aclo dclo (11/45) flustered-ness

Tony Duell ard at p850ug1.demon.co.uk
Fri Mar 30 17:44:44 CDT 2007


> 
> Tony wrote....
> > What you have to realise first is that, like most Unibus signals (because
> > that's what they are, they are carried on the Unibus cables), ACLO and
> > DCLO are open-collector driven. They're pulled high (to +5V) by resistors
> > on the bus terminator cards, and can be pulled low at any point, by any
> > device.
> I was thinking that open collector signal levels were low and float, not low 
> and hi. From my testing, you're obviously right though, see below. I guess 
> the pullups are what take the float to hi?

Right.

An open-collector output is what the name implies. The output pin is 
connected to the collector of an (NPN) transistor, the emitter of which 
is groudned. So if that transistor is turned on, the pin is low, if it's 
turned off the pin floats

You (the desigern) get to provide the collector load for that transistor. 
Here it's just a resistor to +5V. So if the transistor is turned off, the 
pin appears high (pulled up via the reisstor). 

The thing is, you can connect many outputs together without problems 
(unlike with normal 'totem pole' outputs where having 2 outputs connected 
togtehr will cause high currents to folw if they're in different states). 
With opne-collector outputs, if any transistor is turned on, the signal 
will be low. If they're all off, it'll be high. Hence the term 
'Wired-AND' (or 'Wired-OR' if you're using negative logic) for such 
connections of multiple open-collector outputs. I've seen schematics (and 
I think DEC were among them) where there's an AND gate symbol drawn 
around the point where all these signals get joined together just to 
remind you what's going on.
> I got +5v & GND for my logic probe from CA2 & CC2 and powered up just the 
> BA11K. BF1 & BF2 both showed low as I'd expect (since the cpu was powered 
> off). This was a sanity check on my part to make sure AC/DCLO were pulled 
> low. I powered up everything and BF1 & BF2 went hi. I hit load on the RL02 
> and the drive started to spin up. At just the time I'd expect the heads to 
> load, the fault lite came on. BF1 & 2 were still hi. If the RX02 is powered 
> off, the drive will not fault, it will spin up fine. Also, if the RX02 is 
> off and everything else on, the drive loads fine and boots RT11. The moment 
> I plug in the power for the RX02 the RL02 immediately faults (but the cpu 
> keeps on running right where it was). One thing I need to check, I am not 
> positive (because it's all rather loud) that when the RL02 faults that it 
> spins down the drive. I don't recall hearing the noticeable decrease in 
> volume. I would have thought it would spin down upon any fault. Maybe it is 
> and i just didn't hear it.

IIRC, not all 'faults' cause a spin-down.

What this has told you is that the problem is not due to ACLO or DCLO. 
Now that's useful information. When you're trying to sort out a fault 
like this, knwoing where it _isn't_ is avtually very useful. 

> 
> I can start going through the prints on the RL02 as Tony suggested, looking 
> at all the signals that can trip the fault light. Is that the right 
> approach? I mean, the problem obviously isn't in the RL02 or controller - as 

Well, it's what I'd do next, obviously (as I suggested it). The reasoning 
behind doing it that way, is that you have a definite problem (the FAULT 
light comes on). So your next move should be to find out _why_ that light 
is coming on.

> they work flawlessly if the RX02 is turned off. This would make me think the 
> problem is more likely something about the RL02 or M8256.
> 
> At least I'm learning a lot :)

Good!. That's one reason I don't like taking things like this to private 
e-mail, I feel that everyone (including me, I might add) learn a lot from 
such discussions, even if we don't ahve that particular machine. Fault 
tracing is much the same on any hardware, actually.

-tony



More information about the cctech mailing list