Shop OBEX P1 Docs P2 Docs Learn Events
Need help with an SCR circuit...Power saving camper (RV) LED Bulb. — Parallax Forums

Need help with an SCR circuit...Power saving camper (RV) LED Bulb.

BeanBean Posts: 8,129
edited 2020-02-12 12:57 in General Discussion
I've not used SCRs before.

I'm trying to make a circuit such that the LED1 will be on when the switch is closed after a long time (10 seconds or so) of being open (C1 and C2 are discharged).
But that the LED1 stays off if the switch is quickly opened and closed again (about 1 second of being open).

Not sure if C1 is needed or not ? I assume it is.

Is there an easier way to do this ? The LED could be a high power 1W, I would adjust R3 for that.

Any help would be appreciated.

Thanks, Bean.

P.S. I have updated the schematic on 2/12/2020.
«1

Comments

  • kwinnkwinn Posts: 8,697
    C1 is needed. The triac is triggered by the rapid voltage increase on c1/c2 when the switch is closed. Once c1 is charged to 12v opening sw1 will start it slowly discharging. If sw1 is closed before c1 voltage drops below a certain point it will not trigger the scr since the voltage change on c1/c2 is too small.

    When sw1 is opened the triac will stop conducting immediately regardless of the charge on c1/c2.
  • BeanBean Posts: 8,129
    kwinn wrote: »
    C1 is needed. The triac is triggered by the rapid voltage increase on c1/c2 when the switch is closed. Once c1 is charged to 12v opening sw1 will start it slowly discharging. If sw1 is closed before c1 voltage drops below a certain point it will not trigger the scr since the voltage change on c1/c2 is too small.

    When sw1 is opened the triac will stop conducting immediately regardless of the charge on c1/c2.

    Thanks, Kwinn. That was my thinking exactly.

    Bean
  • jmgjmg Posts: 15,140
    edited 2020-02-11 20:11
    To discharge C1, you would need a gate resistor on the SCR.
    SCR gate trigger currents also are not tightly controlled, but you can buy a sensitive gate version. *
    SCRs are not low-drop switches, but that may be ok here ?


    * A quick check, finds
    BT258S-800LT 800V 8A with trigger range : 20 ~ 50 µA = overkill, but has tight Igt and low VF
    EC103D1W 400 0.8A with trigger 3 ~ 12 µA
    NCR100W 1.1A 15 - 50 µA (cheaper, might be ok ?)
    Bean wrote: »
    I'm trying to make a circuit such that the LED1 will be on when the switch is closed after a long time (10 seconds or so) of being open (C1 and C2 are discharged).
    But that the LED1 stays off if the switch is quickly opened and closed again (about 1 second of being open).

    Not sure if C1 is needed or not ? I assume it is.
    With care, you may be able to omit C1.
    In that case you would tune the Rgk resistor to narrow down the Igt, and make it so the dV/dT difference between the short & long breaks is defined enough.
    Spice may help here to centre the trigger point with the tolerances present in SCR1 C2 etc

  • jmgjmg Posts: 15,140
    edited 2020-02-14 00:27
    Quick check with Spice gives a starting point for you (your C1 removed).
    This has 2k2 Rgk, which is roughly 350uA of offset current, that reduces trigger variation % to (350uA +15~50uA) taking NCR100W
    With the values shown
    12V: 1.85s triggers 1.75s no trigger
    24V: 0.825s triggers 0.80s no trigger
    1365 x 938 - 57K
  • If the SCR is not an absolute, you could do the same thing with a 555 that would be continuously set until the time between input pulses get to long and the 555 then resets. Add a latch if the led should not go off if pulses start up again. If the SCR is required, the 555 could be used to trigger the SCR. Once fired, what will reset the SCR? Your source is DC, once on, the SCR will never turn off again without removing power.
  • jmgjmg Posts: 15,140
    .. Your source is DC, once on, the SCR will never turn off again without removing power.

    Yes, I think that is what Bean wants.
    The single Switch becomes dual operation, based on time pressed. Very simple user interface.


  • Tracy AllenTracy Allen Posts: 6,656
    edited 2020-02-12 06:08
    I’m wondering if C2 in series with the gate is necessary. The SCR won’t trigger until the input voltage reaches 1 Vbe = 0.6V, Current from the R1/R2 divider could directly source the necessary current, delayed by C1. But 10 seconds?

    I think I’m not understanding the requirement as stated:
    I'm trying to make a circuit such that the LED1 will be on when the switch is closed after a long time (10 seconds or so) of being open (C1 and C2 are discharged).
    But that the LED1 stays off if the switch is quickly opened and closed again (about 1 second of being open).
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2020-02-12 06:42
    I can't make sense of the sentence either, only the circuit. Certainly C1 is not required if R1+C2 provide the gate trigger current (SCR requires a postive voltage from gate to K) at a fresh power-up. The 10k may be too high though.
    As jmg stated, a gate discharge resistor is needed and the LTSpice model simulates that well (the diode is not required, it's just for the sim). I am guessing that if it is turned off, then it requires C2 to discharge sufficiently before the circuit can be turned on again.
  • jmgjmg Posts: 15,140
    edited 2020-02-12 09:00
    I think I’m not understanding the requirement as stated:
    I'm trying to make a circuit such that the LED1 will be on when the switch is closed after a long time (10 seconds or so) of being open (C1 and C2 are discharged).
    But that the LED1 stays off if the switch is quickly opened and closed again (about 1 second of being open).

    I took 10 sec to mean the recovery time, so that a short break pulse (1s) is OFF, and a longer break pulse (10s) is ON, at release time.

    That's what the spice circuit above checks.
    The 10k may be too high though.
    10k is over 1mA, so that's enough for the 200uA and 50uA Igt SCRs

    You could also make a SCR using DUAL P+N FETS, and those have more compact package choices (not sure if that matters here) and with those you can also wire to avoid the ~1V-loss that SCRs suffer from.

  • BeanBean Posts: 8,129
    edited 2020-02-12 12:57
    Thank you everyone for the info.

    This is for a LED bulb for a camper (RV).
    When only battery power is available, it is important to conserve power.

    The idea is that the bulb has a low-power LED that is will always come on when the switch is closed (not shown in my circuit).
    When you first turn the switch on, the both the main LED (my circuit) and the low-power LED will come on.
    But if you don't need bright lights, you can flick the switch quickly "off and on" to get just the lower power light (main LED will be off).
    If you want bright lights again, just turn the switch off for longer (10 seconds or so) then back on.
    This avoids having any buttons or such on the bulb, which you would not be able to get to anyway. And does not require any rewiring of the camper.

    I hope the above makes sense. Please ask if you have any questions.

    I'm trying to do this as simply and cheaply as possible because I want to make it easy for others to build. Also you will probably need about a dozen or so for each camper.

    Keep the ideas coming,

    Thanks, Bean

    P.S. I have updated the schematic.
  • kwinnkwinn Posts: 8,697
    Bean wrote: »
    Thank you everyone for the info.

    This is for a LED bulb for a camper (RV).
    When only battery power is available, it is important to conserve power.

    The idea is that the bulb has a low-power LED that is will always come on when the switch is closed (not shown in my circuit).
    When you first turn the switch on, the both the main LED (my circuit) and the low-power LED will come on.
    But if you don't need bright lights, you can flick the switch quickly "off and on" to get just the lower power light (main LED will be off).
    If you want bright lights again, just turn the switch off for longer (10 seconds or so) then back on.
    This avoids having any buttons or such on the bulb, which you would not be able to get to anyway. And does not require any rewiring of the camper.

    I hope the above makes sense. Please ask if you have any questions.

    I'm trying to do this as simply and cheaply as possible because I want to make it easy for others to build. Also you will probably need about a dozen or so for each camper.

    Keep the ideas coming,

    Thanks, Bean

    P.S. I have updated the schematic.

    What a great idea Bean, thanks for posting it. I am in the process of restoring an old trailer so I can do a little roaming in the summers and will be making use of that circuit.
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2020-02-12 23:08
    Yes, it's a clever and simple idea. I'd look at using the complementary transistor (bjt or fet) as jmg suggested because it is more compact and efficient although the trigger threshold would need to be raised using a zener I feel.

    Doh! Looking back at your circuit with the LED running directly from the supply through a small resistor means you are not worried about efficiency. Just use an SCR :)
  • jmgjmg Posts: 15,140
    edited 2020-02-13 02:12
    Bean wrote: »
    This is for a LED bulb for a camper (RV).
    When only battery power is available, it is important to conserve power.
    Would it be useful to characterize this also for 24V operation ? I guess some RVs use 24V ?

    Addit - Spice gives these windows for 12V & 24V tests :
    12V: 1.85s triggers 1.75s no trigger
    24V: 0.825s triggers 0.80s no trigger
    Bean wrote: »
    I'm trying to do this as simply and cheaply as possible because I want to make it easy for others to build. Also you will probably need about a dozen or so for each camper.
    Do you plan SMD, or thru-hole(ish) design here ?
    The BT258S-800LT I mentioned above looks a good fit for 'largely unknown' loads. They have sub 1mA holding typicals, and a narrow (for SCRs) Igt span.

    One benefit of this circuit, is users can self-adapt to timing variations :) but it would be nice to limit unit-unit variations.

    The worst tolerance part is likely to be the cap, ±20% is common for Aluminum Electrolytic Capacitors, ±10% are rare in SMD elcos.
    A badge tolerance of ±10% is common in ceramic caps, but their voltage tolerance is not great, tho I'd guess that can be calibrated for.


  • BeanBean Posts: 8,129
    I will be trying to build the circuit this weekend.
    The only SCR I have is a radio shack 276-1662. The only info I have for it is on the package:

    RMS On-State Current IT (RMS): 1 amp
    DC Gate Trigger Current IGT: 100uA

    That is it. I can't find anything on the web about it either. But I guess it will be okay for testing purposes.

    Bean
  • jmgjmg Posts: 15,140
    Bean wrote: »
    RMS On-State Current IT (RMS): 1 amp
    DC Gate Trigger Current IGT: 100uA

    That is it. I can't find anything on the web about it either. But I guess it will be okay for testing purposes.
    Yes, fine for testing. The 100uA IGT is the useful bit. You could measure the actual trigger current, for future selection.

  • You have to determine the gate voltage for the SCR. This way you can figure out the RC charge time to turn it on. I don't see a RS 276-1662 anywhere only a 276-1020 SCR. The 1020 has a 3 volt trigger level 6 Ma.
  • When the circuit is on, and the switch opens, a large part of the current from both C1 and C2 flow through the 10k resistor R1 and back through the SCR leg. The resulting current through the SCR has to drop to below the characteristic holding current of the SCR, before it will turn off. That will on the order of 1mA, but it varies with type of SCR and from one to the next. With R1=10k, and about 11 volts starting across the capacitors, that current will also be on the order of 1mA. Time constant 0.1 second. hmmm, pretty close. You could eliminate the back current by putting a diode in series with R1. That would ensure that it turns off quickly when the switch flicks open.

    Initial triggering is determined by the dV/dt across capacitor C2 when it is in a discharged state. In the existing circuit, dV/dt is limited by R1, which has to have a low enough value to provide the gate turn-on current of ~100µA,. That is another reason to include a diode in series with R1, to avoid the dependence on holding current and free up the value of R1. R1 could be lower resistance and C1 might not necessary. I do think a resistor <1kΩ from the SCR gate to cathode is necessary in order to let C2 discharge completely through the 1MΩ R2.

    Thanks for the clarification of the goal. I like bare to the metal solutions like that!
  • jmgjmg Posts: 15,140
    .... You could eliminate the back current by putting a diode in series with R1. That would ensure that it turns off quickly when the switch flicks open.
    The diode is not needed, as an open switch ensures turn-off. C1 is not needed.
    I just added diode to spice, being lazy to allow a simple pulse generator to emulate a switch.
    I do think a resistor <1kΩ from the SCR gate to cathode is necessary in order to let C2 discharge completely through the 1MΩ R2.
    I chose 2k2 for the gate resistor, as you do not want to lose too much trigger current thru that path, but some current helps narrow the trigger range down.
    2k2 is ~ 350uA so stabilizes the trigger point of a < 100uA SCR.
    You can also think of the 10K : 2k2 as a divider, such that over ~ 3.3V is needed to hit the gate voltage threshold, plus 1V more to hit 100uA.

    Another, somewhat hidden time-constant is R1-C2, that needs to be long enough to still have current flow after the switch debounce finishes.
    10k,10uF is ~100mS, which should be plenty.

  • Tracy AllenTracy Allen Posts: 6,656
    edited 2020-02-14 05:38
    The way I see it, suppose it is in the state where the switch has been closed for a while, the SCR and the bright LED are turned on, and the capacitor is charged to almost 12V. Now you open the switch. There are two current paths that discharge the capacitor.
    Screen%20Shot%202020-02-13%20at%209.14.36%20PM.png
    The red path flows through R1 and around through the SCR. That path drops out as soon as the red current drops below the SCR's holding current.
    The path shown in blue is the slow discharge, and it needs R3, because otherwise the gate-cathode junction of the SCR blocks the flow in that direction.
    Adding the diode shown in yellow in series with R1 would eliminate the fast discharge path and would allow a relaxed selection of values for R1 R2 R3 and C. It may be that the current through R1=10k is already below the holding current of the SCR in question, but there is a lot of variation in SCRs.
  • jmgjmg Posts: 15,140
    edited 2020-02-14 08:50
    Adding the diode shown in yellow in series with R1 would eliminate the fast discharge path and would allow a relaxed selection of values for R1 R2 R3 and C. It may be that the current through R1=10k is already below the holding current of the SCR in question, but there is a lot of variation in SCRs.

    Remember the SCR includes the LED load, which will be a few LEDs in series, so the current fall-off will be zener-like.
    but you are correct that LED load pathway is largely unknown, and if you wanted to remove field unknowns of LED loads (eg some may be SMPS), then a series small signal diode isolates R1, and makes the break-timer elements essentially R2 and C2.

    If the (somewhat unknown) LED load includes a supply cap, that can still disturb things.


  • I may off on this one, but in the redrawn schematic, there is also a discharge path for C1 through R1->R4->LED2 until the voltage is low enough for the led not to conduct. Effectively in parallel with R2. Because R1+R4 is so much smaller than R2, R2 will really not affect the time constant until the LED2 no longer conducts leaving only the path through R2.

  • BeanBean Posts: 8,129
    Here is the circuit that I actually got working.
    I changed it so that when it is in "power saving mode" it just adds another LED to the string instead of having a seperate LED for "dim" mode.

    This works but has the interesting operation that it always starts "dim" for about 1/2 a second before it goes bright. Not sure why that is happening, but I think it is a good thing.

    Uses about 50mA in high power mode and about 16mA in low power mode.

    Comments welcome.

    Bean
    943 x 787 - 49K
  • jmgjmg Posts: 15,140
    Bean wrote: »
    This works but has the interesting operation that it always starts "dim" for about 1/2 a second before it goes bright. Not sure why that is happening, but I think it is a good thing.
    Sounds like something is marginal.
    Did you find C1 was needed ?

    With the C2:C1 split you have there, the current from R1 is going to split between C1 & C2, so the ~ 250uA is going to be maybe 25uA into the gate.

    If vC2 is 0, vC1=0, there is a 50ms delay until C1 gets to Vg, and if vC2 residual is -1V, that is ~ 150ms.
    With no R2 the fast-discharge value of C1 is determined by the led forward drops, but there is a 22 second time constant of R3.C1 to eventually discharge C1 to ~ 0.5v.
  • Tracy AllenTracy Allen Posts: 6,656
    edited 2020-02-17 21:50
    Curious, I breadboarded the circuit using an SCR (S106F) from my junk box. A slightly different configuration from Bean's, but it works great.
    SCR_2mode.png
    It uses 4 green LEDs I had on hand, and turns them on in series at 70mA. If the switch is subsequently turn off but turned on again within 5 seconds, the SCR does not re-fire, and the LED are dim at 13mA through the 330Ω resistor R5 in parallel with the SCR. This uses only the one capacitor, but I did verify my hunch that the small signal diode in series with R1 is necessary to achieve the long refractory interval. With that diode in place to block the reverse current through R1, the choice of the other components is forgiving.
    274 x 257 - 21K
    576 x 432 - 75K
  • jmgjmg Posts: 15,140
    Because the choices for sensitive gate SCRs is quite limited, and because their forward voltage drop here is wasted energy,
    I thought I'd look for a variant design that can use a Power MOSFET instead.

    See attached for a design using Complementary Pair Bias Resistor Transistors R1 = 47k, R2 = 47k (SOT-363 package MUN5313DW1 are 4c/3k)
    This is still SCR in nature, just built using dual biased transistors. D2.R2 and D3.R3 are another MUN5313DW1, wired as series Diode+resistor(B-C).
    BOM is 2 active packages + RC timing element + MOSFET.

    I think there is also another solution using 74LVC1G175 (D-FF,/R) + MOSFET, but that needs a regulator of some sort (LED maybe or zener?)
    1614 x 942 - 63K
  • Tracy AllenTracy Allen Posts: 6,656
    edited 2020-02-27 03:21
    Here is a design that uses a programmable unijuction transistor (PUT) in a circuit that transfers the power handling over to a mosfet. The 2N6028 PUT in the TO92 case could handle the 70mA load on its own, but I'm following jmg's lead.
    PUT_MOSFET_2level.gif

    PUTs differ from SCRs, in that the gate is middle N instead of the middle P in the PNPN structure. It turns out that PUTs are very sensitive, having sub-microamp trigger current and mA level holding current, and that is what makes them useful in timing and trigger circuits. You don's see them used much at all these days.
    Too bad, they can operate over a wide voltage range at low power. The trigger and hold currents and voltages are "programmable" through the choice of the resistors attached to the gate. The unijuction aspect is a misnomer, as it is actually a 4-layer, 3-junction device, but it does act much like a unijunction transistor where the characteristics are fixed in the silicon. Costwise, a PUT unfortunately is a more expensive, being uncommon, than jmg's pair of prebiased transistors.
    718 x 398 - 15K
  • kwinnkwinn Posts: 8,697
    PUT's and SCR's are interesting devices that can be used in some unique circuits such as the ones in this thread. Unfortunately they are not used often these days so they are relatively expensive and often have long delivery times. In some cases it may make more sense to use one of the sub $1.00 microcontrollers instead. I was going to use the SCR/PUT circuit for the trailer I am restoring but have decided to go with the P1 to create a more complete "trailer automation" system to control lights, hvac, battery charging and anything else that will benefit from being automated.
  • Ahh, feature creep! Then, you need 3V for the processor, level shifters up and down, a user interface, documentation for the user interface, ............!
  • jmgjmg Posts: 15,140
    kwinn wrote: »
    PUT's and SCR's are interesting devices that can be used in some unique circuits such as the ones in this thread. Unfortunately they are not used often these days so they are relatively expensive and often have long delivery times. In some cases it may make more sense to use one of the sub $1.00 microcontrollers instead..

    Yes, just looked up UJT on Digikey, and they do come in SMD SOT23, but the price is $2.24053/3k !!
    That must make it the worlds most expensive SOT23 and a reel of those needs an armoured truck ;)


  • The SOT23 version of the 2N6028 PUT is a new thing in the last couple of years, when Central Semi took on the business. But right, yikes, over $2 each for either the TO92 or the SOT23. A few years ago I bought a bag of almost 500 of them for pennies each when Mouser was getting rid of non-rohs stock. Why so many? I was using them in an easy low current circuit to indicate the state of 12V SLA batteries via flashing an LED.

    It used to be that every incandescent lamp dimmer had a PNPN triac as its control element, cheap, cheap commodity items. Then came florescent lighting, and now LED lighting, and intensity control is a hugely more complicated proposition. Triacs are going to fade away.

    Still, $1 for a small sensitive gate SCR such as this one, readily available, that is not bad for a basic circuit in @Bean's RV.

Sign In or Register to comment.