Current VGA adapters are limited to 1920x1200 @ 60 Hz which is < 200 MHz pixel clock. What is the maximum pixel clock of the plugable VGA adapters? Are there other adapters that have higher pixel clocks?
Single Link DVI is max 165 Mhz. So Dual Link DVI would be 330 MHz.
HDMI 2.0 can have a 600 MHz pixel clock. That’s more than enough for 2048x1536 @ 120 Hz.
Graphics cards had 400 MHz (maybe even more than 600 MHz) RAMDAC’s for much higher VGA resolutions or frequencies. How difficult would it be to make an adapter that can equal the capability of a graphics card’s analog DVI-I or VGA connector? I guess cable makers don’t make these because chip makers don’t make the chips? Is it similar to the reason that HDMI 2.0 to dual link DVI-D adapters don’t exist (or vice versa)?
The USB-C to VGA adapter seems to be more capable than advertised. I tried the following resolutions:
800x600@120Hz,83.49MHz
1024x768@120Hz,138.46MHz
1280x960@120Hz,218.31MHz
1400x1050@120Hz,261.63MHz
1600x1200@100Hz,281.15MHz
1920x1440@75Hz,297.82MHz
2048x1536@60Hz,266.95MHz
I haven’t tried to go any higher than the resolutions/frequencies available in Windows.
Thanks for posting these interesting results. VGA pixel clock isn’t something that we’ve attempted to push the limits on, as there hasn’t been much need/ability to collect this information as most VGA displays only support input of 1080p or 1200P at 60 or 75Hz refresh. Since it’s an analog signal, it could certainly scale higher than this, but issues of signal quality, cable quality, and interference become quite problematic.
DVI, then HDMI, then DP were designed to scale to higher modes, which is why these protocols have well-defined ranges and pixel clocks as you mentioned.
Making an adapter equal to the abilities of a GPUs native VGA port might be theoretically possible, but with so few high-end monitors supporting VGA anymore it probably wouldn’t make a lot of financial sense for companies specializing in conversion chips to invest in developing the necessary ICs.
I don’t think there’s a need to collect this information because you made the product and should know the specs of the chips you used already. It might be helpful to the user to know those specs to help them choose which adapters to buy. Sometimes I think there’s a pact between the chip makers and adapter makers to make the reported specs as vague as possible or that might be the fault of the online stores…
The 1080p or 1200p displays you refer to are probably LCD type displays or similar fixed resolution displays. I guess you want to test with those to ensure the pixels from the VGA adapter match with the pixels on the LCD display with minimal blurring into neighbouring pixels. But if you’re going to make a VGA adapter, then testing it on a CRT type display that allows variable resolutions and refresh rates might be a good idea as well.
I’ve done some further testing with the USB-C to VGA adapter. The Intel Graphics Control panel in Windows allows creating custom resolutions without restarting the computer. Pixel clocks as high as 330 MHz worked without issue. There is no signal when attempting to use pixel clocks higher than that. My CRT displayed those resolutions with vary little interference. The intensity of a vertical line test pattern varied (without movement) from left to right but that may have to do with the pixel width shrinking with higher pixel clocks and approaching the dot pitch of the phosphors. I don’t know what the dot pitch of the phosphors is. The EDID of the CRT says 370 MHz max for the pixel clock but I don’t think pixel clock range is meaningful for CRTs unless it relates to the dot pitch of the phosphors?
In Ubuntu, I looked at the DPCD info from the adapter by dumping the contents of /sys/kernel/debug/dri/0/DP-1/i915_dpcd
I believe it says the adapter uses two DisplayPort 1.2 lanes at HBR2 data rate each. This would allow a max pixel clock of 360 MHz (only a little higher than the DAC allows) for 24 bit pixels and 288 MHz for 30 bit pixels but I didn’t try 30 bit pixels. I don’t know how many bits per pixel the adapter will allow as input, and how many bits per pixel are actually converted to analog?
Anyway, I think 330 MHz is quite nice and you guys should mention that in your product specs (if you want to confirm that it’s the limit). Max pixel clock is a very concise way to state the limit. It’s much simpler than choosing arbitrary GTF or CVT or CVT-RB 5:4 or 4:3 or 16:9 or 16:10 30Hz, 60Hz, 75Hz, 85 Hz, 120Hz, timings, resolutions, and frequencies. But I guess you could list a good representative set of those as examples, including ones with large resolutions and others with high refresh rates.
Here’s a correction to my statement where I said that HDMI 2.0 to dual link DVI-D adapters don’t exist. I found an HDMI to dual link DVI-D solution which uses an HDMI to DisplayPort adapter (active) and a DisplayPort to dual link DVI adapter (active). The HDMI port needs to be at least HDMI 1.3 or 1.4 to provide the pixels at 268 MHz in order to get 2560x1600@60Hz.
Some computers may have a single link DVI-D connector which may actually allow pixel clocks that are higher (up to 340 MHz?) than standard DVI-D (165 Mhz) because they are actually an HDMI 1.4 port. To get a dual link DVI monitor to work in this case, a DVI to HDMI adapter (passive) is required to attach the other adapters.
Some OS’s may have problems with those adapters. Ubuntu works well. I think Windows may require 3rd party software to add the desired resolutions because of an erroneous limit in the Intel Graphics Control Panel. Mac OS X may have issues even after the desired resolution is added.
I’ve been looking for a DisplayPort to USB-C female adapter to allow using USB-C alt mode adapters with normal DisplayPort outputs.
I found the “000-USB3-CC-FF” from “PI Manufacturing”. It is a USB-C female to female adapter, but it does not work with USB-C alt mode signals.
I found the “UPD2018” from “SUNIX”. It is a PCIe 3.0 x1 card which adds two USB-C ports (10 Gb/sec). Unlike other similar cards, this card has the added feature of using a DisplayPort input to make a USB-C alt-mode capable port. With this, you can add additional USB-C alt-mode ports to your computer and the DisplayPort signal will not be limited to CPU integrated graphics. This will also benefit users that have a display that uses USB-C alt mode cable but don’t have a USB-C alt mode output. The only drawback is that it requires a PCIe slot (and it’s out of stock).
I would like to see an adapter that takes a USB-A (10 Gb/sec) input and a DisplayPort 1.4 input (HBR3), and creates a female USB-C alt-mode capable output. This should work like the UPD2018 but will use an existing USB port instead of a PCIe slot. Maybe the adapter can use an external power supply for higher wattage power delivery. Maybe this adapter could be a USB-C hub with one or more DisplayPort inputs to create one or more alt-mode capable USB-C ports.
and the UDP2018 and i can’t get any video output, i have the drivers installed and nothing seems to work, display port connected between the UDP2018 and my GPU and the pluggable is connected
I retested it now to be sure it works as expected. This is my setup:
I’m using a Gigabyte GA-Z170X-Gaming 7 motherboard. The UPD2018 is in the PCIEX1_2 slot. A Sony GDM-F520 CRT display is connected to the usbc-vga adapter. The usbc-vga adapter is connected to the USB-C port that is next to the DisplayPort input port of the UPD2018 (because the other USB-C port does not support USB-C DisplayPort alt mode). The DisplayPort input is connected to the second DisplayPort output of an Nvidia Titan X using the UPD2018’s included short DisplayPort to DisplayPort cable.
The first DisplayPort output of the Titan X is connected to a DisplayPort display because it helps to have a known working display to adjust settings for a new adapter. Or you could enable Remote Desktop or Screen Sharing or VNC.
I used a custom resolution of 2048x1536@73 Hz (329 MHz GTF timing) since 330 MHz is the max for the usbc-vga adapter.
For macOS 10.12.6, I used SwitchResX to create and select the custom resolution.
In Windows 10, I used the Nvidia control panel to create and select the custom resolution. The Device Manager shows two USB-Serial devices connected to the “ASMedia USB3.1 eXtensible Host Controller” of the UPD2018. I guess those are used for updating the firmware of the non ASM1142 chips of the UPD2018. The USB-Serial devices have a Code 10 error, but that doesn’t affect the normal functioning of the UPD2018 (DisplayPort or USB data transmission).
In Ubuntu 17.04, I used “Software Updates” to switch to Nvidia binary driver in the “Additional Drivers” tab. Then I used the “gtf” command to calculate a new custom resolution for the CRT, and the “xrandr” command to add the new mode and select it.
You must ensure that the resolution you choose has a pixel clock under 330 MHz, and is also within the horizontal (kHz) and vertical (Hz) limits of your CRT (which should be identified in the CRT’s EDID information). CRT’s usually use GTF timings. CVT is similar to GTF. An LCD may use CVT reduced blanking timings. Reduced blanking means there’s less time between the end of one frame or line and the start of the next frame or line.
UPD2018 is in the second PCIe x16 (x8 mode), the usbc-vga adapter is plugged into the USB-C Port next to the DisplayPort input, the DisplayPort is connected to the 2nd or 3rd DisplayPort on a GTX 1070, all ports are verified to work on the 1070, i am using the UPD2018’s included cable DisplayPort Signal…
I see nothing in Nvidia control panel for the adapter or anything, all i see is my main DisplayPort monitor, and my monitor is showing no input signal (monitor works when connected to a GTX 750Ti DVI-I port.
nevermind, i don’t know what i did pulling it out and putting it back in, but it seems to work fine now, got my CRT monitor working at 1920x1440@75Hz off my GTX 1070.
pulled it out, took of the the backplate put it in, nothing
put backplate back on, put it in, tightened it down well, works (though annoyingly the BIOS screen goes to the CRT)
That’s good news. If you’re CRT supports it, you could even do 1920x1440@82Hz (328.2 MHz, 123.7 kHz) through the adapter.
The UPD2018 DisplayPort function is completely invisible to everything since it is merely a passthrough, so you will always be dealing with the originating DisplayPort connector of the graphics source (your Nvidia card’s DisplayPort output).
It would be preferable to use one of the PCIEX1 slots of the Asus Crosshair VI Hero, but that is not possible if you’re using the PCIEX4_3 slot at more than X1. You can force PCIEX4_3 to run at X1 but if you prefer it to run at X2 or more, then you have to use PCIEX8_2 which turns your first slot from x16 to x8. Well, x8 doesn’t reduce the performance of a graphics card by much anyway.
Actually, you would choose the PCIEX8_2 slot because it is PCIe 3.0 and all the other slots are only PCIe 2.0 which would force the maximum USB throughput of the UPD2018 card to 5 Gbps instead of 10 Gbps if you cared about USB 3.1 gen 2 performance. The slot does not affect DisplayPort performance since that is unrelated to the PCIe slot (and in fact the card will do DisplayPort conversion without connection to a computer’s PCIe slot).
The M.2 Socket 3 can be converted to PCIe 3.0 x4 with an adapter, but it would be awkward trying to connect a PCIe card to that. Same for the M.2 Wi-Fi slot, but I’m not sure if that’s PCIe 3.0, or if the BIOS would need modification to allow a non-Wi-Fi PCIe device to be connected there.
Unfortunately my CRT is limited to 1920x1440@75Hz MAX, it won’t even do 76Hz, i do have 2x CRTs that will do 1920x1440@85Hz, but one of them needs repair, the other had an attempted repair that i have not tested, I used to use both but replaced both with my current 27" 1440p IPS panel and Dell P1110 CRT.
I tested it successfully in a MacPro 2008 PCIe 2.0 slot and a PCIe 1.0 slot (2.5 Gbps max). You have 4 PCIe 2.0 slots to try.
My Sony GDM-F520 also has issues with the display occasionally flickering off momentarily. I guess they’re just not made to last forever. This is unrelated to the connection or adapter since of the onscreen menu will flicker off with no connection. It’s probably a worn out component on the CRT board but I don’t have much motivation to get it repaired while it still mostly works (at least well enough to test adapters). I use LCD displays now which have higher resolutions (2560, 4K, 5K). Their 60 Hz refresh rates are not annoying like on a CRT.
nevermind, part of this whole setup is a flaky piece of trash, just turned my computer back on and it is back to not working at all.
it looks like it is flaky in that i have to take it out and put it back in sometimes, only seems to work in one of the slots, i might try a PCIe riser cable with power and put it outside the computer
do you think i could get a USB-C Female to Female and a USB-C to displayPort cable and hook up the adapter that way?
If it’s flaky inside the computer, then it would probably be flaky outside the computer.
I don’t know of a USB-C female to female adapter that will work. The one I tried only worked for USB 3.0 signals, not DisplayPort.
The document “USB-DevDays-DisplayPort-Alternate-Mode-2016-final4.pdf” says that a USB Type-C to DisplayPort Adapter Cable is supposed to be reversible, but I don’t know of any that have a female USB-C connector. It’s probably not an allowed configuration? It says that Adapters and Adapter cables have a USB-C plug (not a receptacle).
When i first turn my computer on it won’t work, shut it off, just pull the card out and put it back in and it works when i turn it back on, having it external i could kill power at any time, and could keep the card on 24/7
I was thinking a DisplayPort to USB-C cable from the GPU, to a Female to Female USB-C adapter, and then the USB-C to VGA, and flip cables untill it works.