1Mbit Versatile SPI/8-bit Parallel Bus SRAM

2»

Comments

  • panupanu Posts: 10
    edited February 2016 Vote Up0Vote Down
    Has anyone gotten a price for this chip ? Even an estimate ?
    The budgetary price for VS23S010D-L (48 pin LQFP package, with video out) for 1000 pieces is USD 1.90 each. It will be available at VLSI webstore in a few days.

    -Panu
  • Thanks for the info, Panu.
    That seems very reasonable.

    Bean
  • panu wrote: »
    The budgetary price for VS23S010D-L (48 pin LQFP package, with video out) for 1000 pieces is USD 1.90 each. It will be available at VLSI webstore in a few days.

    & the budgetary price on SO8 is ?

    panu wrote: »
    No multi-ic currently exists. Perhaps we ought to take that away from the datasheet so that people will not wait for it...
    I'll clarify my suggestion here.
    If there is no single-package version planned, then yes, remove that mention, but if the chip can be used in a multiple-device situation, then keep & expand that info.

    With a Single DAC, it seems it may be possible to run 3 parts, to give RGB - has that been tried ?
  • panupanu Posts: 10
    edited February 2016 Vote Up0Vote Down
    the budgetary price on SO8 is ?
    Sorry, no idea. But it'll soon be available on the webshop, which implies that at least some price steps will be published.
    If there is no single-package version planned, then yes, remove that mention, but if the chip can be used in a multiple-device situation, then keep & expand that info. With a Single DAC, it seems it may be possible to run 3 parts, to give RGB - has that been tried ?
    I wonder if this is what you mean:

    Our design goal was to not do anything that would make it impossible to have three separate ICs works as one, e.g. share the same SPI bus and the clock input/oscillator; configure similar settings to all chips and then switch all modulators on with the same command simultaneously so that they remain synchronous with each other. That way three chips would form a single entity that would have three DACs to drive RGB and one of the chips would make HSYNC/VSYNC in pins D6 and D7. The die has two address configuration input pins to set which unit of such a multi-unit array it is. Our lab test PCB even allows this configuration, it even has a VGA connector and everything. But I think nobody has had the time to do any testing with this configuration, we're still busy writing examples and documentation for single chip use cases.

    The datasheet actually talks about this:
    If a Multi-IC VS23S010C-L system is used for video generation, it is desired to get all VS23S010C-
    Ls operating in synch to each other. This can be achieved by enabling the PLL and checking
    that in all VS23S010C-Ls in system PLL is locked to incoming clock. After all PLLs are locked
    to VXTAL input, setup and enable the Video Display Controller. This same procedure should
    be used regardless of the selected clocking method (PLL or VXTAL clock)


    -Panu

    PS. A couple years ago I tested a single chip with a VGA monitor; our more simple display daughterboard has VGA connector and an array of jumpers that can be used to route the analog output and the four slow sync digital outputs freely to any permutation of the RGB and sync inputs. Basically I just took two hours to test that it works, the results are here: vsdsp-forum.com/phpbb/viewtopic.php?f=14&t=1798 I never continued that work, it was enough for me to just test that the hardware works. The chip (that one chip I tested) worked with surprisingly high speeds. I think I ran it with up to a 12 MHz crystal, meaning that the modulator was running at 96 MHz. Obviously I was just interested to push it as far as it would go, for testing purposes, no idea how stable it was. But it did show the test picture.
  • panu wrote: »
    the budgetary price on SO8 is ?
    Sorry, no idea. But it'll soon be available on the webshop, which implies that at least some price steps will be published.
    OK.

    Looking at the store, I notice this effect

    VS8053B-L, Integrated Circuit.
    1-99 USD 20.00
    100-249 USD 10.00
    250- USD 6.50
    500 USD 4.00
    1000 USD 2.30
    2250 USD 2.19
    4500 USD 2.09
    Ratio 1:1k 8.7:1 (?!) << Wow

    and compare with Parallax price curve
    P8X32A-Q44
    1-99 $7.99
    100-499 $7.19
    500-999 $6.39
    1000-9999 $5.19
    10000+ $4.47

    Ratio 1:1k 1.54:1 << Rather more sane.

    I can understand selling/handling one of anything is costly, but $20 for 99 pcs is somewhat eye-watering markups.

    To keep away from single piece handling effects, I've seen some companies do either or both of
    a) Add a Packing and handling feed - this helps flatten the price curve.
    b) A SPQ of 5 pcs,

  • panu wrote: »
    I wonder if this is what you mean:

    Our design goal was to not do anything that would make it impossible to have three separate ICs works as one, e.g. share the same SPI bus and the clock input/oscillator; configure similar settings to all chips and then switch all modulators on with the same command simultaneously so that they remain synchronous with each other. That way three chips would form a single entity that would have three DACs to drive RGB and one of the chips would make HSYNC/VSYNC in pins D6 and D7.

    PS. A couple years ago I tested a single chip with a VGA monitor; our more simple display daughterboard has VGA connector and an array of jumpers that can be used to route the analog output and the four slow sync digital outputs freely to any permutation of the RGB and sync inputs.

    Yup, that's the mode I was thinking of.
    Sounds like any breakout/test PCB should have 3 footprints on it, to allow '3 as RGB' operation, as well as Composite choices ?

    The other significant use for this, would be as Character Insertion on Video signals, and there. a PLL would be required.
    Something else to add to a break0out board ?
  • panupanu Posts: 10
    edited February 2016 Vote Up0Vote Down
    Hi!
    I can't really comment that much on our pricing, but especially for small amounts you can get a more reasonable price from a reseller, such as world.taobao.com/item/9255689295.htm. These guys actually speak very good english. The Asian guys can offer good prices because they buy in such huge quantities.
  • panupanu Posts: 10
    edited February 2016 Vote Up0Vote Down
    ie can 8 pin Memory use some scan-playback, or must the host update start-address every line (that looks to cost 16 SysCLKs per scan line ?)
    Yes, the setup cost for reading looks like it's 16 clocks for operation Quad Read, Quad Address (8 bits opcode, 6 nibbles address, 2 nibbles dummy data).

    Of course, if you can stop the clock and keep CS low, you can keep the device "paused" until you continue clocking...

    -Panu
  • panu wrote: »
    Hi!
    I can't really comment that much on our pricing, but especially for small amounts you can get a more reasonable price from a reseller, such as world.taobao.com/item/9255689295.htm. These guys actually speak very good english. The Asian guys can offer good prices because they buy in such huge quantities.

    Nicely edited ;)
    The obvious question is: When will 'the Asian guys' have stocks, allowing access to saner pilot-run prices ?
  • panu wrote: »
    ie can 8 pin Memory use some scan-playback, or must the host update start-address every line (that looks to cost 16 SysCLKs per scan line ?)
    Yes, the setup cost for reading looks like it's 16 clocks for operation Quad Read, Quad Address (8 bits opcode, 6 nibbles address, 2 nibbles dummy data).

    Of course, if you can stop the clock and keep CS low, you can keep the device "paused" until you continue clocking...
    I was really wondering if the 48 pin Video auto-scan modes, can be leveraged in the 8 pin part, but the block diagrams seem to show playback into DAC only ?

    If you want to update some of the RAM during blanking, that's two modes, one Write, one Read, and small MCUs do not have a native Quad mode SPI, which adds more SysCLKs to do SW update.

  • panupanu Posts: 10
    edited February 2016 Vote Up0Vote Down
    I was really wondering if the 48 pin Video auto-scan modes, can be leveraged in the 8 pin part, but the block diagrams seem to show playback into DAC only ?

    If you want to update some of the RAM during blanking, that's two modes, one Write, one Read, and small MCUs do not have a native Quad mode SPI, which adds more SysCLKs to do SW update.

    That's all true... the chip is optimized to do composite video. But hopefully all sad faces will be radiating with smiles when, finally, truecolor VGA is demonstrated with 3 chips acting in unison!

    -Panu
  • panupanu Posts: 10
    edited March 2017 Vote Up0Vote Down
    Hi!

    Just to let you know, 3 chip truecolor VGA is reality! We just got it tested and running. You can see the test picture (yes, it looks exactly as it should :) ) at:
    http://www.vsdsp-forum.com/phpbb/viewtopic.php?f=14&t=1798#p10945

    btw, multichip VS23S040 should be out soon, that means single chip VGA out :)

    -Panu
  • tonyp12tonyp12 Posts: 1,915
    edited March 2017 Vote Up0Vote Down
    So it got 128KByte of RAM and a single channel DAC.

    So lets compare that to a off the shelf ARM that probably could do that stuff in software.
    http://www.mouser.com/ProductDetail/Microchip-Technology-Atmel/ATSAMG53G19A-UUT/?qs=sGAEpiMZZMuoKKEcg8mMKHVqFrYvdykF%2bERsPf%2b7cnsBQhc90KGFEg==
    96KB of RAM
    512KB of Flash (fixed graphics/fonts/lookup tables)
    48MHz Cortex4
    DSP and FPU, probably could render some 3D stuff in real-time.
    price:$2.50

    Add 8 resistors for a 3_3_2 RGB

    for 75cents more, 176KB and 120MHz
    http://www.mouser.com/ProductDetail/Microchip-Technology-Atmel/ATSAMG55G19A-UUT/?qs=sGAEpiMZZMuoKKEcg8mMKKywyJ6nXWnoeVOvONePHZyls2pN542sCA==
    or 1MB flash and 256KB Ram at 40MHz
    http://www.mouser.com/ProductDetail/Silicon-Labs/EFM32JG12B500F1024GM48-B/?qs=sGAEpiMZZMuoKKEcg8mMKCbEKngX/el%2bbT0YehKiAsbttZXbtjb0QA==
  • panu wrote: »
    btw, multichip VS23S040 should be out soon, that means single chip VGA out :)

    I can find no references to VS23S040, re package or price ?
    The part code suggests that may have 4 x vs23s010 die ?

    Also,m the web page http://www.vlsi.fi/en/products/vs23s010.html says this
    "Engineering samples are available in limited quantities."


    I guess that is no longer true, and the VS23S010 part is now in full release ?

  • tonyp12 wrote: »
    So it got 128KByte of RAM and a single channel DAC.
    I'm guessing the 040 has 4 die, but of course, also a higher price ?

    another possible solution device is the ICE40UP5K-SG48 - 128K RAM FPGA, in QFN48
    tonyp12 wrote: »
    So lets compare that to a off the shelf ARM that probably could do that stuff in software.

    and this ? ( I like this detail : SPI Flash controller with 32 KB cache memory, 100 MHz code execution)

    https://www.embedded-world.de/en/ausstellerprodukte/embwld17/product-9863795/numicro-m480-series-microcontroller


  • jmgjmg Posts: 11,790
    panu wrote: »
    btw, multichip VS23S040 should be out soon, that means single chip VGA out :)

    I'll bump this, because this VS23S040D RGB variant looks real ?

    VS23S040D-B, 4 Megabit SPI / 8bit bus SRAM with PAL/NTSC Video Controller RGB DACs 5x5mm BGA24 640+ : US$3.20


    Store shows stock


    No breakout boards, and no working examples showing just exactly what VGA resolutions can be supported yet ?

Sign In or Register to comment.