Hey all,

I've been looking at the power supply section in my 65C816 project, and I would like to add a fuse. I'm wondering what type of fuse would be appropriate, as I'm looking at a tripping power of around 2.5W (5V/500mA). I realize that's not a lot of power to blow a wire.
This fuse is for the CPU board only, so I don't think a lot of current will be consumed in normal operation. I'm guessing 100mA max but it is currently a bit hard to measure due to my breadboards being a mess. The whole computer is currently at 200mA including an LCD etc. One constraint is I don't want to use SMD components, and I would like something resettable, so I don't have to carry spare fuses. The goal is to protect the outputs of the CPU and glue logic from shorts in downstream circuits.

So after a bit of research, besides the plethora of SMD components that would be appropriate, I found these resettable fuses from Bourns, by searching for "low-voltage fuse" https://www.farnell.com/datasheets/2283897.pdf
The most sensitive one has a 5 second trip time at 60V/0.5A according to the datasheet. Assuming this delay is proportional to the power, I calculated a 5 second trip time at 5V/6A. So that's obviously not appropriate for 500mA. Doesn't look like "low-voltage" to me
I'm not finding appropriate ratings either in the non-resettable category.

Anyone has a better component to recommend?

_________________
BB816 Computer YouTube series

It's the current which blows the fuse, I think - there's very little voltage drop. So a 500mA fuse is what you want. The 60V rating is a maximum rating, I would say, which means you shouldn't use that fuse for mains voltages - not a restriction which applies here.

Duh, that makes total sense, thanks!
I since found this article that explains how these work https://www.allaboutcircuits.com/techni ... mentation/. My mistake was using the power supply voltage as the voltage drop across the fuse, which is of course incorrect.

So! Looking more at the datasheet for the Bourns' fuses, I'm surprised at the time these take to shut off. If I'm reading the "Typical Time to Trip at 23 ºC" graph right, the -R050 which has a "trip current" of 500mA takes ~50 seconds to trip at this current, and ~4 seconds at 1A. I would have to go to the -R017 to trip in ~10 seconds at 500mA, and this one can only sustain 170mA in normal use.

I wonder then how to pick a safe fuse value. How much time does it take to ruin the output of a chip at 500mA or 1A? I figure it is hard to tell. I see a maximum current rating of ±25mA for the 74AHCT245 I'm using for the buffer. Could it take 500mA for 50 seconds?

_________________
BB816 Computer YouTube series

Perhaps your voltage regulator can perform the current limiting action?
https://e2e.ti.com/blogs_/b/powerhouse/ ... rent-limit

Interesting!

I don't have a regulator on that board at the moment, it expects a 5V power supply (USB, DC jack or bench power supply). Not sure I want to add the complexity.

The other issue I'm seeing is the resistance of the fuse, at these low current ratings we're looking at a couple of ohms. 3.2ohms for the -R017, that's 0.54V of drop @ 170mA. The 65C816 is specced down to 4.75V

_________________
BB816 Computer YouTube series

If it's the bus buffers you're trying to protect, a better strategy might be series termination resistors on the side of each buffer facing the expansion connectors. Those are probably a good idea anyway to minimize reflections. A 500mA fuse in the VDD line probably isn't enough to protect a 74AHCT245 - in principle that still allows a chip power dissipation of 2.5W, which I would imagine is enough to cook it.

--- oops, deleted this as you've already mentioned polyfuses..

But I am surprised at the trigger speed - the ones on the Raspberry Pi seems fairly swift in my experience!

-Gordon

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

That's actually not a bad idea!

I hadn't given much thought to the expansion connector yet, so I'm just thinking out loud, but I could see having a dedicated buffer for each signal, with termination resistors. I'll have to put bi-directional buffers on the data bus, address bus, R/W, /RD, /WR and VA, to allow for DMA access, with DIR tied to BE. Then the rest can be unidirectional buffers. Haven't researched termination in detail yet either, but let's say with a 75 ohms termination resistor, it caps the output current to 66mA @ 5V. A bit higher than the max of 25mA, but 200 ohms termination maybe wouldn't be advised?

I think I will still put a fuse on the power input for more general protection, but I need to figure out a solution that doesn't add too much of a drop.



Yes! The Bourns devices I'm talking about are of this kind. Interesting that the Raspberry Pi solves the drop issue by bumping up the voltage. I guess this explains why using only official power supplies is recommended

_________________
BB816 Computer YouTube series

This isn't quite my territory, but I did find one $1 low dropout regulator which provides current protection and is available in single-unit quantities:
https://www.farnell.com/datasheets/3118386.pdf
(XC6227 Series from Torex)

Looks like it drops 120mV or so, which is less than the fuse you mention. It's a six pin device but looks like it can simply be used as a three-terminal regulator.

It might be good for experienced people to chime in here!

Ah - you picked up my post before I realised thats what you were in about in your first post and edited it out... anyway...

The very first Pi had a 500mA inlet polyfuse and 2 x 100mA polyfuses, one on each of the USB outlets. within days people were bypassing the USB ones and some people even running wires from the µUSB power inlet directly to the USB socket power pins to enable the Pi to power more power hungry keyboards, hubs and so on. The 2nd revision Pi didn't fit the USB polyfuses, just the main inlet one.

Later Pi's used a dedicated USB power controller (current limiter) device to protect the Pi (and possibly things like PCB tracks) from rogue USB devices... (and 'brown outs' when hot-plugging USB devices).

-Gordon

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

_________________
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?

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

Just mulling over this, and with the idea that the question hasn't come up before, I was wondering whether a fuse is normally added to protect the electronics, or to protect the PSU. And then I had another thought: sometimes it's to protect the user and their home, because fuses are about preventing fire.

(In the case of polyfuses on USB connectors, I think it's about preventing some defective (cheap) peripheral causing damage to an expensive computer.)

So, mostly, we don't see retro designs with any protection. Very careful people with the resources will use a current-limited bench supply to bring up a new board. Everyone else does a quick smoke test - and eventually, perhaps, we learn to check and double-check our designs, and then the bare boards, and then the populated board, before powering up.

The fuse blows after the circuit is destroyed. The fuse is mostly a courtesy to reduce electrical fire.



On a more mundane level, 0402 LEDs are less than 1mm^3 of mostly plastic; typically held with a lower volume of solder. It requires an excess of less than 2 Joules to melt the solder and this may cause the LED to fall onto other electronics.

_________________
Modules | Processors | Boards | Boxes | Beep, Beep! I'm a sheep!