low power IR detection circuit

Discussion in 'Embedded Systems and Microcontrollers' started by bug13, Sep 24, 2015.

1. bug13 Thread Starter Senior Member

Feb 13, 2012
1,514
53
Hi guys

I am looking for a ultra low power (under 50uA) circuit to detect 38KHz IR signal, that one that normally uses in TV etc.. Anyone come across one?

Here is what I am trying to do:
I have a TSOP38438 IR Receiver, which demodulate the 38KHz IR signal. But it draws too much current (~350uA) for my application. Instead of having the IR Receiver on all the time. I would like my low power IR detector wake up my MCU when a 38KHz IR signal is detected, then power my IR receiver, then decode the IR signal in my MCU.

Any circuits? idea? or suggestion?

Thanks guys

2. Papabravo Expert

Feb 24, 2006
11,734
2,482
I don't think your low power micro can run fast enough to detect the difference between signal and no signal at that speed. Especially if it takes a while to come out of sleep mode. Have you found a low power IR detector to work with?

3. bug13 Thread Starter Senior Member

Feb 13, 2012
1,514
53
I am working with NEC IR CODE(page 3), it starts with a 9ms + 4.5ms lead code, followed by the data .My goal is to detect the start of the 9ms lead code, it will have plenty of time to wait up my mcu, then get ready to decode the IR signal.

Most of the IR receivers are made by Vishay, and all of they usually draw ~350uA. At this stage, I don't find other low power IR detector IC.

4. Papabravo Expert

Feb 24, 2006
11,734
2,482
I did read your original question, and yes I am familiar with the NEC code. I thought you would be looking at the raw 38 kHz waveform and detecting the sequence of rising and falling edges which are only a bit more than 13 μsec apart.

$\frac{1}{38,000} = 26.31\mu sec$

What you need to do is find a way to have a signal be low(high) while the IR input is beating up and down, and go to the opposite state high(low) when it stops. The "power hungry" device from Vishay is what does that for you. What you need to perform this function with just a raw photodiode, is a long obsolete part (Signetics 8T20) called a double edge triggered one-shot with a short timeout of 50 μsec or so. It was a bipolar TTL device so probably would not have met your power requirements.

Nov 12, 2008
3,659
2,817
6. Papabravo Expert

Feb 24, 2006
11,734
2,482
Don't know if these are still available, and still probably too much current.

http://www.bucek.name/pdf/4538.pdf

I still think finding a way to decode the raw data is the way to go, if the micro can handle it, to achieve your power consumption goals.

7. AnalogKid AAC Fanatic!

Aug 1, 2013
7,053
1,995
A CMOS opamp is very low power. So is a CMOS 4069 hex inverter. One stage of a hex inverter can be biased to act as an inverting amplifier, and maybe turn a photo diode output into a pulse train. Been way too long since I measured the static current of a hex inverter linear amplifier, but it is low.

ak

8. Papabravo Expert

Feb 24, 2006
11,734
2,482
He doesn't need to create a pulse train, he already has one. The way IR protocols work is that a 38 kHz carrier is modulated with ON/OFF keying. The amount of OFF time determines the difference between a zero, a one, and a start sequence. What he needs to do is replicate the behavior of the expensive and power hungry IR sensors to convert the pulse train into a logic level, high or low, and the absence of a pulse train into the opposite level, low or high. For ultra low power operation an opamp based retriggerable one-shot would do the job. Alternatively if the processor has the chops he could do it in firmware.

9. AnalogKid AAC Fanatic!

Aug 1, 2013
7,053
1,995
From what he wrote I thought he was heading toward the latter. Have a super low power signal-present detector, wake up the uC, and let it do all the signal processing.

OTOH, the CMOS hex inverter described above will deliver the 38 kHz bursts nice and square, with 4 inverters left over to missing-pulse detect that into the modulating square wave.

ak

10. Papabravo Expert

Feb 24, 2006
11,734
2,482
Assuming you have the demodulated output, it is easy to use an input capture module to measure the elapsed time between rising and falling edges. With that information you can accumulate the required 16 bits of data and it's complement. Using the input capture module on the unprocessed carrier would be tough to do, with the edges only 13.15 μsec apart.

11. bug13 Thread Starter Senior Member

Feb 13, 2012
1,514
53
I think I need to clarify what I want to do.

I want to have a low power analog circuit that can detect 38Kh IR signal. I don't need the circuit to demodulate the signal, or replacing an power hurry IR receiver.

• At sleep, the Vishay IRreceiver is not powered, so it doesn't use any battery.
• If an 38K IR signal is detedted by my yet to be designed low power 38K IR circuit. The circuit can wake up my MCU.
• My MCU can then apply power to the Vishay IR receiver, the Vashay IR receiver can demodulate the signal, and my little MCU can decode the Signal
• When finish, the Vashay IR receiver is switched of, MCU do something according to the decoded signal, then goes back to sleep.
That is what I intend to do, but I am open to any other suggestions.(as I know that there is so much that I don't know)

I am guessing my MCU needs wake up like 10 -20 times a day at the most.

Last edited: Sep 25, 2015
12. AnalogKid AAC Fanatic!

Aug 1, 2013
7,053
1,995
What kind of time delay can you tolerate between the onset of IR transmission and the output signal from the IR detector circuit to the uC? Do you need to capture the data represented by the first IR carrier burst?

ak

13. BR-549 AAC Fanatic!

Sep 22, 2013
4,119
1,085
Why not just detect IR, and not a modulated IR?

14. bug13 Thread Starter Senior Member

Feb 13, 2012
1,514
53
I am working with NEC IR code, there is a 9ms lead code(9ms 38KHz burst), ideally I want to be able to detect the start of this lead code within first 6ms.

In this first 6ms, I don't need to capture the data represents by the IR carrier burst. And the first 9ms lead code doesn't have any data.

In the later 3ms, I would like to be able to wake up the MCU, apply power to the Vashay IR receiver, wait a bit for the power to stabilise, start capturing the data in the carrier burst.

The full 32 bit data need to be capture, then my application will do something according to the data received.

It's basically an IR remote control application, but the receiver end needs to be ultra low power.

15. bug13 Thread Starter Senior Member

Feb 13, 2012
1,514
53
Form what I understand, detecting IR only works in an enclosed environment. (No other light source)

In an open environment (in a room, or out door), IR is everywhere. It's very hard to isolate the target IR signal.

16. peter taylor Member

Apr 1, 2013
106
5
What about two detectors with one low power IR diode and one high power Vishay detector / decoder.

Tune the IR diode to detect a 38kHz signal, which wakes up your MCU, which reads the Vishay detector etc.

17. Alec_t AAC Fanatic!

Sep 17, 2013
8,816
2,048
Could you use a CMOS (= low power) low duty-cycle multivibrator to switch power to the Vishay IR detector? That would reduce its average current draw. Once IR is detected its power supply could then be latched on for as long as necessary for any decoding.

18. bug13 Thread Starter Senior Member

Feb 13, 2012
1,514
53
That's what I am doing now, I am hoping I can find a low power 38KHz detection circuit.

See post #11

19. bug13 Thread Starter Senior Member

Feb 13, 2012
1,514
53
I have tried similar thing with a timer in my MCU.

I need a long duty cycle to get a reliable detection, but a long duty cycle have a average standby current ~200uA, too much for my application.

A short duty cycle can achieve low standby current, but not reliable. (This can pick up 1 out of 2-3 transmission)

20. Papabravo Expert

Feb 24, 2006
11,734
2,482
Can you verify that the MCU can wakeup from low power sleep in a couple of milliseconds? Which MCU are you using?