Skip to main content ↑ ↓
Topic: Transmission range - the answer=42? (Read 4916 times) previous topic - next topic

Transmission range - the answer=42?

Well maybe not 42, but it looks like there is more than one thing to tweak to get a good transmission range and angle using the [s:]current[/s:] existing designs (not the new constant current designs discussed in the forum).
There has already been plenty of discussion about changing R4, the collector resistor, and the IR transmit LED to improve range and coverage angle. Changing these allowed me to go from 0.3 to 3 meters, which was alllllllmost enough range at a wide enough coverage angle.

The excellent post http://dangerousprototypes.com/forum/index.php?topic=1199.msg13082#msg13082 by Samtronic based on commercial experience, led me to review the Transistor base resistor and the transistor specs, and to conclude that in addition to reducing R4 to 39 ohms or lower and changing the LED (for wider angle coverage) that:
   - reducing R5, the collector resistor can, for some IRToys increase the range, possibly doubling it.
   - using a transistor with a higher gain and max collector current can increase the range many times

I have been able to measure the received signal strength. This shows that reducing R4 and R5 and using an LED from a commercial remote does improve the range, but in my case, only to 3 meters and that the received signal strength was much lower than the 12 commercial remotes I tested.

My findings & recommendation to make the IRToy transmit strength similar to commercial remotes two changes need to be made:
   - R4 to 22 Ohms or even down to 10 Ohms
   - replace T1 with higher gain & preferably collector current.
   I used a pinout compitable MMBTA13 that was the only option at my local electronics shop.

Now I am making an assumption that the average power of the collector resistor is below it's specs - I am too tired and lazy to try an average calculation as IR signals are not "regular" signals. I did try and find information on peak current/power handling, but can't see any info.  I have not noticed any issues with R4, but it is hard to tell after my very poor SMD soldering.

Does anyone have any ideas on this? I know Ian has discussed this as an issue, but I am hoping that for IR remote signals we can just use the average power rating and that this is OK :)
Alternate approaches might be to use the:
   - through hole PCB version which can use higher watt resistors than SMD versions.
   - hack old remotes to use their transmit circuit
   - make your own PCB, or if there is any interest I can release my circuits.


Some  long winded background info on how I came to the above is shown below, it also include discusses possible areas for improvement that I have not fully explored:

Samtronic's post is summarised in my words below:
* The LED - power output for a given coverage / angular range at different currents, especially high current, short PWM.
* Peak LED current is important, which can in turn lead to a lower pulse width (PWM).
* Ability of the power supply to handle the current, especially short peaks.
* R4, the collector resistor helps control the transmit IR LED current.
* R5 the base resistor sets the current flowing into the base, and thus the maximum collector/LED current (in conjunction with R4).

So this led me to review the transistor base resistor.
CURRENT IRToy DESIGN:
   Transistor base current = (5 - 0.2 - 0.6) / 1,000 = 4.2mA
   So Ic = I LED max = Hfe * 4.2mA = 100 * 4.2,A = 420mA
   - ignoring for the moment Hfe variations, effect of Rc (R4) and LED Vf variations.

The above is based on:
   Rb (R5)    = 1,000
   Nominal +V    = 5.0
   PIC high output   = +V - 0.2
   Transistor Vb   = 0.6
   Transistor Hfe   = 100   As best as I can tell this is the minimum gain.

Thus the maximum LED drive current will be limited to around 400mA, even if you have reduced the collector resistor R4, unless you have a transistor that has a higher gain than this minimum. Rather than do yet more soldering on my IRToy, I bread boarded a circuit using a 2N222 transistor, and validated there, that reducing the base resistor R5 also increased the range. However this bread board circuit had a shorter range, even with both R5 and R4 reduced significantly.

Now I foolishly suspected the bread board setup was limiting the peak current, so I learnt Eagle and made my first PCBs in 20 years and was very disappointed to get the same result as the breadboard.
More research and hair pulling highlighted that there have been a few changes with different versions of the board and even suggestions of different parts depending on your supplier. THIS MAY BE IMPORTANT FOR YOUR PARTICULAR BOARD VERSION!

I also remembered a very old IR project of my own and dredged it out. Now this was very useful as the IR receiver LED (TIL413) did not include all of the auto gain and filtering circuits that are built into ‘modern’ LEDs, but provided a raw output that showed the received signal strength. So hooking this up to the computer line input and using some oscilloscope software allowed measurements of received signal strength. I could not see the carrier wave ~38kHz, as the sound card is limited to around 20kHz, but I could see the strength of the signal.
So I compared 12 on hand commercial remotes (6 for my Home theater setup, 3 old VCRs etc, 3 failed multi-remotes - so you see why I want the IRToy to work :) with various configurations, this confirmed that all the commercial remotes had a much higher signal strength than all the variations I tried with the IRToy. So to prove the point, I connected the IRToy to drive one of the remotes and yes, it now gave the same great range and coverage. Now about this time my brain eventually realised what the calculations above were telling me - the last hurdle to overcome was the transistor gain and not issues with breadboard limiting current!
So I added a high gain darlington transistor pair to my PCB and now even SFH480 has a similar signal strength to the 12 commercial remotes.  The base resistor can remain unchanged a 1,000 ohm due to the very transistor high gain.

So finally success, solid, reliable range. For me though, the SFH480 narrow angle is an issue. I need about a 1 meter diameter coverage circle at four meters, and this will not provide that with my hardware, so I will use a scavenged LED. This is especially an issue for a fixed installation - instead of using it as a handheld remote where you automatically aim at EACH DEVICE and adjust the aim if required.

So my updated IRToy now is configured with:
   Rb (R5) = 1,000 Ohm (unchanged)
   Rc (R4) = 15 Ohm
   T1   = MMBTA13   (High gain Darlington pair Hfe minimum ~ 5000)
   Transmit LED = scavenged.
Now I am making an assumption that the average power of the collector resistor is below it's specs - I am too tired and lazy to try an average calculation. I did try and find information on peak current/power handling, but can't see any info.
Does anyone have any ideas on this? I know Ian has discussed this as an issue, but I am hoping that for IR remote signals we can just use the average power rating and that this is OK :)

Alternate approaches might be to use the:
   - through hole PCB version which can use higher watt resistors than SMD versions.
   - hack old remotes to use their transmit circuit

Things I have not fully explored with my version IRToy:
   - IR LED pulse width - have not got to it yet!
   - power supply volt/current issues due to USB limits and PCB traces -eg add a higher value capacitor across power rails.
      - no obvious impact for basic test
   - Capacitor across LED to supply higher peak current.
      - shows promise, but I have not yet sourced appropriate value ESR caps with known ESR values, especially at 38kHz.
         Tests show received signal strength can be 2-3 times, even with dogy caps.
   - PIC output pin - current drive - ignore if using high gain transistor or darlington pair.

I can see a very noticeable "droop" of 25% or more in the level of the received signal. This is particularly evident in the first wide pulse, but often impacts the entire sequence.

This also affects the strength of repeat signals, or if the gap between sending signals is short.


* 42 = a reference to Hitch Hackers’ Guide to the Galaxy. My rather tenuous "link" is 4 changes (R4, R5, led, Transistor), of which two make the major improvement (R4 and T1).

Re: Transmission range - the answer=42?

Reply #1 –
Fantastic writeup!

Quote
I know Ian has discussed this as an issue, but I am hoping that for IR remote signals we can just use the average power rating and that this is OK :)

Average power it is, as far as I know. The 0805 SMD resistors are only rated for 1/8 or 1/10watt, which is several times smaller than what you need when you swap down to 22ohms, but it doesn't seem to be an issue (I'm not comfortable selling them like that though).

It's a 50% average over the transmit phase, and the transmits are in short bursts, so there is a lot of room to work with. However, as the IR Toy is intended to be a development platform for uC amateurs, I want to make sure that even if the LED is on solid nothing is going to (hopefully) cause an issue.

Also keep in mind the 500mA USB limit. Which could also be averaged with a sufficiently large capacitor.

We can also increase the current and decrease the duty cycle, I understand 30% or 40% at really high amperage are common (1A).

Some of this is a good thing to consider in the updated design BOM. The updated design has been dogged by issues, but I think I finally got it right. The PIC 18F2550 used has tripped in price, and now the wholesale cost of the IR Toy update is the same as the retail price of the v1. That meant it had to go back for one more revision, this time using a PIC 18F25J50 (a 3volt part). It also has the constant current driver with an 1206 (1/4watt) base resistor.
Got a question? Please ask in the forum for the fastest answers.

Re: Transmission range - the answer=42?

Reply #2 –
Quote
Most IR-receivers (at least the ones I have encountered) are sensitive to the peak optical power (I think it was a HP/Agilent/Avago or Telefunken/Vishay application note pointing this out for me).

That means it is a bad idea to drive the IR LED with a 50% duty cycle constant current.

It is a much better idea to use a lower duty cycle and higher current, preferable peaking at Ipeak for the LED (typically 1-2 A), but not exceeding Iavg on average.
I have used this method with Vishay TSAL6200 (5mm 940nm +-17deg 1.5Amax 100mAavg) and TSHA4401 (3mm 875nm +-20deg 2Amax 100mAavg) IR LEDs.

A simple driver I have used consists of a low Vce,on transistor (e.g. FMMT489 or even FMMT617 from Zetex/Diodes inc. or just a BC337) a couple of resistors, a low ESR capacitor and the IR LED:


     +3V
       |
       R (current limiting resistor)
       |
       +---+
       |   |
      LED  C
       |   |
       c   |
  --R-b    |
       e   |
       |   |
       |   |
       +---+
          GND


The capacitor needs to be sized for the correct Ipeak of the LED and the duty cycle can be reduced to 20-30%.
The current limiting resistor should be sized both to allow the capacitor to recharge and control the minimum current through the LED (something like Iavg is probably fine).
This is relatively easy to simulate in Spice (e.g. LTSpice from Linear), the tricky part is to model the IR LED - I just tweaked the parameters for a normal diode until the I/V curve looked like the one in the data sheet.

In one design I used TSAL6200 IR LED, 4.7 Ohm current limiting resistor, 4.7 uF low ESR (<2 Ohm) tantalum, FMMT449 tansistor, 100 Ohm base resistor and a PIC16LC505 driving 8 us on, 18 us off (30% 38 kHz).
The power supply was 3V (2xAAA).
The receiver was a Vishay TSOP1838.
The range was well over 10m (30 feet). I seem to recall it was almost 20m...

Another project (a car central locking remote replacement) used TSHA4401, 4.7-10 Ohm, 1-2.2 uF (Sorry, I can't remember the exact values - it was a one-off 5 years ago), FMMT449, 100 Ohm and PIC12C629 with 8 us on, 24 us off (20% 32 kHz).
This was powered by a CR2032 (3V).
The range was almost doubled of the original remote which used 3xLR44 for power.


Hope you can use this lengthy post for something.

Yes, this was a great post :)

I think we have to do two things in the point version update of the ir toy redesign:
1. Change to a tantalum cap on the IR TX and move it to after the resistor
2. Add a second 1206 base resistor footprint so people can populate a value for other currents.
Got a question? Please ask in the forum for the fastest answers.

Re: Transmission range - the answer=42?

Reply #3 –
[quote author="ian"]

Quote
It also has the constant current driver with an 1206 (1/4watt) base resistor.
[/quote]  Is this referring to collector resistor, not base?

Thanks for the feedback, and I'm glad this is helping for next release. If you are interested I can post details of how I measured the received signal strength and some pictures.

I also wanted to re-emphasise that the goal of this post was help those who already have an IRToy with range issues.

Re: Transmission range - the answer=42?

Reply #4 –
Pictures would be great, it sounds interesting.
Got a question? Please ask in the forum for the fastest answers.

Re: Transmission range - the answer=42?

Reply #5 –
The test setup:

The circuit of the old project I am using to help measure the received signal strength is at http://books.google.com.au/books?id=zQW ... ed&f=false, the article also gives some interesting background on IR that is still relevant.

Initially I was looking at the signals at pins 12 and 14, but I may have killed the SN76832 as it got very hot - I suspect a ground issue [or I shorted something :( ] as I have had a similar issue with a different project.

So just to simplify, I thought what about removing the IC and with NO power supply, just measure the TIL413 output directly. This turns out to give highly repeatable measurements of the signal strength, although without amplification, the signal can quickly disappear into the noise ( especially when just using a PC Sound card as the oscilloscope!).

One of the challenges was to get repeatable test results. The major factor was physical alignment of the transmit and receive LEDs. This was the second reason why I abandoned the bread board setup - it was to hard to change configuration without moving or bumping something.  The circuit board with jumpers allowed easy changing of the configuration without movement (with the aid of a few clamps!).

Here is my breadboard test congfig:


Here is my test PCB:



edit - I'm having "finger' trouble getting the pictures to show - so splitting this post in two !

Re: Transmission range - the answer=42?

Reply #6 –
Example signals:

One of my 12 Commercial remote used to benchmark signal strength at 2 meters. With the same oscilloscope settings and distance, the received signal strength of these varied from about 1/4 to 1 and a bit vertical divisions at two meters.


IRToy driving test board with the original SFH led via a BD139 darlington pair transistor (Rb=1,000 Ohm, Rc = 10 Ohm).


This transistor was overkill in terms of gain (min Hfe=500, and current), but it was quickly available from my local electronics shop and helped prove that the transistor gain and current does severely limit the range. Using an "as is" IRToy with SFH LED transmit led - the signal is lost in the noise, even at 1/4 of the 2 meter test distance used for most of the tests.


The SFH led, a 2n222 transistor and Rb = 10 Phm and Rc = 1 Ohm also has the signal lost in noise at 2 meters. Switching to the scavenged LED results in a clear signal - but I forgot to save that picture.


The IRToy driving a commercial remote transmitter and supplying the power.



IRToy with MMBTA13 transistor, Rb = 1,000 Ohms and Rc = 15 Ohms showing the SAME signal strength as 'average' commercial remotes.

Re: Transmission range - the answer=42?

Reply #7 –
and finally ....

Power rail voltage changes during transmission shows that filtering will probably make some improvement. This was done with Rc = 10 Ohm, Rb = 100 Ohm and a scavenged LED and a distance of 4cm (everything mounted on old DVD case to try and avoid more shorts).

Re: Transmission range - the answer=42?

Reply #8 –
Very interesting, thank you. I'll post it up on the blog.
Got a question? Please ask in the forum for the fastest answers.

Re: Transmission range - the answer=42?

Reply #9 –
This is all very great stuff, but it seems disappointing that you're using a sound card in place of a 'scope.  Those all undoubtedly have AC coupling on input, meaning that it's not terribly conclusive when you test the power supply droop.  Too bad the analog DSO wing for the OLS is not complete!

Hopefully you can borrow a 'scope for some of the more critical tests, or perhaps another IR hacker can take up where you left off.  I have the USB IR Toy, but have run out of time lately for working on improvements.  Thanks for your efforts!