4

I'm using copper tracks laid out on an FR4 board as a heater device. It's a serpentine design with the copper 0.3mm wide and 0.2mm spacing. I have three sets of tracks which I intend to heat up at different temperatures, one of which at 95°C.

At about 85°C, the board near the tracks starts deforming(see image), with smoke emanating out, from the tracks. Deformed Cu tracksDeformed Cu tracks_2

My question is, does increasing the width of the tracks help in increasing the thermal stability? Also, does the spacing play a major role in it as well?

Does using a thicker board help with dealing with the thermal stress? If not, how can I add in the heatsink capability for it?

FWIW, I'm using a TIP122 transistor with a PWM input to heat it up. The current consumption is about 1.5 ~ 2 A.

Koushik S
  • 65
  • 2
  • 7
  • What is the power loss of your device? Is copper thickness 2-oz 0.07mm and 1.5mm the FR-4 board? – GR Tech Nov 05 '15 at 07:55
  • How are you measuring the temperature? – DoxyLover Nov 05 '15 at 08:03
  • The FR-4 board is 1 mm. I'm not sure about the copper thickness though. Not sure about the power loss of the device. I was running the TIP122 with a 12V supply and the device was taking less than 2A before it started smoking. – Koushik S Nov 05 '15 at 08:06
  • I'm measuring the temperature with an LM35 glued to the tracks – Koushik S Nov 05 '15 at 08:08
  • Related: This excellent reference - [**TI Analog Engineer’s Pocket Reference - 4th edition**](https://www.ti.com/seclit/sl/slyw038b/slyw038b.pdf) provides some useful information on PCB track current/ voltage drop / heat / fusing issues. Especially pages 55-68. – Russell McMahon May 05 '16 at 05:31

2 Answers2

2

Thermal stability is not your problem. Either you are running your traces hotter than you think, or your substrate is not FR4. Your temperatures are high enough that the epoxy is failing and the board is delaminating.

Since you have a temperature sensor attached to the traces, I'll assume your temperature measurement is accurate. So I'll focus on the substrate. Unfortunately, "FR-4" says nothing about the actual materials used in making a PCB. It refers only to flammability - "FR" stands for "Flame Resistance", after all. Worse, FR4 has become a generic term, like Kleenex, and there is no guarantee that any substrate labeled FR4 actually conforms to the standard. This site, for instance, lists continuous operating temperature for FR4 as 285 F (140C) minimum, so a real FR4 substrate ought to survive your situation.

I'd suggest finding another PCB supplier.

WhatRoughBeast
  • 59,978
  • 2
  • 37
  • 97
  • 1
    You were right on both accounts! The LM35 had a slow response to the thermal changes, AND the board wasn't really FR4, but was a bakelite-made one - I didn't know about the difference before. Thanks for the help! – Koushik S Nov 24 '15 at 05:08
0

You mix materials with very different thermal resistances. Typical common copper plane of PCB (2-oz 0.07mm) has a lateral thermal resistance of 40oC/W and 1.5 FR4 board 2,400oC/W!

So the lateral heat flows through copper i.e. from the sides of PCB. This can be solved using large area and short paths (refers to the thermal flow). So even using thicker board the heat still flow through copper.

Also wide gaps between copper tracks adds to temperature rise.

A double side PCB will help, because in this case the heat will flow through the board.

With careful design you can reduce the overall thermal resistance down to 8oC compared to 160oC.

GR Tech
  • 2,208
  • 2
  • 15
  • 17
  • Would thicker copper tracks make any difference in terms of temperature stability? Also, would having some sort of a heatsink at the bottom of the board (aluminium?) help, in any case? – Koushik S Nov 05 '15 at 08:48
  • I think I've seen some custom heatbeds for 3D Printing, made using the copper track technique, where I believe they've used a single sided PCB. I'm unsure of the other parameters though. – Koushik S Nov 05 '15 at 08:51
  • Thicker copper it is inversely proportional to thermal resistance but adding a little. The heatsink in the bottom it is equal to the second copper layer (double side pcb). Above all is the surface area of the device. If for example the dissipation is 24W and the target temperature rise is 95oC then the area of the copper should be around 25sqcm distributed in large area with short paths. – GR Tech Nov 05 '15 at 09:03
  • @KoushikS: Are you perhaps referring to http://reprap.org/wiki/PCB_Heatbed ? – davidcary Nov 07 '15 at 04:25
  • @davidcary Hadn't come across that (thanks!). Was referring to [this](http://www.printrbottalk.com/forum/viewtopic.php?f=113&t=6879) and [this](http://www.raphnet.net/electronique/pcb_heater/pcb_heater_en.php) – Koushik S Nov 07 '15 at 13:52