DVI-to-HDMI converter for HD4650

[chrisr]

Hi,

I own a Radeon HD4650 graphics card with 2 DVI ports and an on-board HDMI audio device. Unfortunately, this card did not come with a DVI-to-HDMI adapter, and I now understand that I will need a special DVI-to-HDMI adapter if I want to receive the HDMI audio signal. Does anyone know if this DVI "specialness" is proprietary to ATI, please? What are the distinguishing characteristic of the adapter, so that I can know that I will be buying the correct one?

Thanks,
Chris

[eslrahc]

AFAIK all dvi to hdmi adapters are the same and will provide both video and audio.
The adapter that came with my gigabyte 3870 is generic.

The more important question to ask is does your your card support either/both if using an adapter.
Mine does but only if connected to dvi 1, but there are some 4670 models which do not support using an adapter such as the Powercolor PCS AX4670 1GBK3-P or the Diamond
4670PE31GDT.

Charles

[chrisr]

AFAIK all dvi to hdmi adapters are the same and will provide both video and audio.

That's strange, because DVI doesn't typically doesn't carry audio. That's why I assumed that ATI had "enhanced" the DVI connectors in the HD4xxx cards in some non-standard way. There are certainly adapters for sale which explicitly state that they do not support audio.

The more important question to ask is does your your card support either/both if using an adapter.

I checked that first: the box states that HDMI support is built in, and "enabled with an intelligent DVI to HDMI adapter". And it is a Powercolor card. Although I would have thought that a generic DVI-to-HDMI adapter would always be able to provide video.

[droidhacker]

Read this: http://www.hardforum.com/showthread.php?p=1031650868

Basically, it *should* be implemented to use the link negotiation to determine audio capabilities, but at least in the RHD2000 series, TMDS signal 3, 4, 5 pins (4, 5, 12, 13, 20, and 21 on the DVI plug) need to be connected to ground in order to enable audio-over-dvi.

I'm not sure about the RHD3000+ -- it may work with a regular adapter, maybe not. They are *definitely* not the same adapter as for the RHD2000 series.

A few points;
1) The cable must be of sufficient quality to carry the necessary bandwidth for audio+video (it will disable stuff if it doesn't all carry),
2) There is virtually no difference between DVI and HDMI. They are pin identical except that HDMI lacks a few connectors (i.e., HDMI = single link DVI -- note that HDMI-DL is spec'd but nobody anywhere uses it).
3) It *may* be possible to software-enable audio-over-dvi where the special plug is not available.
4) There is nothing fancy in the adapter itself that multiplexes signals -- the ONLY thing that the adapter might do is connect things that aren't normally connected in order that the card can detect that it is present.
5) Did I mention that doing this is completely redundant? I think its a scam intended just to separately sell the adapter plugs.
6) You can avoid the need for a special plug (at least on RHD2000 series) by connecting/soldering pins 4, 5, 12, 13, 20, and 21 on the video card's DVI plug to ground. Doing this will permanently identify the plug as supporting HDMI audio, and will also permanently disable dual-link DVI (if so equipped). Note: If you do this, do NOT NOT NOT EVER plug in a dual-link DVI cable. It probably won't fry anything since the pins will be pulled to ground, but probably better not to try it.

You might note this product: http://www.hisdigital.com/un/product2-52.shtml
If you look at the pins, you can see that it HAS all the pins required for dual-link DVI-D. Usually, if it only needs single-link (as per HDMI cables), then it will actually not physically HAVE pins 4, 5, 12, 13, 20, or 21. So why are they there?

One of the things I have been considering is the possibility that part of the reason for the enabler/adapter might actually the the CEC pin -- for those who don't know, this is basically a wired remote control allowing one device to send control signals to another device, i.e. change volume or channel.

Would be nice to have one of the AMD guys clarify the requirements for enabling audio-over-dvi for RHD3000+.


***What would be REALLY nice is if someone who HAS one of these 3000 series adapters would trace the way it is WIRED and post the scheme, i.e., which pins are connected to ground, which pins are connected to each other, which pins are not connected.

[droidhacker]

I've got a little bit more interesting information from this thread: http://www.avsforum.com/avs-vb/showt...2#post13911462 post #320 -- it says that the "old" adapter for 2000 series works for 3000 series *driver dependent* -- which suggests to me that this is something that can be enabled in SOFTWARE.

[droidhacker]

I'm *almost 100%* certain that the adapter is redundant for audio-over-dvi for the open source drivers. The main clue is in the radeonhd driver having option "HDMI", which indicates to the driver that the device is HDMI and thus enabling audio. Combined with evidence that audio support varies depending on the driver used, I would say that this is proof that the switch that is triggered by the "special" adapter is in the driver itself. In the open source driver, this can obviously be overridden to enable audio regardless of the presence of the "special" adapter.

[Hephasteus]

They just interleave the audio and video signals and up the clock rate to compensate. Hence the cable quality issue. Or they hijack some of the grounding pins?

[droidhacker]

They just interleave the audio and video signals and up the clock rate to compensate. Hence the cable quality issue. Or they hijack some of the grounding pins?

No pins are hijacked. That would be stupid since it wouldn't be compatible with your TV like that.