General Notes

Floppy

Termination is 150 Ohms, so 33mA at 5V. I’ll have to be careful not to have more than five signals driving low at once on each port/regulator.

22 signals total, depicted in the APC System Reference Guide, figure 3-13 “FDD Signal Connector Interface and Pin Assignments”. But I don’t need three of the drive select signals. Though it would be nice to just yank the cable off the controller end and plug in to that, not have to also unplug one of the drives, or not be able to use both.

Might be able to ignore the “two sided”, “disk change”, and “write protect” – hook both to LEDs instead. “file unsafe” and “file unsafe reset” could go to an LED and a push-button, probably only important for debugging to indicate the controller screwed up. MFM, Window

I wonder if I could share the read and write signals, switching direction depending.

Signal	Name	Pin	Direction	Description	Implementation
Drive Select A	DS1	25	Out		PP driver
Drive Select B	DS2	23	Out		PP driver
Step	STP	15	Out		OC driver
Direction	DIR	17	Out		OC driver
Write Data	WDT	13	Out		OC driver
Write Gate	WGT	11	Out		OC driver
Head Load	HDL	33	Out		OC driver
Side Select	SSL	37	Out		OC driver
FM / MFM	MFM	3	Out		OC driver
VFO Sync	VFOS	27	Out		OC driver
Low Current	LWC	49	Out		OC driver
File Unsafe Reset	FLR	47	Out		Push button
Index	IDX	31	In		PU 150R
Write Protect	PRT	7	In		PU 150R, LED
Track 0	TK0	9	In		PU 150R
Read Data	RDT	5	In		PU 150R
Ready	RDY	29	In		PU 150R
Dual Side	TSD	41	In		PU 150R, LED
File Unsafe	FUS	45	In		PU 150R, LED
Data Window	WIO	1	In		PU 150R

Seems there’s some disagreement about pin numbers! Be careful what the APC thinks is pin 1 and what the NEC drive thinks is pin 1. I also suspect non-NEC drives will have a different opinion!

NEC APC Reference Guide depiction of connector on the back of the floppy drive (NEC FD1165-A):

  +---+
> |. .|  1  2
  |. .|  3  4
  |. .|  5  6
  ~~~~~
  |. .|
   . .|
   . .|
  |. .|
  ~~~~~
  |. .| 45 46
  |. .| 47 48
  |. .| 49 50
  +---+

NEC APC floppy controller connector, from pictures I’ve taken:

  +---+
> |. .|  1  2
  |. .|  3  4
  |. .|  5  6
  ~~~~~
  |. .|
   . .|
   . .|
  |. .|
  ~~~~~
  |. .| 45 46
  |. .| 47 48
  |. .| 49 50
  +---+

They’re the same. But wait! Not so fast… Earlier suspicions confirmed! The photos I took when I hooked up a GoTek to the controller suggest that the pin numbering isn’t that simple.

Repair Log

20190917

Disassmbled. Removed 8” floppy drives to make access easier for Gotek attachment. De-gooed the CRT and case and got the permanent marker off with some 99% alcohol.

While hunting for a useable 5V supply on the floppy drive power cables, I shorted two pins on the connector and blew fuse F1. F1 is labeled on the PCB as being 5A. Fuse reads “NEW GGS 125V 5A” and measures 20mm long, 5mm cap diameter, 5.5mm cap length.

I got some (very expensive) replacement fuses, and proceeded to blow them instantly upon turning the power switch on. What a lovely green (and very expensive) flash they made! This happened even if the power supply was not plugged into the rest of the computer. Oh dear.

20191002

After much power supply disassembly…

First, I attempted to reverse-engineer the power supply schematic, in order to understand what the various components were doing, and what was the topology of the supply. That took a long time, but I devised some interesting techniques.

After I had a picture of supply topology, I decided to inject DC from a bench power supply into places just upstream from linear regulators. I chose points just after diodes from the secondary side of the transformer – I figured those would be good places to inject DC voltages at low current. The -5V and -12V regulators were fine (as were the associated capacitors). The +12V and +24V supply scared me a bit, as I had to put nearly 400mA in to get them to come up to voltage. But then I remembered that there’s a giant 33 Ohm resistor between the +24V and +12V regulators, and that would account for all that current! How wasteful…

So the “left” (linear) side of the supply seemed happy enough. Now to look at the right side, which was the switching elements of the supply… I pulled out all the transistors and tested them. The power transistor that switched into the primary side of the transformer appeared to be a dead short every way I measured it. That can’t be right, or good.

I spent a bunch of time shopping for a replacement transistor, since the original 2SC2751 isn’t available now. I settled on a 2SC4140, which seems to match the old part in almost every way, but might be just a touch under-spec in a couple of places. Fingers crossed. :-)

Upon replacing the power transistor (Q1?), everything powered up OK! Whew.