Shop OBEX P1 Docs P2 Docs Learn Events
Sensing CCTV baseband video? — Parallax Forums

Sensing CCTV baseband video?

austin_48111austin_48111 Posts: 7
edited 2010-11-02 11:06 in BASIC Stamp
I have project in which I'm attempting to use CCTV video over a pair of baluns (balanced/unbalanced converters) to convert from coax to unshielded twisted pair into a BASIC Stamp. This is an analog signal theoretically at 1Vp-p. I'm receiving utterly inconsistent results with this, likely because of it's nature as an analog signal. Though it is 1vP-P I sense almost zero voltage over a multimeter, in the range of .374v at most. All I see on the BS2 is floating results (high, high, high, low, low, low, high, high, high, etc). If you have any experience working with baseband video (RGBHV, component, non-broadband composite) in any fashion you would likely have some experience that applies here.

Not knowing what I'm doing and putting a full wave bridge rectifier in between gets me 5.74v but again I acheive utterly inconsistent results. I thought that may convert it all to a rounded off DC voltage but clearly I'm incorrect. I have no oscilliscope to determine what the root cause of this issue is.

1. Do I need to amplify the signal?

2. Do I need to convert it from analog to digital signaling?

3. Would I be better off using a third party IC designed to sense or convert this? Any clue where I could find one?

4. Would I be better off avoiding a balun and staying with CCTV over coax?

I'm about the greenest rookie on this forum so please keep it slightly above laymans terms.

Comments

  • LeonLeon Posts: 7,620
    edited 2010-11-01 08:33
    Extract the sync signals with an LM1881 or something similar.
  • Mike GreenMike Green Posts: 23,101
    edited 2010-11-01 08:47
    What do you want to do with this signal once it travels from its source over the cable and its baluns?

    Remember that a Stamp is too slow to do anything with a video signal. A multimeter is much slower. The video signal does jump all over the place normally and you're sampling it (with the Stamp and multimeter) effectively at random.

    I assume you want to do some kind of analysis on the video image. If so, the Stamps are way too slow and have way too little memory. A Propeller could handle it using an external analog to digital converter, but it doesn't have enough memory to do anything useful with the information. Sampling a video signal at low resolution (like 160 x 120 in B&W) would work.
  • austin_48111austin_48111 Posts: 7
    edited 2010-11-01 09:03
    Without going into an overly long spiel about the end goal I'm going to have 4 CCTV inputs that need to be switched dynamically based upon whether they're active or not to avoid having multiple feeds directed to the same cable. This is for a railroad project in which cars will be switched around frequently so feeds further back on the line may have cameras already on them and we do not want to rely upon the railroad staff to remember to manually switch feeds to unused cables. What I want to do is simply sense that the signal is connected, then fire a relay or something similiar to switch the feed to a different output.
  • Mike GreenMike Green Posts: 23,101
    edited 2010-11-01 09:37
    Leon's suggestion may be the best. You'll need some kind of amplification and filtering and the LM1881 can do that with only a few parts. If the video signal is absent, the vertical sync output of the LM1881 should remain high (absence of vertical sync). You could use a PBasic PULSIN or COUNT statement to look for vertical sync pulses over a few milliseconds and use that to decide whether the video signal is present. Use reed relays (like these) to switch one video signal onto your video output. Pick a reed relay with a coil current of 20mA or less and you can drive it from a Stamp output directly. If you need more coil current, you'll need a driver transistor. Remember that you'll need a protection diode across the relay coil as shown here on the last page.
  • austin_48111austin_48111 Posts: 7
    edited 2010-11-01 09:41
    Wow, excellent, thank you! That helps a lot and saves me weeks trying to reinvent the wheel.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2010-11-01 10:42
    The LM1881 is a good idea, and it should handle the 2V P-P to 1V P-P envelope swings as each signal is coupled, in turn, to a 75-ohm load. The chip is designed for unbalanced inputs, though. So you will have to convert your balanced signal back to unbalanced to use it.

    Regarding whether to use baluns or not, they do offer the advantage of using CAT5 cable, which is cheaper than coax over long runs. However, over very long runs, there can be some signal degradation, which can be ameliorated with adaptive equalizers.

    -Phil
  • Peter KG6LSEPeter KG6LSE Posts: 1,383
    edited 2010-11-01 11:12
    austin_48111 wrote: »
    This is an analog signal theoretically at 1Vp-p. I'm receiving utterly inconsistent results with this, likely because of it's nature as an analog signal. Though it is 1vP-P I sense almost zero voltage over a multimeter, in the range of .374v at most.

    FYI most meters are horrid at anything over 100Hz .

    use a scope . :) or a RF volt meter ( harder to find )
  • austin_48111austin_48111 Posts: 7
    edited 2010-11-01 12:10
    Phil Pilgrim (PhiPi) wrote: »
    The LM1881 is a good idea, and it should handle the 2V P-P to 1V P-P envelope swings as each signal is coupled, in turn, to a 75-ohm load. The chip is designed for unbalanced inputs, though. So you will have to convert your balanced signal back to unbalanced to use it.

    Regarding whether to use baluns or not, they do offer the advantage of using CAT5 cable, which is cheaper than coax over long runs. However, over very long runs, there can be some signal degradation, which can be ameliorated with adaptive equalizers.

    -Phil

    It sounds like we will need to convert back to unbalanced. I definitely need UTP but think the LM1881 fits the bill perfectly. It's not so much the distances involved that will be the primary concern but rather the wiring methods we will need to resort to in this particular environment. Being a locomotive items have to be coupled between cars and coupling a 25 pair cat3, cat5, cat6 or (should we be crazy enough to pay for it) cat6a cable is a lot easier than having a dozen coax to figure out how to couple. We're not homebrewing the UTP baluns but are using products like NVT or Nitek that we've used in the past with good results, and we tend to do a lot of CCTV and IP surveillance. They have passive and active units and can get up to around 7000'-8000' with active units on both ends with a decent cable. I'm going to assume that, besides just amplification, the "active" units make use of the adaptive equalizers you mention.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2010-11-01 15:08
    Just out of curiosity, how many of those inter-car couplings could you have on one circuit?

    -Phil
  • austin_48111austin_48111 Posts: 7
    edited 2010-11-02 11:06
    No more than 4 total cars and the engine itself. The total distance should be roughly 500', worst case, from the most distant camera to the DVR located in the engine so we will try to get away with passive baluns for now but I'm not certain how much, if any, attentuation will occur with these units I'm working on sensing video with.

    Honestly, It's a commuter rail project in a state where there's nearly no commuter rail to speak of so I'm not certain if by the time it's fully active there will be even that many cars to worry about. I think the local populace may have difficulty switching from their automobiles to a more economical but limited transportation system.
Sign In or Register to comment.