EMI problems with the HP palmtops

Archived 2013-12-08
This page has been archived and some links may not work correctly anymore. No updates of this page will be done anymore.

(EMI = Electromagnetic Interference: Communication problems when the HP Palmtop cannot communicate via InfraRed anymore when a mobile phone is near, e.g. in setups where the Palmtop should use a mobile phone via IrDA for Internet access.
Contents of this page:


NEWS:

I have bought a used 700LX to study how HP solved the EMI problem there (since the 700LX is designed to work with a GSM phone on top, it must be somehow shielded against the EMI, and tests with ring.com and a seperate GSM phone so the dehaviour of the 200LX and 700LX regarding EMI coulr directly be compared, showed that indeed the 700LX is only disturbed marginally by a mobile phone). The results can be viewed on the separate 700LX page: Click here!


Introduction:

The problem appeared, when the first people wanted to use their HP 100/200 LX to connect to the Internet using the IrDA interface of a mobile phone in the summer of 1999. It is nice to have the ability to simply put the palmtop upon a table, put the cellular phone besides it, and surf the web, download email etc.
Actually it should be possible without problems, because the palmtop has got an IR port, and more and more cellular phones also have an IrDA port which looks to a PC just as a standard modem. The ´internet suite´ WWW/LX by D&A Software is the only software which is able to speak the IrDA protocol over the palmtop´s IR port. Actually the IR port of the palmtop speaks the "HPSIR" protocol which cannot be handled by modern mobile phones. So you have to use WWW/LX if you want to go online via IR.
But you will run into the following problem: The connection is either terribly slow, or it isn´t possible at all to get a working connection.

The people who ran into that problem first were Andreas Garzotto (programmer of WWW/LX), Stefan Peichl and I, Daniel Hertrich. Later Mark Willis, Joeseph Buford and Jerome V. Healy joined us to help. With time I got many requests for help regarding that subject, but a real solution isn´t found yet!

The problem is caused by the following:

Everytime we want to establish a connection to the internet with the mobile phone via infra red (IrDA-protocol by WWW/LX), the palmtop receives strange disturbance signals. These signals are caused by voltages induced by the electromagnetic field surrounding the antenna of the mobile phone.

Actually these signals must have a frequency of 900MHz, 1800MHz or 1900MHz (GSM, PCS), depending on the frequency of the cellular network. But these modulation frequencies are pulsed with a frequency of about 1,6 kHz (maybe that´s only true in my case - I don´t know if the pulsing frequency differs between cellular networks!).
And these pulsing is probably the frequency which causes the problems.
I don´t think that the very high frequency of 1-2GHz causes problems. because this frequency is so high that it cannot be decoded by the decoder hardware of the palmtop (and even is not shown by a normal oscilloscope) - see the pictures below!
But the 1,6 kHz signal is shown and decoded! And it is decoded into many, many characters that prevent the IrDA protocol from working properly.
The signal that actually comes through the photo diode has a pulse frequency of around 10 kHz (depending on the character that was sent by the sending device).

If you would like to test these interference by yourself, you can use Stefan´s software ´RING.COM´. It displays every character received by the IR receiver circuit (also the induced signals) and every character causes a click in the speaker. And the average of ingoing characters per second is measured. So you can see how bad these interferences are in your case. Maybe send as a reference a vcard over IR from your telephone. That will not disturb the IR port, but only the actual IrDA data will be arrive in RING.COM (because the antenna pf the phone doesn´t send anything). Then establish some kind of radio connection (i.e. do a call) and place the phone near the palmtop. You´ll see and hear the disturbances arriving in RING.COM!
You can download RING.COM with documentation  here.

Another experience is that with Siemens phones (S25, S35i), the disturbances are so bad, that most of the times a connection isn´t even possible, but with Nokia phones (8810, 7110, 6210...) you get a connection in most cases. You have to find the ideal distance between palmtop and phone in order to suppress most of the disturbances but keep the IR link working. For Siemens phones, this distance is about 15-20cm, for Nokia phones you can place the phone with a much bigger distance from the palmtop. About 50cm should still be possible. Several people worked out little tricks: For example with the Siemens phones (probably also with others) it helps if you take the phone in your right hand during the internet connection: touch the display with your palm, lay your fingers arund the upper edge of the phone and touch the cak of the phone (where the antenna connection is) with your third and fourth finger. That seems to bundle the emitted electromagnetic disturbances in a way that it doesn´t reach the palmtop anymore.

But of course we want to find a durable working solution!
First I thought, the right solution would be to build a high pass filter into the circuit of the IR Receiver to filter out all high GSM frequencies.

I thought we should replace the component no 7 (see picture below) by a properly dimensioned high pass filter. I tried this. But without any luck! :-(
I tried to replace component no 7 with different capacitors of values 150 pF up to 330 pF (My calculations showed that  220pF  is the right value to drop 1,6kHz signals by approximately 15dB and leave the 10kHz sinals untouched. With a filter of 1st grade, a higher attenuation of the 1,6 kHz signal is not possible -- the filter of 1st grade is built by the inserted capacitor and the resistor to ground (no 12 and 13) with a value of 100kOhm + 6.81kOhm = 106,81 kOhm).

Before I did these tests I thought that the GHz-frequencies directly caused the problems and inserted several low pass filters into the receiver circuit. None of them worked)

One solution that already works in some palmtops is to connect pin 2 of the photo diode (the right one, if you watch the diagram below) to ground. It´s not such an elegant solution, because it drains extremely much power and we don´t know if it can hurt the circuit. And it doesn´t work in every machine. About every third machine we tested that ´low cost solution´ in wasn´t affected by the EMI anymore. The other machines couldn´t at all use the IR port anymore. We don´t know why. But Stefan used this solution for about half a year successfully and his palmtop survived it.


Drawings (schematics and board layout)

The circuit board layout with component numbers (description of the components below):

100lx_irda_circuit
schematic

The components of the IR receiver circuit:

PD: the photo diode
1:    R = 21,5 k
2:    R = 21,5 k
3:    Transistor KST 5179 or 2N 5179 by Samsung (pins see below) (Data sheet (PDF, 50kB) downloadable here!)
4:    Transistor KST 5179 or 2N 5179 by Samsung (pins see below) (Data sheet downloadable above (=part #3)!)
5:    C = xx xF (value unknown)
6:    R = 215 R
7:    C = 0,01uF (marking "KA4")
8:    C = ´685´
9:    C = ´105´
10:  R = 10 k
11:  R = 21,5 k
12:  R = 6,81 k
13:  R = 100 k
14:  IC ´TLC372C´ by TI  description here (HTML with llink to PDF data sheet)

The pins B,C and E of the Transistors 3 and 4 are placed this way:

  ------
B O    |
  |    |
  |    O C
  |    |
E O    |
  ------
 

The circuit diagram of the IR receiver (self-explored by Joeseph Buford and me):

circuit


My measurements

I made the measurements under the following conditions: An opened 100LX lying on the table, the Siemens S25´s IR port and the 100LX´s IR port are facing each other with a distance of about 3 cm.

You can see the arrangement here (on this picture the S25 is a little bit more away from the 100LX than 3 cm):

The following picture shows my measurements of the signal voltage of an IR-received character (measured at pin 3 of the 372 chip no. 14) without any interference.

Horizontally, one division is 0.2ms, so the frequency of the pulses is approx. 10 kHz
Vertically, one division is 2 V, so the peaks are approx. +/- 1V
 

The following picture shows my measurements of the voltage induced by the S25´s electromagnetic field:

with the same scalings of both axes as the picture above.
The signal has got a frequency of about 1.6 kHz (period cycle is abt. 0.6 ms).


If you would like and/or feel able to help, ideas are greatly appreciated by: Andreas Garzotto, Stefan Peichl, Mark Willis, Joseph Buford, Jerome V. Healy, me and many, many other users of the HP 200LX!
Please email me! I´ll forward your idea to the others.
Thanks to all who help(ed) us and to all who will help!