Safety "on board"

[newxito]

I’m redesigning my clock board and I would like to make some safety improvements. My actual board has two fuses, one for the 12V DC input and one for the HV output. In order to prevent overheating (fire), I also would like to monitor the temperature inside the case adding a DS18B20 to the board. If the temperature exceeds a predefined limit the idea is to shut down the HV. In normal operation with 6 x IN-18 the only component that gets a bit warm is the HV coil but you never know.

Is temperature monitoring an overkill? Any other ideas to improve the design safety?

[Bill van Dijk]

This is basic risk assessment. In order to establish the need for action, the likelihood of the event is correlated with impact.

 

questions:

1-How many clocks have caught fire due to a problem of some sort?

2-how likely is this sort of failure to occur?

 

to be correlated with:

1-what is the impact if this sort of failure were to occur?

2-what effort (cost, time, etc.) is required to prevent this from occurring?

 

Once the answers are found, the decision on required action can be made.

 

Bill

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/07888f27-c387-4a96-bf8b-e51c8cabd04d%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[gregebert]

In my opinion, safety is #1 regardless of cost.

Fuses - If you are bringing line-voltage into your clock, use a UL-listed power-entry module with an IEC connector (for the power cord), and an integrated fuse.

If you simulate your design, you can calculate the RMS line current under max-load conditions.

'Kill-A-Watt' devices are fairly inexpensive and will measure RMS current for you on a finished project.

From there you can select a fuse; be sure to review the manufacturers graphs for current vs time; you dont want too large a fuse because that wont protect you.

Too small a fuse will blow over time, or when powered-on when capacitors charge-up.

I usually add some small series resistance to limit the peak charging current. This is where SPICE simulations are very helpful.

On my first clock, which had no transformer, the series resistors act as backup fuses.

If you are using an external supply, think about adding a fuse right as the power comes in, even if it's soldered onto the PC board.

Heat- 3 things to do here:

#1 - Avoid generating it in the first place.  If you can use a switching regulator instead of a linear one, do that. There are pin-compatible switching regulators that replace the ubiquitous  TO-220 LM7805/LM7812, etc. Anode resistors (if you use them instead of current regulators) will generate heat, and need to be appropriately sized. Only my first clock design uses anode resistors, and I chose 2W resistors to dissipate 600mW.

#2 - Get rid of the heat. If you need to heat-sink a part, do the thermal analysis (not difficult) and only choose devices and heat sinks that provide theta JA/JC, theta SA values. Calculate (or simulate) the power dissipated in your device, then pick a suitable heat-sink that will keep the junction temp well-below the typical 150C and zero airflow. I start with an ambient temp of 50C and max temp of 100C, and work backwards to get a reasonable heat sink.

Lastly, dont trap heat inside your case; leave vent holes towards the top, and inlet holes at the bottom. Air will circulate thru the case as the temperature rises.

#3 - Keep electrolytic caps away from heat sources, otherwise they can dry-out and fail. I use EPCOS caps rated for high-temp applications in solar-energy inverters.

Dont forget that electrolytic caps in power supplies will self-heat; you can calculate the power from the RMS ripple-current and the ESR rating. Remember - The RMS ripple current will likely be higher than the DC load current.

I had a small electrolytic dry-up and short-out after 30 years, so be aware they dont last forever. Also, *dont* use electrolytics from your junkbox, or from surplus dealers. Buy fresh ones.

Surge Protection

After the fuse, I have a varistor to absorb line-transients, and a 0.01uF cap to absorb very short transients that are too fast for the varistor. If something really bad happened on the power grid, the fuse will hopefully blow.

Other things

Be sure to use appropriate-sized wiring or trace-widths for the fuses behind it. (Dont use 30 gauge wire with a 10 amp fuse). There are PCB-trace-width calculators for currents; I suggest this for anything over 100mA.

Trace-spacing is also important for higher voltages; I review rules for anything over 40 volts.

[John Rehwinkel]

> #3 - Keep electrolytic caps away from heat sources, otherwise they can dry-out and fail. I use EPCOS caps rated for high-temp applications in solar-energy inverters.

I'm becoming fond of the polymer electrolytics in apparatus I build myself. While they don't have quite the heat ratings of traditional electrolytics, they seem to be more robust for
long life at modest temperatures as they're not as subject to drying out.

- John

[Allen Dutra]

Newxito,

If heat is a concern I generally invest in cooling and changing the design to reduce heat instead of a thermal shutdown. Providing extra cooling and investing in parts the generate less heat ensures long component life and reduces the likelihood of fire.

When starting fires electrically, there are generally two options: create a spark the lights small tinder or heat any fuel to the point of spontaneous ignition. Physically the process is the same in both cases but the former is a faster process then the latter.

Assuming you have a device that creates more than 1000 volts and sparks a fire, a temperature activated power supply shutdown isn't helpful when the power supply is already on fire. At which point nearby fire alarms need to be the next line of defense. Alternatively, if you have a device/power supply that is handling 100s of watts, the amount to power being used can approach the point of spontaneous ignition when things go wrong. In these high power situations, thermal measurements can detect when power is being used in dangerous and unhelpful ways and a thermal shutdown will protect the device. 

Nixie Clocks tend to be somewhat exotic devices but don't generally need 1000s volts or 100s watts to work. Hence Nixie Clocks don't have unavoidable fire/heat concerns that come with devices like large power supplies, electric stoves or electrical discharge insect control systems (bug zappers) and Greg's advice will creating electrical safe where it matters. 

[newxito]

I started thinking about heat and fire because I recently used a wooden box as a case for a clock.

Very useful advices and explanations, thank you all