| ??? 04/07/08 15:38 Read: times |
#153017 - about those decoupling capacitors Responding to: ???'s previous message |
Andrew Murton said:
Hooray! Thanks for your help, mainly Richard. I took your warning about solderless breadboards to heart and rebuilt the circuit on stripboard with real soldered joints and it works. I just downloaded a simple LED counting program and it's running.
I probably had some poor connection or other problem before. I'm still a bit hazy on decoupling capacitors but don't let me take any more of your time, I can find somewhere with diagrams. I'm still on my 15MHz crystal and using the buffers shown in Fig 15-2 of the DS89C420 Ultra High Speed Flash Microcontroller Users Guide, controlling EA, PSEN and RST. I replaced the HC125 with a HC245 because that's what I've got to hand. The EA line doesn't go through a buffer but I could as I have six unused. Anyway it's working. Thanks very much everyone. The most important thing about decoupling capacitors on the power supplies to your circuitry is that there be a very-low-impedance return (GND) path to the supply and a very-low-impedance distribution path to each active component. The capacitors must be distributed through out the circuit such that the path from each active component to its decoupling member is very short, and that can't be overemphasized. On most of my wire-wrap boards, I have a contiguous Vcc plane on one side and a contiguous GND plane on the other, so that I can solder bypass members directly to the Vcc plane at the Vcc connections to the active components. This makes the connections to the bypass capacitors on the order of 1-2 millimeters in length. With today's CMOS technology, high frequency clocks and fast signal rise/fall times of less than 5 ns, there is lots of power-to-gnd noise that can be minimized by proper supply bypass. Large capacitors don't help with this. Only a number of small (0.01 uF or thereabouts), at least one per IC will serve most circuits properly in this respect. It is also a good practice to isolate GND paths used for reference from those used as supply returns. If you are fortunate enough to have a contiguous ground and/or power plane available, that's good. In most cases, particularly on 2-layer boards, one generally hasn't got that feature, so care must be taken to keep supply bypass members close to the source of switching noise. It's a good idea to keep in mind that noise can travel through such capacitors, and that's why it's important to have a very-low-impedance supply return (GND) path, so noise travels into but not back out of your decoupling capacitors. If you do that, your capacitors serve as high-pass filters to GND. Keeping those return paths short and low-impedance can't be overemphasized. RE |
| Topic | Author | Date |
| Take this up with the Maxim/Dallas support guys | 01/01/70 00:00 | |
| Why ? | 01/01/70 00:00 | |
| meaty PSU | 01/01/70 00:00 | |
| Mke sure you've got the right software | 01/01/70 00:00 | |
| Dallas MTK | 01/01/70 00:00 | |
| You've got to use exactly the parts they recommend | 01/01/70 00:00 | |
| no scope | 01/01/70 00:00 | |
| I'm not sure this is a deal-breaker ... | 01/01/70 00:00 | |
| same with DS89C420 | 01/01/70 00:00 | |
| decoupling | 01/01/70 00:00 | |
| I have no confidence at all in "breadboards" | 01/01/70 00:00 | |
| stripboard circuit works! | 01/01/70 00:00 | |
| about that bootloader ... | 01/01/70 00:00 | |
| that will not help | 01/01/70 00:00 | |
about those decoupling capacitors | 01/01/70 00:00 | |
| In case you're curious ... | 01/01/70 00:00 | |
| my DS89C450 experience was... | 01/01/70 00:00 | |
| What did you do? | 01/01/70 00:00 | |
| 8052.com SBC | 01/01/70 00:00 | |
| How considerate! | 01/01/70 00:00 | |
| progress ? | 01/01/70 00:00 | |
| programming circuit variations ... FIY | 01/01/70 00:00 | |
| DS89C430 boot loader | 01/01/70 00:00 |



