Harun Scheutzow 1998-03-14 (html version 2000-10-13)
The sender is a MC1488P from ST (SGS-Thomson), probably manufactured in 1991. The supply voltage is +12V and -12V. There is no extra capacitive load at the transmitters end.
The speed is 115200 bps.
The load is a approx. 4 m long shielded 9-wire cable, one shield around all the wires. In the middle of the cable there is a self-made mechanical switch. At the far end there is a 1489 receiver. But I'm too lazy to open the computer now to find out the manufacturer.
The 1488 output reaches it's maximum and minimum level without any problems. Max and min level are approx +10V and -10V, that means 2V below the supplies.
The rise and fall time is measured between near the 0% and 100% point, not 10% and 90%, because this is to expensive to be done "jff" (just for fun).
The time for a max or min voltage is measured for one bit time in a maximum change pattern (101010...).
The rise time is 1.5 us. The rising ramp is an exponential function with the fastest rise speed at the minimum level. The -5V to +5V range is crossed in 0.56 us.
The fall time is 2.2 us. The falling ramp is an exponential function with the fastest fall speed at the maximum level. The +5V to -5V range is crossed in 1.00 us.
Both maximum and minimum output voltage are present for 6.6 us.
The waveform and timing is the same like at the transmitter, with less than 5% tolerance.
Another test is made with an approx. 10 m long cable of the same quality like the first, with the same switch. The load at the far end is a Lasat 33600 Modem, switched on.
The rise time is 3.2 us, the -5V to +5V time is 1.4 us.
The fall time is 4.6 us, the +5V to -5V time is 2.2 us.
The maximum output voltage is present for 5.0 us and the minimum output voltage is present for 4.6 us.
The next test is done without the Lasat 33600 Modem, that means without any RS232 receiver at the far end.
The -10V to +10V rise time is 2.8 us, the -5V to +5V time is 1.3 us.
The +10V to -10V fall time is 4.0 us, the +5V to -5V time is 2.2 us.
The range above and below +/-10V is very round if going to the maximum/minimum voltage.
The maximum voltage is 11.0 V, present for approx. 4.0 us. The minimum voltage is -11.5V, present for approx. 3.0 us.
The highest still reliable speed of the 10 m cable and Modem load is probably 230400 bps. At this speed the +/-10V is still reached but the output immediately reverses it's transition direction. Above 230400 the output will not maintain the 20V swing and slip to an average voltage determined by the transmitted pattern.
Another test is performed with an extremely short cable, only 0.3 m, and the Lasat Modem as load. The scope probe capacity is probably 15 pF and might become important here.
The voltage swing is -10V to +10V.
The rise time is 220 ns.
The fall time is 280 ns.
The fall time gives a slew rate of 71 V/us and the rise time gives 91 V/us.
The maximum allowed slew rate for RS232 is 30 V/us. The data sheets of bipolar 1488 transmitters say that additional load capacitors are necessary at the output to limit the slew rate of 1488 to the RS232 requirements.
The above measurement was done at 230400 bps. I'm to lazy to implement 460800 and 921600 bps in my serial driver now. That's why I will not verify whether the ordinary 1488 does really 1 MBit/s. But the measured slew rate would allow reliable operation up to 1 MBit/s with a 0.3 m (30 cm) cable.
During all measures there was never a waveform which looked like a reflection or other irregularity. All rise and fall was monotonic.
The above results will be compared with the RS232 driver portion of a MAX209, which is under test now. The MAX209 is operated from +5V and +12V with an 100 nF capacitor (SMD, size 1206) at every place (+5V to GND, +12V to GND, self-made -12V to GND, transfer).
The speed is 153600.
The first load is the 10 m cable _without_ the Lasat Modem.
The maximum voltage is 11.4 V, to minimum voltage is 12.0 V.
The rising slope looks strange. The rise starts at -12.0 V. It is nearly straight for 1.3 us reaching -2.0 V. There it stalls for 2.5 us. The rise starts again and reaches 6.0 V 2.3 us later. It stalls again for 1.6 us. The last step takes 1.7 us up to 11.4 V. The start and stop phase of stalling is soft (round), softer than the start at -12.0 V. Each new slope is slower than the one before. The total rise time is 9.4 us in this case.
The falling slope looks ok. It takes 1.6 us from +11.4 V down to -8.0 V and additional 1.6 us to reach it's final -12.0 V with a soft curve. The maximum slew rate is around 0.0 V, not at the start of the falling.
The next two with the Lasat Modem at the far end.
The maximum voltage is +11.0 V, the minimum voltage is -10.0 V.
The rising slope looks less strange than without Modem, but it is still extraordinary. The rise starts at -10.0 V and reaches +6.0 V after 2.0 us. The slope is a S-curve. Rising stalls here a short time, approx. 0.4 us, and goes again into a S-curve which takes 2.2 us to reach the final maximum voltage. The total rise time is 4.6 us.
The falling slope looks ok. It takes 1.7 us from +11.0 V down to -8.0 V and additional 1.2 us to the final -10.0 V.
It was verified that the +12 V supply remains stable, that means the spikes are much shorter than 1 us and they to not exceed 0.5 V.
With only a 40 cm cable and the scope probe, no other load, the rising slope is very straight and only 1.6 us long. This is probably too fast for any slew-rate limiting circuit in MAX209. ??? ######### The falling slope is 1.2 us long and very straight too.
With the 4 m cable and the 1489 type receiver at the far end the maximum voltage is 10.0 V and the minimum voltage is -10.5 V.
The rising slope is straight and 1.9 us long.
The falling slope is a bit S-like. It takes 1.3 us from +10.0 V down to -8.0 V and additional 0.6 to 1.0 us down to the minimum voltage of -10.5 V.
####### It might be a current limiting circuit in MAX209, not a slew rate limiting one. The current limiter is probably only in the +12 V path, not in the self-made -12 V path, because the latter has an automatic (but much higher!) current limit due to the charge pump.
An RS232 receiver has a resistor of some kOhm to GND. This lowered the current necessary for the minimum voltage to GND transistion and the current limiter did not bite. But for the GND to maximum voltage way the resistor has the opposite effect and the current limiter switches on.
The MAX209 short circuit current from an output which should drive minimum voltage to GND is 25 mA. The scope connected in parallel to the current measuring multimeter does not show any significant ripple. The self-made -12 V supply of MAX209 drops from -12 V to an oscillation between -10 and -9 V.
A maximum voltage to GND short circuit drives only 15.7 mA. But even there is no ripple on scope, even not if using the 20 mA setting of the multimeter instead of 200 mA. The 20 mA lets an voltage of approx 200 mV across the short circuit.
A ripple during max to GND short circuit would explain the behavior at the long cable. But there is no such ripple.