existing flasher without low load detection

The circuit above is an existing flasher without low load detection. I am trying to develop electronic 2-wheeler flasher with low load detection. While Googling, I got a circuit with relay in it, one relay used for flashing and another one for load detection, I mean when there is a load changes then the flash rate doubles. Specifications which I am looking for:

flasher with relay

  • 21W or 45 W flasher,
  • operating voltage- 9 to 12V,
  • Flash rate-85 c/m,
  • Indication of the bulb failure- flashing frequency doubles at single lamp failure,

Can I replace the flash relay with MOSFET? Load sense relay by sensing resistor and comparator? I actually don't know how I have to implement the low load detection in the circuit, since I am a beginner.

  • 1
    \$\begingroup\$ Are there constraints on cost or using an IC or using a microcontroller or ...? This can be done easily with a '555 or with a 74C14 (used as a timer circuit) or in various other ways. || What are the connections - I assume 31=ground, 49=battery 49a = load. Yes? \$\endgroup\$
    – Russell McMahon
    Jun 18 '14 at 10:37
  • 1
    \$\begingroup\$ What is the end target - are you trying to replace an existing part (sounds like this from your spec) and is this for one or a few or many. We can help regardless but knowing things like this help guide the answers. [I was in India in MArch and planned to travel Chennai-Bangalore-Goa-Pune but got sick in Chennai (as one does :-) ) and so went Chennai-Pune direct by train. Sleeper class! - lots of fun. ] – \$\endgroup\$
    – Russell McMahon
    Jun 18 '14 at 10:47
  • \$\begingroup\$ yes i do have a cost constraints. i made a flasher with RC oscillator but it dint had low load detection and its the main feature which people are looking for,but since i am a beginner,facing problem to achieve it .yeah i am trying to add this feature to an existing discrete flasher. @RussellMcMahon . [ its nice to know you liked India that too by travelling in train in sleeper class,i really cant imagine , if you come india again visit Kerala(My native)most beautiful place i believe,u will surely love it] \$\endgroup\$
    – yasmi
    Jun 19 '14 at 3:58
  • \$\begingroup\$ with micro controller is it possible to make ?? @RussellMcMahon \$\endgroup\$
    – yasmi
    Jun 19 '14 at 4:02
  • \$\begingroup\$ Yasmi - see Schmitt trigger addition to my answer plus comments re microcontroller. You could add something like my load detection circuit to the analog circuit at top. \$\endgroup\$
    – Russell McMahon
    Jun 19 '14 at 11:50

Adding blown-bulb 'double flash rate' feature to your original circuit:

Here is a suggested modification to your circuit.
As with all such analog circuits involving somewhat variable associated products you may need to fine tune it for best results . (This is just as true for IC based products BUT in some cases no means of adjustment is provided by the IC maker).

As you can see, the circuit is a cut and paste using your cct as the main material.

  1. The first thing to check is whether R2 controls flash rate as I think it does. Make R2 = 100k or place a 220k across the existing resistor.
    What happens to the flash rate?

IF the flash rate ~~= doubles then procede to 2. Place a

  1. There may be polarity issues around C2. Try this and see.

  2. Rs senses flasher current.
    Size Rs such that when current is high (2 bulbs) V_Rs > 0.6V and Rs turns on.
    R1 protects Q3 base against excessive voltage on Rs
    R2 turns Q4 on using Q3.
    C3 provides off time delay to maintain Q3 on between flashes.
    R3 can be large - just ensures Q4 off if Q3 off and no stored voltage in C3.
    Q4 adds R4 in parallel with R2 to increase flash rate.


enter image description here

  • \$\begingroup\$ i tried this circuit its flashing but no changes when load reduces Russel ,it may be because i am not good at electronics . i dont know where i am going wrong. thank you so much for guiding \$\endgroup\$
    – yasmi
    Jun 20 '14 at 11:01
  • \$\begingroup\$ @yasmi - First test to see if I have got timing resistor correct - place a 100k say across R2 and see what effect it has on flashing rate. IF this works then the circuit should be usable. Rs resistance must be set so voltage drop is > 0.6V with 2 bulbs and < 0.6V with one bulb. | Whoops - I just realised that this works backwards with slow flash for 1 bulb and fast flash for two. BUT this is able to be corrected. | Please try placing 100k across R2 and then we can go from there. \$\endgroup\$
    – Russell McMahon
    Jun 20 '14 at 13:40
  • \$\begingroup\$ @yasmi - OK- TWO tests - In your original circuit please do these and report. (1) Connect 100k across R2 and report the effect on flashing. Remove 100k then (2) Connect 10 k across R3 and report effect on flashing. \$\endgroup\$
    – Russell McMahon
    Jun 20 '14 at 13:43
  • \$\begingroup\$ Hi Russel, Flash rate is increased when i connected 10K across R3, when i connected 100K across R2 there is no flashing,light is just on . \$\endgroup\$
    – yasmi
    Jun 21 '14 at 7:37

In the old days of thermal flashers used in automotive applications, the faster flashing when the load was reduced (due to a burned out bulb) was a by-product of the way they worked (less current meant that the bimetal temperature overshot less, so they flashed faster). This is considered a useful feature now, but it's not "natural".

One method would be to use a microcontroller and measure the current (very crudely is good enough). Something like a 6-pin or 8-pin PIC would do the job. Timing accuracy will be more than good enough using the internal clock.

Here is an example of a commercial flasher that uses a current measurement shunt. You can see the DIP chip hiding underneath the relay. The current measurement is by measuring voltage across the red-circled "M"-shaped stamped piece of metal (which is a very low value resistor). The author of this web page modified the flasher so it would not interpret the lower current of an LED retrofit bulb as an incandescent bulb that had failed.

enter image description here

Edit: If you want to use a PRC ASIC, you should do a proper market search, but here is an example one:

enter image description here

You should use about 110K for R1 to get the desired 85cpm flash frequency. The threshold for current detection is typical 51mV, so the voltage drop is quite low.

  • \$\begingroup\$ How to develop with micro controller i have no idea.I know programming but very weak in electronic circuits. \$\endgroup\$
    – yasmi
    Jun 19 '14 at 4:12
  • 1
    \$\begingroup\$ If this is a serious volume application I would probably recommend you use an ASIC from the PRC. \$\endgroup\$ Jun 19 '14 at 10:19
  • \$\begingroup\$ See edit regarding ASIC. \$\endgroup\$ Jun 20 '14 at 0:36
  • \$\begingroup\$ Cannot use ASIC because of cost and Dimension constraints.and no relay also. Thank you @spehro \$\endgroup\$
    – yasmi
    Jun 23 '14 at 3:09
  • 1
    \$\begingroup\$ @yasmi ASIC should be ~Rs.15 or better, which will be hard to beat. It could drive a BJT or MOSFET. \$\endgroup\$ Jun 23 '14 at 3:41

NOTE: Tutorial (of sorts) alert.
The hex Schmitt trigger based flasher circuit towards the end of this answer is as much an example of what can be done with a simple "digital" IC by using somewhat unusual methods. A RTL OR gate is used, a diode OR gate and a few other strange techniques.
Once it is seen what can be done in this manner 'all sorts' of useful circuits can be implemented at low cost. The usual downside is a higher parts count of low cost parts and some interim head scratching.

You can replace the flasher relay with a P Channel MOSFET with gate driven by Q1 and a resistor from gate to source (to allow oscillator to work as before.) They show what appears to be 400 Ohms across coil implying this is in parallel with coil resistance. If so then actual value will be lower.

MOSFET source to terminal 49.
MOSFET drain to terminal 49A.
MOSFET gate to Q1 directly.
400 Ohms or less gate to 49.

Load sense relay operates when current is high enough and shorts 200 Ohm Q2 collector resistor. Effect obscure without further head scratching but affects time constant of Q1 Q2 100 uF oscillator feedback loop.

The load sense relay could be replaced by a comparator and FET BUT as long as about 0.6 Volt drop was acceptable across the controller, you should be able to use simply a small transistor with a sense resistor across its base-emitter. If I_load caused > 0.6V drop in the sense resistor the transistor would turn on and you could use it to operate another small and cheap bipolar transistor to short out the 200 Ohm resistor instead of using relay contacts. The resultant circuit would use no relays and only one MOSFET to provide flashing action.

Below is an example only circuit based on a modified version of the one you provided. Q3 is probably about right - the 400 Ohm resistor that was across the relay coil will need to be lower. valued. If this was the coil resistance then an approximately 400 Ohm resistor will be needed. Q4 sensing probably would work OK but the drive of Q5 is inadequate as shown. A little more thought would produce a working result.

enter image description here

However, if you can provide a good specification of how it should behave then a completely new circuit may be more satisfactory. A 555 or 74C14 or LM358 or LM339 based circuit would probably work better.

Questions from my comments above:

Are there constraints on cost or using an IC or using a microcontroller or ...?

This can be done easily with a '555 or with a 74C14 (used as a timer circuit) or in various other ways. Would a cheap IC be acceptable?

I assume 31=ground, 49=battery 49a = load. Yes?

Are you trying to replace an existing part.

Is this intended to be a one off, or for a few devices, or are many to be made?

Hex Schmitt trigger example (74C14 40106 4584 ... ):

This is as much an example of what can be achieved with a single package of 6 x Schmitt triggered inverters. A solution along these lines could be made to work well (with some playing probably being needed to deal with inrush current when 1 bulb is present (see text) and tuning of Rs to allow reliable high/low current differentiation.
BUT I'd probably use a microcontroller if at all possible.
BUT if you understand what this circuit does or tried to do then writing a microcontroller program would be much easier.

Below is a circuit diagram (that Olin may not approve of) intended only as an example.
This is "out of my head" and untried but may even work :-).

IC1 - hex Schmitt triggered inverters.
NB !!!!!!! MUST be suitable for automotive supply voltages. If used on 12V then 15 or 18V rating preferred.
Note that eg 74HC14 is usually 6V max Vdd.
This Digikey search provides possibly suitable parts.

20V - STM 40106

18V TI 40106

18V ON-Semi 4584


Note that some hex Schmitt trigger inverters have lower hysteresis ranges than normal so oscillators run faster with the same RC values.

Circuit operation:

Rz, Z1, Cz at top left provide a Vdd supply for the IC slightly below Vbattery min to help keep flash rate somewhat stable.

There are two flash oscillators, fast and slow. While it would be possible to alter the rate of a single oscillator, because there are 6 inverters available, this is an easy approach and allows independent setting of fast and slow rate. It has good and bad points.

R1 C1 and one inverter form an oscillator and R2 C2 and another inverter form another. This is a very standard circuit when using a Schmitt triggered gate. It will not work with a non Schmitt-triggered gate.

The heart of the current sense for slow/fast swapping is RS and Q1. When bulb current causes a drop across Rs of more than about 0.6V Q1 turns on. If current is too low to cause 0.6V drop OR if lamp is off then Q1 will be turned off.
So as R = V / I, for Ihigh Rs > 0.6/Ihigh and for Ilow Rs < 0.6/Ilow. A value of Rs should be able to be found that allows Q1 to turn on only when the high current load (2+ bulbs) is being powered.

When Q1 turns on point B (circled) is pulled high (to about Vin). As Q1 cycles on and off with flashing capacitor Cd bold point B high during off halves of flasher cycles. Rd discharges Cd when Vin is removed in preparation for next poweron or to allow rapid response if a bulb fails.

Diodes D1 and D2 are oscillator gating switches. When the voltage to the left of these diodes is low the diodes conduct, Capacitor C1 or C2 is discharged to low and the related oscillator is disabled. The oscillator gate outputs are high when disabled (as input is clamped low) and as will be seen, the gate cannot then drive the output.


Consider the 2 bulbs condition after a few flashes.
Q1 turns on on each flash.
Cd charges high so point C is low so the upper (fast)oscillator is disbaled and point E is high. Diode D3 is blocked.
BUT As B is high, point D is also high so D2 is reverse biased so slow oscillator with R2 C2 is enabled and runs. When point F goes low during oscillation D4 conducts and turns on MOSFET Q2 to drive bulbs. This provides an on pulse for Q1 which keeps point B high and holds system in slow flash mode. The slow oscillator turns Q2 on when output is low and the FET is truned off by Rg when oscillator output is high.

D3, D4 and Rg + Q2 form a DTL OR gate. (Or NOR gate depending on your perspective).


If a bulb fails Q1 never turns on as drop across Rs is too low (by design).
Point B is taken low by Rd discharging Cd.
Point D is now low, disabling the slow oscillator.
Point C is high, enabling the fast oscillator.
Operation is as above but now with the fast oscillator running.

enter image description here


Diode DP3 (top left) provides reverse battery protection IF WANTED.
Diode DP1 provides protection against inductive loads. This should not be necessary with lmp lods but is shown for completeness.
Diode DP2 is an alternative of sorts to DP1 - if excessive voltage is supplied to the output it will shunt it to supply if DP3 is not used. This is optional and not liable to be wanted.

Rs needs to be wattage rated high enough to take flasher current. As Vrs <=~ 0.6V usually the Pr_Rs = V x I = 0.6 x Iload. For 2x 20 W bulbs Imax ~= 40/12 = 3.33 A. Inrush current will be higher than this. Current is present only 1/2 the time. Allow for say 5A x 0.6V = 3W x 50% on = 1.5 W. A 5 Watt resistor is cheap and probably sensible.

Inrush current MAY cause problems due to the peak current at turn on even with a single bulb and you may have to add a resistor between Q1 and Cs to slow the charge enough to avoid inrush charging when a bulb is blown. There are other ways to handle this but a simple RC delay is probably good enough.

See IC data sheets for oscillator frequency but frequency of VERY ROUGHLY 1/(R x C) applies.

Q2 should have enough current rating to handle inrush currents when bulbs are cold.

If IC1 (gates) Vdd is > about 0.6V below Vin then FET gate will be driven slightly positive when off. This is not a problem as long as drive voltages are understood and designed.


Using Asian (China) prices - which should be close to what can be obtained in India in volume:

1 Rupee ~~~= 2 cents US.

IC1 - 10 cents or less. (5 Rupee)
Q2 - 6 cents (3 Rupee)
Z1, 1,D1,2,3,4 1N4148 or similar < 1 Rupee
Other components are "glue" minimal price.
Say 30c - 40c (15-20 Rupee) all up for components ?


Microcontroller = uC.
Needs programming per item and needs program written.
An Arduino would be able to do this at a cost per uC = IC cost. Main advantage of Arduino would be ease of programming for people with minimal uC experience. (Olin will not like this suggestion!)
If the uC has a low voltage comparator it should be able to be used for Rs sensing.
If not (VERY low cost IC) then the arrangement shown for Q1/Rs in above circuit would work.
A basic power supply to denoise Vdd is necessary.
Rz, Z1, Cz from above cct may be enough. A 2 stage zener power supply is cheap and extremely good at noise reduction (repeat as shown with 2nd stage fed by first.
If Vdd is well below Vin then an N Channel MOSFET on ground side of flasher would be useful BUT needs a connection change to standard wiring. So an extra transistor (1 cent or so) to translate drive to MOSFET gate MAY be needed.

"Hanging" uC supply off upper power rail works well and allows direct P Chanllel MOSFET drive. ie Vdd = Vin. uC_ground = say Vin - 8V.

A uC with good anti-brownout and undervoltage reset is a very good idea.

Almost any uC will do. $US0.20 or less in volume.

PIC0F204 - $US).38 / 3000 at Digikey - no doubt cheaper when arms are bent privately, is very suitable. It has a comparator with 0.6V bandgap reference. Put uC in lower leg with say 5V supply and use one transistor to drive high side P Channel MOSFET. Olin would approve :-).
3 terminal regulator can be used for PSU but 2 stage zener adequate and probably cheaper and maybe better.

Questions? Ask ...

  • \$\begingroup\$ Thank you so much .I already developed a flasher with IC VN1160 which is a direction indicator driver for Motorbike, its bit expensive and also looking to develop flasher with easily available components, even i do have a dimension and cost constraints,if the circuit can be simple by using 555 and 74c14 i can try for it,again i have no idea how i have to achieve it. \$\endgroup\$
    – yasmi
    Jun 19 '14 at 4:04
  • \$\begingroup\$ Is it possible to add low load detection circuit to existing flasher circuit?? if not how flasher circuit can be achieved with 74C14? \$\endgroup\$
    – yasmi
    Jun 19 '14 at 6:39

Here is an application note that does exactly what you want.

Design of a Low Cost, 45W Flasher with Short Circuit Protection Using LM2902 Quad Op-Amp and CSD18534 MOSFET
Texas Instruments Application Report SLVA650–April 2014. They say:

  • This Application note presents the design of a low cost, flasher circuit with short circuit protection.

    The design incorporates the entire recommended design feature set for two wheeler flashers and includes low/high voltage operation, half load frequency doubling, and short circuit protection.

While the circuit may appear somewhat complex at first glance, most of the components are compact and readily available and low cost, and circuit complexity could be somewhat reduced.

U2 is an LM2902 - a basic quad opamp - available for about $0.12 in 1000 quantity from Digikey and for somewhat less in China (and probably in India). An LM324 could be used here.

Their chosen MOSFET seems to have a rather higher rated current than I'd have expected (100A) buit other similar parts are readily available eg IPP80N06S2L - also having logic level gate drive and similar V and I ratings. Either of these FETs is around $0.60 in 1000 quantity, but a lower current lower cost part may suffice. (See below re cold-filament inrush current). Apart from the opamp and mosfet the other parts are 3 cheap bipolar transistors and a handful of Rs Cs and Ds.
R12 provides bulb current sensing to trigger the 1 bulb / 2 bulbs flash rate change.

The circuit is extremely poorly drawn with it not always being evident whether lines join or cross. Accordingly, care will be needed to recreate the circuit correctly. In most cases it will be obvious which applies, and when a "dot" has been shown at an intersection a join can be assumed, but at other locations it is not so obvious. For example, C2, C3 & R4 all almost certainly connect at a single point, the top of R2 probably does NOT connect to U2A output (but no guarantees). I've advised TI about these issues and hopefully they will provide an update "sometime". May be a while :-).

enter image description here

Incandescent bulb filaments have a very much lower resistance when cold than when hot, so when full voltage is first applied to a cold filament the current draw is many times what it will be when the bulb reaches operating temperature.

Accordingly, a key factor in the use of a high current switch is the massive inrush current of a cold incandescent bulb filament. While for eg 2 x 21 Watt indicator bulbs in a 12V system the operating current when warm = Power/Voltage = (2 x 21W)/12V =~ 3.5A, the initial "cold filament" inrush current may be 10 to 20 times as high - so in this case 35 to 70A. Operate 2 x 21W bulbs on 6V (as in some motorcycles) and theoretical cold-filament current could notionally be over 100A. In practice wiring and connector resistance and possibly battery capability will (hopefully) usually reduce this to merely "many tens of amps".

A circuit simulation and reference schematic are also provided.

  • \$\begingroup\$ Hi @Russel i did it in another way.It worked fine :) & thanks allot for the support. \$\endgroup\$
    – yasmi
    Aug 18 '14 at 5:48
  • 1
    \$\begingroup\$ Hi @Russel i did it in another way,similar to the circuit you provided above.It worked fine :) thanks allot for the support.I have lots of doubts in electronics but this stack exchange banned me from asking further questions :(. \$\endgroup\$
    – yasmi
    Aug 18 '14 at 6:11
  • \$\begingroup\$ @yasmi It does not look like you are banned in any way. I may be wrong but everything looks normal. At one stage I think they temporarily closed one of your questions aand 'people' complained and it was reopened. They sometimes close questions when they need improving and theoretically open them again when they have been improved. | You could try asking another question to see what happens. | I'd be interested in seeing the circuit that you ended up using and what components it uses. \$\endgroup\$
    – Russell McMahon
    Aug 18 '14 at 11:31
  • 1
    \$\begingroup\$ i donno why,i still not able to post any questions,i am blocked.Is there any other way to contact you??? any other social networks?? you are more supportive, and your way of explaining is just awesome,very good professor for a beginner.I got many ideas from you. \$\endgroup\$
    – yasmi
    Aug 20 '14 at 10:43
  • \$\begingroup\$ @yasmi - (1) I will ask the moderators if you have been barred and if so why and what can be done. Or if not, why you may be having problems. (2) My email address is on my profile page. (3) An excellent site that encourages question and answers of all sorts is the "PICList". It covers many other subjects as well as PIC microcontrollers. Many members with a vast range of experience. Read about it at www.piclist.com and then follow instruction re joining at bottom of page (on another site). (4) "... professor ..." -> :-). No. A term reserved for people in the academic world. But thanks :-) \$\endgroup\$
    – Russell McMahon
    Aug 20 '14 at 11:20

enter image description here

The circuit suggested earlier has the right concept to control hyperflashing. However, the logic is incorrect. When the 2 bulbs are present, hyperflash will happen and when 1 bulb is present, normal flashing happens.

  • 2
    \$\begingroup\$ Welcome to EE.SE. It would be nice if you explained why this circuit operates in the manner you have described. Note that "suggested earlier" isn't a good pointer on this site as posts float up and down depending on votes and user sorting preferences. You can edit to mention the author's name instead. \$\endgroup\$
    – Transistor
    Apr 10 '18 at 8:00

To correct the logic, it has to be inverted. Please refer to the second circuit which has an inverted logic and it is tested to work with 85C/min normal and hyperflash to 120C/min with low load. Hyperflash Trigger current is set to about 0.7A. enter image description here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.