| ??? 06/15/08 16:04 Read: times |
#155926 - Resistance in the supply rails |
In order to avoid contaminating another thread,
http://www.8052.com/forum/read.phtml?id=155921 I'm starting this one. At issue is the placement of small-value resitors in the supply connections for the purpose of localizing noise generated by the charge pump in MAX232 devices to the serial interface, rather than allowing it to permeate the remainder of the circuit. I proposed the insertion of low-value resistors in the supply rails for the MAX232, with carefully designed bypass/filtering in the supply rails, between the noisy charge-pump-based MAX232 and the rest of the circuit, with half as much resistance in the Gnd path as in the Vcc path. While I would not recommend this for most circuits, I find it appropriate for the MAX232, and not necessarily for later versions of it, such as the MAX232A or MAX3232. The basic theory is that, as the MAX232 imposes noise on the supply by periodically cauing current surges on Vcc and Gnd in order to charge its cap's. When this happens, the resistor voltage due to IR changes will be offset by current flow into and out of the cap's. My position is that this mitigates the impact on Vcc and GND outside the MAX232's circuit, as there is meaningful bypass inside and immediately outside that circuit, using high-frequency-response (low ESR, ESL, meaning, essentially, NOT electrolytic) capacitors on either side of the series resistors. Because the receivers of the MAX232 operate against a deadband of +/- 3 volts to either side of Gnd and I submit that worst-case load changes on Vcc and Gnd will not have meaningful impact on incoming signals. Further, since the input thresholds for logic levels are pretty well separated, too, they, likewise, will be essentially unimpacted by the small voltage changes on Vcc and Gnd, particularly since those will be significantly mitigated by a properly designed notch filter on Vcc and Gnd. The impact the MAX232 has on the supply rails, after all, is primarily due to the operation of its charge pump, which relies on 1 uF cap's to develop the +/- 10 volts it claims to provide. At issue is the impact of at < 50 ohm resistance would have on the ability of the MAX232 to detect signals on both input and output. RE |
