Thank you for reading this. This post has four sections: purpose; parts, values/models, quantities, and datasheets; schematic, calculations, and other relevant info; and questions.


  • To create a circuit that can oscillate three IR LEDs at 37kHz.

Parts, Values/Models, Quantities and Datasheets

Schematic, Calculations, and other Relevant Info Schematic and TLC555 Resistor Calculations enter image description here

  • Regarding the tolerance for the output frequency: it should be 37kHz ±2kHz
  • Regarding the value of C1: that was the smallest ceramic capacitor I already had in stock.
  • Regarding the value If I1: in a separate circuit—consisting only of a 5V supply, 1 IR LED, and 1 470-Ohm resistor—the current value was 8mA. Thus, for three parallel IR LEDs in a similar situation, total current draw should be 24mA.


  • Is this design expected to work (oscillating the IR LEDs at 37kHz)?
  • If it does work, how sensitive is this design to parasitic capacitance (I work around a lot of electronics in my office)?
  • Lastly, any other suggestions/ ideas?

Thank you. If anything's unclear, please let me know.

EDIT: edited 36kHz to 37kHz

EDIT: added tolerance for output frequency.

  • \$\begingroup\$ Have you tried running a simulation? That would be a good check of your design. \$\endgroup\$ May 18, 2021 at 20:35
  • 2
    \$\begingroup\$ @ElliotAlderson no I haven't—thank you for the suggestion. What software would you recommend on that? \$\endgroup\$
    – orjün
    May 18, 2021 at 20:36
  • \$\begingroup\$ 200 ohms sounds a bit low for a CMOS 555, MLCC is not ideal if you want stable timing, a polymer film one would be better. \$\endgroup\$
    – Neil_UK
    May 18, 2021 at 20:41
  • 1
    \$\begingroup\$ The transistor would most likely get damaged if base is connected directly to 555 output. But which frequency you want, you talk about 37 kHz and 36 kHz? And at what accuracy? Are you sure you reach your accuracy level with a 555? Also, before even considering feasibility, what is this circuit supposed to do to assess if it can do it? What would blinking IR LEDs with constant 30-something kHz carrier achieve? \$\endgroup\$
    – Justme
    May 18, 2021 at 20:46
  • \$\begingroup\$ @Justme thanks for the response. (1) I edited the 36 kHz back to 37 kHz; that was a typo, my mistake. (2) My intended accuracy is 37 kHz ± 2kHz. (3) I have an IR receiver that sends either a HIGH or LOW output depending on if the IR beam is established or broken. This way I can test to see of the circuit's working. \$\endgroup\$
    – orjün
    May 18, 2021 at 20:59

3 Answers 3


No, this is not feasible, for many reasons.

The 555 transmitter parts would need to be extremely precise and expensive for getting within 5% tolerance for the continuous infrared carrier, but it's just that the infrared receiver you are using is not compatible with continuous infrared carrier.

You are using the TSOP38238 which in the datasheet it specifically says that continuous signals are filtered out. They are not suitable for continuous reception of carrier wave, but for bursts of carrier wave.

The most likely outcome is that TSOP38238 output will go active low for few milliseconds, and then the AGC starts to filter out the continuous tone, and the output will return to idle high.

That is how most infrared remote receivers work, and a different kind of receiver is required to receive continuous carrier wave for beam break detection, or the carrier wave must be sent in bursts for detection with this receiver.

Another thing is that the TSOP38238 receiver sensitivity drops to half when the carrier frequency is 5% off. Do note that the 555 itself has some initial tolerance, likewise all the timing resistors and capacitors. You would need much more accurate components than 5%, and random amazon-bought capacitor set with bad reviews of out-of-tolerance capacitors will not help to achieve that. At least you would need a trimmer potentiometer for adjusting the frequency, and still it could drift due to temperature and moisture etc.

  • \$\begingroup\$ Got it. Quick question: Specifically regarding the IR receiver I'm using, would the feasibility of this design change if I were to use this one instead: vishay.com/docs/82485/tsmp58000.pdf. It's carrier frequency is 20 kHz to 60 kHz. \$\endgroup\$
    – orjün
    May 18, 2021 at 21:49
  • \$\begingroup\$ @Arian it depends what you are after. That module will provide the actual carrier out, not a demodulated signal. Whether that suits your system is your decision. \$\endgroup\$
    – Justme
    May 18, 2021 at 21:54
  • \$\begingroup\$ I see. The goal is to detect IR breaks between the IR LED and the receiver. Just to make sure I'm getting this correctly, if I fix the outright flaws of my current design (resistor for base current of the 2N2222A, etc.) and given the rather crappy tolerances of my design, would you say that the TSMP58000 receiver would "respond" to this system (or are the tolerances still wildly off)? \$\endgroup\$
    – orjün
    May 18, 2021 at 21:59
  • \$\begingroup\$ @Arian Had a look at TSMP58000 data sheet. Looks like it would give you a steady 38kHz output signal while the transmitter beam is NOT BLOCKED. Reflections off other objects could still generated a 38 kHz output signal while the main beam is blocked...I can change channels with my remote pointed at a wall opposite the TV. As Justme suggests, the TSMP58000 output may require further signal processing to suit your needs...you likely want a final digital signal: blocked vs. not_blocked \$\endgroup\$
    – glen_geek
    May 19, 2021 at 0:37

Your resistor values are much too low. Use a much smaller capacitor and much higher resistor values. You have 0.25\$\Omega\$ for the Rt so the TLC555 will attempt to sink 20 amperes from the 5V supply. Not good. Absolute maximum is 150mA and you really don't want to be anywhere near that high.

You can go down as far as about 100pF for the timing capacitor and still have predictable timing, lower still if you don't mind tweaking.

You need a series resistor on the transistor base. The 15mA is a limit you must ensure, not a current the chip will limit it to.

Note: You should have some idea of what accuracy you are shooting for. You may not get within 10% of the predicted center frequency if you use wildly low or high components. Here is what I think is similar to you are trying to interface with (a Vishay part), but you should use the actual datasheet for the device you are using:

enter image description here

You really should be within 1% or so of the nominal center frequency to maximize the sensitivity.

  • \$\begingroup\$ Got it, thank you! Should I edit the same post or make a new one with the edit? \$\endgroup\$
    – orjün
    May 18, 2021 at 21:06
  • \$\begingroup\$ @Arian Leave the original schematic and edit a new one below it, so that comments already made make some sense. \$\endgroup\$ May 18, 2021 at 21:08
  • \$\begingroup\$ Makes sense. Thank you. Quick follow up question: is the information regarding the accuracy of the output frequency of the TLC555 component (at given resistor and capacitor values) provided in its datasheet? \$\endgroup\$
    – orjün
    May 18, 2021 at 21:10
  • 1
    \$\begingroup\$ Yes, it is, for a specific range of resistances, one capacitor value and a range of voltage. +/-1% typically, +/-3% maximum. Add the resistor and capacitor tolerances and you get another 6% or more. You can adjust with a trimpot if you have a frequency meter. Most of us probably would use a microcontroller with a crystal or resonator timebase. \$\endgroup\$ May 18, 2021 at 21:17
  • 1
    \$\begingroup\$ I would certainly suggest the microcontroller for stability and an eye to the future, though if you're willing to trim the TLC555 it will work. There's no waste in using an 8-pin MCU to replace an 8-pin 555 if it does the job 100x better. \$\endgroup\$ May 18, 2021 at 21:23

There's a couple of things to note:

  • No base resistor on the NPN - need to limit base current
  • Your RC values aren't reasonable (cap too big, R's too small.)
  • Don't ground pin 5! (Control Voltage)
  • The LEDs will be happier/more efficient if you drive them with a low duty cycle (this is typical for an IR remote.)

In the spirit of using the 555, I worked out an astable oscillator that gives about 10% duty cycle at 38KHz.

Simulate it here

enter image description here

What's different:

  • 1k base resistor
  • 1nF cap / 22k ohm basic RC
  • Feedback from the output instead of DISCH
  • bypass cap on Control Voltage
  • diode/resistor pair sets ~10% duty cycle for LED
  • much higher current drive (about 250mA peak) for LEDs

This circuit uses the OUT to charge and discharge the cap. With the addition of the diode and resistor as shown, the charge and discharge times are independently set: 'charge' time is about 1/10 of discharge.

The LEDs get hit with a high peak current (about 240mA) which the IR receiver can see better, with less power used in the process. With typical T1-3/4 LEDs this should work as far away as 10 meters. In reality, just one LED would do it. (speaking from firsthand experience designing remotes.)

A word of caution: the LEDs will be very bright, as in, don't-shine-in-your-eyes bright, since your eye reflexes (iris, blinking) don't react to IR like they would visible light. I'm not kidding around about this: the IR LEDs will be intense enough that they could damage your eyes if you stare into them long enough.

That all said, have a look at this gizmo: https://learn.adafruit.com/tv-b-gone-kit/download

  • \$\begingroup\$ Thank you! Quick addition and question: I plan on using this circuit to establish an IR beam with an IR receiver (TSOP38238, farnell.com/datasheets/2049183.pdf). I then plan on detecting IR breaks (so the beam is going to be on for a long time, until physical IR-opaque entity cuts the beam). Is this circuit feasible with that? \$\endgroup\$
    – orjün
    May 18, 2021 at 22:10
  • \$\begingroup\$ Quick related follow-up: Would you mind sharing with me which software you used to design that circuit? \$\endgroup\$
    – orjün
    May 18, 2021 at 22:11
  • \$\begingroup\$ It's an online simulator called Falstad. It's a simplified SPICE analog sim that uses Javascript. You can edit the sim that I provided in the link and play around with it. \$\endgroup\$ May 18, 2021 at 22:21
  • \$\begingroup\$ And yes, you could use the TSOP type sensor to detect the beam presence, even a continuous one. \$\endgroup\$ May 18, 2021 at 22:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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