I always planned to include it in some form in the next iteration, and I might just drop it into an ATF1508(along with the glue), to try and circumvent the macrocell limitations that Mr. Rictor has been having. It is brute-force, but I think I'll need the extra macrocells anyway.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

The famous Bob Pease once described a circuit for reverse polarity protection using an N-channel Enhancement Mode MOSFET. The circuit was something like that one I sketched in the attached figure. Search EDN Design Ideas/Electronic Design for the article.

_________________
Michael A.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

Initially, I had considered putting something in parallel more appealing than series, but now that I think of it, I can get hold of a servo connector, and put a diode in series with that. The supply I'm using puts out 5.44 volts, so I think I can get away with a series diode there. It also saves crowbarring the supply I'm using, since I have no idea what overcurrent protection it has, if any.
I'll put the diode on the next board revision; I do not want this happening again.
Speaking of which, I really need to start a checklist/to-do list for that.

A reasonable strategy would be to use a polyfuse with a crowbar. Then you have your own OC protection and negligible voltage drop in service. The polyfuse will automatically reset itself when the power is removed and it's allowed to cool.

Over the years, the Raspberry Pi has gone through a number of power supply protection iterations - however one thing there is that they're almost disposable, unlike our one-off and old chip designs here, so I feel we need to be a bit more careful.

On the Pi, wiring up the PSU is almost impossible as the power input is USB which is almost impossible to put in the wrong way. (although I've seen otherwise). People do power the Pi via the GPIO inputs though which is no different to our 6502 case here.

The initial (Pi) strategy has been a polyfuse and reverse biased schottky diode and today they still use a polyfuse (self-resetting fuse) but they've also tried "perfect" diodes and nothing at all. If anyone is interested then (partial) schematics can be found here: https://www.raspberrypi.org/documentati ... /README.md

Polyfuses are good, but they do drop a few mV - todays Pi PSU recommendation is for a 5.2v PSU, although there are reasons other than the on-board polyfuse for that.

Personally, I try to use polarised power connectors if possible, but constant vigilance, and all that...

-Gordon

_________________
--
Gordon Henderson.
See my Ruby 6502 and 65816 SBC projects here: https://projects.drogon.net/ruby/

Servo connectors are polarised. The main issue is finding a male connector for one, since those are usually on the receivers. The female ones are common enough. Failed servos and motor controllers have sent a few my way. The common sort of 2.45mm pin header fits them, and that's often all that's used, along with moulded cases on some of the more expensive receivers.
I have access to other polarized connectors, but those are mainly battery connectors(XT60s and the like), and I don't want to risk plugging (comparatively) high voltage straight into my mainboard, because I think 11v will probably destroy every last bit of silicon there.

I'm currently using a 2-pin 2.45mm socket on the board, and a matching pin header on the motor controller. I do need some form of protection, and ideally a polarized connector.

For battery-powered applications, putting in the battery backwards is an easy mistake to make, even if it only makes contact momentarily. The variable voltage also has to be accounted for, even if only with a conservative choice of clock speed. Being able to recharge the battery in-situ is also potentially valuable, but leads to yet more complication.

If there isn't already a comprehensive thread about power supply arrangements, perhaps there should be one.

I power my POC units from an old PC ATX power supply, which of course, does have fault protection—there is no input fault protection on any of these units. On POC V1, I used a Molex 5-1/4" floppy drive connector for power input, which cannot be plugged in backwards. POC V2 has a 3-1/2" floppy drive connector (aka Berg connector), which theoretically could be plugged in backwards. However, doing so would be difficult and immediately obvious. I have not had any reverse polarity contretemps with either of these setups.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Man, I hate those Molex connectors.

I completely appreciate their design, but they're so hard to get out. Feature, not a bug, for most cases, but not when just futzing around.

But, yea, I was thinking more like those Berg connectors. Seems more usable.

You could also use simple headers, just use a 4 pin header with 3 pins with a gap, and clog the hole for the missing pin (stick a wire in it and cut it flush, put some epoxy in it, something like that).

I quite like RCY (often referred to as JST) connectors since they're small, polarised, 2 position (rather than 3 for servos) and available cheaply on eBay in packs of 10. I even get the ones with silicone insulation for that quality feel if I'm going to handle them often. It's handy that 3S LiPo (~12V) batteries are available OTS with RCY though I check the voltage carefully before connecting because the same connectors are wired to power sources with all sorts of voltages. At least I don't have to check polarity. I can even connect them blind by trying one way around and then the other. I like to add a bit of heat shrink between the insulation and the housing for some strain relief. They can be tough to disconnect hence the RCY (JST) connector handle, which works well. It's nice to standardise on a connector because now I have all sorts of plug-n-play modules like switch mode supplies with RCY in and out, in-line switches and shunts for current measurement.

That's why I switched to the Berg connector for POC V2.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

It would be nice to use standard connectors for things. However, me being me, I would want different connectors for different voltage levels so I don't mix 'em up.
Battery power(variable DC) already comes in JST and XT60, and I'm already using JST for that.
A Molex connector from an ATX power supply would probably be good for both 5 and 12 volts, since they usually have both anyway. I even have a couple of old ATX power supplies, although I can't get the silly things to stay on. I have to hold the power-on line active or they just shut off(with or without a load resistor), and I feel like doing that isn't such a good idea. Molex connectors are also really flippin' hard to disconnect, which is something I do a lot with Ittiara when I break it down to put it away.
I'll have a look at the Berg, see what it is.

The ATX PSU command wire isn't latching - you *are* meant to hold it shorted to keep the PSU on.

Berg connectors are the ones used for PC fans. As such, they're readily available in 3-4 pin variants, so just put a different voltage on each pin. However, it is still quite easy to misalign or reverse them, unless you choose a shrouded socket.