Rick Jenkins wrote: > > On Saturday 16 October 2004 21:28, Garst R. Reese wrote: > > > I have two boards, one with the analog stuff and a digital board with > > the msp, an mmc card, an irda chip, and an LCD. Both have ground planes. > > They are tied together with a molex .5mm flex cable. > > Sounds good so far. I try to use alternate ground conductors in a ribbon > cable, which gives a low-impedance ground link and does wonders for signal > quality. You need extra positions on the connectors, bet ther're cheap. > There are 8 gnd leads between and around analog leads. > > I bring out AVss > > and DVss and tie them together on the analog board, which supplies AVcc. > > I don't quite understand this. If the supplies are really tied at this point, > presumably both run from the same regulator, which removes the isolation. > I'd separate the supplies on both boards.
AVss and DVss are the respective ground leads. Now they are just meshed together as Steve suggested. > > > I put a .1uF across AVss and AVcc on the digital board and put the 10uF > > + another .1uF cap from AVcc to analog ground on the analog board at the > > connector. > > You really have to get those 0.1uF capacitors in as tightly as you can to the > chip. Putting them "at the connector" is a poor choice. Begrudge literally > every last millimeter of trace from the capacitor to the Vcc pin, and from > the capacitor to the via connecting it to the ground plane. The same applies > to the trace from the Vss pin to its ground via. With a bit of effort and > low cunning it is usually possible to keep the total trace length of the > capacitor well below 5mm. > > In connecting the 10uF, aim for at least a 10mm trace, but no more than 30mm. > This short trace provides inductive isolation between the capacitors, > allowing each to decouple its own band of the spectrum. OK, done, but I do not have room for your suggested AO-CAPs. > > I'd want each board to have its own decoupling, for several reasons. First, > it's nice to be able to test the boards separately. Second, you may want to > re-use just one of the boards for another project. Third, the cable > introduces connections which are too long, even for the 10uF. Fourth, the > connectors can deteriorate over life and develop a few ohms to a few tens of > ohms of resistance. This tends to matter rather little for signals, unless > they are very fast or characteristic-impedance terminated (unlikely with the > MSP430), but is a disaster for decoupling. > > You don't mention a 10uF on the DVcc, but I assume it is there somewhere. I just was not clear. There is a 10uF from DVcc to the ground plane and the .1uF. At the msp there is a 4.7uF and a .1uF from AVcc to gnd. All spaced according to your suggestions. On the Analog board, which has the AVcc supply there is another 4.7uF to ground and several .1uF decoupling caps on micropower op-amps. AVcc is a reference supply and is only spec'd for a 10uF load capacitance, unlike most supplies, it does not require an output cap for stability purposes. I can place it either next to the supply chip or next to the connector. > > Bad decoupling produces some of the most difficult bugs to trace. I normally > find that, unless production is going to be very large, it is better to be > really generous and overdo decoupling, so as to be quite certain that you > don't spend hours tracing a decoupling problem, not to mention the time and > cost for a board revision cycle. For high production, make a prototype board > with lots of decoupling, and try the effect of de-populating decoupling > sites. > > So I give evey chip a 0.1uF as close in as I can possibly squeeze it, and I > lay out these decoupling capacitors first. Then I throw on at least a bulk > capacitor for each eight or ten chips if they are low-powered, or every five > if they are not. Then I look for special cases such as the A/D converter, > the reference, and low-level amplifiers, and give them their own bulk > capacitors, isolating their supplies with a resistor of 10-100 Ohms. > Regulators always have bulk capacitors very close to their terminals. Since > developing these habits I've had astonishingly good analogue performance and > no decoupling bugs. > > For bulk decoupling, particularly in the case of analogue, I have come to > prefer aluminium organic capacitors such as the Kemet A700 series. They are > cheap, not much larger than ceramics, and have much lower microphony, which > can be important in analogue applications. > > > DVcc is supplied by a TI TPS79730 on the digital board. AVcc > > is supplied by a MAX6129EUK30 .04% 3V reference on the analog board, so > > they should be pretty close. The peripherals have separate supplies. > > This scheme should be fine, as long as the two regulators share the same > primary supply, so that they come up and down more or less together. They do. Thanks for this excellent tutorial. I am sure it well serve others on the list. Garst > -- > Rick Jenkins <r...@hartmantech.com> > Hartman Technica http://www.hartmantech.com > Phone +1 (403) 230-1987 voice & fax > 221 35 Avenue. N.E., Calgary, Alberta, Canada T2E 2K5 > > ------------------------------------------------------- > This SF.net email is sponsored by: IT Product Guide on ITManagersJournal > Use IT products in your business? Tell us what you think of them. Give us > Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more > http://productguide.itmanagersjournal.com/guidepromo.tmpl > _______________________________________________ > Mspgcc-users mailing list > Mspgcc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/mspgcc-users
