The introduction of HDMI into the home has been a rather problematic one. Unlike the analog forebears that came before it, DRM came baked into the interface in the form of HDCP, and demands on the quality of the cable were stringent, meaning longer lengths of cabling often caused issues. There were also device to device incompatibilities which were due to variances in HDMI signal transmitters and receivers in devices. Finally, recording HDMI signals was not a possibility until very recently.
That being said, the market has more than “caught up”, the convenience of a one-plug and one cable connection has made it the one port most demanded on new TVs. HDCP compatibility issues are now rare, low cost cables generally “work” and longer lengths are beginning to be less of an issue with special “adaptive” equalization on transmitters and receivers. Despite this, HDMI continues its march with new versions offering higher bandwidths for 3D applications, and the next generation of 4k resolution screens, which may lead to new troubles once again.
That being said, one thing users might find the desire for is to run the signal from one HDMI source to several HDMI sinks. This is useful, say for lecture theatres where you might want to play the same material from a laptop to several projectors or screens simultaneously, or for running things like signage across a room. This could also be useful if you were using an HDMI capture card that had no “pass through” facility, meaning that you couldn’t otherwise monitor what you were recording. Because HDMI is essentially built upon DVI, this solution is equally applicable to repeating DVI signals – say if you want to have two cloned monitors without using two ports on a graphic card.
With the high frequency digital signals, it’s not as simple as analogue buffering of the signal to make a copy of it and because of bi-directional HDCP DRM handshakes and EDID exchanges to set up display timing, it’s not feasible to reliably do this in a “dumb” passive manner like it was with analogue signals. Complying with HDCP requirements require further intelligence, which seems to be the case with the Avermedia Live Gamer Portable capture box I reviewed earlier.
As a result, when I saw a few splitters on sale, I decided to grab two different models to see what they were made of, and how they operated.
The unit was ordered from eBay, and is the more expensive unit, selling for about AU$16. This is still considered inexpensive, especially because I opted for a unit bundling in a power supply. The seller advertises this unit as being 3D 1080p capable, supporting HDMI 1.3 + HDCP 1.0/1.1/1.2 and resolutions 480p/576i/576p/720p/1080i/1080p.
It came in a relatively generic colour box, but the specifications already seem to conflict with the advertised specs. This instead claims HDMI v1.4, which means 4K support, but then it goes to claim 1080p and 3D. I suspect it’s a case of bumping up the number to try and capitalize on a less well educated buyer who might think a bigger number is necessary or “better”.
The rear of the box features some basic specifications, not that it is very meaningful (e.g. analog video bandwidth numbers which seem astronomically out of touch with reality) – some of the text at the bottom especially doesn’t make good sense, along with the common spelling mistakes found on Chinese packaging. For example, “approved by HDMI attached lab” – never heard of those guys. “Cable fixing bolt for long cable, rack mounting kit and wall mounting kit version” – I didn’t find any.
The box also alludes to the unit being used as an amplifier – and I can see its utility there. Even if you didn’t want to split the signal, it’s quite a cheap way to get a signal “regenerated” for another distance. This can be useful if you want to send video a long distance, as often, cheaper cables generally get unreliable around 10m. If you wanted to go 15m, you could risk a cheap 15m cable that won’t work, a very expensive 15m cable that still might not work, or split the difference and go with two 7.5m cables with a “mid span” splitter acting as a “regenerator” which would have a much higher chance of working.
In order to help you make the decision, there seems to be a guide as to the maximum cable length for no distortion – curiously, it’s stated for up to 15m in using the thickest gauge, but only 10m out, which may tell us something about the quality of the output driver in this unit.
The unit comes in a metal shell, as most quality video adapter equipment should for shielding reasons. The exterior seems to be a painted finish, but with a distressed look due to lots of random scratches. This product was sold as “new” – it certainly didn’t look quite like it.
Inputs and outputs to the unit are placed on opposing sides and aren’t particularly cryptic. There is one HDMI input, two HDMI outputs, along with blue link-active indicator lights. Finally, there is a barrel jack for 5v DC, along with an indicator light. External power is required for this unit to operate. Already, the construction is off to a questionable start with a misaligned power jack and erratic LED alignment.
The top shell of the unit is secured to the bottom by four screws – two on each opposing side. The cover “flap” of the lid seems to have a good bend in it too. Hmm. At least it came with a 5V 2A power supply, which we’ll get to later.
Teardown and Analysis
So how does this unit “split” an HDMI signal? We can find out by taking the lid off.
The PCB is marked SP-HDMI 1TO2-A V1.3A, so it’s likely an HDMI v1.3 product.
Lets analyze the HDMI signal path, which relies on two main chips. The signal is fed into the top IC, a Silicon Image PinnaClear SiI9187 HDMI port processor. This chip is intended for use in televisions, to handle the HDMI inputs for the TV. As a result, it has four HDMI 1.3 inputs, an integrated HDCP 1.1 decryption engine, with pre-programmed HDCP keys at manufacture, long-cable adaptive equalization and five EDID RAMs for HDMI and VGA ports.
As this product was intended for integration into TVs, where the signal isn’t “leaving” the sealed unit, it is allowed by HDCP rules to do the decryption. However, the key is in the application diagram, where the port processor has a TDMS output to a DTV SoC. This indicates that the received video is being output practically as HDMI/DVI without encryption.
This chip itself acts as the HDMI sink in the transaction with the source, so as far as your input is concerned, it’s talking to the Silicon Image chip. No mention of 3D support is made though, so it might not really be 3D compatible.
The TDMS output seems to be routed to the other side of the board, and then back to the top to connect to the Pericom PI3HDMI412AD 1:2 Active HDMI Compatible De-Mux with Advanced Re-Driver Functionality. This chip controls both output ports, and is supposed to be used uni-directionally to switch one input to two outputs. It doesn’t seem it was intended to simultaneously drive two ports, however, it may have somehow been abused in the configuration to do so.
Looking at the I2C Truth table, it seems that Byte 2, S1 and S0 are used for port selection, where 01 is for Port A active, 11 is for Port B active and x0 is for Port A = Hi-Z. However, Byte 3 seems to show output port select 01 as Normal, 11 as Test Mode, and x0 as Port B = Hi-Z. Maybe engaging Test Mode on the Pericom chip allows both outputs to be active at once and is an “easter egg” used in this product.
Looking at the PCB, it seems the Pericom chip is being controlled over I2C, maybe if I bothered, I could analyze the bus for more information. This brings us to the last major chip – the ST Microelectronics 8S003F3 Value line 16Mhz 8-bit MCU. This particular chip seems to be present mainly for controlling the Pericom demux, and has a JP2 connection which appears to be some programming interface possibly. It does have a large set of caps and resistors near it – so maybe it has something to do with EDID data somehow as well, but I’m not sure.
As a result, the device is operating like this:
Source <==HDCP==> Silicon Image PinnaClear Receiver A Unencrypted (?) TDMS V Pericom Demux (? Test Mode) V V HDMI 1 HDMI 2
This will likely result in a product which, while seeming to operate with HDCP, doesn’t actually comply with HDCP requirements.
The construction seems to follow a cost minimised approach, with only two PCB hold-down screws used, of the four the casing “expects”. They did use polymer surface mount caps instead of electrolytics, and there is a 3.3v linear regulator, so I suppose the quality of the 5v supply doesn’t need to be too great, which is encouraging. The soldering on the board seems to be of very limited quality, with crooked SMD components suggesting hand-construction. However, there aren’t any filter inductors on the HDMI input or output, which seems to be contrary to good design practices.
The underside of the PCB shows it to be an uncomplicated design, with very little actual need for the secondary layer. However, it seems the through hole components in the HDMI port “hold down” legs are not soldered down and neither are the other ends of the LEDs. This makes the ports fragile. If you are intending to connect/disconnect regularly, I suggest you modify the unit to solder down the ports from top and bottom. The power jack is also soldered but in a crooked alignment.
Also evident are solder splashes which are left behind and best cleared to prevent any possibility of shorting out components, causing failures. Construction quality is definitely lacking somewhat.
The supplied adapter isn’t particularly encouraging as well, with its generic outer case, and “brand-less” label. The case is held together with one screw. It claims to be a 5v 2A adapter.
Yet again, it seems a cost minimised approach is taken, with a paper PCB and a small sprinkling of components with no “real” sophisticated controller ICs in sight. No brand electrolytic capacitors are used on both primary and secondary, and a full wave bridge is built using four discrete 1N4004 diodes, which is pretty much “bargain basement”. Interestingly, the use has been replaced with a fusible resistor instead – maybe for “soft start” reasons … *chuckles*.
The soldering on the power supply is marginal to good, with one hole at the bottom distinctly lacking in solder. I have doubts that the power supply provides good quality power, but I suppose this of limited importance due to the downstream regulation. At least this unit seems to have a clear primary and secondary isolation, along with optoisolator usage, which should mean the risk of electrocution is limited to a failure of insulation in the capacitor bridging the two sides, the optoisolator, or the transformer itself.
My first experiment was to try and determine the graphics modes supported by the unit. To begin, I connected the unit directly to the HDMI output of a laptop using an Intel integrated graphics controller, which resulted in the port LED lighting up. However, this didn’t turn up any detected display, with or without the power supply. It seems there is some intelligence as to the EDID broadcasted by the unit, and it may depend on the connected downstream device to ensure best compatibility.
That being thwarted, I hooked up an HDCP protected source – a Bluray player, to test the HDCP negotiation. As it turns out, while hooking up sources and sinks to the device will cause port LEDs to light up, the unit cannot function without its external power supply. As it turns out, HDCP was successfully negotiated, and the output was successfully seen on the attached TV with no notable degradation of signal.
Now that we know HDCP works on the sink side, I wanted to know if HDCP was active on the outputs. My hypothesis, based on the way the unit is constructed, is that HDCP would not be active. I didn’t have any easy way to verify this, without many non-HDCP capable displays available, so instead I decided to use my Avermedia Live Gamer Portable to see if it will capture from the splitter output.
To my shock, it did actually decide to capture, which it wouldn’t do if the unit was connected directly to the HDCP protected source. This validates my hypothesis on the unit and its construction, and validates this message from Avermedia’s website, which was very cryptic to me.
Just to make it clear – I do not condone the breaking of DRM and violation of copyright laws. Using this product for this purpose may be illegal – check your local laws. This was just a chance discovery, reported in the name of factual research, and is not to be misused! I take no responsibility for any consequences of this information.
A proper, HDCP compliant device should instead follow something similar to the Live Gamer Portable’s design – by having a dedicated HDMI transmitters for each port with integrated HDCP encoder and keys. If the input receiver detects HDCP, then it should “flag” the content, so that the HDCP encryption can be enabled on the transmitter so that both output ports are protected again, albeit with different keys. I believe this is the intention of the HDCP system design, although, from looking at how HDMI and HDCP were implemented in the early days, I think they would rather get rid of HDMI recording or devices with both source and sink behaviour altogether.
One of the two “HDCP” supported splitters I ordered from eBay has been analyzed, and the unit uses a Silicon Image receiver with Pericom demux and ST microcontroller to implement a sink, which has a TDMS output, which is rebroadcasted across both ports. The build quality is somewhat lacking, but the product does work, with some strange EDID behaviour. The unit’s output seem to lack HDCP protection which would make it non-compliant with HDMI standards, and could lead to abuse. I don’t condone this, and a proper design should instead use dedicated HDMI transmitters with integrated HDCP encoder and keys, along with a flagging system to reapply HDCP to outputs when necessary.