Alternatives to RC circuit for simple variable delay up to 2 minutes

Question

I'm looking for a simple and cheap way of creating a variable delay of up to about 2 minutes.

schematic

^{simulate this circuit – Schematic created using CircuitLab}

I want to know if there's an alternative to using an RC circuit that is simple and cheap. Other concerns are secondary, within reason.

The problem with this circuit is it just seems a bit 20th century.

Firstly, the capacitor needs to be large physically because of leakage current and the small current required to turn Q3 on. This means the cost possibly leaves room for improvement. The size, weight and lifespan can probably be improved. The delay consistency can certainly be improved. (Since the capacitor's internal leakage and other leaks will vary significantly with temperature, humidity etc). Plus there's the surge when the cap gets charged.

Are there any tricks I could use? I'm wondering if there are any off-the-shelf ICs which are dirt cheap that perhaps combine a small RC circuit and some way of configuring the delay that is perhaps implemented by a binary counter.

I'm still playing around with ideas so I'm not sure what there will be in the way of power rails; if necessary I could make a 5v rail available, but it would be useful to know if there's a solution that could work on as little as 1 volt. So this question is less about solving my specific problem, but checking that I'm not oblivious to some general purpose alternatives.

[Re: duplicate: This is not a duplicate question. The answers given for the question identified as a possible duplicate offer no improvement as requested by my question. A 555 timer still uses an RC circuit without a solution for a large capacitor!]

I've done this with a Schmidt trigger inverter/ buffer (74HC14 or similar) — George Herold, Sep 27 '16 at 16:05
Outdoor motion detectors for floodlights use the simplest circuit which use high CV rated plastic caps (polyester or polystyrene film) and leakage < 0.001CV leakage do trigger a diac which triggers a charge to fast turn on a Triac switch with timer values in many minutes for the cheapest AC timer switch. That's what the industry uses. not PIC's with PD sensor IR emitters 2 pots and production BOM costs of a couple bucks. — Tony Stewart EE75, Sep 27 '16 at 17:24
Linear makes some time delay ICs that run in the minutes to days range — Voltage Spike, Sep 27 '16 at 18:35
@Jodes It was a (brief) suggestion, not a question, but now you've pointed out that it doesn't solve the problem anyway. — user253751, Sep 28 '16 at 03:02

score 11 · Answer 1 · answered Sep 27 '16 at 16:04

Depending on how you measure "simple", a small microcontroller is a good answer. Even the tiny PIC 10F200 can do this easily, and comes in a SOT-23 package. That's the same package single transistors come in, although it has 6 leads instead of 3. The only other part you need is the bypass cap. This circuit will be physically smaller than the one you show, and in volumes probably not much different in price considering the large cap yours requires.

The 10F200 comes with a internal oscillator good to a few percent, which is better than a cap you can get that wouldn't cost more than this micro. This PIC can also have internal passive pullups on some inputs, so the switch connects directly between a pin and ground without any other parts required.

The delay is done digitally by counting cycles. Delays done that way can be arbitrarily long, and have the accuracy of the internal oscillator.

Depending on what you are trying to drive, you might be able to do so directly from a output pin. Otherwise, you still need a transistor with base resistor, or a "logic level" FET with no other parts.

One advantage of using digital logic for the delay is that the transistions will be crisp and essentially instantaneous. In your example, the output transistor will fade out slowly.

@Tony: True. It is one valid solution, but that might or not apply in this particular case. This is exactly how I would do it, and probably most professionals would do it. Any professional designing this sort of stuff has to be able to use a microcontroller easily when the criteria call for it, as they do in this case. Digital is really the way to go here. As your own answer points out, leakage is a big problem at these times for analog parts. It can be overcome, that adds cost and isn't as "simple", depending on what you find simple. — Olin Lathrop, Sep 27 '16 at 17:34
An 8 bit binary counter can effectively multiply any RC timing circuit's time constant by 128 and doesn't need a development environment and coding skills. It will also provide digital crisp transitions. — Ian Bland, Sep 27 '16 at 23:45

Tony Stewart EE75 · Accepted Answer · 2016-09-27T16:55:30.607

The problem with big caps is leakage current and the equivalent time constant possible for long term delays may not support 2 minute delays.

Also the bias current of the switch adds to this leakage current and further reduces the possible time delay, T.

Let's examine some random components.

Panasonic HA series, Type V Alum. Elect. SMD Vdc = 6.3 to 100V

Spec:
- I < 0.01 CV or 3 (μA) After 2 minutes (Whichever is greater)
- We know that Ic=C*dV/dt and Ir=V/R
- Using 470uF @6.3V, CV=3000u or Ic<30uA which is greater than 3uA
- using 5V Rleak>=5V/30uA >= 167 KOhm and RC=T=78 sec,
  - so that's no good since you want T to be >> 120 sec in order to add a fixed R and reduce that to 120 sec, choose T =200 sec or 3x bigger than this cap minimum requirement
    
    this means your cap leakage must be <10uA @5v or CV> 10,000u so,if we choose a >= 25V rating we are OK
- But it means our switch must also use much less than 10uA or we have another leakage problem and your latest schematic with two 470k in parallel, one being a pot.
- is (5V-0.7Vbe)/245k=18.2uA which looks like a problem.
- but wait, T is the exponential estimate where 1/e = 0.367879441 times its initial value. but here you are decaying from 5 to 0.7Vbe or 14% of the initial value, so this ore than doubles but not triples the time before the transistor switches off,
- but wait again, the transistor is saturated so current gain is reduced to around 10 , but if we come out of saturation it will restore hFE.
  - So when does the relay come out of active duty? We call this holding current the " MUST Switch off" voltage threshold which is always lower than "may switch off which is never specified.
- Let's say your relay coil needs 50ma and your NPN has an hFE of 100 but "Might" turn off at V/2 or 25mA
- this means your base current must be 250uA to keep the coil active
- ** but we know that exceeds the leakage spec on our Timer cap, so it turns off faster.**
  - so you can see there are a few variables that affect the time delay
to make matters worse , if your relay load has any inductance, the contacts will,arc for a longer time during very slow switch off and may chatter and buzz from stray hum ingress into the driver.

This is certainly not the best 2 minute timer delay design, but an analysis of what can go wrong with your current design.

Don't forget inductive coils switched low will release a high voltage spike on the semiconductor switch (Collector ) so,standard practice is put a reverse biased diode across the DC coil to shunt the turn off spikes.

Some designs for AC timer relay use a digital CD4060 timer chip and FET and a plastic cap with > 5 minute decay from self leakage then choose any binary /2^n count out with Reset . Others rely on leakage and use a 555 CMOS timer.

Typically the leakage is not as much as the spec sheet says. I've used 100uF tant's without any problems (room temperature only) and have charged up Al electro's and come back the next day to still find it with ~90% of it's charge. (oh and hardly any current is drawn by 74HC logic.. except when switching. :^) — George Herold, Sep 27 '16 at 17:16
It would be great if you could expand on the comment you made under the question; techniques like those of optimised, mass produced goods are hard to come by — Jodes, Sep 27 '16 at 18:12
ok Jodes, I'll see what my fat fingers can come up with , on an iPad;) — Tony Stewart EE75, Sep 27 '16 at 18:59
wouldn't you rather have a battery operated RED LED solution that gets dimmer after 2 minutes when the lights turn out. for a darkroom... — Tony Stewart EE75, Sep 27 '16 at 19:10
Lol I suppose my circuit does look like it might be used like that :-/ the BJT was a poor choice to explain the problem — Jodes, Sep 28 '16 at 02:44

Simon · Answer 3 · 2016-09-27T17:04:22.017

0

Use an NE555 in monoflop configuration in the following way. That should mimic the behaviour of your circuit. Choose R1 properly or chain resistors and potentiometers for time window, coarse and fine.

schematic

^{simulate this circuit – Schematic created using CircuitLab}

edited Sep 27 '16 at 17:04

answered Sep 27 '16 at 16:32

Simon

312
1
6

-1 which 1uF ceramic cap has a leakage time constant of > 2 minutes? many Murata parts are only 50 seconds or 50 Ohm-Farads – Tony Stewart EE75 Sep 27 '16 at 17:00
@TonyStewart: Your totally right on that. – Simon Sep 27 '16 at 17:08

Pete Kirkham · Answer 4 · 2016-09-27T23:21:47.790

-1

I built this to hold some LEDs on for a few seconds when a vibration switch is triggered to make a light-up origami wand for my daughter. The NPN and NFET are fairly generic SOT 23 devices and the passives were 0805. As the FET has a high input impedance the time is determined by the resistor. The values below give about 2½ seconds, so multiply R by ten and C by five to give about two minutes. Although it's still a simple RC it doesn't have to be large physically.

schematic

^{simulate this circuit – Schematic created using CircuitLab}

edited Sep 27 '16 at 23:21

answered Sep 27 '16 at 16:16

Pete Kirkham

1,976
12
16

-1 Which caps have a time constant of 36 minutes do you suggest? – Tony Stewart EE75 Sep 27 '16 at 17:03
@TonyStewart 220uF times 10MΩ = 2200 seconds = 36minutes, 220uF being the largest value 0805 in a quick search on farnell so I haven't checked their leakage, but the point was that there a much larger capacitors in small form factors than are required to give the two minutes the OP requires. – Pete Kirkham Sep 27 '16 at 23:05
I would advise you to check if you can find one more than 2 minutes leakage rate – Tony Stewart EE75 Sep 28 '16 at 00:24

Alternatives to RC circuit for simple variable delay up to 2 minutes

4 Answers4

Don't forget inductive coils switched low will release a high voltage spike on the semiconductor switch (Collector ) so,standard practice is put a reverse biased diode across the DC coil to shunt the turn off spikes.