On Feb 27, 2018 9:34 PM, "John R. Hogerhuis" <[email protected]> wrote:



On Tue, Feb 27, 2018 at 6:24 PM, Brian White <[email protected]> wrote:

> Yes, except most generic level shifters (like that sparkfun one) don't
> have enough channels to do all the flow control lines, so you might get
> rts/cts but not dtr/dts.
>

AFAIK RTS/CTS is irrelevant when using TPDD other than that CTS must not be
found deasserted. That can be handled via loopback. Or not! If it needs the
RTS/CTS line to power level conversion, maybe it has to pass through.

DTR/DTS is obeyed by TS-DOS and may be used by the drive.  It's used as a
cable detect, at least. Don't know what lines or how they are used to do
the level conversions.

I did hunt up some special ones that do have enough channels, but I haven't
> tried to actually make an adapter out of it yet. For one thing it needs a
> power source which might not be doable just from parasitizing the serial
> port itself.
>
>
That is what we think happens... a PC serial port generally provides 12v .
A Model 100 provides 5V signals which is marginal for RS232 but in-spec.

The TPDD is at TTL levels so the presumed circuitry does the conversion.


> The original real cable and the Marty Goodman one does not require a power
> source, because it does not bother to do full proper level shifting, it
> just relies on a trick which happens to work on a M100.
>


I don't know about that. As long as it adapts TTL level signals to RS232 at
at least 5V, it's doing its job and is in-spec.

Also of interest is that I've never had any problem with real TPDD cables
when connected to a PC instead of a Model T. In other words, TPDD cables
don't just work with the Model Ts. So any good replacement design should
work as well.



That's what I meant by one of the complications is that "often you can get
away with a lot" interfacing with rs232. A 0v or floating line should not
actually count as a low, but a lot of times it works anyway. Depends on the
tranciever/uart being used, or other implementation details like those
pull-downs in the M100. That muddies things amd makes you think you have
something right when it's really not yet.



If the M100 outputs only -5v to +5v, that is in spec for rs232, and then
the M100 could power a 5v device directly, but the cable would still have
to be prepared to handle the case where it's plugged in to some other rs232
port which could have as much as 25v (or really, even more to allow for
some other device to be out of spec itself) on it's signal lines. So you'd
really ideally want a voltage regulator that can handle 30v and output 5.

Come to that, I guess the diode cable doesn't, can't actually guarantee to
deliver safe 0-5 to the pdd. It has neither a real voltage regulator nor a
real rs232 shifter, and so if the computer has a 25v serial port, it must
surely hurt the pdd. It's only "working" by luck, that the m100 just
happens to run at +/-5v and most pcs just happen to run at +/-13v or less.

A cable with only diodes is probably fine for use exclusively with a M100.
A cable with diodes and resistors is probably fine for most pcs where the
port only goes to 12-13v, and maybe still works on a M100 by luck that the
inputs on the pdd still probably work somewhat below 5v.

But if you don't actually *know* for certain what a given port runs at,
like say a port on some other odd/old machine (connect a pdd to some s100
box or something), then you can't actually know it's safe to even plug in.
It might end up hitting the pdd with 24 1/2 volts or something.

This is a perfect example of how in many cases "I saw it work with my own
eyes." or "It's been working for everyone for years already." actually
doesn't prove a thing.

-- 
bkw

Reply via email to