110 Bd USB/Serial

[Michael Katzmann]

I have a USB to serial converter that I'm using to drive a Teletype
but it seems incapable of 110 Bd. If I stty to 110 it comes out at
~260 Bd.
I presume this particular unit is incapable of such a low baud rate.

Is anyone aware of a USB serial device capable of 110 Bd ?

[david_...@brown.edu]

Well, the most distinctive “feature” of 110 baud is that it uses two stop bits, for a total of 11 bits per character. Most other speeds only use one stop bit. This gives your teletype and paper tape device the phenomenal speed of ten characters per second. I do remember the joy of replacing my old 9600 baud modem for a telebit 33.6k. 

Whether the USB to serial can manage as low as 110 and support 2 stop bits might depend on the device you have. So sorry I can’t answer your question directly, just noting an additional setting detail you need to account for. 

  -- ddj

Dave Johnson

On Feb 13, 2021, at 1:07 PM, Michael Katzmann <vk2...@gmail.com> wrote:

I have a USB to serial converter that I'm using to drive a Teletype

--
You received this message because you are subscribed to the Google Groups "[PiDP-11]" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-11+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-11/CAMyt5ORFGG-VbvrEpcwPze%2B3%2B%2BVYy7aLiU-ZdCGXmNR%2BHk6WYA%40mail.gmail.com.

[David Johnson]

Found this discussion -- https://altairclone.com/forum/viewtopic.php?f=2&t=142

[Clem Cole]

Which chip does it use?   Frankly, the only ones that seem to work reliably from the system SW standpoint on the UNIX/macOS/Linux side USB have been based on the FTDI chips set ... all other chips like the CH34x series have had lots of different driver issues and frankly, I pretty much don't mess with them at this point.  But ... that said of standard baud rates, the FT232x chip can only go down to 300 baud reliably according to the datasheet DS_FT232R and the baud rate app note: AN232B-05_BaudRates.  And I have personally not tried anything slower than 9600 baud with it.    IIRC it can generate 183 baud with some driver work, but 110 is not a possible match with their crystal.

So another solution may be to use a software-based UART from something like an Arduino or an ESP chip between you and the RPi.  Set the Arduino up on a network port coming into the RPi, note you'll need to be careful of network buffering, but then use the SW based UART in the Arduino IDE out to your model 33.

ᐧ

[Michael Katzmann]

I can use the built in UARTs of the Raspberry Pi 4 (I confirmed it will do 110 bd) but I have to build a 3.3V to RS232 converter.... just not as neat as the USB dongle. The USB/Serial devices with 'Prolific' chips will go down to 75 bd but 110 is not listed in the datasheet (and the Linux driver quantizes the requested baud rate to the nearest known rate (so 110 will end up at 150)). I might get one and hack the drive to see if it can actually do it.

Michael

[Clem Cole]

https://www.amazon.com/MAX3232-Connector-Converter-Equipment-Upgrades/dp/B07PFB4MHR/ref=pd_vtp_3

But you still need to go RS-232 to 20ma with that.

Another solution is something like a Broadcom HPCL-4200 chip (https://docs.broadcom.com/docs/AV02-2353EN) but that generates TTL or 5V CMOS levels not 3.3v and 2 470 ohm resisters.      Typical usage is:

You can replace the MAX232 with a 3.3v to 5V bidirectional level shifter made out of a BSS138 and a couple of 10K resister like this:

an issue of course, is multiple supplies.

ᐧ

[Jonathan Morton]

On 14 Feb, 2021, at 5:47 am, Michael Katzmann <vk2...@gmail.com> wrote:

I have to build a 3.3V to RS232 converter...

The MAX232 is a long-time favourite for interfacing to 5V logic, but the MAX3223 seems to be the better choice for 3.3V logic, eliminating an extra level-shifting stage:

It produces ±5.5V output on the RS-232 lines, is specified to run up to 230 Kbaud, and powers itself down to a very low-power quiescent state when the RS-232 device is not itself powered or connected.  It is also capable of running at 5V on the logic side.

And it comes in a convenient PDIP package.

 - Jonathan Morton

[Jonathan Morton]

> On 14 Feb, 2021, at 5:10 pm, Clem Cole <cl...@ccc.com> wrote:

> Typical usage is:
> <20ma_to_RS232.jpg>

I'm having some trouble seeing how that circuit could interface to an identical copy of itself. It would probably work if the receiver were connected directly across the pair. But are we assuming that the terminal at the other end of the loop is an "isolated" type, and thus requires an active, non-isolated transmitter *and* receiver at this end?

Level-shifting these two devices to 3.3V CMOS is probably easier than you think. The input of the HCPL-4100 is TTL compatible, which means it will accept a 3.3V CMOS output directly. The output of the HCPL-4200 can be shifted down to 3.3V using a simple 2:1 resistor divider. To reduce quiescent power consumption, these resistors can be quite high value if they are paralleled with a stack of three identical (small) capacitors, but in the context of a current loop that itself consumes 20mA per direction while quiescent, that probably isn't worthwhile.

- Jonathan Morton

[Clem Cole]

Cool, actually I figured they had one, but did not look.   I happen to have a ton of these in a junk box (and 1488/1489 pairs) but a bunch of level shifters so for a one shot, that's what I would do.

ᐧ

[Clem Cole]

On Sun, Feb 14, 2021 at 1:13 PM Jonathan Morton <chrom...@gmail.com> wrote:

> On 14 Feb, 2021, at 5:10 pm, Clem Cole <cl...@ccc.com> wrote:

> Typical usage is:
> <20ma_to_RS232.jpg>

I'm having some trouble seeing how that circuit could interface to an identical copy of itself.  It would probably work if the receiver were connected directly across the pair.  But are we assuming that the terminal at the other end of the loop is an "isolated" type, and thus requires an active, non-isolated transmitter *and* receiver at this end?

What is scribbled at as +12 thru a 470 ohm  is the 20ma  power from model 33 side (separate from the PI).  The whole idea was isolation - we liked to keep the model 33 very separate from the computer.   In the old days, we used 4N27 or 4N33's actually and the circuit is a little more complex.   Remember 20ma and 60ma were industrial equipment from the really old days.   RS-232 does not arrive until the mid/late 60s.

Check out: http://www.kavina-systems.com/articals/current_loop_app_note.html

ᐧ

[Jonathan Morton]

> On 14 Feb, 2021, at 11:20 pm, Clem Cole <cl...@ccc.com> wrote:
>
>> I'm having some trouble seeing how that circuit could interface to an identical copy of itself. It would probably work if the receiver were connected directly across the pair. But are we assuming that the terminal at the other end of the loop is an "isolated" type, and thus requires an active, non-isolated transmitter *and* receiver at this end?

> What is scribbled at as +12 thru a 470 ohm is the 20ma power from model 33 side (separate from the PI). The whole idea was isolation - we liked to keep the model 33 very separate from the computer. In the old days, we used 4N27 or 4N33's actually and the circuit is a little more complex. Remember 20ma and 60ma were industrial equipment from the really old days. RS-232 does not arrive until the mid/late 60s.

I'm at least vaguely aware of the history.

What wasn't clear before reading those extra notes was that the connections go RX+ to TX- and TX+ to RX-. This is a very unusual polarity convention, which is not described in the HCPL-4x00 datasheets. From previous experience with battery-powered circuits and modern differential signalling, the natural assumption is that like polarities should connect.

I suspect this relationship would be clearer to future readers if you explicitly put the relevant parts of the ASR-33 into the diagram, on the far side of an obvious dividing line representing the cable to it.

- Jonathan Morton

[andy]

Yes - Amazon (genuine) Plugable USB Serial converters are great at such a low speed. No drivers necessary using a Pi.

If you need more than a single port, the old Keyspans do great work, but require the correct (easily obtainable) drivers.

If you want to go totally nuts, the old Equinox Terminal servers run happily at 110 baud.

cheers!

Andy