circuit. The 1 watt resistor connected the base of the chopper transistor to one of the high voltage rails. From the schematic, I'm struggling to make sense of the 5 watt resistor, but it ultimately functions as the bleed down resistor for the input filter caps. It appears the 5 watt resistor fell some time before the unit arrived at my door. With it missing, the 1 watt resistor was somehow doing double duty, and ultimately cooked itself and failed open, which is what killed the output.

The 1W resistor was a weird value, 510kΩ, that I didn't have in my stock of resistors. I also don't have many 1W resistors, so I went to the local electronics store and in the intervening time confused myself into thinking I wanted a 5.1kΩ 1W resistor, contrary to what I had written in the note to myself. They only had E series resistors too, so I came home with a 5.6kΩ 1W resistor that I happily soldered in place. I put on my flash protection gear, kevlar underpants and nomex toque, and flipped the switch.

The Plot Thickens

The chopper transistor didn't seem to appreciate suddenly being connected to 160 VDC at 100 times the current it expected, promptly turned into a 3 way wire, which popped the fuse.

It has already been established that the chopper transistor (Q5) is a mystery device, so I thought that I'd pull the supply from my second unit to see if maybe it would give me a clue. It didn't. What I did discover was that almost every single electrolytic capacitor on the power supply had managed to leak through to the underside of the PCB, and make a big mess out of everything. So off to Digikey I went, and hunted down mechanically and electrically compatible replacements for every single electrolytic capacitor. I also looked for possible replacements for Q5 and added a few of those to my order. Oh, and of course, 4 fuses. Given that mosfets are used elsewhere on the board, I assumed that a mosfet would have been used for the chopper transistor. This clouded my judgement. I started to trace out the PCB schematic:

Based on some convincing documentation from a SGS databook, I convinced myself that Q5 was most likely a mosfet. I selected a few candidates, and ordered them. When everything arrived, I recapped both power supply boards. The electrolyte created a heat impervious lead oxide all over board #2, and it took a few days to finish that one. Board #1 took probably 20 minutes. Then I moved on to Q5. I plugged in my most likely mosfet candidate, powered the board up, and still nothing. No smoke, no fuse popping, just nothing. I hooked the working board up to my oscilloscope, and quickly realized that I didn't know what I was looking for.

Most of the information online about switch-mode power supplies covers more modern designs with PWM controllers and opto-isolated feedback. This supply had used a linear regulator combined with a discrete oscillator to drive the chopper, and magnetically coupled feedback. I learned a lot from Ken Shiriff's excellent article, "Apple didn't revolutionize power supplies; new transistors did". I also ordered a couple 80's vintage books on switchmode power supplies and set to learning in my spare time. In the mean time, summer kicked in and I had to take a break from this project.

Epiphany

4 months having passed since starting this project, it became apparent that I needed to get everything put back together, or I would never be able to remember what went where. So I thought, why not just ask Power-One what this transistor is? I contacted their technical support, and while they were very willing to help me with my quest, they reported they had nothing on file for the SP584 or SP594.

So close, yet so far!

But then I remembered that while searching eBay for Power-One power supplies, I came across units that looked very similar to the SP584/SP594, save for the goofy outboard circuitry. I went back on eBay, refined my searches, and found out that the SLP53 series were not only almost identical as far as the input and switching circuitry was concerned, but that they were also quite common. So I went back to Power-One and asked them if they had data for those units on file. In short order, I received a screen-shot of the line from a BOM that looked like this:

TSTR NPN 850V 10A TO-220 AZ EA 1.0000 Q5

Finally! I checked this data against a 1988 vintage SGS databook, and they definitely sold a transistor that matched this description in the 80's. This had to be it! That model was long since discontinued, so my next stop was the Digikey parametric search tool, where I found a close match with the ST BUL510. While it's also obsolete, Digikey still has a stock of them. I ordered 4. They arrived the next day, I installed one, and it worked.

While I had the ear of a friendly engineer, I figured I'd ask about the other mystery device, the transistor connected to the feedback transformer. I received this:

BF1 TRSTR;PNP;60V;800mA;hfe 300;TO92;;;1.000 EACH 1 Q4

The closest modern match for this one is the BC32740TA.

Replacement Parts

For those that are following along at home, here are all of the part numbers for the replacement electrolytic capacitors, the fuse and the two mystery transistors:

Manuals

• 930A Operating Manual Revision 3.2
• 930A Operating Manual Revision 5.03