content/opentally.rst

Sat, 07 Nov 2020 15:45:09 +0100

author
David Douard <david.douard@sdf3.org>
date
Sat, 07 Nov 2020 15:45:09 +0100
changeset 130
56ffb9577050
parent 128
aba381b2bac9
permissions
-rw-r--r--

Fix tags for HP 34970A posts

================
 OpenTallyLight
================

:Author: David Douard
:Category: Electronics
:Tags: ITC100, Tally Ligth, Arduino, Firmata


A month ago, a friend of mine called me to ask for help with a technical device
that seemed not to work properly.

He was testing with a friend some video recording setup. They use vMix_ to mix
(live) video streams from 4 (or more) vide cameras (Panasonic GH5 if I'm not
mistaken). They also have a DataVideo ITC100_ intercom system. Now the problem
was: how to control tally lights via the ITC100_ from vMix_?

Surprisingly there are not so many affordable solutions on the market.
Eventually, they found one device, made by a small US company, that would fit
the needs. It's the setup of this very device that was not working properly
when they called me.

TL;DR;
======

Go to the `project's gitlab page`_ if you want the kicad project files.

_`project's gitlab page`: https://framagit.org/douardda/opentallylight


The non working tally light controller
======================================

The device itself is a small-factor plastic box with a USB B connector, a LED
indocator, plus a pair of HD15 connectors (for the connection with the
ITC100_).

The first unsual fact was the LED indicator was not on. But that not the only
bizarre stuf:

- the USB device was sometimes identified by the PC, sometimes not (and no it
  was not a bad contact),

- when plugged to the ITC100, some tally lights immediately lighed up but quite
  dimly,

- and no attempt to change the current and preview streams did not change
  anything on the tally light side.

So I grabbed the device home, opened it up and tried to identify a possible
problem. At that time, we knew the PCB had a problem (because of the LED
staying off), and the seller sent a new PCB (without the case to save a bit on
the shipment), so I was not too worried of hacking the dead PCB.

I discovered a short circuit between the pads of the LED, and this was
disturbing the 5V USB supply. In fact, depending on the hability for the USB
ports used on the PC to deliver enought current to actually power the
microcontroller or not, the device was seen by the PC or not.

I though this was the problem, so I just removed the LED and it's limiting
current resistor: no more over current on the 5V rail, the device was now
properly recongnized by the PC every time weplugged it in.

I also tried to understand a bit the circuit which was in fact dead simple: a
PIC18F4553 microcontroller is presented as a HID device to the PC and control
directly (via a resistor) the pins of the HD15 connectors.

So I tried to 'reverse engineer' the "control" protocol and since it was a
simple HID device, I just wrote a small Python script to try all the possible
addresses on the first 2 bytes, using pyhidapi_. And I quickly discovered how
to control individual pins connected controlling the tally lights (via the
ITC100_). On my Linux machine with a simple LED connected directed to the HD15
connector, it was looking like it was working.

.. _pyhidapi: https://github.com/apmorton/pyhidapi

But once plugged on the ITC100_, it was still not working. Same erratic
behavior, same total absence of actual control of the tally lights.

This time I had a multimeter with me, so I took measurments on the ITC100_
side: each tally light signal was at 12V! and had to be pulled down to ground
to light the corresponding tally light on.

Since the device was attempting to control directly these signals from the PIC,
no way this could have ever worked. And this was sold something like 650US$
plus shipment...

A first prototype
=================

So I explained to my friends there was no possible way of making this device
work as intended: it was just not designed correctly. Meanwhile, one of them
told me that vMix also have support for controlling tally light via
Arduino-based setups using the Firmata_ protocol.

.. _Firmata: https://firmata.org

Since the fiasco with the "professional" device they bouight, they were really
on the urgent need for a working solution, like for the day after, so I spent
the evening building a very rough prototype.

I wanted to provide them a better solution, besides a working one. Like with
proper galvanic isolation between the microcontroller and the ITC100_ part of
the circuit.

Since I had no time nor enough optocouplers in my stash of junk parts, so I
just used regular non isolated 2N2222 to control the ITC100_ tally light lines.

.. image:: {static}/images/opentally/proto.jpg
   :alt: A very rough protoype of OpenTallyLight

It's a very crude protoype, but i worked! Believe it or not, but they have been
able to record several events with this board!

A (slightly) better protoype
============================

Since we validated the Arduino and Firmata based approach, I proposed them to
design and build a better device.

At first, I kept the idea of using an existion arduino compatible board and
only design a daughter board for the optocouplers and the connectors.

I chose the Sparkfun Pro Micro: it was one of the smallest board with enough
pins to drive the 8 tally lights (each with 2 LEDs, the green for preview
signaling, and the red for signaling the recording in progress).

So I designed a first PCB (in fact several of them, but meh) using Kicad_.

In the mean time, I also spent quite some time looking for a nice aluminium
enclosure for the PCB. I finally found a small extruded one for a 80x50mm PCB
on aliexpress. It was perfect, so I ordered 5 of them.

I also decided to replace the HD15 connectors by RJ45 ones, then assemblng
custom cables from shielded RJ4 Cat6 ones.

So I ended up with this design:

.. image:: {static}/images/opentally/opentally-pcb-1.jpg
   :alt: A slightly better design of OpenTallyLight


Looks neat, but... I did not realize I put the RJ45 socket backwards! Doh!

.. image:: {static}/images/opentally/proto2.jpg
   :alt: A (not so) better proto of OpenTallyLight

In order to test is nometheless, I put the Pro Micro on the back side of the
PCB (otherwise I was not able to plug the RJ45 cable in).

Unfortunately, if this board worked fine on Linux (I mean the Firmata protocol
part, since vMix does not work on Linux), the Firmata protocol on Pro Micro is
not working on Windows. vMix was not able to establish the connection with the
Pro Micro, neither did the Firmata test program available on the Firmata_ web
site. I'm not sure why it did not work on Windows, but I had no time nor the
desire to debug this Windows evil misbehavior.

I looked in my tray of development boards, and found a few Nucleo ones that are
supported by the Arduino environment and have almost compatible pinout with the
Pro Micro. I had to budge a pair of tracks on the PCB, but I could make it work.

OpenTallyLight
==============

So I designed a new version of the board. This time I stick to a good known
working pair: an ATmega328P + FT232RL based solution (the Pro Micro is based on
the ATmega32U4 which embeds the USB controller).

I wanted to use a USB B connector rather than a micro USB one (too fragile for
this kind of environment), and since I also wanted the design to stay small
enough to fit in my cute little enclosures, I decided to implement the arduino
on the PCB itself.

The PCB design looks like:

.. image:: {static}/images/opentally/opentally-v3.jpg
   :alt: A better design for OpenTallyLight

Which gives, in real life:

.. image:: {static}/images/opentally/front.jpg
   :alt: Front side of the PCB

.. image:: {static}/images/opentally/back.jpg
   :alt: Back side of the PCB

And the result once installed in the enclosure:

.. image:: {static}/images/opentally/complete-unit.jpg
   :alt: The completed OpenTallyLight

.. image:: {static}/images/opentally/enclosure.jpg
   :alt: The completed OpenTallyLight (enclosure front)

.. image:: {static}/images/opentally/enclosure-2.jpg
   :alt: The completed OpenTallyLight (enclosure back)


The good news is it worked immediately. The bad news is I forgot to put an ICSP
connector on this PCB. So I had to burn an Arduino bootloader on the ATmega
chip before I soldered it on the PCB.










.. _vMix: https://www.vmix.com/
.. _ITC100: https://www.datavideo.com/eu/product/ITC-100
.. _kicad: https://kicad.org

mercurial