Open Source/Freeware Budget USB DAC

Quote:

---

Originally Posted by j4cbo

Don't get on an Airbus A380, Boeing 787, or Airbus A350 (when it is finished), then. :P

---

They all use ARINC-based buses which are ICAO certified for flight use and have specific safety features (and controlled modes of failure) along with manual or mechanical backups for the flight controls. Try again.

On the A380, the Ethernet network is only used for the in-flight entertainment system. And that is fairly prone to crashing as it uses commercial Linux-based embedded systems, but is completely separate from all flight related circuits. It even delayed the 787's flight certification since there was a vulnerability in the passenger side network that allowed access to the flight systems through an exploit.

Hi

I was following this thread since the beginning and maybe I can add my 2 cents comment about transmitting information safely over ethernet.
One way to get safe and reliable transmission is using RTP protocol and a mix of TCP and UDP packet.
You basically send your normal packet using UDP (fast but unreliable) but since RTP packets contain a timestamp in every packet, in case you are missing one packet you can always request it using TCP.
This method requires a bigger buffer on the client side and a circular buffer on the server side but in most case it allow you to sync multiple system using the timestamp and have a reliable transmission.
This is used in some IPTV system in order to get reliable transmission of multicast UDP packet.

I don't say it's easy to put in place but that a way to achieve the reliability and allw the client side to "know" when the information is missing and possibly avoid any "dangerous" behaviour.

The simple way to get "safe" operation is just to shut down the laser if the TCP stream breaks for whatever reason, and have a keepalive ping with timestamp or timestamp-equivalent buffer management, which the etherdream standard supports.

It's not like we're trying to move megabytes per second across the internet here.

Quote:

---

Originally Posted by heroic

The simple way to get "safe" operation is just to shut down the laser if the TCP stream breaks for whatever reason, and have a keepalive ping with timestamp or timestamp-equivalent buffer management, which the etherdream standard supports.

---

I'm not arguing whether it would be possible (albeit not with 100,0% guaranteed reliability, but that's not necessary here). I am curious whether the FDA/CDRH would approve a system with such a mechanism in the chain, instead of a hard-wired remote emergency stop button.

If they don't, you'd still need to run a separate cable to your projectors anyway, and having a mains cable connected to the emergency stop directly is just as effective. It's not like our modern diode lasers need a considerable cool-down period like the old argon heads did, so shutting them down by a hard power cut isn't going to do any damage and is as safe as it could possibly get.

However, in that case, having a DAC on the Ethernet is kind of pointless, since it would only mean moving the challenges to a different location. Instead of having a simple USB DAC with X/Y on differential and R/G/B on single ended wiring over a CAT5, you'd have a remote DAC which runs off your system's built-in Ethernet interface. Effective gain: Nothing, since you can use the same infrastructure (8-conductor twisted pair copper) for either, unless the Ethernet-based DAC offers additional features, like per-projector transforms on frame data or multiplexing several ILDA data streams on a single connection.

I still don't think using an IP stack is the way to go, though, but that's my personal view, since it adds unnecessary complexity, vulnerability and lag when compared to other solutions that have a more direct approach on top of Ethernet.

The ethernet DAC gives you the ability to use an ethernet switch to 'split' the signal, meaning you can rationalize the wiring substantially. You can also use whatever TCP/IP interconnects you need for a given installation- if your booth is thousands of feet away from the projectors in a stadium, for example, you can run over fibre. TCP/IP makes all that stuff Just Work, so it's a big win.

Etherdream also supports transforms and colour lag compensation, done on the DAC. As well as projector management via add-on minicards (thermal control, etc.) and the ability to query the projector to find out what its status is- the ILDA analogue system has *no* sense-back at all, which is one of its major deficiencies.

Quote:

---

Originally Posted by heroic

The ethernet DAC gives you the ability to use an ethernet switch to 'split' the signal, meaning you can rationalize the wiring substantially. You can also use whatever TCP/IP interconnects you need for a given installation- if your booth is thousands of feet away from the projectors in a stadium, for example, you can run over fibre. TCP/IP makes all that stuff Just Work, so it's a big win.

---

The major problem I see with a 'standard' tech like Ethernet is when people other than yourself are doing the wiring. Of course, if you do all the cabling yourself, you know which Ethernet cable goes where and what switches/connectors to plug. However, if your're doing a big stadium show, for example, there's a good chance that your show will be built by roadies who have only basic knowledge on what signals are on what cables, and if a plug fits in a socket somewhere, they'll stick it in.

So what could happen in the Ethernet case is that your laser show DAC-only network gets connected together with the sound or light crew, the admin network or even (chuckle!) bridged over a wireless link. Roadie think: "It's a network after all, what could possibly go wrong, right? Maybe some guy over on the 5th floor loses his internet, big deal."

If the Ethernet network stays confined to a few DACs linked together on a switch, I see little trouble there, but there's no mechanism to guarantee that it will. Even putting big stickers on all of your boxes with "LASER NETWORK ONLY" or "DO NOT TOUCH" won't prevent roadies from treating it as any other network device. If they treat it as a network device and don't try to cram any random RJ45 plug into it. (Oh, sorry, was that a phone line?)

So my concern is not whether or not Ethernet will work, but whether or not it can be made absolutely idiot-proof.

Quote:

---

Originally Posted by Stoney3K

there's a good chance that your show will be built by roadies who have only basic knowledge on what signals are on what cables, and if a plug fits in a socket somewhere, they'll stick it in.

---

And yet you suggest running analogue signals over cat5 is just as good? At least if you're using ethernet it won't actually break anything.

Quote:

---

Originally Posted by heroic

And yet you suggest running analogue signals over cat5 is just as good? At least if you're using ethernet it won't actually break anything.

---

Well, Ethernet doesn't have any voltages that would be damaging to ILDA equipment, neither would the -5V/+5VDC from an ILDA system damage any equipment if it was plugged into Ethernet.

Even more so, when Ethernet is done right, it doesn't have any signaling on two of the 4 pairs on a CAT5 cable (which is deliberate, to allow co-wiring with telephone cables), so if the ILDA variant is wired with the X/Y pairs on the unused wires, it won't break anything either. Of course, an ILDA-over-CAT5 will break if you plug it into a PSTN telephone circuit, but how many of those would you find on stage setups? ;)

Whether or not it would create potentially unsafe situations (unexpected laser power-up) is a different matter altogether, but ILDA was never intended as a safe laser control protocol. It was only designed to control projectors attached to laser heads.

As I said, I am very curious whether the FDA/CDRH would allow any laser safety protocol over a digital link, piggy-backed on the display information. It may not be a very issue in Europe, but it could be a problem in the USA.

I'm not arguing that ILDA hard-wired to a CAT5 is a good idea. I'm just saying that it would be just as useless for a "one wire for everything" solution as Ethernet would be, if the safety/emergency stop information has to be transmitted separately anyway.

Quote:

---

Even more so, when Ethernet is done right, it doesn't have any signaling on two of the 4 pairs on a CAT5 cable

---

Hate to tell you but Gigabit Ethernet (1000base-T) that's now becoming more mainstream in laptops and switches uses all 4 pairs, not only that but with Power over Ethernet that’s becoming more prevalent in enterprise and upmarket switches can supply 48-56 volts at 15-30 and now 60 watts.

Quote:

---

Originally Posted by loopee

Hate to tell you but Gigabit Ethernet (1000base-T) that's now becoming more mainstream in laptops and switches uses all 4 pairs, not only that but with Power over Ethernet that’s becoming more prevalent in enterprise and upmarket switches can supply 48-56 volts at 15-30 and now 60 watts.

---

PoE and 1000Base-T are both extensions on the original Ethernet (10BaseT / 100BaseTX) specification and will only assert their signals once they're 100% sure the remote party supports them. At least, that's how it should work according to the specs, but devices operating beyond specification (or through proprietary extensions, yuck!) are a different issue altogether.