Thermistor sizing

A friend has an HP laser printer rated at 750W that trips out his inverter and/or generator when it powers on. The inverter is rated at 2k and his standalone generators up to 3k. Fortunately, I have a box of NTC thermistors. But unlike my previous inrush experiments with 50-75W laptop and TV, this will be closer to the limits.

Datasheet: https://www.elektronik.ropla.eu/pdf/stock/avx/ntc.pdf

INITIAL INRUSH UNDER 2000W

The easy problem, I'm sizing the pigtail to handle the initial in rush:

• 2000W / 110VAC ~= 18A :: maximum amps
• 110VAC / 10ohm ~= 11A :: initial current max

SUSTAINED LOAD 750W

• 750W / 110VAC ~= 6.8A :: steady state current
• 10ohm @25C -> 0.37ohm at max T :: the reduced resistance at max temp
• (6.8 * 6.8) * 0.37 = 17.1W
  • 2.2W NF08AA0100M <SMOKE!>
• (6.8 * 6.8) * 0.28 = 12.9W
  • 5.0W NF20AA0330M <smoke>

So either part is too small alone but the larger part is 'less bad.' So let's start with parts rated to handle the steady state current:

• 6.4A - 0.13ohm NF20AA0409M 4.0ohm cold, 5W
  • 110VAC / 4.0ohm = 27.5A <trips> :: single
  • 110VAC / 8.0ohm = 13.75A works, ~5A headroom :: two in series
  • (6.8 * 6.8) * 0.13ohm = 6W <HOT>, 6/5W 120% of rated heat load :: two in series
• 7.8A - 0.08ohm NF20AA0409M 2.5ohm cold
  • 110VAC / 2.5ohm = 44A <trips> :: single
  • 110VAC / 5.0ohm = 22A <trips> :: double
  • 110VAC / 7.5ohm = 14.6A, works, ~3A headroom :: three in series
  • (6.8 * 6.8) * 0.08ohm = 3.7W < 5W

So it looks like three, NF20AA0409M in series will work. Best practice, use thermal grease and mount each thermistor as far apart as possible to the sides of the junction box. The box will run warm, 11W, but should have a reasonable duty cycle versus being plugged in and powered all the time. But one load is missing, the heater.

Without the printer specs and a schematic, I can't design for the heater load. The reason is this is a soft-start setup for the main power that handles both the control electronics and the heater. Trying to reduce the input current during heater warm-up would kill the control electronics.

There is an alternative design that uses a current limiting, power resistor and time-delay relay. The circuit has the current limiting resistor in series but the open contacts from the time-delay relay are across the leads. When power is applied, the relay waits "n" time to allow charging current to the load. Then the time-delay closes the relay contacts and shunts the power around the current limiting, power resistor.

• 110VAC / 15A = 7.3ohm current limiting, power resistor (**notice three, 2.5ohm thermistors!)

Next I'd need to find a time-delay, shunt relay ($$). But unlike the three thermistor solution, the overhead would be just the trip-coil load ... much cooler. Thanks to the mechanical action to close the contacts, there is a delay already built-in that can be enhanced with ... a thermistor. However, the printer has its own ON/OFF switch and the inrush may be closing its power switch.

So this is what I sit around thinking about on a Friday night but I look forward to any suggestions and recommendations.

Bob Wilson

 

So this is what assembly looks like. First sizing and positioning the parts:


Next looping the leads to make sure everything fits in the junction box:


Now soldering the leads to put the termistors in series. I would have used heat-shrink but you can't really get complete coverage:


Everything together, the strain-relief clamp ensures no mechanical motion inside the junction box. The plastic cover helps keep the critters out:


There will always be some variablity in the parts. The initial resistance between the hot (small flat) plug points are 7.1 ohms. Not great but good enough, 110VAC/7.1ohm = 15.5A. Still low enough to prevent an instant trip.

Bob Wilson

 

Nothing a good engineer enjoys more than to find the requirements have changed:

Brother MFC-8710DW. To use the printer, I cut off everything except the PC, and low-wattage monitor and two 5-watt room lamps.
. . .
Power Consumption: 1200W peak at 77º F, 702W copying at 77º F​

I'll meet with my friend, Ron, with the part and an IR thermometer and propose a potentially destructive test. But first, we need to verify it works at the design limit, 750W.

Now I've looked at the resistance-temperature profile and there is a chance that over-stressed, the part may handle it briefly. As it heats above 100C, the resistance continues to fall-off, reducing the rate of power heating the thermistors. If at peak load, the heat is less than the 5W maximum, it may work. Otherwise, the magic smoke escapes.

So what happens if it smokes, pull the plug and no problem. I'll order some SCK304R7 (4.7 ohm, 15A) or SCK20-0512 (5.0 ohm, 12A), a better and cheaper part. Two in series will easily handle everything with an appropriate soft-start and full current operation.

Bob Wilson

ps. Once upon a time I drew some 4-wire, telephony units, that were in 'dead stock' because no one in their right mind would use (or could order) such 1960s technology. But I soon learned we needed a 4-wire switch. So I designed the circuit and bought the parts at Lowes and local electronics store. I tested and shipped it ... and it worked. Happily the 4-wire voice was replaced by Voice-over-IP technology which I also implemented at another site.

The sad thing about retirement is finding interesting problems to solve under time and budget limits. Of course there is always PriusChat. <GRINS>

 

I agree which is why used a metal junction block. The faceplate can be removed and I'm planning a 'smoke test.' I need to confirm the initial, 7.5ohm assumption works. We can also remove the face plate and open the other knock-out blocks.

But we still have a problem with the intermittent heater load.

I've already found 12A rated parts that should easily handle the printer 'heater' load without overloading the thermistors. Unfortunately, if the thermistors have already warmed up for the 700W load, there won't be as much of a resistance change when the heater comes on.

In this case, it may make sense to open the printer and find a wire powering the heater for a separate thermistor. Many heating elements have a positive temperature-resistance slope, low when cold and higher when hot. So a thermistor can easily stretch out the heater warm-up and if the controller uses the heater current to measure its temperature, no problem.

We always have the option of a double-wide junction box and mounting the thermistors with thermal paste on the inside of the case. Ideally with a small fan.

Bob Wilson