Re: Improved LIRC serial receiver circuit

[eric p. <par...@ho...>]

Just ran across this posting and it looks like this would be the solution to
my problems.  I am guessing that my laptop (IBM 600x) is not putting out a
constant enough RS232 serial port voltage and that is the issue.
Does anyone have the schematic that is mentioned in this post from 2002? Or
a similar IR schematic that has been used with a laptop successfully?
I have the parts for the advanced schematic mentioned on the lirc site but
not sure that will work for me.

Here is the post from Thomas Breuer:


Hi everybody, here are some notes about the commonly used LIRC serial
receiver and a suggestion for an improved but still very simple circuit.
Caution ? technical stuff to follow ;-) Some people reported strange
problems using the current circuit (http://www.lirc.org/receivers.html).
This circuit relies on the fact that the threshold level of most modern
RS232 line receivers is somewhere around +1...2 volts. Therefore, a simple
asymmetric supply of 5 volts is sufficient and the usual RS232 signal levels
of +5V...15V (Space) and ?5V...-15V (Mark) can be ignored (at least for a
direct connection without long cables). But there is a problem related to
the pullup resistor connected to the output of the IR receiver chip. The
datasheets (I will only refer to the SFH5100-xx and the TSOP17xx here)
specify external pullup resistors to be larger than 10k. This might be too
small to pull up the signal as the line receivers have around 7k input
impedance. Therefore, a 4k7 resistor is used. But this value is already
critical (especially with the SFH5100) as now the internal output transistor
may not pull the signal fully down anymore. Some RS232 receivers fail to
read this as a valid low level so that it works on one PC while failing on
another. Also, temporary failures may occur due to the influence of
temperature and/or power load changes on the output and threshold levels. In
other words: it may work, or it may not ;-) Another problem with this
resistor is that a lot of power is wasted when the output of the IR chip
goes low. This lowers the available output voltage swing. As mentioned
above, the value of this resistor can not be increased as this would also
lower the output voltage swing. So choosing the correct value is quite
difficult and depends on the IR receiver chip and the RS232 line receiver of
the connected PC. The following circuit suggestion eliminates this problem.
It may even work with EIA-562 interfaces used in modern notebooks where the
output voltage swing can be as low as +/- 3.7 volts. I couldn?t test this as
my notebook doesn?t even have a serial interface :-( However, I simulated
such a low voltage interface by reducing the supply voltage down to +/- 3.7
volts. The circuit was still working with an output swing of +/- 3 volts.

 The new circuit --------------- My suggestion for the improved circuit
requires only three additional transistors. Here is a link:

 http://home.t-online.de/home/tb_electronic/vdr/lirc/lirc_rx.html

The lirc_serial driver obviously sets RTS positive and DTR and TXD negative.
So we can get a positive (as used in the standard circuit) and a negative
supply. The circuit draws only about 1.5mA (SFH5110 used) out of these pins,
causing only a slight voltage drop. Details: - Using a LP2950CZ5 instead of
a normal 78L05 (pin-compatible). This draws only 0.1mA quiscent current and
has less than 0.1V dropout voltage. - Using a Schottky diode (e.g. BAT46)
for protecting the regulator. The voltage drop is only 0.2V at 1.5mA. BAT46
and LP2950 are available at http://www.reichelt.de - The voltage regulator
is connected between RTS and DTR instead of GND. Therefore, even with only
+/- 3.7 volts there is still enough input voltage. - A simple push-pull
output stage (T2/T3) is added. This saves a lot of power (well, compared to
the limited power available from the driver lines). This would otherwise be
wasted in the pullup resistor, lowering the available supply voltage
significantly. Now only R1 (100k) of the level shifter T1 adds 0.2mA
quiescent current. Additional 1.4mA (@ +/-10V) will flow into the RS232
receiver input (usually around 7k input impedance). Notes: * The two
capacitors at the input *and* output of the LP2950 regulator are mandatory
(applies also to the 78L05). Without, voltage regulators will oscillate. *
The labels with the voltage levels in the schematic are based on +/- 10
volts at RTS resp. DTR (with the circuit connected). * As T1 inverts the
signal, the output of the receiver is active high. LIRC will autodetect this
anyway. * For clarification, the internal output transistor and pullup
resistor of the SFX5100-xx/TSOP17xx chip are shown in the schematic. *
Anxious people may add a zener diode (27 volts) across C1 to protect the
LP2950 (maximum 30 volts input) as the RS232 specs allow up to +/- 25 output
level. However, no modern interface will exceed +/- 12 volts. With the
suggested components the circuit should work under most conditions (please
test it on notebooks with EIA-562 interfaces!). You may use a 78L05 instead
of the LP2950 and a normal diode if the circuit is connected to a desktop
PC. The output voltage swing will be reduced by 2 or 3 volts, but this is
still more than enough. Hopefully, this suggestion is useful to some of the
desperate LIRC users who couldn"t find any explanation for the strange
reaction of their receivers. I myself encountered this problem when starting
to build my VDR box. Meanwhile, I have developed a small IR Controller for
this project (www.tb-electronic.de/vdr/ir_controller.html) where this is
also fixed by using a simple level shifter. Thanks to all the LIRC
developers for the great software! Thomas