I need a usb3 to rs232 adapter that WORKS RIGHT

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Carlos E.R. wrote:
On 08/08/2019 23.53, Paul wrote:
I have no direct experience with designing in RS232,
just the usual kooky contact with people who know
how to do that stuff. (You know what engineers are like.)

We had to create an rs232 from discrete components at college ;-)
Not transistors, but chips. Ie, doing the job of an uart with plain ttl
or cmos chips.

Now that'd be something interesting to see.

Speaking of "built out of chips" -- did you ever watch the "Breadboard
Computer" series on YouTube? It was great seeing his little blinken-led
machine come together.

Although I have the feeling he was over-simplifying things in some cases
(but if he didn't, I don't think I'd have followed along nearly as
well).

-----BEGIN PGP SIGNATURE-----

iQEzBAEBCAAdFiEEBcqaUD8uEzVNxUrujhHd8xJ5ooEFAl1NRw sACgkQjhHd8xJ5
ooHQuwf+NZ4tjNBDOFiXen9qGR7wIy36VAHuxaPoHhtLFOlGo0 fE8ysorbll1WYf
g5/NdntD2ueIdtUJ9XuyN77YasLVSV/gdem/Mh7o4QwWePG66nC5fLDUdqXLLH/Q
sgwd9u6qUu9vfJQyIuVoHBse09jzig+J204oNxepm4FekqDERo 7BGKlk2WWCPsz2
8tt495UUGeiLeAK+QBkrAzE+E1k+e6chGQ6xxTeB7/JQyDFkurg6HV2PeeR94t7R
tdQpthn9lEZ2BqGA8OZVOhP1stjsCzOI8iVSJ4JlCGzDQxGlEi EWwYmcw7dC1G62
g6XWw1RYNuJlirIDNHKbFpCUqUQP8w==
=6km5
-----END PGP SIGNATURE-----

--
|_|O|_|
|_|_|O| Github: https://github.com/dpurgert
|O|O|O| PGP: 05CA 9A50 3F2E 1335 4DC5 4AEE 8E11 DDF3 1279 A281

Char Jackson wrote:
On Thu, 08 Aug 2019 16:08:33 -0400, nospam
wrote:

In article , Frank Slootweg
wrote:

I have a customer with a Mac and a w10 laptop and a CNC machine
with an RS232 (only) interface. He wants to be able to send
text files to the CNC machine. The Mac has USB3 and Thunderbolt
interfaces on the back. The W10 laptop has usb3 ports.

Now I could go the the various accessories sites and get
such an adapter, but they are typically trash and don't
work for beans. The customer did state this was his
experience. (Mine too.)

Any one know of a USB3 to RS232 adapter THAT ACTUALLY
WORKS RIGHT?

Very likely the problem is not that the adapter does not work, but
that such adapters are normally intended to drive peripherals such as
printers, not CNC machines.

the usb-serial adapters i've used provide a standard rs232 port, which
i used to talk to the console ports on old upses, a managed switch and
a gps, among other devices.

+1

My team and I use usb-serial adapters every day to administer
Enterprise-grade networking equipment via their Console ports. The brand
of the adapter doesn't seem to matter at all. They all just work. I've
never picked one up that didn't work. It's just a standard rs232 port.
There's no special protocol that the adapter itself needs to understand
or support.

Of course there are protocols, unless data transfer is purely
uni-directional and the CNC device can handle the whole transfer in one
go without any delay.

Some devices may be in that category, but one can't assume that all
are, nor that the OP's customer's are.

Since there are several different flow control protocols (see my and
Paul's post), one will have to select which one to use. That selection
is a software one and the USB-RS232 adapter's software - whether
supplied, part of Windows or otherwise - will have to have provisions to
make that selection.

These days (hardware) buffering - in both the adapter and the
connected device - might minimize the need for flow control, but without
specifics about all involved components, it's 'unwise' to assume it's
not needed.

In article , Carlos E.R.
wrote:

At high baud rates, there are two flow control
schemes, and you have to select one of them.

flow control exists at all bitrates and no need to select anything.

Flow control always exists, but which one you choose is crucial. Has to
be the exact same one as the machine at the other end uses. The "no need
to select anything" comes from the usual assumption that all machines
(modems) came with the same settings - except the one that didn't.

no, it comes from the assumption that the cnc software knows what the
cnc hardware needs.

In the past, high speed rs232 would require hardware flow control: the
"in-band" method uses bandwith, after all. I wonder how well can an usb
converter handle it. Badly if is a three wire thing.

don't use a 3 wire cable.

In article , Carlos E.R.
wrote:

Any one know of a USB3 to RS232 adapter THAT ACTUALLY
WORKS RIGHT?

Very likely the problem is not that the adapter does not work, but
that such adapters are normally intended to drive peripherals such as
printers, not CNC machines.

the usb-serial adapters i've used provide a standard rs232 port, which
i used to talk to the console ports on old upses, a managed switch and
a gps, among other devices.

+1

My team and I use usb-serial adapters every day to administer
Enterprise-grade networking equipment via their Console ports. The brand
of the adapter doesn't seem to matter at all. They all just work. I've
never picked one up that didn't work. It's just a standard rs232 port.
There's no special protocol that the adapter itself needs to understand
or support.


That's low speed use, hardly over 9600.

It's 19200 bps, to be exact. What's your point? Did someone add a
requirement for any specific speed to this discussion? The OP is still
quoted above, so it's easy to see that no speed requirements were
provided. The question was, do these adapters work? The answer is yes,
they do.

The point is you have not ascertained those USB-RS232 converters work at
high speed (115000).

they do.

Dan Purgert wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Carlos E.R. wrote:
On 08/08/2019 23.53, Paul wrote:
I have no direct experience with designing in RS232,
just the usual kooky contact with people who know
how to do that stuff. (You know what engineers are like.)
We had to create an rs232 from discrete components at college ;-)
Not transistors, but chips. Ie, doing the job of an uart with plain ttl
or cmos chips.

Now that'd be something interesting to see.

Speaking of "built out of chips" -- did you ever watch the "Breadboard
Computer" series on YouTube? It was great seeing his little blinken-led
machine come together.

Although I have the feeling he was over-simplifying things in some cases
(but if he didn't, I don't think I'd have followed along nearly as
well).

This is my breadboard.

https://www.picclickimg.com/d/l400/p...-evaluator.jpg

Whether breadboarded circuits work,
depends on the edge rate of the logic.

For example, you would not use 74S (Schottky) logic
on a breadboard like that. Too much noise and reflections.

Whereas 4000 series logic would not be a problem.

And I have breadboarded an entire computer on that thing.
It uses a multiplexed calculator display for output viewing.
And this was the processor, a TMS9900. The scary part
was, the computer was *stable*. So the edge rates and noise
level on this, couldn't have been all that bad.

https://en.wikipedia.org/wiki/Texas_Instruments_TMS9900

When I tried to build a frequency counter on the same
breadboard, I had noise problems. (That was using
jelly bean logic.)

Paul

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Paul wrote:
Dan Purgert wrote:
Carlos E.R. wrote:
On 08/08/2019 23.53, Paul wrote:
I have no direct experience with designing in RS232,
just the usual kooky contact with people who know
how to do that stuff. (You know what engineers are like.)
We had to create an rs232 from discrete components at college ;-)
Not transistors, but chips. Ie, doing the job of an uart with plain ttl
or cmos chips.

Now that'd be something interesting to see.

Speaking of "built out of chips" -- did you ever watch the "Breadboard
Computer" series on YouTube? It was great seeing his little blinken-led
machine come together.

Although I have the feeling he was over-simplifying things in some cases
(but if he didn't, I don't think I'd have followed along nearly as
well).
[...]

And I have breadboarded an entire computer on that thing.
It uses a multiplexed calculator display for output viewing.
And this was the processor, a TMS9900. The scary part
was, the computer was *stable*. So the edge rates and noise
level on this, couldn't have been all that bad.

Oh, the YT videos were all discrete chips. Calling it a "Breadboard
CPU" is probably a better description of what the YT guy did.

As I recall, it was something like

- Clock
- PC register
- A & B registers
- ALU
- Some RAM
- EEPROM storage
- Display (7-segment displays)

As I recall, most of thing was built around various iterations of
quad-gate logic ICs (e.g. the 74LS0000 NAND gates).

-----BEGIN PGP SIGNATURE-----

iQEzBAEBCAAdFiEEBcqaUD8uEzVNxUrujhHd8xJ5ooEFAl1Nen 4ACgkQjhHd8xJ5
ooFUwwf/YK5YJ3bBw9B9MXG7ee9wEMKKeYwVPIEQF/aPYMQ1vzWKIeKtzzoyLuhR
qzSQbs4M9lqcADwEHOlecQmSB/i8UReeh9afL1idXuwwMP86LQ2tA/jwv8YZxWRf
9zWJRGob/bdis5bYEImXUeBldjF4AerE1KLlmeoKJ1c2cIUL4ODJ597oh0W jt203
tKegmMXOO/vLyP53/HJ5EBQcu/XwzMgQr6PG3+ZTEJaUOmMXn9DaFP4XjaR4fIhI
lmnJrnFqJ9AGCzSmg46ekQfUvNPBNNCb+XSOrP0bh0ts8rBrKY 0iDvQ+PR/X6sCX
C9WzhN9HxHbAbRA+jAUlaUxaCMqaUQ==
=JDto
-----END PGP SIGNATURE-----

--
|_|O|_|
|_|_|O| Github: https://github.com/dpurgert
|O|O|O| PGP: 05CA 9A50 3F2E 1335 4DC5 4AEE 8E11 DDF3 1279 A281

On Fri, 09 Aug 2019 20:15:44 +1200, Eric Stevens
wrote:

On Thu, 08 Aug 2019 20:32:40 -0500, Char Jackson
wrote:

On Fri, 09 Aug 2019 12:00:56 +1200, Eric Stevens
wrote:

On Thu, 08 Aug 2019 16:26:28 -0500, Char Jackson
wrote:

A CNC machine is essentially a plotter. You're sending a series of
commands that ends with a command to execute. After that, the machine
goes to work. There's no real time control, at least not that I've ever
seen. Once executed, the machine goes to work and lets you know when it
has finished. It's not like you're sitting there, trying to guide it
with a joystick.

That's not always the case.

What's not always the case?

It would help to know what the machine is
and whether the problem is the downloading of a control program or the
direct control of the machine.

If it's a CNC machine, as it was described up thread, there is no
'direct control'. Is there a CNC machine that isn't controlled by a
computer? If so, would it still be considered to be CNC?

In other words you don't know.

Those last two questions were rhetorical.

FYI many (most) cnc machines incorporate their own computer which
controls the machine. The machine can be controlled from the built in
control panel or (frequentlY) from an external computer connected via
variously RS232 or ethernet. Such external control is never direct but
always via the machine's own software. In serious industrial machines
the machining program is always loaded in from outside.

Quoting me: "...there is no direct control..."
Quoting your c/p: "...Such external control is never direct..."

It looks like you're catching on.

All of the discussion around connection speed, latency, and 'direct
control' is misguided because CNC machines don't work that way. You
don't tell a CNC machine to 'move along the X axis until I tell you to
stop". You tell it to move in a specific direction for a specific
distance and it stops by itself when it completes that request. You're
not driving the thing in real time, whether from its own control panel
or remotely. You're simply giving it a set of instructions.

On Fri, 9 Aug 2019 10:20:30 +0200, "Carlos E.R."
wrote:

On 09/08/2019 09.30, Char Jackson wrote:
On Fri, 9 Aug 2019 08:53:14 +0200, "Carlos E.R."
wrote:

On 08/08/2019 23.13, Char Jackson wrote:
On Thu, 08 Aug 2019 16:08:33 -0400, nospam
wrote:

In article , Frank Slootweg
wrote:

I have a customer with a Mac and a w10 laptop and a CNC machine
with an RS232 (only) interface. He wants to be able to send
text files to the CNC machine. The Mac has USB3 and Thunderbolt
interfaces on the back. The W10 laptop has usb3 ports.

Now I could go the the various accessories sites and get
such an adapter, but they are typically trash and don't
work for beans. The customer did state this was his
experience. (Mine too.)

Any one know of a USB3 to RS232 adapter THAT ACTUALLY
WORKS RIGHT?

Very likely the problem is not that the adapter does not work, but
that such adapters are normally intended to drive peripherals such as
printers, not CNC machines.

the usb-serial adapters i've used provide a standard rs232 port, which
i used to talk to the console ports on old upses, a managed switch and
a gps, among other devices.

+1

My team and I use usb-serial adapters every day to administer
Enterprise-grade networking equipment via their Console ports. The brand
of the adapter doesn't seem to matter at all. They all just work. I've
never picked one up that didn't work. It's just a standard rs232 port.
There's no special protocol that the adapter itself needs to understand
or support.


That's low speed use, hardly over 9600.

It's 19200 bps, to be exact. What's your point? Did someone add a
requirement for any specific speed to this discussion? The OP is still
quoted above, so it's easy to see that no speed requirements were
provided. The question was, do these adapters work? The answer is yes,
they do.


The point is you have not ascertained those USB-RS232 converters work at
high speed (115000).

As you now know, they do, but it's still a made up requirement. The OP
didn't say anything about that.

On Fri, 9 Aug 2019 10:22:08 +0200, "Carlos E.R."
wrote:

On 09/08/2019 09.37, Char Jackson wrote:
On Fri, 9 Aug 2019 08:44:50 +0200, "Carlos E.R."
wrote:

On 09/08/2019 03.32, Char Jackson wrote:
On Fri, 09 Aug 2019 12:00:56 +1200, Eric Stevens
wrote:

On Thu, 08 Aug 2019 16:26:28 -0500, Char Jackson
wrote:

A CNC machine is essentially a plotter. You're sending a series of
commands that ends with a command to execute. After that, the machine
goes to work. There's no real time control, at least not that I've ever
seen. Once executed, the machine goes to work and lets you know when it
has finished. It's not like you're sitting there, trying to guide it
with a joystick.

That's not always the case.

What's not always the case?

It would help to know what the machine is
and whether the problem is the downloading of a control program or the
direct control of the machine.

If it's a CNC machine, as it was described up thread, there is no
'direct control'. Is there a CNC machine that isn't controlled by a
computer? If so, would it still be considered to be CNC?

It depends if it is programmed by the computer, then the machine runs
the program on its own, like a plotter would do, or if its directly
controlled by the computer.

Either way, there is no 'direct control'. The distinction you're making
is the difference between sending all of the movement commands before
the program is executed or whether the commands are sent and executed
one at a time. Are we still talking about CNC machines or has the
discussion morphed?


The point is that direct control of a CNC machine over USB-RS232
converter _might_ be problematic.

There is no 'direct control', so there is no problem.

On Fri, 9 Aug 2019 10:31:11 +0200, "Carlos E.R."
wrote:

On 09/08/2019 10.15, Eric Stevens wrote:
On Thu, 08 Aug 2019 20:32:40 -0500, Char Jackson
wrote:

On Fri, 09 Aug 2019 12:00:56 +1200, Eric Stevens
wrote:

On Thu, 08 Aug 2019 16:26:28 -0500, Char Jackson
wrote:

A CNC machine is essentially a plotter. You're sending a series of
commands that ends with a command to execute. After that, the machine
goes to work. There's no real time control, at least not that I've ever
seen. Once executed, the machine goes to work and lets you know when it
has finished. It's not like you're sitting there, trying to guide it
with a joystick.

That's not always the case.

What's not always the case?

It would help to know what the machine is
and whether the problem is the downloading of a control program or the
direct control of the machine.

If it's a CNC machine, as it was described up thread, there is no
'direct control'. Is there a CNC machine that isn't controlled by a
computer? If so, would it still be considered to be CNC?

In other words you don't know.

FYI many (most) cnc machines incorporate their own computer which
controls the machine. The machine can be controlled from the built in
control panel or (frequentlY) from an external computer connected via
variously RS232 or ethernet. Such external control is never direct but
always via the machine's own software. In serious industrial machines
the machining program is always loaded in from outside.

Ok. Thanks.
Then the speed (the latency) of the connection is of no consequence.

We're making progress. Thanks.

Char Jackson wrote:
[...]
All of the discussion around connection speed, latency, and 'direct
control' is misguided because CNC machines don't work that way. You
don't tell a CNC machine to 'move along the X axis until I tell you to
stop". You tell it to move in a specific direction for a specific
distance and it stops by itself when it completes that request. You're
not driving the thing in real time, whether from its own control panel
or remotely. You're simply giving it a set of instructions.

Which needs to be processed or/and buffered. If that processing - by
the receiving device - is slower than the sending speed (baud rate)
or/and the (hardware or/and software) buffering is not capable of
accomodate all the sent data in one go, some kind of flow control is
needed. Which is the point we (YTIW) are trying to get accross.

On 9 Aug 2019 12:11:41 GMT, Frank Slootweg
wrote:

Char Jackson wrote:
On Thu, 08 Aug 2019 16:08:33 -0400, nospam
wrote:

In article , Frank Slootweg
wrote:

I have a customer with a Mac and a w10 laptop and a CNC machine
with an RS232 (only) interface. He wants to be able to send
text files to the CNC machine. The Mac has USB3 and Thunderbolt
interfaces on the back. The W10 laptop has usb3 ports.

Now I could go the the various accessories sites and get
such an adapter, but they are typically trash and don't
work for beans. The customer did state this was his
experience. (Mine too.)

Any one know of a USB3 to RS232 adapter THAT ACTUALLY
WORKS RIGHT?

Very likely the problem is not that the adapter does not work, but
that such adapters are normally intended to drive peripherals such as
printers, not CNC machines.

the usb-serial adapters i've used provide a standard rs232 port, which
i used to talk to the console ports on old upses, a managed switch and
a gps, among other devices.

+1

My team and I use usb-serial adapters every day to administer
Enterprise-grade networking equipment via their Console ports. The brand
of the adapter doesn't seem to matter at all. They all just work. I've
never picked one up that didn't work. It's just a standard rs232 port.
There's no special protocol that the adapter itself needs to understand
or support.

Of course there are protocols,
snip

Of course there are the standard serial protocols, but no special
protocols that the adapter itself needs to understand or support. Once
you load the software, what you see in Device Manager is a standard
serial port on ComX, where X is usually 1-9, and you make all of the
settings there. The adapter itself is transparent. They have no DIP
switches, LEDs, etc. IME, the software for one adapter will control
almost any adapter. When needed, we grab an adapter off of the nearest
desk or shelf, never the same one used previously or the one that came
with the software that was installed, and it just works.

In article , Char Jackson
wrote:


Very likely the problem is not that the adapter does not work, but
that such adapters are normally intended to drive peripherals such as
printers, not CNC machines.

the usb-serial adapters i've used provide a standard rs232 port, which
i used to talk to the console ports on old upses, a managed switch and
a gps, among other devices.

+1

My team and I use usb-serial adapters every day to administer
Enterprise-grade networking equipment via their Console ports. The brand
of the adapter doesn't seem to matter at all. They all just work. I've
never picked one up that didn't work. It's just a standard rs232 port.
There's no special protocol that the adapter itself needs to understand
or support.

Of course there are protocols,
snip

Of course there are the standard serial protocols, but no special
protocols that the adapter itself needs to understand or support. Once
you load the software, what you see in Device Manager is a standard
serial port on ComX, where X is usually 1-9, and you make all of the
settings there. The adapter itself is transparent. They have no DIP
switches, LEDs, etc. IME, the software for one adapter will control
almost any adapter. When needed, we grab an adapter off of the nearest
desk or shelf, never the same one used previously or the one that came
with the software that was installed, and it just works.

agreed.

people are greatly overcomplicating things, most of whom never actually
used an adapter.

the cnc software (or whatever else it is) wants to see a serial port
and the adapter provides one.

On 09/08/2019 17.32, Frank Slootweg wrote:
Char Jackson wrote:
[...]
All of the discussion around connection speed, latency, and 'direct
control' is misguided because CNC machines don't work that way. You
don't tell a CNC machine to 'move along the X axis until I tell you to
stop". You tell it to move in a specific direction for a specific
distance and it stops by itself when it completes that request. You're
not driving the thing in real time, whether from its own control panel
or remotely. You're simply giving it a set of instructions.

Which needs to be processed or/and buffered. If that processing - by
the receiving device - is slower than the sending speed (baud rate)
or/and the (hardware or/and software) buffering is not capable of
accomodate all the sent data in one go, some kind of flow control is
needed. Which is the point we (YTIW) are trying to get accross.


You need flow control way earlier than you hit the internal buffering
problem :-)

The way rs232 hardware works is, the receiving end gets one byte, it
tells the sending side to wait till the CPU inside has read that byte.
This is done in hardware - and originally, with a cable going end to
end. If that line does not exist, then it is done in-band, impacting
speed, but still does not need CPU intervention IIRC.

Ie, it should be handled by the firmware inside the USB-RS232 firmware,
if it is not crap.

--
Cheers, Carlos.

Char Jackson wrote:
On 9 Aug 2019 12:11:41 GMT, Frank Slootweg
wrote:

Char Jackson wrote:
On Thu, 08 Aug 2019 16:08:33 -0400, nospam
wrote:

In article , Frank Slootweg
wrote:

I have a customer with a Mac and a w10 laptop and a CNC machine
with an RS232 (only) interface. He wants to be able to send
text files to the CNC machine. The Mac has USB3 and Thunderbolt
interfaces on the back. The W10 laptop has usb3 ports.

Now I could go the the various accessories sites and get
such an adapter, but they are typically trash and don't
work for beans. The customer did state this was his
experience. (Mine too.)

Any one know of a USB3 to RS232 adapter THAT ACTUALLY
WORKS RIGHT?

Very likely the problem is not that the adapter does not work, but
that such adapters are normally intended to drive peripherals such as
printers, not CNC machines.

the usb-serial adapters i've used provide a standard rs232 port, which
i used to talk to the console ports on old upses, a managed switch and
a gps, among other devices.

+1

My team and I use usb-serial adapters every day to administer
Enterprise-grade networking equipment via their Console ports. The brand
of the adapter doesn't seem to matter at all. They all just work. I've
never picked one up that didn't work. It's just a standard rs232 port.
There's no special protocol that the adapter itself needs to understand
or support.

Of course there are protocols,
snip

Of course there are the standard serial protocols, but no special
protocols that the adapter itself needs to understand or support.

Sigh! I'm talking about *flow control* protocols. Did you even *read*
the part you snipped?

Flow control protocols - such as Xon/Xoff, RTS/CTS, etc. - are
implemented in the adapter and possibly its - driver, etc. - software.

But don't take my/our word for it, take for example the USB-RS232
adapters from FTDI (which Paul already mentioned).

https://www.ftdichip.com/Support/Documents/DataSheets/Cables/DS_USB_RS232_CABLES.pdf

"Future Technology Devices International Ltd
USB to RS232 Serial Converter Range of Cables
Datasheet
...
3.2 Features
...
* Fully assisted hardware (RTS#/CTS#) or X-On / X-Off software
handshaking."

(This datasheet is pointed to by
https://www.ftdichip.com/Products/Cables/USBRS232.htm)

Once
you load the software, what you see in Device Manager is a standard
serial port on ComX, where X is usually 1-9, and you make all of the
settings there.

Yes, most adapters will come with VCP (Virtual COM Port) drivers. But
like with real COM ports, selecting if flow control is to be used and if
so which type of flow control, is part of the API of a COM port.

The adapter itself is transparent. They have no DIP
switches, LEDs, etc.

Some do have LEDs and some might have switches, but that's mostly
irrelevant for the discussion, because flow control selection should be
possible by software,

IME, the software for one adapter will control
almost any adapter. When needed, we grab an adapter off of the nearest
desk or shelf, never the same one used previously or the one that came
with the software that was installed, and it just works.

If your devices "just work", they must fall in the category I
mentioned (and you snipped) - "unless data transfer is purely
uni-directional and the ... device can handle the whole transfer in one
go without any delay" - or/and your adapter - and your device - use(s)
flow control, without you realizing it.

Don't get me wrong: If things work for you, then all the better, but
don't assume flow control isn't relevant for anybody/anything.

FYI, I've been involved in this stuff since the early 70s and have
written driver code for (RS232) flow control.