Shop OBEX P1 Docs P2 Docs Learn Events
Looking for an IC (that hopefully exists), but I don't know what it's called — Parallax Forums

Looking for an IC (that hopefully exists), but I don't know what it's called

Bobb FwedBobb Fwed Posts: 1,119
edited 2011-12-02 16:18 in General Discussion
I'll tell you what I need, and maybe someone can just tell me the name of what I need; I can look up the specific specs and model myself.

What I need is programmable output pins (high and low) that can run at high(er) voltages (like 12-16V). The rate at which the outputs can change are inconsequential, for the most part, the outputs will be programmed once at start up and left as is. Output current is not important, they will be driving MOSFETs. Is this called an IO expander? Is there high(er) voltage versions of these, or cheaper output only versions?

I setup a test circuit that mimics what I need with an OpAmp, but I am going to be wiring up 36 of these MOSFETs, and I don't want to use two Propellers and 36 amps to do it.
«1

Comments

  • Bill HenningBill Henning Posts: 6,445
    edited 2011-10-21 14:27
    UNL2003 and friends...
  • tonyp12tonyp12 Posts: 1,951
    edited 2011-10-21 15:33
    How about 4 or 5 of these IC, combining 8 mosfets with a 3 pin ser-to-parallel shift register (daisy capable) .
    http://www.mouser.com/ProductDetail/STMicroelectronics/L9822EPD013TR/?qs=riOa31DHTCp3r82F%2fPshGw%3d%3d

    it's a 5V chip, so I used this non inverter HCT to interface to it:
    http://www.mouser.com/Search/ProductDetail.aspx?R=74AHCT125D%2c118virtualkey66800000virtualkey771-AHCT125D118
    i choose non-inverted so the reset line will match props reset line.

    attachment.php?attachmentid=86202&d=1319236829

    If you still want to use your own mosfets and they have a 5v gate
    use 5 of these HCT595 (5v out with a low 2v input thresholds)
    http://www.mouser.com/ProductDetail/NXP/74VHCT595D118/?qs=jquClx72t9Ax%2fGnvIRSoIA%3d%3d
    MosfetPCB.jpg
    320 x 218 - 37K
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-21 16:43
    Bobb,

    What is want is called -- surprise! -- a MOSFET driver. Check here for the specs, then look up the part on DigiKey:

    http://www.micrel.com/page.do?page=product-info/mosfets.shtml

    -Phil
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2011-10-21 17:01
    Phil Pilgrim (PhiPi) wrote: »
    Bobb,

    What is want is called -- surprise! -- a MOSFET driver. Check here for the specs, then look up the part on DigiKey:

    http://www.micrel.com/page.do?page=product-info/mosfets.shtml

    -Phil
    Woh! The creativity of naming blew me away.
    There is a massive number of variations on those ICs (as with most ICs).

    Any pointers for a MOSFET driver that is programmable (SPI/I2C) and will drive N-MOSFETs straight-up (nothing unusual that I'm doing, but no H-bridge fanciness is needed)?
    I don't have time to look too much more right now (I'll have time Monday), but maybe at least a pointer to internal vs. external switch, etc.

    Thanks.
  • kwinnkwinn Posts: 8,697
    edited 2011-10-21 17:25
    The MIC4467/8/9 should do what you want and it comes 4 to a package. The ULN2803 or TPIC6595 would work as well based on your stated requirements. The ULN and TPIC have 8 drivers, and the TPIC also has a serial to parallel converter built in if you are short of pins.
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2011-10-24 13:13
    Thanks, I think I'm going to end up using the TPIC6595 class. As I might be bumping it up to 48 MOSFETs I'll have to drive. I would have liked the IC to be 3.3V compatible, but not the big of a deal, and not require external components, like resistors to drive the MOSFETs high, but again, not that big of a deal.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-24 13:31
    The TPIC is definitely not the part to use for this. For one thing, it only sinks current and does not source voltage. For another, the amount of current it sinks is way overkill for driving a MOSFET anyway. MOSFET gates do not draw continuous current. They're a capacitive load only, requiring high currents only when being switched rapidly. Since you're not doing rapid switching, even a low-current-capacity voltage driver with a small series resistor would be adequate and a much better match for the task at hand.

    -Phil
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2011-10-24 13:41
    I'm having problems finding anything that is programmable though. Probably a better way to do it would be to use a set of SIPO shift registers, feeding a more traditional set of logic MOSFET driver.
    I've never used a shift register, but I think this would be proper use of such a device.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-24 13:59
    Well, then how about a string of CD4094's running with a Vdd of 15V? You can then use a voltage translator on their clock and strobe inputs, and the data input of the first one in the string. You will have to use series resistors on the outputs to limit switching currents to the MOSFET gates, since CD4000-series CMOS will source or sink only a few mA.

    -Phil
  • PJAllenPJAllen Banned Posts: 5,065
    edited 2011-10-24 16:34
    http://www.micrel.com/page.do?page=/product-info/products/mic5891.shtml

    http://search.digikey.com/us/en/products/MIC5891YN/576-1310-ND/771779
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-24 17:47
    The Micrel part is pretty much the opposite of the TPIC part. IOW, it sources current only but does not sink any. It also requires a 5V logic supply and 3.5V on the input pins to register as a "high." With level translators on its inputs and pulldowns on its outputs, it would probably work, though.

    -Phil
  • PJAllenPJAllen Banned Posts: 5,065
    edited 2011-10-24 18:00
    Don't know what Bobb's got connected to the MOSFETs, but that maybe he could use these in lieu thereof? They latch ("programmable"). Serial (SR) input.
    Don't know an equiv. with 3V3 logic, mayhap there is such. (Must be, nowadays.)

    3V input, latching mosfet drivers?
  • kwinnkwinn Posts: 8,697
    edited 2011-10-24 22:31
    Based on the stated requirements there is no reason the TPIC could not be used. It may not be the best choice for driving mosfets but it can do it, and I am not aware of any chip with a serial input and 8 mosfet drivers.

    >> programmable output pins (high and low) that can run at high(er) voltages (like 12-16V).
    >> rate at which the outputs can change are inconsequential, for the most part, the outputs will be programmed once at start up and left as is.
    >> Output current is not important, they will be driving MOSFETs.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-24 23:03
    The TPIC6595 would have to be used with pullups, since it only sinks current.

    Comparing costs:

    CD4094: $0.52 ea. + eight series resistors per package
    TPIC6595: $3.58 ea. + eight pullups per package
    MIC5891: $2.50 ea. + eight pulldowns per package

    All three solutions can run at 15V and require level converters for the clock, latch, and first data input pin.

    -Phil
  • PJAllenPJAllen Banned Posts: 5,065
    edited 2011-10-25 07:24
    (...but, with the MIC5891s, he might not need the FETs. I saw that Bobb stopped by late yesterday afternoon, though he didn't check in over here. Huff.)
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-25 09:01
    PJ Allen wrote:
    ... he might not need the FETs.
    That's a possibility. It would be nice to know what he's using them for. If this has anything to do with his other recent thread, I doubt that any of the solutions suggested here would be optimum. It's hard to make good recommendations from incomplete information.

    -Phil
  • jmgjmg Posts: 15,185
    edited 2011-10-25 21:42
    Phil Pilgrim (PhiPi) wrote: »
    The TPIC6595 would have to be used with pullups, since it only sinks current.

    Comparing costs:

    CD4094: $0.52 ea. + eight series resistors per package
    TPIC6595: $3.58 ea. + eight pullups per package
    MIC5891: $2.50 ea. + eight pulldowns per package

    All three solutions can run at 15V and require level converters for the clock, latch, and first data input pin.

    -Phil

    There are variants in the PICx595, with the lower drive ones being cheaper.
    http://focus.ti.com/paramsearch/docs/parametricsearch.tsp?family=analog&familyId=356&uiTemplateId=NODE_STRY_PGE_T
    TPIC6C595 is cheapest, or STPIC16D595 etc

    Then there is a HEF4794/4894 from NXP, an open drain version of the 4094, can work to 3V, with likely lower drive.
    All of these are open drain, but if you want to drive above the rails for a HiSide N-FET driver, you will need a smarter part.

    Allegro have some nice devices,
    Something like the (new) A3944, has load feedback, so you get diagnostics and protection.
    ( or Infineon have devices like TLE7232GS etc )

    and LED drivers are an expanding family, that can include Serial expansion in the simpler variants.
  • tonyp12tonyp12 Posts: 1,951
    edited 2011-10-25 22:29
    Maybe Bob can enlighten us with what type of power driving he needs and currents.
    Here is a Highside with 8 outputs (12-35v) at around 0.3-0.4Amp each.
    3.3v compatible inputs.

    http://www.onsemi.com/pub_link/Collateral/AMIS-39100-D.PDF
    mouser got 6 in stock: http://www.mouser.com/ProductDetail/ON-Semiconductor/AMIS39100PNPB3G/?qs=ve9n7aGWr8mhRgXB9KAY9A%3d%3d
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2011-10-26 10:03
    PJ Allen wrote: »
    ...I saw that Bobb stopped by late yesterday afternoon, though he didn't check in over here. Huff.)
    I feel like I'm being stalked. :-P

    Sorry, I forgot to check this thread, I haven't been receiving emails about replies on it...guess I need to fix that.

    Thanks for all the help so far, and to help things out, I will give you guys more information.

    Yes, as PhiPi deduced, it has to do with the other thread I started at about the same time as this one: http://forums.parallax.com/showthread.php?135388-Looking-for-some-math...is-this-even-possible

    This is my two current prospects, I put them together this morning, so it may not be fully flushed out. Anyone see any inherent problems it? The "IN" is 12-16V, and the maximum voltage coming in on "Anode" is 9.5V, and "Cathode" will usually be somewhere near ground, but always above. I already see I'm missing a decoupling cap on U5.

    wiring.jpg


    wiring2.jpg
    (that part on the left is the TPIC6595)
    wiring.jpg
    908 x 699 - 61K
    wiring2.jpg
    952 x 686 - 149K
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-26 10:42
    Bobb,

    Don't forget that the conductance of a MOSFET depends on the gate voltage relative to that of the source, not to ground. In the circuits you've shown, you're relying upon an absolute voltage to gate floating MOSFETs. It won't work. That's why I suggested using photovoltaic optoisolators, since one can be connected between the gate and source of each MOSFET to drive it.

    -Phil
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2011-10-26 11:12
    That's why I have 12-16V driving the gates and only 9.5V (with respect to ground (WRTG)) at the "Anode" (thus, potentially upwards of 9.5V on the source). As I said, the "Anode" will be close to ground, but probably no higher than 1.5V WRTG. That's a 2.5V (worst case) difference, which is enough to open the IRF7301 to 11A (and I only need 500mA).

    I do see a flaw, I need to choose a different MOSFET (I just had the 7301 from a previous project), because its only rated to +/- 12Vgs, and I could potentially have +16Vgs. There are others that will go to 20Vgs.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-26 11:35
    I would still go the photovoltaic isolator route. That way you can use standard 74HCT595's to drive them, your rheostat circuit will be completely isolated from your control circuit, and you won't need a separate gate supply. Take a look at this device, for example:

    -Phil
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2011-10-26 12:37
    could you give me an example on how to wire that up? So far I've only found one example of a circuit like that in a google search, but it uses two MOSFETs.
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2011-10-26 12:45
    In this example: http://www.chinaicmart.com/uploadfile/ic-circuit/2009716232314688.gif
    I'd just need two diodes for each MOSFET. That look right?
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-26 12:56
    Like this:

    attachment.php?attachmentid=86356&d=1319659743

    The shift registers' Vdd can be 3.3V (74HC595) or 5V (74HCT595). Vdd in the above schematic will be whatever you use for the shift registers.

    -Phil
    mosfet_isolators.gif
    407 x 440 - 5K
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-26 13:02
    I'm not sure what the SCRs are for. Maybe to pull the gates down when turned off. In that circuit, they might be there to prevent shoot-through, since it's driving a push-pull stage. You could could use high-value resistors instead (gate-to-source), since you're not changing states at a high rate of speed.

    (I've added shunt resistors in the above schematic, since the photovoltaic output doesn't sink current.)

    -Phil
  • jmgjmg Posts: 15,185
    edited 2011-10-26 14:37
    Bobb Fwed wrote: »
    The "IN" is 12-16V, and the maximum voltage coming in on "Anode" is 9.5V, and "Cathode" will usually be somewhere near ground, but always above. I already see I'm missing a decoupling cap on U5.

    Since this is an analog switch problem, here is a single part that will do everything, and replace a lot of parts...

    http://www.analog.com/en/switchesmultiplexers/analog-switches/adg1414/products/product.html

    SPI chainable, 3V to +/-15V (+25V max single ended) 0.55 ohms matching. - and you get 8 switches.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-26 14:52
    jmg,

    That's an interesting part, alright, but it's not rated for the current that Bobb needs it to be.

    -Phil
  • jmgjmg Posts: 15,185
    edited 2011-10-26 15:56
    Phil Pilgrim (PhiPi) wrote: »
    That's an interesting part, alright, but it's not rated for the current that Bobb needs it to be.

    I did not see an explicit current, but the 9.5V 'anode' mentioned, and R given on the SCH, gives 1.9mA to 28mA, which should be ok on a 10 ohms sw ?.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2011-10-26 16:13
    He said he needed 500mA (post #22). The analog devices part is rated for about an eighth of that.

    -Phil
Sign In or Register to comment.