logajan@ns.network.com (John Logajan) (08/08/89)
<8908070940.AA25858@ucbvax.Berkeley.EDU>, AWINFHAH@HMARL5.BITNET writes: > The address lines have a 33 ohm series resistor between the MMU and the > RAM array. Somebody noted that they are there to limit the current drawn > from the MMU. This only partially true. Another (and probably the main) > purpose of the resistors is to avoid ringing on the address lines. They are indeed to suppress ringing, and they do it by restricting the transition current. You see, when a voltage changes (a logic level shift) at one point, it moves down the wire at the speed of light. So for those first few nano-seconds, the driver is pumping current into the wire, not by the load that is in on the wire further down, but by the limiting factor of the wire itself. Only after the signal propogates to all parts of the wire and reflects back (caused by electron pile up on dead ends) does the wire calm down to supplying the "real" load current. If the wire is pretty low impedence (a term than takes into account resistance, capacitance and inductance) a lot of current gets sent down it. Eventually the current will hit the end of the wire and, being only lightly loaded by the dram chips (that is one of the problem/features of drams, they make them light loads so they can stick a zillion on the same address line without bogging it down) the current has nowhere to go, so it reflects back toward the source. This oscillation will die down, and a resistor in the circuit would help disapate the oscillation more quickly. BUT, if you stick the resistor right next to the driver, in series with the wire, then the drive will see that the wire/ resistor combo has higher impedance, so it will not send as much current down the line in the first place, and so reflections will be smaller, and hopefully not cause signals to jump in and out of their logic level ranges. > Now there are three possibilities. > 1. Atari picked the right value for 1 bank installed. In this case the > resistors are a bit too large when you install the second bank. This > can cause address setup time problems (causing unexplainable crashes). > > I think number 1 is true for the old 520s (in which case they probably > have 68 ohm resistors), while number 2 should be true for the 520+, the 1040 > and the Mega. Since the drams are such a light load in comparison to 33 ohms, there is no need to adjust the resistors. The resistance is not critcal, any resistance helps, but too much resistance could eventually attenuate the signal. I'm not sure what the value would be, to be too high, but probably a couple hundred ohms. (since the drams don't need much current, but other chips on the board might need more, especially if they are TTL type loads, which are hefty compared to drams.) -- - John M. Logajan @ Network Systems; 7600 Boone Ave; Brooklyn Park, MN 55428 - - logajan@ns.network.com / ...rutgers!umn-cs!ns!logajan / john@logajan.mn.org -
logajan@ns.network.com (John Logajan) (08/08/89)
> <8908070940.AA25858@ucbvax.Berkeley.EDU>, AWINFHAH@HMARL5.BITNET writes: > > The address lines have a 33 ohm series resistor between the MMU and the > > RAM array. Somebody noted that they are there to limit the current drawn > > from the MMU. This only partially true. Another (and probably the main) > > purpose of the resistors is to avoid ringing on the address lines. Oh, I fogot. Another reason to use such resistors is to cut down on system generated EMI (electro-magnetic interference.) All this electronic equipment has to meet FCC EMI standards -- and every little bit of attenuation helps. It stands to reason that if less current initially flows, and ringing is less, then less EMI will be radiated. -- - John M. Logajan @ Network Systems; 7600 Boone Ave; Brooklyn Park, MN 55428 - - logajan@ns.network.com / ...rutgers!umn-cs!ns!logajan / john@logajan.mn.org -