| ??? 06/23/08 00:42 Modified: 06/23/08 00:46 Read: times Msg Score: +2 +2 Good Answer/Helpful |
#156110 - No disagree, but ... Responding to: ???'s previous message |
... I still don't understand the need of adding impedances in the GND line.
Richard said:
At issue is the fact that he was using the original MAX232, which collects charge from Vcc and pumps it up, drawing current from Vcc, then discharging it into Vcc, and then collects charge from GND and then dumps it into Gnd. These currents don't have to impact the global Vcc and Gnd nets. They can be kept circulating through the cap on the Vcc and Gnd terminals on the MAX232. Ok, let's discuss this nasty charge pump in detail. In the following you see a simplified schematic of the charge pump section of a typical MAX232 (many of its todays clones contain nearly the same): 
The most important cap isn't shown in this scheme, by the way, I mean the obligatory decoupling cap across Vcc and GND. In the left you see the voltage doubler and in the right the inverter. There are two steps of operation: 1. phase - S1/S3 and S6/S8 are closed for a half period of fCLK: C1 is connected to Vcc and GND and becomes charged up to Vcc. A charging current is flowing into the Vcc pin and out of the GND pin. The currents are exactly equalling and flow from and to the voltage regulator. (We omit the decoupling cap, for a while.) C2 which formerly was charged up to V+ (about Vcc x 2) discharges into C4. From this process there's no current flowing through Vcc and GND pins, because C2 is isolated by the open switches S5/S7. 2. phase - S2/S4 and S5/S7 are closed for the next half period of fCLK: C1 discharges into C3. No net current is flowing through Vcc and GND pins, because C1 is isolated by the open switches S1/S3. C2 becomes charged up to V+. The current flowing into C2 comes from the paralleled C1 and C3, which are partially discharged. As consequence a current of same height must flow into the Vcc pin and out of GND pin. Again, the currents are exactly equalling and flow from and to the voltage regulator. (We omit the decoupling cap, again.) So, in both phases a charging current is flowing into the Vcc pin and at the same time a current of exactly the same height is flowing out of the GND pin. Up to now, we omitted the decoupling cap across Vcc and GND pins, what also many newbies do, when they don't understand the issue in full detail. The consequence is a contamination of Vcc and GND line with noise, which is nothing else but the voltage drops of charging current across supply line impedances. The naked voltage regulator usually isn't capable of delivering the fast and steep charging current pulses, so the Vcc and GND noise can be considerably high, just as you mentioned it lately. Now, supply line filtering shall be discussed: What, if we add a RLC-filter to the supply pins of MAX232, similar to those shown in this example: 
Then, the charging current can circulate within the decoupling cap and must no longer flow across the whole board from and to the voltage regulator. Concretely spoken, only a small part does "leave" the decoupling cap. In the frequency domain, the impedance ratio of ZL+R to ZC determines how much current still must be delivered by the voltage regulator: If (at a certain frequency) the impedance of L and R is 100 times higher than the impedance of decoupling cap C, then about 99% of charging current (at this frequency) of MAX232 is delivered by the decoupling cap and only about 1% has to be delivered from outside. This means: Only 1% of the current is flowing through the choke and exactly the same current is flowing from the GND pin of MAX232 back to the voltage regulator. So, by using the L and R in the Vcc line, not only the current through the choke is heavily decreased, but also the GND current leaving the GND pin of MAX232! There's no need to add an impedance also in the GND line!! The law of conservation of charge does the trick... Kai |
