Subject: Re: 1401 TAU debug status & 2N2894 
Date: Wed, 12 Apr 2006 ... 
From: Ron Crane     rccrane at pacbell dot net
To: [ those involved ]

The IBM 102 and IBM 117 transistors are fast PNP germanium switching transistorsin
a TO-18 package. The cross reference for them on the SMS cross reference sheet is 
a 2N705 with a Vceo of 15 V.

Because the 2N705 was not available, we tried the 2N1303, which worked, but
is much slower, 800 nS (2N1303) vs. 100 nS for the IBM 102 and 2N2894 in the logic 
gate circuit. In the flip-flop circuit it works for set and reset, but fails completely
on clock input because the delay is much longer than the RC time constant of the AC 
coupled clock input circuit. The 2N1303 got used because it passed the "try it 
and see if it works" test done by me, not because any of the characterization or 
measurement described in this message was done at that time.

In the flip-flops, the emitter is tied to ground and the collector tied to - 12 V through
a 1.5K resistor and clamped to -6 V with a diode giving a maximum measured 
collector-emitter voltage of about 6.7 V. If the diode gives out, or the -6V supply goes
open circuit (inserting a card with power on), there is a maximum of 12 V through a 
1.5K resistor to collector.

In the gate circuit, the collector is tied to -6V through a 560 ohm resistor. 

In all cases, the 12 V rating on the 2N2894 transistor is adequate. The actual breakdown
on the 2N2894 units purchased is higher. The lowest was 14.5V with most of them 
above 18 V.

I believe we should remove the remaining 9 or 10 2N1303's from the 102 positions and 
replace them with 2N2894's. This will avoid problems in the future if the cards get 
swapped to a location where the original design speed is needed.

- Ron