Shop OBEX P1 Docs P2 Docs Learn Events
It can't possibly be done, can it? — Parallax Forums

It can't possibly be done, can it?

LoopyBytelooseLoopyByteloose Posts: 12,537
edited 2011-10-06 21:18 in General Discussion
Of late, I have been deeply impressed by the nature of latching relays (aka pulse sequencing relays). But all these devices seem to require an electro-mechanical approach - none are solid-state.

Are there solid-state versions or is it just impossible to make one?

What's so good about latching relays?
1. They don't consume additional power to hold the relay in a powered coil state.
2. They remember their configuration even if the power goes off.

I can think of all sorts of situations where you want micro-controller control, but don't want the circuit to default to another state if power goes down. Oddly, I tried to buy some locally and here in Southern Taiwan they have never heard of them.
«1

Comments

  • Mike GreenMike Green Posts: 23,101
    edited 2011-10-04 07:54
    Latching relays rely on mechanical memory, sometimes with the help of a little magnetism, to hold their setting. They also require more power to activate (change state) than an equivalent conventional relay ... like any other kind of persistent memory. It would be easy enough to build a solid-state version if there were enough demand for it, but there's not. It's too simple to take an existing cheap microcontroller with a little data EEPROM and couple that with a SSR to make a latching SSR.
  • ercoerco Posts: 20,261
    edited 2011-10-04 08:31
    @Loopy: I'm also a huge fan of latching relays, especially these DIP units that a Stamp or Prop can drive directly:

    http://forums.parallax.com/showthread.php?132685-Latching-Relay-Goldmine&highlight=latching+relay

    A simple BS1 can control 28 relays without any other hardware. Goldmine has those Omron relays for $1.50, and they go on sale regularly for 99 cents.
  • Dave HeinDave Hein Posts: 6,347
    edited 2011-10-04 08:33
    It seems like you could build a solid state latching relay with a MOSFET and a capacitor between the gate and source. As long as the leakage current between the gate and source is low the capacitor should hold the on/off state for a long time after the power is turned off. This would require that the control signal be in a high impedance state when the power is off. You might need additional circuitry to drive the gate to ensure this.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2011-10-04 08:49
    @Mike
    Oh, I didn't consider SCRs. That is something I need to think about. But how does one latch in an open position? It seems like I'd still need a mechanical relay for that side.

    @Erco
    I'd love to get some small ones, but so far nobody in Kaohsiung knows what I am talking about. It seems the electronics shops here have never heard of them.
  • ercoerco Posts: 20,261
    edited 2011-10-04 08:56
    Those Omron relays have 2 coil pins, and you simply reverse the polarity to latch or unlatch the relay. I have seen others that use 3 pins: one to latch, one to unlatch & common.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2011-10-04 10:45
    @Erco
    Those at Electronic Goldmine are a real bargain. $1.50 instead of $10USD. I'll have to work up an order. Mike just doesn't understand how handy these are.
  • Dave HeinDave Hein Posts: 6,347
    edited 2011-10-04 13:41
    Loopy Byteloose wrote: »
    Oh, I didn't consider SCRs. That is something I need to think about. But how does one latch in an open position? It seems like I'd still need a mechanical relay for that side.
    Loopy, I assumed you wanted a solid state latching relay that remembers it's state when there's no power. An SCR won't do that. There are lots of ways to implement a solid state latching relay if you're not concerned about retaining it's state when the power is off.
  • lanternfishlanternfish Posts: 366
    edited 2011-10-04 17:09
    Are these (.pdf) what you are looking for?
  • Beau SchwabeBeau Schwabe Posts: 6,568
    edited 2011-10-04 17:19
    lanternfish,

    I saw those, but they still require power to retain the state....

    "Reset: Remove and reapply control voltage.
    Reset is also accomplished by removing output
    voltage."
  • kwinnkwinn Posts: 8,697
    edited 2011-10-04 18:38
    You could make a solid state equivalent to a latching relay by combining a depletion mode fet and an scr in series. The scr requires a signal to start conducting while the fet requires a signal to stop conducting. AFAIK depletion mode fets are mainly used for rf circuits so the current handling of such a combination would be pretty small.
  • lanternfishlanternfish Posts: 366
    edited 2011-10-04 19:00
    Beau Schwabe (Parallax) wrote: »
    lanternfish,

    I saw those, but they still require power to retain the state....

    "Reset: Remove and reapply control voltage.
    Reset is also accomplished by removing output
    voltage."

    Oops. Didn't read original post closely enough.
  • Beau SchwabeBeau Schwabe Posts: 6,568
    edited 2011-10-04 19:42
    Somewhere there is probably some oddball SSLR (Solid State Latching Relay) that uses a 'bit' of non-volatile memory with the same number of maximum lifetime cycles comparable to it's electro-mechanical cousin.
  • LawsonLawson Posts: 870
    edited 2011-10-04 21:24
    How about a battery backed flip-flop? With a low leakage low voltage FET and a opto-isolators on the input to the flip-flop, I could see the battery keeping the relay active for >3 years. If the battery was rechargable, the SSR could be configured to leak a little power when the load was off to recharge the backup battery. A super-capacitor would also work if several months memory was enough and some parasitic power was available. Aw heck, Dave Hein's suggestion of a capacitor and a FET could likely hold state for a day with careful design and shielding. (the input isolator in particular)

    Lawson
  • ercoerco Posts: 20,261
    edited 2011-10-04 22:02
    ... or, just spend the 99 cents and get a proper electromechanical latching relay with over a million cycles. If it wears out, PM me and I'll mail you a new one! :)
  • lanternfishlanternfish Posts: 366
    edited 2011-10-05 01:49
    And that really is the answer: even in this age of (micro)electronics the best solution is sometimes (electro)mechanical
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2011-10-05 02:04
    @Lanternfish
    I agree with you. There just isn't one that will retain info without a capacitor or battery providing a memory of state.

    If one can really make one, I suspect that it is worthwhile to patent and exploit. The electro-mechanical ones remember forever!
  • Beau SchwabeBeau Schwabe Posts: 6,568
    edited 2011-10-05 07:18
    Not to beat a dead horse... but technically a Hobby Servo motor, such as a Parallax Servo Motor, coupled with a few optic sensors could be used as a latching relay. :-) ... Even a multi-position latching relay :-)
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2011-10-05 10:23
    Consider the dead horse beaten.
    The Hobby Servo is NOT a solid-state solution. It is electro-mechanical. The position of the rotation determines what is on or not. And the lack of rotation with no power to the switch or optic sensor is the memory.

    Nonetheless, as usual with "Beau Schwabe solutions", there is a certain elegance.
  • Tracy AllenTracy Allen Posts: 6,666
    edited 2011-10-05 10:54
    While not at the same power level, a buried gate eeprom cell has that characteristic. It effectively latches the underlying state of the mosfet. So when you apply a voltage across mosfet, it either conducts or not as the case may be. The buried charge can last for many years--Owners of flash drives hope so!.

    Memristors have been realized at the nano level as a fundamental circuit element that can store a resistance value. It is a laboratory curiosity at this point in time, but researchers are already fabricating these into dense crossbar memories and combining them with conventional mosfets to form logic cells that can retain on the nano scale the kind of flip-flop memory you may be happy about. It is pretty impressive!
  • ercoerco Posts: 20,261
    edited 2011-10-05 19:59
    Latching relays on sale for 99 cents: http://www.goldmine-elec-products.com/prodinfo.asp?number=G2011
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2011-10-05 20:47
    Seems Tracy Allen may be on to a true 'all solid-state' approach to a latching relay. Meanwhile, we will have to use the mechanical for memory retained whether power goes off or remains on.

    I still think someone might patent their solution and make some money. The mechanical latching relay is a rather sophisticated mechanical device and usually costs $10 or more. Besides, they tend to be rather large.
  • Beau SchwabeBeau Schwabe Posts: 6,568
    edited 2011-10-05 22:11
    erco,

    I hear ya... 99 cents...

    There is a discrete component option that could be a close runner up with the old but not forgotten CD4049 ....

    $0.52 http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=296-2055-5-ND
    $0.45 http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=1277PH-ND
    $0.09 http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=P330BACT-ND

    $1.06 ... <-- Rats! if I shopped around or purchased in bulk I could get it for much cheaper.

    The attached circuit uses a 0.022uF cap as a memory element. When the main power (+V) is removed this circuit will hold the last digital state for several minutes/hours (.. <-- I'm not sure yet it's running on my desk as I type this). Following up on the initial question an N-MOS connected to the Output would essentially be a Solid State "pseudo"-Latching Relay in this configuration.

    Note: The leakage from the Capacitor or 'memory element' is in the order of nano-amps according to the data sheet of the CD4049... but since this isn't a typical configuration it's hard to determine without empirically measuring it over time. When re-powered the charge is re-enforced to the state of the previous holding position.
    SSL.JPG
    627 x 561 - 95K
    SSL.JPG 95.4K
  • ercoerco Posts: 20,261
    edited 2011-10-05 23:39
    Great work, Beau! Love your sense of experimentation.

    I liked your idea about using a servo as a memory device. I got a smokin' megadeal on Ebay last month. I bought 40 9-gram servos for $35. That's under a buck each, so right in line with the dollar-store targets we're shooting for.

    PS: Can the Propeller 2 support 40 servos? :)
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2011-10-06 03:37
    Sorry Beau, but all is lost if the V+ is removed from the CD4049. Not so with a mechanical relay.

    I know I am being stubborn about this, but that is the whole point. For indefinite periods - including complete lack of power and a return of power; it can't be done. Mechanical trumps solid-state.

    Please forgive me. I am not doing this to upset people. It is just an interesting observation.
  • Dave HeinDave Hein Posts: 6,347
    edited 2011-10-06 04:36
    Loopy, I posted a solution two days ago in post #4.
  • Beau SchwabeBeau Schwabe Posts: 6,568
    edited 2011-10-06 06:37
    Dave Hein,
    "Loopy, I posted a solution two days ago in post #4." - Right, that's essentially what the CD4049 is doing, functioning as a 4T memory element. The Capacitor used is pretty low (0.022uF) and can hold the state for at least a couple of hours. A larger capacitor would hold the state much longer.

    Loopy Byteloose,
    "Sorry Beau, but all is lost if the V+ is removed from the CD4049. Not so with a mechanical relay." - Yes, this is why I said "Pseudo"-Latching. The CD4049 circuit with the stated values can hold the state for at least a couple of hours before a refresh is required.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2011-10-06 08:16
    I'll will leave it here- no more posting by myself. I really don't want to annoy or harass, but close is not equivalent functionality.

    For instance, a mechanical latching relay can be pulsed by a microcontroller to completely lock out power indefinitely as a failsafe - including power to the micro-controller - until a repair person examines the unit and submits a pulse to reinstate use. Sure, it is a bit of a suicide pill, but that is a valid use.

    I just don't see how solid-state can offer that degree of certainty.

    And of course, there is still the question of how to do this with AC power as well as DC.

    I just thought that it might be an interesting new product as there certainly appears to be continuing demand for latching relays.
  • wjsteelewjsteele Posts: 697
    edited 2011-10-06 09:29
    Interesting... I haven't seen anyone mention a Memristor (http://en.wikipedia.org/wiki/Memristor)! It'll save state when the power is off.

    Bill
  • ercoerco Posts: 20,261
    edited 2011-10-06 10:03
    Tracy Allen wrote: »
    Memristors have been realized at the nano level as a fundamental circuit element that can store a resistance value. It is a laboratory curiosity at this point in time, but researchers are already fabricating these into dense crossbar memories and combining them with conventional mosfets to form logic cells that can retain on the nano scale the kind of flip-flop memory you may be happy about. It is pretty impressive!

    They do sound useful. Hope the "Shack" stocks them in my lifetime. Better hurry!
  • kwinnkwinn Posts: 8,697
    edited 2011-10-06 16:28
    Lets not forget the MRAM chips (see http://www.freescale.com/files/microcontrollers/doc/data_sheet/MR2A16A.pdf).

    If they can be used as data storage I am sure they could be used in conjunction with some sense and drive circuit along with a fet, scr, triac, etc. to produce a solid state version of a latching relay.
Sign In or Register to comment.