Shop OBEX P1 Docs P2 Docs Learn Events
switches / latches / mosfets or any ideas? - Page 2 — Parallax Forums

switches / latches / mosfets or any ideas?

2»

Comments

  • Hal AlbachHal Albach Posts: 747
    edited 2014-01-30 07:56
    The only aspect that really bothers me is the rocker switches. For 32 rockers, they seem to need 64 independent debouncing circuits provided.
    In the circuit I show there is no need for debouncing since the momentary switch operates the latching relay directly.


    .
    Power center for Reach.jpg


    The amount of leakage in an off Mosfet will be in the datasheet for the particular device. For a IRF5210 P-Chan 40 Amp device drain to source leakage is from 25 to 250 uA, depending on applied voltage and device temperature.
    1024 x 773 - 56K
  • Duane C. JohnsonDuane C. Johnson Posts: 955
    edited 2014-01-30 09:12
    Hi Reach
    Reach wrote: »
    If a MOSFET is off how much current is leaking by? How do I calculate this number? I know its a small amount but i need to add this all up so that I can consider other options.
    It's really quite small.
    Select the MOSFET and take a look at the data sheet.
    I might suggest the IRF4905.
    The P-Chanel IRF4905 spec.
    IDSS of -25µA at -55V and 25°C.
    IDSS of -250µA at -44V and 150°C.
    This may help in the design. This graph was personally measured by me:
    MOSFET_IRF4905_Vgs.gif

    Plus any currents needed to drive it.
    The driver circuit snippet I posted in message 9 has a maximum OFF current
    for the 2N7000 of about
    IDSS of 1µA at 48V and 25°C.
    IDSS of 1mA at 48V and 125°C.

    For normal temperatures the OFF current would be less than 27µA for the 3 MOSFETs.

    The IRF4905 has an RDS(ON) resistance of 20mΩ
    as compared to the IRF5210 which is 60mΩ which would
    result in 1/3rd the heat dissipation for the same current. And its cheaper too.

    Duane J
    824 x 460 - 10K
  • ReachReach Posts: 107
    edited 2014-01-30 10:07
    Thanks Duane I needed that. So Ill post the schematic soon its very cleaver how I have managed to conserve power. I designed it so that the human interface (buttons) can override the propeller control and the propeller is not consuming power scanning the button array. The propeller can also override the buttons for remote control via iPad. All without the use of latching relays too.

    The only unknown presently is how to wake up the propeller when a button is pressed.
  • ReachReach Posts: 107
    edited 2014-01-30 10:41
    Here is the circuit I am going to employ - It basically toggles the LED when s1 is depressed sort of no need for a denounce as it triggers on the first hit - then charges // discharges the cap. The relay is small 30ma so the prop pin can energize it without too much trouble.

    Basically when the circuit is first turned on Q1 is pulled high meaning J2 off or ALL OFF.

    When S1 or RY1 are shorted briefly Q1 is turned on sending current to Q2. Q2 is arranged as a pull down switch and in such a way that the node between R3 and R2 is an on / off region.

    C1 begins charging once S1 or RY1 are released so that the next state (off) is ready (again no de-bounce needed) .

    L1 shows the circuits state and J2 will provide output power and feedback for circuits down the chain (so to speak).

    R6 is a pull down for C1 to pass current through on a ON state It also is a value of the RC time constant when charging on the OFF state with R5.

    Again my only issue at the moment is that the propeller will be in sleep mode most of the time to conserve power. Once this circuit is triggered by S1 I need to notify a "WAKE - UP" situation to alert the propeller. Again there will be 30 elements of this circuit (if that's the correct way to put it) or this circuit X 30 in other words.


    Untitled.png
    813 x 498 - 28K
  • Duane C. JohnsonDuane C. Johnson Posts: 955
    edited 2014-01-30 12:35
    Hi Reach;

    I don't understand how your circuit operates.

    I was about to put your circuit into my simulator.
    However.
    The Collector of the NPN Q2 is biases negatively from the Base and Emitter.
    That is not normal for an NPN circuit.

    R6 appears to do nothing, except possibly to bleed the output to ground

    You are apparently trying to implement a "Toggle Flip Flop", not a Set/Reset Flip Flop.
    If so, the controlling circuit can't tell the state of the FF.
    With a S/R FF the state after the controlling input will be known.
    With a Toggle FF the state after the controlling input will be unknown as it will be dependent on the original state.

    My circuit is a S/R FF so the state will always be known.

    Duane J
  • ReachReach Posts: 107
    edited 2014-01-30 13:12
    Duane you are right please excuse my mistake. Q2 is backwards. Please run it as you will discover it preforms quite well. R6 is important ill leave more hints after you run it.

    The thing I enjoy about this circuit is its low power and use of discrete parts.
    Untitled.png
    813 x 498 - 22K
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-01-31 04:18
    Hi,
    It is very interesting... nothing wrong with a toggle flip-flop and discrete parts. Since MOSfets can be used in parallel to provide more power, allowing for additional power MOSfets in parallel could be very handy.

    I do recommend to build boards in banks of 4 circuits, just to make installations easier to manage. One board per circuit creates unnecessary mayhem, while one board with 30 circuits makes for a failure to be complete.

    I have doubts that even the Propeller input requires a relay. It could be isolated with an opto-isolator and skip the relay.
  • GadgetmanGadgetman Posts: 2,436
    edited 2014-01-31 05:22
    Is this a 'toggle' system?
    (One pulse wsitched on, the next switches it off again, and so on)

    The only item I know of that works properly with that kind of system is ball-point pens...

    How will the Propeller 'know' that the switch has been toggled?

    What level of 'override' is this supposed to deliver?

    'Computer overrides' - The computer can take control if it detects the need to. (ex. fly-by wire systems in the Airbus)
    'Pilot overrides' - Pilot/captain can completely override the computer.
    both have their uses. A cooperative 'solution' where neither can overrule the other usually doesn't work all that well and really shouldn't be used in anything deemed critical.

    What kind of equipment will it control?
    Anything that requires more than On/Off?
    (There I would have used 'OFF/Computer/ON' where the computer is allowed control while the switch is in 'neutral' )
  • evanhevanh Posts: 15,935
    edited 2014-01-31 05:48
    Reach wrote: »
    Duane you are right please excuse my mistake. Q2 is backwards. Please run it as you will discover it preforms quite well. R6 is important ill leave more hints after you run it.

    Now I understand, you are wanting manual control of valves/contactors without the controller.

    The MOSFET's gate will need better protection. It might be advised to use a BJT instead, they're hardier.

    Gadgetman raises a good point about feedback to controller. You'll likely want to have an optocoupler in series with the LED to provide this. I also recommend replacing the relay with an optocoupler as well. This makes it all solid-state while still providing full isolation.

    And final recommendation is fully isolate the controller's power supply from that driver circuit supply, ie: Don't have a common ground.
  • ReachReach Posts: 107
    edited 2014-01-31 07:32
    I will try to answer some questions.

    The logic is sort of last pressed takes control. So the button is pressed and thus turns on or off a device, the computer does the same. Almost a OR configuration.

    The feedback to the propeller will be another transistor attached to J2 but this may change yet currently its cheap. I would love to use a optical feedback device but again prices are approaching the limits. The LED in the circuit is actually mounted inside the stainless vandal proof switch - non latching btw.

    I insist on Mosfets for many reasons but mainly for higher power capabilities and without the need of heat sinks - try buying a cheap 30 amp BJT.

    I was thinking about using TO220 mosfets that could be inserted into a socket for removal / replacing but I may cover the entire PCB in a brick of epoxy. Perhaps a conformale coating will be the real life solution. Banks of 3 was my suggestion to the client.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-01-31 10:42
    Well, I mentioned an opto-isolator in lieu of relay from the Propeller. Another opto-isolator in parallel with the indicator LED could provide feedback to the Propeller.

    And since it is a Toggle, the feedback seems necessary to have the Propeller know what the current status is. Having the feedback will allow the Propeller to evolve more sophisticated applications for the control circuit.

    +++++++
    Even if you don't initially desire to use opto-isolators, including them as an alternative construction in the printed circuit board may be useful if you later change your mind. A lot of manufacturers provide added features in their printed circuit board and then decide which ones are really useful. And don't forget to allow parallel MOSfets for greater power needs. You may never populate the empty slots, but they are also handy if a field repair is required to replace a damaged MOSfet. Lightning does strike sailboats.

    If you are covering the board in epoxy, it can be a softer epoxy or even a thin layer and then covered with silicon gel. In that way, it can be cut away for a repair.

    Socketed TO220 seems foolish at 30 amps. 30 amps really needs a strong, low-resistance connection that solder offers. And it may need heavier copper than circuit board traces can provide.... just look at 30 amp household wires, solid 12 guage copper. Amps dictated the physical size of the conductor; volts dicate the quality of the insulation.

    +++++++++
    I realize you have to compromise on cost, but try not to compromise on good quality. That will assure you future customers.
  • ReachReach Posts: 107
    edited 2014-01-31 14:16
    loopy,

    How can an opto-isolator work in lieu of the relay. It was my impression they would allow curent to flow in one direction. See the circuit rely's heavily on the cap discharging and charging thus current flowing both directions across the switch - relay.
  • evanhevanh Posts: 15,935
    edited 2014-01-31 15:21
    A quick look finds TLP222 or TLP227 are your friends.

    Keeping with the MOSFET you want to at least add a zenor diode (Transient suppression versions are good) across the Gate-Source pins to help absorb spikes. MOSFET gates get destroyed too easily otherwise.

    If you want a meaty MOSFET then I came across one that was built tough for on/off control of lighting. It had a huge 13 nF gate capacitance, which can be a benefit, although a later variant had that notably reduced. The IRFP2907.
  • Mark_TMark_T Posts: 1,981
    edited 2014-01-31 17:40
    What a dishonest datasheet - Figure 9 (or note 6) is where they admit the 209A rating is utter nonsense as the package is
    limited to 90A... Hate that. That 90A limit means that the die bonding wires are at the point of failure too I suspect, ie
    in practice 50A is the upper limit for reliable use.

    Incidentally IRLS3036 is a better performer, package limit 240A, Rds(on) typical = 1.5mOhm and its logic level
    (Rds(on) typ = 1.7mOhm).

    If you really want to handle high currents though you'd go to a bolt-on ISO-TOP package or similar, there's no
    easy way to connect 100A wiring (ie bus-bars) to anything with pins.
  • evanhevanh Posts: 15,935
    edited 2014-01-31 19:01
    It wasn't really a recommendation for using this exact part. It is quite old now. It was just an example of something heavy duty. You've highlighted an impressive TO-220 sized part of similar target market.

    However, I've happily used the IRFP2907 at 150 Amps so the 90 Amps is conservative and it's a lot better than an ordinary TO-220 package. I was careful with the legs though; I soldered the heavy wires flat on the legs all the way up to the package, the straight and widely spaced legs are beneficial for this. The PCB is not part of the high current circuit. I hand assembled the boards of course.

    The difference, as usual, between the max currents is a trade-off verses max voltage. I note the similar Gate capacitance.

    Logic level switching is not that interesting when you are dealing with isolated drivers, and isolation is rather useful at these power levels.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-02-01 00:47
    Reach wrote: »
    loopy,

    How can an opto-isolator work in lieu of the relay. It was my impression they would allow curent to flow in one direction. See the circuit rely's heavily on the cap discharging and charging thus current flowing both directions across the switch - relay.

    Sorry, I missed that requirement of having the current flow in both directions. There might be a MOSfet opto-coupler that can do the job. Mouser seems to list some.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-02-01 06:15
    Here is one possibility.. these guys will send you up to 20 Free Samples of an a bi-directional opto-coupled MOSfet..

    http://www.avagotech.com/pages/en/optocouplers_plastic/solid_state_relay_photo_mosfet/

    There may be others out there as well.
Sign In or Register to comment.