QBUS voltage and drive question

[jonathanjo]

Hi Friends ... feel free to tell me it doesn't matter! I'm just starting to learn about QBUS and want to make sure to avoid casualties on my old equipment. Thanks!

For context I'm working on the power-good signals BDCOK etc for an H9275A backplane for a QBUS KDJ11J-D/S (main story over here).

Voltage?

I'm confused by Diane showing a 120R resistor on her TTL debounce circuit, and saying:

The resulting output has to be converted from TTL +5V logic to DEC Qbus +3.4V logic

My understanding is that QBUS is TTL, 5VDC, and that signals such as BDCOK are designed simply to be pulled low by an open-collector output:

Supermicrosystems Handbook says (p A-35):

• TTL Inputs Low/High=0.8V max/2.0V min, Outputs 0.4V max/2.4V min
• Devices driving the 120-Ohm bus must have open collector outputs and meet the following [sink/source specification].

Drive?

Everybody seems to drive this with ordinary TTL chips such as 74LS04, not open collector ones: I'd have expected 74LS05 for example.

Except Frain which has what seems to be a good NPN drive but with an unnecessary pullup:



And Joe Lang who has the only one which looked right, an open-drain 2n7000.

 

[Steve Toner]

Yeah, well, it's complicated...

DEC treats the Qbus as a transmission line, and it's terminated (at least in the case of the 9278 backplane) by a 330 ohm resistor to +5V and 680 ohms to GND. This gives an impedance of approx. 222 ohms (the resistors are in parallel for AC signals) - they call it 220, and a steady-state voltage of approx. 3.36V - they call it 3.4. Bus receivers are specified as Input High >= 1.7V, Input Low <= 1.3V. So the bus is biased to a High value and all you need to do to get it to a Low value is pull it down. And so an open collector output is perfect for that. Adding a pullup resistor will raise the 3.4V and improve noise immunity, but is not strictly necessary. But note that DEC does use 2K pullups on the front panel signals



As far as driving the bus lines with active outputs: You would never do it if there is something else on the bus that might pull that particular bus line low (essentially shorting the TTL output to ground), but (and this may be somewhat controversial) if you are certain that nothing else will pull the line to ground, it'll work. As with the pullup resistor, it'll want to pull the 3.4V a bit higher and thus increase the noise immunity. The only concern is that TTL outputs don't like to source a lot of current, and it's got that 680 ohm resistor to ground that it's got to drive. 5V into 680 ohms is about 7mA, which is more than a TTL output can provide. The 120 ohm series resistor mentioned above would limit the current, but also increase the Low voltage to not less than 1.3V, so it's probably not a good idea. If you look at a typical TTL totem pole output, there's already a series resistance between +5V and the output, so it is limiting the current and the only issues are whether the power it dissipates (less than 5mW) is too much and whether the traces on the die can handle the current or turn into a fuse...

 

[leegleason]

Isn't the resistor in the Frain design serving to limit the current through the transisistor, as well as serving as a pullup? Maybe that's why it's there.

 

[Steve Toner]

No, it actually increases the current that the transistor has to sink because it's in parallel with the 330 ohm bus terminating pullup resistor. If you're thinking that without that resistor the collector would be connected directly to +5V, that's just not something you'd do in this case. The resistor(s) provide the high logic level and the transistor serves to pull the line down to a low logic level...

 

[daver2]

Transmission line theory and bus termination = magic !

Dave