Remote IR on-off switch

[pimpom]

I've come across this DIY IR remote on-off switch design more
than once with only minor variations.
http://img109.imageshack.us/img109/6399/irswitch.png

It looked OK at first, but then it seemed to me that R3 and D1
are not really needed. Why not just omit R3 and short out D1 as
on the right side of image?

And is such heavy filtering necessary? I'd think 38kHz pulses
fired in ~1kHz bursts would be adequately filtered with a much
lower RC combination and will have faster response too.

 

[pimpom]

Do you mean burst of 1000 cycles, or about 25 ms at 38 kHz?

No. Datasheets indicate that, for noise suppression, most IR
receiver modules are optimised to receive 38kHz pulses in bursts
roughly 0.5 msec on, 0.5msec off. This is where I derived the
~1kHz figure from.

The 4013 changes state on a positive clock, or after about 13
ms (27k
and 0.47 uF). The discharge time constant prevents retriggering
for
about 1/2 second (1 M and 0.47 uF). Do you want it to toggle
faster
than once per second? The danger is that you could get multiple
pulses
from a buttonpush (if that's what the input is coming from) and
have
the toggle end up in a random state.

I _am_ building something using an IR receiver module, but using
another technique. It's just that I came across this circuit more
than once and got curious about the validity of the design.

I get your point about unintentional toggles. Anyway, I tossed in
the matter of filter time constant as an afterthought. My
attention was mainly on why the designer(s) considered it
necessary to use R3 and D1. The pulsed output from Q1's collector
is already unidirectional and does not need rectification. The
collector, even without a reverse blocking diode, is essentially
an open circuit (megohms) in the off state and will have
negligible effect on the filter efficiency.

 

[pimpom]

Or you could add a pair of uProcs and trasnmit specific codes
which
would make it much more robust. I couln't get the Freescale
68HC908
series to modulate 38KHz directly - possibly a lack of
programming
skills on my part but with a little added glue logic it works
fine.
Receiving is not at all difficult. TV remotes (Samsung and Sony
for
certain) transmit 32 bits total with mirrored 8 bit blocks.
It's
pretty reliable. You can do it too if it's important.

Thanks for your interest. It's not that I want to build this
design. I'm using a different approach in something I _am_
building. It's just that I came across this circuit and the
inclusion of R3 and D1, while apparently logical at first, seems
superfluous on closer inspection. Please also read my reply to
John O'Flaherty.

 

[pimpom]

The diode gives a different time constant for discharge than
for
charge. If the diode is shorted out, the 100 k collector
resistor will
shorten the discharge time constant to 0.13 sec or so, by
paralleling
the 1 Megohm.

Omitting the 100k resistor will obviate that.

However, if you can get by with a 1 megohm collector
resistor, based on the transistor leakage and the IR receiver
turning
the transistor fully off, it should work as you've drawn it.

BC5xx transistors have very low leakage - Ices 4uA max at Tj 150
C. Actual leakage encountered in practice are probably much less
even at 150 C. And Tj in the target hobbyist use is unlikely to
exceed 50 deg C. So we can expect sub-100 nA leakage.

I haven't seen leakage figures for the output of an IR receiver
module. Anything less than 10uA will be shunted away by the 27k
paralleling b-e and won't cause Q1 to conduct.

If I were to adapt this circuit for my own use, I'd probably use
a higher capacitance for C1 (an electrolytic should be OK) and
correspondingly lower values for R4 and R5. Say 2.2uF, 4.7k and
220k respectively.

 

[pimpom]

I have no idea what the output circuit of that IR thing looks
like,
but as long as it reaches nearly to the positive rail, it
should be
fine.

Block diagrams on manufacturers' datasheets usually give the
output as the collector of an NPN BJT, like the output of popular
comparators but with a 20-30K internal resistor to Vcc. This is
probably meant to enable driving the base of an external NPN BJT
without any external component.

 

[Spehro Pefhany]

I've come across this DIY IR remote on-off switch design more
than once with only minor variations.
http://img109.imageshack.us/img109/6399/irswitch.png

It looked OK at first, but then it seemed to me that R3 and D1
are not really needed. Why not just omit R3 and short out D1 as
on the right side of image?

---
Even better: (View in Courier)

+5
|
+-----+-----------+
| | |
| [10k] |
| | E
| +--[10k]--B PNP
| | C
[IRRX]---+ |
| | +-----+---->4013 CLK
| | | |
| | [470nF] [1M]
| | | |
+-----+-----------+-----+
|
GND

That way the cap would charge up very quickly and discharge slowly,
making an excellent debouncer for the 4013's clock.

CD4013 does not have a ST input on the clock, and that will violate
the max clock rise/fall times. Would be okay with a HC14 or other ST
gate before the clock input.

 

[Jamie]

I've come across this DIY IR remote on-off switch design more
than once with only minor variations.
http://img109.imageshack.us/img109/6399/irswitch.png

It looked OK at first, but then it seemed to me that R3 and D1
are not really needed. Why not just omit R3 and short out D1 as
on the right side of image?

And is such heavy filtering necessary? I'd think 38kHz pulses
fired in ~1kHz bursts would be adequately filtered with a much
lower RC combination and will have faster response too.

Lets see, Diode to prevent the 100K from draining the REF down
faster than the 1M that is there.

Resistor to remove the in rush currents on the pulse from the
transistor? Which also may lead to parasitic abnormalities.

Just my guess.