VGA, RS232@1Mbps on optical fiber

[Berni]

Well in that case you are forced into using a raw hardware solution.

Even the professional video over network solutions likely at the least run a RTOS inside it to handle the network stack. Those USB video capture dongles also tend to run a fair bit of firmware inside of them, fully possible there is a RTOS in there too.

[berke]

It's a complex question to explain, basically I got a written permission to conduct a couple of personal experiments in the laboratory, but you are not allowed to introduce *anything* that runs an unauthorized operating system. Even the laptop I can take with me to the lab has neither USB stick (1) nor hard drive, it boots from the LAN and can't access anything other than the local infranet. I'm not even allowed to take it out of the lab for the lunch/coffee breaks.

With these kinds of security concerns I'm surprised that you don't seem to have qualms about using random PLA-sponsored electronics off of Aliexpress.

Anyway, what kind of length are we talking about and are you willing to use multiple fibers?

[Yansi]

You could  convert VGA to SDI video over fiber.  Chinese guys sell even modules with some 485 sidechannel - not sure about features, never touched that thing.

[DiTBho]

what kind of length are we talking about and are you willing to use multiple fibers?

approximately ~<1200m of cable

[berke]

approximately ~<1200m of cable

With a 200€ budget??

[DiTBho]

approximately ~<1200m of cable

With a 200€ budget??

yup, I don't have to buy the optic fiber cable, there are already several "spools" (1000m, 2000m, different connectors) in the laboratory, and I can use them.

[DiTBho]

I'm tempted to buy one of those toys on Amazon (224 Euros shipped from Shanzen) with the written promise at the time of purchase that I can "return" it if not satisfied because it doesn't conform to specifications.

However, I will definitely buy some optical modules, at least to make a synchronous fiber optic serial, just to learn.

[Yansi]

Still, for 200€ you will barely buy the two SFP modules and accessories, not talking about other components, if you want to build it yourself.

For the VGA <--> SDI variant, it should be buildable on your own, but i would at minimum double your budget, maybe even tripple it. (And still not counting the hourly rate for developing it).

So, I really not see an issue building such a thing, the issue may be your expectations of price. 200€ is a price you could possibly buy a mass market solution (if one would exist), but you will definitely not buy the whole development for it.

Do not buy "some modules", buy the ones you could build the whole setup including the video. It will be cheaper than running two separate links, occupying two or more fibers.

As I've suggested, chinese folks sell the SDI<-->SFP fiber converters with some kind of RS485 support, all bidirectional over a one single-mode fiber (WDM).
For example this: https://www.aliexpress.com/item/1005005901467364.html
Plenty of solutions to convert SDI<>VGA (or other interface).

I am still not sure what u r trying to achieve here (have not followed that carefully), but have a feeling you are overthinking the required solution.

[DiTBho]

have a feeling you are overthinking the required solution.

so, we have to transmit VGA and RS232 over 1000 meters away, and it must be done on optical fiber because it is an environment with very high electromagnetic pollution. There are big motors, coils and things that even require lead shielding.

How can you say - overthinking the required solution - when I clearly said that I have ZERO knowledge about the cost of optic things, as in this specific case, becasue I am paying with my own pocket for the whole things? (freelancer contract)

[Nominal Animal]

Within the stated constraints, I do believe a video ADC like TI TVP7002 (HTQFP-100, 7.71€ at Mouser in singles, uses 3.3V and 1.9V supplies) combined with an FPGA is a must.  It is what is used in OSSC, the Open Source Scan Converter (low-latency video digitizer and scan conversion board used by many retro video game enthusiasts), and the video quality should be quite good.

VESA 1024x768 60fps mode uses a 48kHz VSYNC and 60MHz pixel clock (1344 pixels per scan line).  Basic 640x480 60fps has a 31kHz VSYNC and a 25MHz pixel clock.  These are all well within the TVP7002 capabilities.  TCP7002 is controlled using I2C, so you also need a small microcontroller to adjust configuration and parameters, while an FPGA selects the parallel RGB data to be sent, mixing in the UART channel(s), and uses a serdes or PHY (XAUI/SFI) to interface to a SFP+ fiberoptic module.

The serdes/PHY is a problem.  It converts parallel data to the GHz-range serial (differential) pair the SFP+ module uses, and vice versa.  Some FPGAs like Lattice ECP5 do have built-in serdes for 3.2Gb/5Gb for up to four lanes; you need two lanes, one in each direction.  Only the ECP5UM and ECP5UM5G series have the serdes, and they're BGA only, but they might be able to drive an SFP+ module directly (see here, XAUI mode).

Using a 10GBase-SX SFP+ module and layer 1 (I believe!) Ethernet (jumbo) frames for the data, for 38 bytes of overhead per 46-9000 payload data bytes, the data would be proper 10G Ethernet stuff you could route within a LAN.  If a few first bytes identify the payload, you can use the same format on both legs of the transmission.  Note that on the other direction, you do need to buffer some UART bytes, as the overhead is a bit much (8400%) otherwise; i.e. burst, instead of send/receive individual bytes.  I'd include a few header bytes in the payload, identifying whether it contains a full scanline (and if so, which scan row and frame/field number), and whether one or more UART payload bytes follow.  Or you could use UDP/IPv4 datagram(s) within the Ethernet frame, so different ports would be used for different UART connections, making the format IPv4-routable, and easier to extend (for example, more than one UART link).

Let's say you limit to a 1Gbits/second bandwidth instead.  The practical data bit rate including Ethernet frame overhead is about 800 Mbit/s.  This suffices for 1024×768 24bpp 30 fps (a little under 567 Mbit/s, plus framing overhead, plus the UART data).  This would require you to have a full framebuffer of about 19 Mbits (2.4 Mbytes), because the VGA input data rate is about twice that.  The other option is to transfer 60fps interlaced, every other scan line per field, so that only a few scan lines need to be buffered at a time, also leading to much lower latency; but also leading to tearing, since the odd and even scan lines would be from different consecutive display frames.  It would really be 1024×384 60fps, in other words..  The FPGA would act as the MAC, reading the TVP7002 parallel output selecting and buffering only the visible part of each scan line, while pushing the previous one a byte at a time at 125 MHz to a gigabit GMII PHY, followed by a media converter to fiber (1GBase-SX).

I think, that is.  I am not absolutely certain of any of the above, not having done it myself before; the FPGA part in particular is new to me.  The state diagram isn't complicated, unless you include stuff like mode autodetection; using a separate microcontroller to measure HSYNC/VSYNC and control and configure both the video ADC and the FPGA should make things much simpler, modular.

[Yansi]

First, could you please tell us what is the end application?  Why do you need to transfer VGA video over such distances? This is rarely being done, as there are often better solutions around that problem. Extending video over such distances is rarely needed, hence the lack of quality mass produced solutions. Moving video over such distances is primarily only done in the video broadcast industry (which I am slightly familiar with, hence my quick SDI suggestion).

The other issue you have is your attitude dismissing all offered solutions, without giving a reason. So again, could you please be more specific about the application? We may then very well be more specific with suitable solutions to your problem.

For remote access to computers, there are much better solutions like remote desktop, to access remote PCs over the LAN/internet. But again, we do not know what is the source of the VGA (XGA 1024x768) video. Also, plenty of mass produced HW and SW solutions exist for transferring serial data over the LAN/internet.

Nobody is being angry, about your zero knowledge, people just may be irritated that you are denying offered help without providing further detail as of what is needed.

If this is a freelance contract, why are you paying with pocket money? Does not make much sense to me. Just make your contract in a way, they will pay for all including your time spent. I would never agree doing any contract work without being paid. Or did you mean a "hobby personal project" instead?

[Yansi]

Within the stated constraints, I do believe a video ADC like TI TVP7002 (HTQFP-100, 7.71€ at Mouser in singles, uses 3.3V and 1.9V supplies) combined with an FPGA is a must.  It is what is used in OSSC, the Open Source Scan Converter (low-latency video digitizer and scan conversion board used by many retro video game enthusiasts), and the video quality should be quite good.

VESA 1024x768 60fps mode uses a 48kHz VSYNC and 60MHz pixel clock (1344 pixels per scan line).  Basic 640x480 60fps has a 31kHz VSYNC and a 25MHz pixel clock.  These are all well within the TVP7002 capabilities.  TCP7002 is controlled using I2C, so you also need a small microcontroller to adjust configuration and parameters, while an FPGA selects the parallel RGB data to be sent, mixing in the UART channel(s), and uses a serdes or PHY (XAUI/SFI) to interface to a SFP+ fiberoptic module.

The serdes/PHY is a problem.  It converts parallel data to the GHz-range serial (differential) pair the SFP+ module uses, and vice versa.  Some FPGAs like Lattice ECP5 do have built-in serdes for 3.2Gb/5Gb for up to four lanes; you need two lanes, one in each direction.  Only the ECP5UM and ECP5UM5G series have the serdes, and they're BGA only, but they might be able to drive an SFP+ module directly (see here, XAUI mode).

Using a 10GBase-SX SFP+ module and layer 1 (I believe!) Ethernet (jumbo) frames for the data, for 38 bytes of overhead per 46-9000 payload data bytes, the data would be proper 10G Ethernet stuff you could route within a LAN.  If a few first bytes identify the payload, you can use the same format on both legs of the transmission.  Note that on the other direction, you do need to buffer some UART bytes, as the overhead is a bit much (8400%) otherwise; i.e. burst, instead of send/receive individual bytes.  I'd include a few header bytes in the payload, identifying whether it contains a full scanline (and if so, which scan row and frame/field number), and whether one or more UART payload bytes follow.  Or you could use UDP/IPv4 datagram(s) within the Ethernet frame, so different ports would be used for different UART connections, making the format IPv4-routable, and easier to extend (for example, more than one UART link).

Let's say you limit to a 1Gbits/second bandwidth instead.  The practical data bit rate including Ethernet frame overhead is about 800 Mbit/s.  This suffices for 1024×768 24bpp 30 fps (a little under 567 Mbit/s, plus framing overhead, plus the UART data).  This would require you to have a full framebuffer of about 19 Mbits (2.4 Mbytes), because the VGA input data rate is about twice that.  The other option is to transfer 60fps interlaced, every other scan line per field, so that only a few scan lines need to be buffered at a time, also leading to much lower latency; but also leading to tearing, since the odd and even scan lines would be from different consecutive display frames.  It would really be 1024×384 60fps, in other words..  The FPGA would act as the MAC, reading the TVP7002 parallel output selecting and buffering only the visible part of each scan line, while pushing the previous one a byte at a time at 125 MHz to a gigabit GMII PHY, followed by a media converter to fiber (1GBase-SX).

I think, that is.  I am not absolutely certain of any of the above, not having done it myself before; the FPGA part in particular is new to me.  The state diagram isn't complicated, unless you include stuff like mode autodetection; using a separate microcontroller to measure HSYNC/VSYNC and control and configure both the video ADC and the FPGA should make things much simpler, modular.

Too complicated.  VGA ADC yes (use whatever of your liking), but instead of FPGA which is a huge development pain, use a suitable serialization chip, like GS2970 + GS2972 to convert video into SDI bitstream and then shove the data into an SFP module. You could use the ancillary packets to transmit your serial line within your video data stream.
The only drawback is you may need to scale or crop the XGA video resolution into something SDI compatible like HD (720p or 1080p). But you can already get off-the-shelf mass produced converters for that, heck even to convert then SDI into fiber. Again, depends on what the OP really needs. ((//EDIT: Not talking about cheap converters from Amazon. Plenty of secondhand broadcast grade equipment over ebay for reasonable prices).


//EDIT: those Semtech serializers are also BGA, but just 100 balls. Easy stuff, compared to the 381 ball FPGA.

[DiTBho]

For remote access to computers, there are much better solutions like remote desktop

In this topic I never said that I have to access a remote computer!
I clearly wrote that I need to remotely export the VGA video output (of an industrial device, which is evidently NOT a computer), and a serial!

[DiTBho]

The serdes/PHY is a problem.  It converts parallel data to the GHz-range serial (differential) pair the SFP+ module uses, and vice versa.

Very interesting post, thanks 

I think the PCB is *very* problematic for me with Ghz-Range differential stuff.
So, for me is, it's a red-flag, which sounds "forget it, cannot be done".

Instead, I will try some products from Amazon and Aliexpress, with the policty "try, keep or return".
Including some VGA-to-SDI-optic; SDI is new for me, it sounds interesting, but never used, so ZERO know/how, experience, etc.

For sure, I will play with some optic modules on a custom PCB, but for simpler stuff, like multiplexing some serial devices 9600..19200bps over an optical cable.
With 1-2MBps as final bitrate, it will be a simpler PCB, so 100% feseable for me, and also *very* useful as I seriously need something similar for an other future task.

[DiTBho]

The other issue you have is your attitude dismissing all offered solutions, without giving a reasond you mean a "hobby personal project" instead?

Attitude .... what OP really needs ....well, in the first post I clearly wrote what I was looking for, and I specified that I was looking for someone who has already used these products or something similar.

The only reply dismissing the offered solution was agaist "use a RPI", which I then explained in details, replying to Nominal Animal.

I don't know SDI, and I didn't even reply to you "yes/no/maybe" becase in the meanwhile I am documenting about!

[Yansi]

I think the PCB is *very* problematic for me with Ghz-Range differential stuff.
So, for me is, it's a red-flag, which sounds "forget it, cannot be done".

What is so difficult to route 2 cm (or 4/5" if you want) of two traces near each other? Even if it would not be impedance controlled or what not, it would more than likely work (although not recommended).

Your biggest enemy is the BGA package routing and soldering, not high-speed diff pairs. Those are easy.

For remote access to computers, there are much better solutions like remote desktop

In this topic I never said that I have to access a remote computer!
I clearly wrote that I need to remotely export the VGA video output (of an industrial device, which is evidently NOT a computer), and a serial!

No you did not state that. You did state you've had some "feel that you need" and that  you need to extend VGA in a "very noisy environment". Neither have you stated, what the signal source is, nor that it is an industrial application.

Than you claimed nothing can run unauthorized OS, so SBCs are forbidden - does not make any sense, unless it is a safety critical application, in which case I seriously doubt you should be building it with cheap chinese converter boxes or other non-safety-certified commercial HW or on your own. How are you then sure, that the off-the-shelf HW will not contain unauthorized SW, operating systems, or anything else of concern?

...and then you run scared from hearing about Gbps data pair on a PCB. 

Not sure how better we could help, as you are being very secret about everything. A bunch of solutions were already thrown.

[DiTBho]

No you did not state that. You did state you've had some "feel that you need" and that  you need to extend VGA in a "very noisy environment". Neither have you stated, what the signal source is, nor that it is an industrial application.

That's all you need to know!
And I didn't put any secret on the specifications and constraints!

[DiTBho]

Than you claimed nothing can run unauthorized OS, so SBCs are forbidden - does not make any sense, unless it is a safety critical application, in which case I seriously doubt you should be building it with cheap chinese converter boxes or other non-safety-certified commercial HW or on your own. How are you then sure, that the off-the-shelf HW will not contain unauthorized SW, operating systems, or anything else of concern?

I seriously doubt there is a GNU/linux SBC in a cheap Chinese converter box, and even if it was, for sure it's not easily re-programmable (hackable?), so they can turn a blind eye.

But they can't turn a blind eye to any GNU/linux SBC that even a housewife would know how to use to save data on its internal SD card, used, among other things, to boot the SBC itself.

Something like this DOES not even enter the laboratory, they see it as a serious threat.

I found a Chinese MPU in the HDMI video capture device I opened last week, followed by a physical 100Mbps Ethernet. The MPU has enough power and ram to compress H26* and has a large internal flash-ROM to include the HS26* algorithm as well as a full TCP/IP stack. Not bad, unfortunately there is no jtag exposed, so it's not even worth wasting your time on, you couldn't do much with it anyway.

In case of need, they don't see it as a threat, they can exceptionally turn a blind eye.

I don't want to tell you too much, just, they made a logistical *BIG* mistake, and it takes time for them to fix it, in the meantime I don't want to sit around doing nothing. I need a temporary solution, which I am willing to pay out of pocket even to avoid finding myself overwhelmed by work commitments in the months immediately preceding the holidays, when they have solved the problem and managers who know nothing about what happened will expect nothing but the deadlines to be met.

you know, ... big companies ... anyway, for sure the final solution will be safety-certified (it must be), but I am the one who has neither to manage it nor to pay for it. It's not my business (and technically, I write software, I don't do hardware); what I need is to read a damn remote instrument, leaving it where it is at the moment, away from my desk, without the need to get close and damage my equipment, or worse still, hurt myself.

Cannot say a word more than this 

[berke]

I don't know what kind of job/contract you are in and I understand you can't give full details, but it sounds like it's not a technical problem you have.  My advice to you would be to forget the price, shop around and get a quote for a serious solution that will work (maybe $5k? $10k?) and will be available in a timely manner, and put it in a small proposal explaining what you want to do, how much money/hassle it will save and why.  This will cover your ass and you won't have to pay out of pocket, will allow you to buy equipment from "places that don't sell to individuals" and will make you good and shit will get done.  $10k should be a minor expense if not a rounding error in a place with 1200 m of linear space. heavy EMI, radiation and security requirements even if it's some kind of underfunded public lab.  I understand the temptation of just paying a couple hundreds out of your pocket to get something fixed especially in a technically interesting manner but it really doesn't sound like 200 is gonna cut it.

[Nominal Animal]

If I were you, I might take a risk with Primeda Telecom PMD-VGA-KVM-SC-N (Amazon UK, eBay UK, AliExpress); Shenzen vendor/manufacturer).

You do need to add an USB microcontroller to provide RS-232 or UART ports, likely via custom USB HID packets, which can handle 64000 bytes/sec per endpoint, but a Teensy 4.0 could handle that easily (with RS-232 drivers if you need proper RS-232 ports) for up to 7 UART/RS-232 ports.

[DiTBho]

@Nominal Animal
Yup, good idea! I think it's worth a try 

[DiTBho]

but it sounds like it's not a technical problem you have

It depends on how we look at it.

In the first posts, I asked for direct experience with products in the cheap area, so in case it's not a request to solve a "technical problem", it's more "social engineering". Then I also considered the (hypothetical? (1)) case of making something, giving constraints and specs.

(1) which may make sense, if you have some idle months and want to make good use of (part of) the time learning something new.

My advice to you would be to forget the price, shop around and get a quote for a serious solution that will work (maybe $5k? $10k?) and will be available in a timely manner, and put it in a small proposal explaining what you want to do, how much money/hassle it will save and why. 

I've never dealt with administration, so I have no idea of the costs and how much money/hassle it will save, and how they will react to a proposal, but I can find out. It might even work, so it's worth trying it immediatly as "plan A"; at worst they tell me "no dice", and we stay on the "plan B".

Thanks 

[DiTBho]

I take this opportunity to study and experiment with SDI. In this case it's stuff I will play at home, so hobby/learning on a low budget.
May be also some SDI <---> fiber optic adapters, with some short, 2..4m, LC/LC or SC/SC patches.

I will probably buy on Amazon, "try: keep or return" policy, some HDMI-to-SDI, and SDI-to-HDMI adapters, to be used with BNC cables.
I already have an HDMI LCD made by NEC, and a VGA to HDMI converter, the one I mentioned, product/vendor, in one of my previos posts.

- - -

If they will agree to buy professional equipment (planA), it will be interesting to compare it with low-cost equipment 

[DiTBho]

If I were you, I might take a risk with Primeda Telecom PMD-VGA-KVM-SC-N

Talking about other stuff, I found a colleague who had tried it before returning it on Amazon because - full of problems, especially the USB part, I don't reccomend it - so he summarized.

(this is the kind of information I expected to read when I opened this topic)

At this point I will avoid the  Primeda Telecom PMD-VGA-KVM-SC-N and move forward with other candidates.

[Nominal Animal]

full of problems, especially the USB part

It is likely limited to HID devices only, which is why I said you will have to do your own RS-232/UART to USB (HID) to RS-232/UART conversion.

I expect none to work with say a hub and/or USB-to-serial converters.

If I were you, I'd find out if the 'full of problems' means it did not work like your colleague assumed it would/should (with USB-to-serial converters and/or hubs), or if it means it does not work with a keyboard+mouse+vga combo.  If the latter, then fine; but I suspect it is the former, and nothing will work according to that criteria, I believe.