Philosophy: The better of two evils

[losikid]

Hey,

Working on a new product and I can't decide between two evils. The product is intended to switch on and off RC accessories (Light bars, etc).

Evil Number 1:
~The GND wire is switched to kill or resurrect the circuit's power.
~When no power, circuit is killed, switch is open. (Like a Normally-Open relay)

Evil Number 2:
~The HOT wire is switched to kill or resurrect the circuit's power.
~When no power, the circuit is still connected, switch is closed. (Like a Normally-Closed relay).

The BUTS:
Evil Number 1:
~RC electronics usually run a ground wire back to the source. So there's less chance of the hot wire shorting to the chassis if the insulation fails.

Evil Number 2:
~There is no power, does it matter if the circuit is still closed?




I can make it so that it can be normally open and switch the high side. But this is going to add $10-20 to final price of an already over-priced device. So I'm debating picking an evil.

Anyone have additional insight, opinions, or comments they want to share so I can ponder on this abit more?







SPOILER ALERT:


I'm back onto my Multi-Toggler project. I'm giving it powered outputs that can handle a few amps per switch.

[rockhugger]

Are you considering what is absolute? From my understanding, Interrupting a circuit of this nature via the positive can allow an uncontrolled back flow through (Neg/Gnd) to a short.

Interrupting the Neg. Would mitigate such potential hazards, or did I miss simething?




Blue 42?

[TWS]

I would go with toggling the ground wire, seems the safest way to me? If the "switch" fails then the light would just stay on?

[losikid]

Originally Posted by rockhugger

Are you considering what is absolute? From my understanding, Interrupting a circuit of this nature via the positive can allow an uncontrolled back flow through (Neg/Gnd) to a short.

Interrupting the Neg. Would mitigate such potential hazards, or did I miss simething?




Blue 42?

In most wiring, Negative is the common and is grounded to the chassis. Negative is common because it is at the lowest electric potential (lowest electric pressure). All fusing and (usually) switches needs to be placed at the source of voltage (positive) so if any part of the circuit is shorting to ground, then you can kill power to the whole circuit. It's not safe to fuse or switch the ground lead because if the circuit is shorting to ground somewhere else, then the fuse and switch can't kill the circuit.

Think of a garden hose with a spray nozzle on the end. The positive side is connected to the house (source of water pressure) and the end of the hose is connected to the spray nozzle. Outside is grounded, negative.

When the spray nozzle is turned on, the water flows and there is no pressure (aka voltage) between the end of the hose and outside. When the spary nozzle is off, water doesn't flow, and there is alot of pressure between the end of the hose and outside.

Now, you are controlling the circuit/water with the spray nozzle. But if the hose ruptures you create a short to ground and your spray nozzle is no longer controlling the flow of water. You have to turn the faucet off at the house to kill the water flow.

I'm trying to choose,

Do I control the spray nozzle? And shorts can't be controlled.

Or control the faucet on the house, but the faucet is spring loaded to normally be open. I have to have power to turn the faucet off. If there is no power(water pressure) or the controller breaks, then the faucet springs open.


For full size cars, I would design it to be super safe and the circuit cuts all power, fails in the off state, and requires power to turn the circuit on. But for RC cars, the chances are you'll disconnect the battery if something goes wrong instead of trying to get to a little tiny toggle switch inside the cab. So I'm trying to decide which evil is the best to save the customer 10-20 bucks.

[losikid]

Originally Posted by TWS

I would go with toggling the ground wire, seems the safest way to me? If the "switch" fails then the light would just stay on?

In the event of switch failure, the light would turn off.

But if the circuit fails (shorts out), then you have an uncontrollable short and the light may stay on, but can't be controlled.

You can't just turn the circuit off and fix it later if there is a short.

[E. Bryant]

Why not just find the appropriate "smart" high-side driver IC and call it a day? You can get a fairly nice quad-output part from ST for a few bucks.

[Hydrocarbon92]

You could use a p-channel mosfet and have the best of both worlds. You'd be switching the positive wire, but it would be turned on/off by grounding it's gate. The high-side driver could control a n-channel mosfet in the same spot, but work with a more typical positive input. Looking at your project, I realize that probably won't help.

If this is a mechanical switch, you only really worry about which wire to switch when you think the load could be shorting and the chassis is grounded. Switched ground means if a wire shorts and the switch is on OR off, sparks will fly and batteries could fry. If you switch positive, then any shorted wire when the switch is off results in zero excitement. Shorting a wire to a grounded chassis if the switch is on, regardless of which you switch, sparks will fly. Switching either wire when there's no exposed ground usually means no harm if there's a short, unless it's positive to a negative wire.

Honestly if you're worried about failed insulation, you could just incorporate a tiny glass fuse. I think I'm having an issue with your wording because your project looks to be all mechanical switches, but you're wording it like it's using relays or mosfets.

[losikid]

Originally Posted by E. Bryant

Why not just find the appropriate "smart" high-side driver IC and call it a day? You can get a fairly nice quad-output part from ST for a few bucks.

Adding high-side drivers will increase the final cost 10-20 bucks. I can't use a multi-output driver because I plan on each switch being able to supply a bus voltage or use an independent input voltage. This requires individual gate drivers and each driver has other associated components so I'd have to add roughly 20 components on a board that is just 1.25 sq.in big. I'd probably have to upgrade to a 4 layer board at that point and assembly costs will go up as well.

Originally Posted by Hydrocarbon92

You could use a p-channel mosfet and have the best of both worlds. You'd be switching the positive wire, but it would be turned on/off by grounding it's gate. The high-side driver could control a n-channel mosfet in the same spot, but work with a more typical positive input. Looking at your project, I realize that probably won't help.

If this is a mechanical switch, you only really worry about which wire to switch when you think the load could be shorting and the chassis is grounded. Switched ground means if a wire shorts and the switch is on OR off, sparks will fly and batteries could fry. If you switch positive, then any shorted wire when the switch is off results in zero excitement. Shorting a wire to a grounded chassis if the switch is on, regardless of which you switch, sparks will fly. Switching either wire when there's no exposed ground usually means no harm if there's a short, unless it's positive to a negative wire.

Honestly if you're worried about failed insulation, you could just incorporate a tiny glass fuse. I think I'm having an issue with your wording because your project looks to be all mechanical switches, but you're wording it like it's using relays or mosfets.

I'm debating p-channel high side switching and n-channel low side switching...I'm wanting to avoid a high-side driver for an n-channel fet to reduce costs.

P channel fets are normally conducting. You have to apply voltage to the gate to turn them off. If the gate oxide becomes damaged or there is inadequate voltage to turn off the fet (Possible in a short scenario), then you can not turn off the fet.

So you can switch high side, but there's a good chance you can't turn it off during a short. If the fet fails, then the circuit is always on.

Or you can switch low side and you're screwed during a short. If the fet fails, then the circuit is always off (unless a short bypasses the fet).

I'm designing a MOSFET board to plug into the back of the toggle switch board to increase the power handling capabilities from 80mA to several amps. The toggle switches will be controlling the gates.

[losikid]

For reference. The Sub-Atomic Multi-Toggler

[Hydrocarbon92]

A short is very unlikely - very few parts of a scale crawler are grounded, unlike a real crawler. ESC's generally use n-channel and 99.9% of the time they fail simply from excessive current, not a short. A resetable fuse, replacable fuse, or even a fused solder link would suffice for over-current or short. A mosfet failing during typical use is extremely rare unless it's a chinese clone, so I always recommend buying from a USA supply house.

If it does get shorted, it really depends on where. If you short a n-channel mosfet gate, it won't turn on but may fry the wires/traces between gate and switch when powered. If you short negative to positive and the mosfet is off, it won't do anything til it turns on. When it does, you'll fry whatever is the weakest link in the short - wiring between the short to the battery lead. If it shorts the negative lead on the accessories to ground, the accessories wil turn on & stay on.

One more thing regarding tiny switches controlling the gates. Even modest mosfets can pull considerable gate current, causing switch arcing. They effectively act as a big high-ESR capacitor. Some decent mosfets can see a peak of >10A at the gate! Since you're not doing high speed switching, a current limiting resistor of a few hundred ohms will be sufficient.

One interesting mosfet I've found is the ST Omnifet, VNL5090N3 & VNL5090S5. THey can support up to 13 amps and have an integraged driver with all sorts of protection. They were designed for automotive apps and are very cheap.


All in all I'd suggest the n-channel and offer a simple warning to have people disconnect the battery when not in use & inspect any damaged wiring. That is BASIC lipo safety, after all.

[losikid]

ESC's use N-Channels in a bridge configuration, they can switch the high side. So if a motor lead shorts out, the fet can either actively isolate the circuit or fail and still isolate the circuit.

I want to use N-channels with a high side driver, it is the only "safe" option. But it's not a cheap upgrade. So I'm debating selling something that doesn't mitigate risk during failure. But neither option seems better to me. Switching on the high side at least follows proper switching convention but that is the convention because you can isolate power through the whole circuit. But a P-Channel will only isolate when the circuit is working correctly.

I'm an EE in aerospace, we don't compromise and always consider failure scenarios. Then we figure out how to mitigate damage during a failure. Which is why everything is so damn expensive. I don't think the extra 20 bucks is worth it on an RC car accessory. The servo wire should melt before the battery becomes dangerous.

Those are cool little FETs, although at 90 mOhms it's not going to be easy to dissipate 2.25 watts of heat if someone wants to run 5 amps through it. Just the case alone will reach 260* C. If I was to only use a gnd plane on my board as a thermal heat sink I bet the FET would still get upwards of 100*C. Probably closer to a 2 amp fet without heat sink. (2 amps thru 90 mOhm = 0.36 watts. That mosfet has a case to ambient thermal junction resistance of 102*C per Watt. So a 36* temp rise plus a 30* ambient temp. Will run at 76*C (170*F)).

Compared to this P-channel (I just grabbed a random one. I would try to find something higher than 20V), http://www.mouser.com/ds/2/427/si7655adn-348543.pdf , at an Rds of .0048 ohms and an ambient thermal junction resistance of 21 *C/W. It can handle 14.4 amps per 21*C of temp rise. So I wouldn't have to worry about heat dissipation at all at the 5 Amp levels I'm aiming to set the product specs at. (2.52*C Temp rise at 5 amps).

And yes, I'm implementing gate resistors, debounce caps, flyback diodes, tvs diode, the works...But that is all stuff to eliminate warranty on my end. I'm just not sure if it's necessary to be able to protect the user's stuff during a short if they flip the switch off...because I'd pull the battery if something started shorting long before considering flipping the switch. But I also understand some batteries are hard to get to so the switch may be quicker.

[Hydrocarbon92]

So is this circuit going to power the whole panel or will you have individual mosfets activated by each panel switch?

As far as the complexity, remember the KISS rule. Since you're using them like solid state relays, there's really no need for much beyond a fuse and/or a fat diode IMO. My bet is you'll end up having a higher rate of random component failure than customers shorting or reversing the wires. It's also going to start getting big, potentially rivaling a bank of HD signal relays. The relays would have the advantage of giving users that solid click you feel from a heavy duty switch on a real rig.

At the very least I'd suggest maybe having a few mid-power switches and 1-4 high-power switches. You could cut cost and complexity that way, and I doubt many if any crawlers need multiple 5A switches. Just a thought.

[losikid]

Originally Posted by Hydrocarbon92

So is this circuit going to power the whole panel or will you have individual mosfets activated by each panel switch?

As far as the complexity, remember the KISS rule. Since you're using them like solid state relays, there's really no need for much beyond a fuse and/or a fat diode IMO. My bet is you'll end up having a higher rate of random component failure than customers shorting or reversing the wires. It's also going to start getting big, potentially rivaling a bank of HD signal relays. The relays would have the advantage of giving users that solid click you feel from a heavy duty switch on a real rig.

At the very least I'd suggest maybe having a few mid-power switches and 1-4 high-power switches. You could cut cost and complexity that way, and I doubt many if any crawlers need multiple 5A switches. Just a thought.

One for each switch.

It'll all fit on my 1.45"x0.85" circuit board, Including the 20 pin header. The p channel fets I linked to are only 3mmx3mm and I can use 0402 components on the protection circuitry (Might do a bigger package for the flyback diode).

The circuit is really easy to design, I have both n-channel and p-channel prototypes sitting on my desk. I just can't decide which way I want to go....I'm leaning towards p-channels though, wiring it makes more sense to me. This thread is mostly aimed toward seeing how people feel about switching the low side vs high side. But I made it a much more complicated question like I usually do.