Comparators for LVDS use? / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Comparators for LVDS use?

Move Thread

LAN_403

Conundrum

Mon May 03 2010, 05:10PM

Registered Member #96 Joined: Thu Feb 09 2006, 05:37PM
Location: CI, Earth
Posts: 4061

Hi all.
Trying to locate a quad comparator here that will work with LVDS.

the plan is to "clone" the LVDS signal, taking the differential and converting it to a single drive suitable for an LED, then demodulate it at the far end of the fibre.

the problem here is that even an MC34004P isn't fast enough, the signal is distorted beyond hope even though there is a clear change in the output with different colours on the screen (handy if you want an ambilight clone but not particularly useful).

looking at the datasheet it seems that the chips i have are nowhere near fast enough (200ns versus 20nsec based on 50MHz max)

any ideas for a better chip?
regards, -A
any ideas?

Steve Conner

Mon May 03 2010, 06:29PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

If it helps, you can connect a LVDS signal straight into those GBIC fibre optic transceivers, and they're cheaply available on Ebay.

LVDS for flat panel displays runs at hundreds of megabits per second. For each tick of the pixel clock, each LVDS channel has to transmit 6 or 8 bits. The new embedded clock LVDS is even worse, it rams everything down one channel.

So some random comparator chip probably won't be fast enough. Even if it was, some random LED probably won't be fast enough to transmit it faithfully: as far as I know the GBICs use laser diodes.

mikeselectricstuff

Mon May 03 2010, 11:14PM

Registered Member #311 Joined: Sun Mar 12 2006, 08:28PM
Location:
Posts: 253

..erm why not use a proper LVDS receiver chip...?

klugesmith

Mon May 03 2010, 11:55PM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1716

I'll second what Steve said.
Commercial fiber-optic transceivers with low voltage diff inputs and outputs are getting cheap on the surplus market.
Form factors include 1x9 and GBIC for dual SC fiber connectors, and SFF and SFF-Pluggable for dual LC connectors.
(In a previous job I had to qualify these parts. Today I know people working in labs with literally buckets full of GBICs and SFFP's ).

You would have trouble finding a optical datacomm transceiver rated less than 1 gigabaud.

I still think of LVDS broadly as a voltage signaling standard, not narrowly as a display interface with a certain number of lanes, a serial multiplexing protocol, etc. How many LVDS lanes do you need, over what distance, with what voltage/power/space available for fiber interface? Maybe I can get you some free gbics and sockets.

-Rich

Wolfram

Tue May 04 2010, 09:05AM

Registered Member #33 Joined: Sat Feb 04 2006, 01:31PM
Location: Norway
Posts: 971

Steve McConner wrote ...

If it helps, you can connect a LVDS signal straight into those GBIC fibre optic transceivers, and they're cheaply available on Ebay.

Yes and no. Even though the interface on the GBICs is very much like LVDS, the differential voltage swing is a bit more like LVPECL. Depending on which standard you read the minimum differential input swing is from 500 to 650 millivolts. The differential output swing of LVDS transmitters is typically 200-400mv. Directly connecting them would probably work in practice, but it's not guaranteed.

Anders M.

Conundrum

Wed May 05 2010, 07:04PM

Registered Member #96 Joined: Thu Feb 09 2006, 05:37PM
Location: CI, Earth
Posts: 4061

interesting stuff, thanks guys.

I seriously underestimated the amount of data being transferred. In some ways its amazing the opamp did anything; presumably its still acting as a differential amplifier even though most of the high frequency stuff is being "lost"

That said, this technique of comparing raw data could allow a direct VGA to RGB LED converter to be built (display low res picture on second VGA channel and "decode" it to LEDs by comparing with syncs) would be sucky resolution but might give maybe 32*32 on a good day.
all that is needed is a suitable eeprom on the DDCA channel to make the pc output vga as many don't now unless monitor is plugged in and working 100%.

EDIT:- does anyone know if the LM1881 will work at near VGA speeds?

-A

"Bother" said Pooh, as the laws of physics smacked him with a clue by four...

Bjørn

Wed May 05 2010, 07:38PM

Registered Member #27 Joined: Fri Feb 03 2006, 02:20AM
Location: Hyperborea
Posts: 2058

The LM1881 can be programmed for different frequencies by a resistor. It is not the most stable solution unless you are willing to filter the video signal before feeding it to the chip.

Conundrum

Thu May 06 2010, 06:20PM

Registered Member #96 Joined: Thu Feb 09 2006, 05:37PM
Location: CI, Earth
Posts: 4061

hmm, ok.

please note. see my thread in the "General chatting" forum...
regards, -A

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob