DEC bus interface chips

DEC's Unibus and Qbus have similar electrical characteristics. They are roughly open collector TTL, but with tighter specifications:

parameter minimum maximum condition
Driver Vol 0.7 V 70 mA
Leakage current 25 uA 3.5V (even when Vcc = 0V)
Output pin capacitive load 10 pF
rise and fall times (10% to 90%) 5 ns
prop delay to low state 25ns
prop delay to high state 35ns
Receiver input pin capacitive loading 10 pF
Vil 1.3V
Vih 1.7V
Leakage current 80 uA 3.8 Vdc (with Vcc from 0.0V to 5.25V)
Leakage current 10 uA 0V DC

DEC used several different receiver, driver, and transceiver chips. One of the most commonly used quad transceivers was the DEC8641, which was a National DS8641 selected for tighter specs. Later they used the DS3662, an improved spec replacement for the DS8641, but they still apparently selected them, since the DS3662 specs still don't meet the Qbus requirements.

National has end-of-lifed both the DS8641 and DS3662. They are now extremely difficult to obtain.

I don't know of any current production (non-EOL) parts that fully meet the Qbus specs.

There are some parts that get close, but don't quite meet the specs. If I had to use new parts today, I guess I'd probably use one of these:

You can get close for drivers using the 74ABT126, tying the inputs to ground and using the enables. To really guarantee meeting the Qbus specs, you'd have to parallel two drivers.

The receiver specs are more troubling. I'm not sure what to do other than using a comparator. In practice, normal TTL inputs will probably work, if you don't worry about temperature extremes and large configurations with multiple backplanes.

Posted to Info-PDP11 on 11-apr-2005 (my response here has been edited somewhat since it was posted):

Peter wrote:

> But this brings to mind the following:
> For Unibus tranceiver chips, I have a small stash that will keep me
> going for awhile (8881's, DEC380's, etc.). But if I make a PCB that
> other folks think is useful, would they be able to get those chips?

I suggest using only the quad transceiver chips, so that people don't have to find multiple types. The most common of those were the DS8641, and the improved-spec DS3662.

> Has someone identified alternative chips that are easy to find, and
> definitely work on a real Unibus?

The Unibus driver/receiver specs are almost the same as the Qbus driver/receiver specs.

The hardest specs to meet seem to be the recevier Vih spec, the leakage specs (especially leakage when unpowered), and the maximum slew rate (minimum rise/fall times).

There are two quad transceiver chips in current production that I think might work in practice, though they don't actually meet the DEC specs on the input side: the TI AM26S10 and SN75138. The Am26S10 specs are closest to the DEC specs, and the AM26S10 is less expensive than the SN75138. They are pin compatible with each other, but not with the DS8641/DS3662. A PC board could be dual patterned to accept either footprint.

parameter DEC spec National DS8641 spec National DS3662 spec TI AM26S10 spec TI SN75138 spec
driver Vol max 0.7V at 70mA 0.7V at 50mA 0.9V at 100mA 0.7V at 70mA 0.45V at 100mA
leakage current max 25 uA driver, 80 uA receiver 100 uA max 100 uA max 100 uA max 25 uA typ, 300 uA max
leakage current max, Vcc = 0V 25 uA driver, 80 uA receiver 100 uA max 100 uA max 100 uA max 1500 uA max
capacitive load 10 pF max ? ? ? ?
driver rise time min 5 ns (Qbus), min 10 ns (Unibus) typ 17 ns typ 20 ns min 4 ns, typ 10 ns ?
driver fall time min 5 ns (Qbus), min 10 ns (Unibus) typ 17 ns typ 20 ns min 2 ns, typ 10 ns ?
receiver Vil 1.3V min 1.3V min 1.5V min 1.75V min 1.8V min
receiver Vih 1.7V max 1.7V max 1.9V max 2.25V max 2.9V max

Note that both the AM26S10 and SN75138 have a Vil(min) above DEC's rated Vih(max), which could definitely cause problems. The driver slew rates are also faster than the DEC spec, which could cause ringing.

The other possiblity would be to use separate drivers and receivers. The 74S38 would probably be OK as a driver (rated Vol 0.5V at 60 mA), and the 74HCT14 or 74LVC14 would probably be OK as a receiver (Vih of around 2.0V), though neither would meet the unpowered leakage spec.

Last updated November 10, 2008

Copyright 2000-2005, 2008 Eric Smith

Best Viewed With Any Browser Valid HTML 4.01!

check now