Steve M. Taylor wrote:
-------------------------------

Whats the problem with SMT ? The beasties you need are only 8 pin ones.

I have no problems with SMT parts. But this project may turn into a kit, so I would like to avoid SMT...

MAX701 CSA is nice - PB input, and complementary reset outputs too.

It look fine - and it's even available in DIP ;-)

But as Peter wrote a reset won't actually solve the problem. But a reset chip could be used for the output enable pin (to save a pin on the µc and an extra wire).

Use a diode - 1N4148 - over each relay, to kill the noise at the source.

1.) 1n4148 are SIGNAL diodes - the transient voltage from the inductive kick can blow them apart (really - been there done that. Got laughed at.

I use 1N4148 diodes for that application also... But if they are likely do die, I guess I should look for a more powerful replacement. Is the problem only likely to happen with large relays, or with all relays?

When boss is also father that really stings....

I can tell my father whatever I like - he hasn't got a clue what I'm talking about anyway :-)

but father has even better relay inductive kick tale, from when he was in training)

Do tell :-)

use at least 1N4001, I prefer 1n4007.

I hope I can find a diode that fits my existing PCB designs (3 units and 0.7mm holes).

The internal diodes are a way away from the relay. The inductive currents will flow in fast long loops BACK to the ULN2803 and generate more crud: The diodes (and RC filters) MUST be on the pins of the relay....

I can easily fit the SMD version of the 1N4007 underneath the relays. It should even fit the existing solder pads.

Incidentally, Allegro do some neat little serially addressed relay drivers that do the 595 and the 2803 in one box.

Yes, the UCN5821 and UCN5841 look nice. They work almost exactly like the 595/2803 pair, and they have a nice pinout (all the outputs in a row).

Too bad they cost 5 times as much as the 595/2803 pair from Farnell (almost $6)...

Best regards,

Mikkel C. Simonsen