Yes that 0v from the tpdd is exactly one of the things that seems really wrong to me, regardless that it apparently "works".
I mean, it's OK in the particular case of the M100, because the M100 contains pull-downs internally that will pull a floating line down to -5. Floating or open, or tied to 0v only through a resistor which some other drain or source could override, not 0v tied fully to gnd with a short or low ohm resistor. (I think. I'm not stating this as a fact like I measured and witnessed it, just my understanding from past conversations I've read.) So for connecting to a M100, assuming this understanding is correct, that seems fine. But when the same cable that relies on those pull-downs is plugged in to some other serial port that may or may not implement such pull-downs (and pull-ups for that matter), then really those 0v signals should technically be in violation of the spec and shouldn't be relied on to work, even if they happen to work in most cases. That's mentioned right in the same Marty Goodman doc, about the lapdos cable. They (apparently, I've never seen one) made a special cable that does produce proper low signals that are -3v or lower, not merely passing through the 0v, because it was meant to connect to a regular serial port. On Thu, Mar 1, 2018 at 6:41 PM, Anthony Coghlan <[email protected]> wrote: > Regarding the TPDD cable and what’s “in” it to handle the RS232 to TTL > level conversions passively, isn’t it possible that the little 3 terminal > devices Marty described are just what he ended up using, i.e., > diode-resistor dividers in which the diode has a high enough reverse > breakdown voltage not to blow when it’s biased at -5v (most diodes would be > OK with that, I think), and the resistor (22K) is large enough to limit > current (to about 200uA) but also small enough not to act with parasitic > capacitances to “dull” the data signal from the Model 100? At 19.2kbaud, > the parasitics would need to be quite high to impact the signal much, I > suppose, though a long cable would contribute to this risk. > > Something like this (ASCII art, anyone?): > > TXD: +/- 5v from RS232 M100 side > ——— > I > I > ——- > \. / Diode > V > ——- > I > I———-0 TTLrx {~4.4v if TXD=5v, 0v if -5v} > Z > Z Resistor > Z > I > I > /////// Ground > > > Is this too simplified? Maybe I’m overlooking something. Seems like a > plausible albeit ingeniously simple approach that might describe what > happens to work at least for this specific application. I’m more surprised > that on the return line from TTLtx to RS232 RXD, the 0v signal still seems > to be interpreted as a logical “low” for the M100. > > Best wishes, > Anthony > > -- bkw
