Fast Small FPGA or other device?

Discussion in 'The Projects Forum' started by business_kid, Oct 14, 2015.

  1. business_kid

    Thread Starter New Member

    Feb 5, 2013
    22
    1
    I am resuming my university optical transceiver project privately. That was limited, because to quote Hal Abelson:"If I have not seen as far as others, it is because giants were sitting on my shoulders!"

    Nobody on my back now. I can transmit over 500 metres over POF. I want to do it without all the crazy restrictions of that project year. I need suggestions for a fast small fpga but maybe could use a PAL or other device (This was forbidden me in Project time).
    Specs are:

    400~500 Mhz clock
    Less than 8 inputs
    One or 2 outputs (depending on config).

    The device application is adjustable fast period measurement. Coding is minimal. No cpu exists.

    Please suggest devices, or possible candidates.

    I had looked at the Xilinx Artix 7 (fast, & with free linux friendly software) but they're huge behemoths, and i only need a small thing. Achronix in India are shipping 1.5Ghz devices(!) but the price & size rules them out.
     
  2. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,531
    1,248
    Your request is astonishing in its lack of detail, information, data, context, and just about anything else needed to make a considered response. Since I was not invited to be a part of the university project, I have no clue as to what you are talking about, and I suspect that I am not alone in that position.

    ak
     
  3. Papabravo

    Expert

    Feb 24, 2006
    10,138
    1,789
    The standard method for gaining the required information is to read the datasheets on the manufacturer's website. This is not the sort of information you're liable to find on a hobbyist website, but good luck.
     
  4. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,386
    1,605
    I too have been disappointed in the lack of a small size FPGA. Sure, if you need 144 I/O pins you can find plenty.


    Finding one with just 14 or so pins has escaped me.

    Do look into CPLD types, I believe I sawone with 20 something I/O pins.
     
  5. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,531
    1,248
    Lattice has a 44 pin TQFP CPLD with 32 macrocells rated for 400 MHz, and free design and programming software. Whether or not it will do the job is a secret.

    ak
     
  6. business_kid

    Thread Starter New Member

    Feb 5, 2013
    22
    1
    Thank you, AnalogKid.

    Detail is it's for fast period measurement. No bells/whistles. . It's part of a long range optical transceiver

    That 44 pin cpld frpm lattice sounds the business. I'll hunt it down.
     
  7. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,531
    1,248
    How many counters, how many bits in each counter, BCD or binary, parallel or multiplexed outputs? Etc.

    ak
     
  8. business_kid

    Thread Starter New Member

    Feb 5, 2013
    22
    1
    I implemented it in 74AUC @ 250 Mhz using gates and 13 & 25 D flip flops daisy chained, because 74AUC is a very limited family. The 13 & 25 counters. I had to use 74AUC16374 (16 flip flop) chips. FM generation was 12 or 13 8nS clock cycles (12=high; 13=low). FM decoding doubled the clock frequency, and sampled after12.5 8nS clocks (25 @ 4nS). So on a nominal frequency of 10 Mhz, I was getting full bandwidth, although comms issues meant I had to derate 10x. It was also full duplex.

    The design is being overhauled, and |I intend to have configurable as to cycle numbers, get out the crap I had to leave in (because it was worth an 'A' to me) and put in my choices - not mentors, or sponsors. I also want to do an analog version which will not have a cpld in all probability. I have had a stroke, am crippled atm, and am on an alternative health cure. I have 6 - 9 months at home doing little.

    I looked at the device you mentioned(ispMACH4000V). Very good, except I'm alone and it needs a JTAG programmer. Is that complex?
     
  9. business_kid

    Thread Starter New Member

    Feb 5, 2013
    22
    1
    Oh, the output was intended as a single channel input, and output. Data bandwidth is the limiting factor at long distances. No requirement for multiple channels. These devices go in industry, and POF is immune to noise, & crosstalk. It also goes through spark free zones.
    The analog version is obviously stereo and not even thought about yet. But being able to send a stage mike noise free for a few hundred metres must be worth something - if it's done really well. Depending on distance, one might be 4 or 8 channel with a shorter distance spec.
     
  10. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,531
    1,248
    Still not a lot of clarity, but it sounds like this is a digital audio project. If so, 400 MHz seems like an very high carrier frequency for audio, no matter what the modulation scheme. And yes, you can get a boatload of FM carriers on a single single-mode fiber. Broadcast-quality video cameras used to use triax, single-conductor coax with two shields, to carry power, video, and audio out to the camera, and video and audio back to the control room. Hi-def cameras do it with fiber.

    The thread title mentions an FPGA, so I thought you knew about how such devices are implemented. All programmable logic devices need a programming port or method. It used to be a printer port dongle; now they are USB.

    ak
     
  11. business_kid

    Thread Starter New Member

    Feb 5, 2013
    22
    1
    Thanks. I finished a degree in 2014, so I met FPGAs. Usually some eeprom and stuff to loads the program at boot up. In place of the standard design (clock featuring 555 IC:-///), I did a remake of the HP5006A Digital Signature Analyser in 2011, and used one. It was a badly timed product by HP, and they used old tech when they needed new.

    I'm not explaining my idea in detail because
    1. I'd be writing half the night (with one hand paralysed).
    2. It's probably patentable, so I have to shut up
    3.It's metamorphosing as I speak.Yesterday, I had one possible customer - digital. Today it's two applications and half a dozen products at least - Analogue & Digital. And the Analogue one takes priority.

    400 Mhz is a little overkill for sampling single channel audio, although I may need that speed in some iterations. BUT, there's clock speeds, sampling speeds, tuned circuit speeds, Optical fibre speeds, and data speeds. All can be different, and some have complex limitations, The rest is the potential of the notion. There's also a bevvy of possible applications that could be work for years.
     
Loading...