LM317 Misfunction

Thread Starter

Ravaner

Joined Apr 7, 2008
12
Hello,

I used LM317 regulator in an adjustable power supply connected to a filament ( working under vacuum ). This circuit worked correctely for several weeks and now is dead. My printed circuit board has been checked and is OK. Normally this IC is thermaly protected and also protected against short circuit. Has someone met same kind of problems with this IC ?
 

beenthere

Joined Apr 20, 2004
15,819
No, they are usually very reliable. What kind of voltage and current was the filament using? Did the filament float on a high voltage? Was there a heat sink on the 317?
 

Audioguru

Joined Dec 20, 2007
11,248
I read somewhere that if an IC (voltage regulator or power amplifier) goes into thermal cutoff too many times then the thermal stress of heating and cooling over and over will kill it. They say to use a heatsink large enough to prevent too many thermal cutoffs.
 

hgmjr

Joined Jan 28, 2005
9,027
Hello,

I used LM317 regulator in an adjustable power supply connected to a filament ( working under vacuum ). This circuit worked correctely for several weeks and now is dead. My printed circuit board has been checked and is OK. Normally this IC is thermaly protected and also protected against short circuit. Has someone met same kind of problems with this IC ?
This sounds like a good time to look at the thermal stress to which the part is being subjected. If the part is not sufficiently heatsinked as beenthere and audioguru have pointed out, there is a high likelihood that it is being thermal cycled to death.

hgmjr
 

Papabravo

Joined Feb 24, 2006
21,094
You do know there is a huge difference in the resistance of a cold filament and a hot one. Why on earth would you be using a regulated DC voltage on a filament. It doesn't sound like a very bright thing to do -- so to speak. In all the tube circuits I ever built I always used AC from a transformer on the filaments.
 

Thread Starter

Ravaner

Joined Apr 7, 2008
12
Many thanks for your answers. First my filament is installed in an homemade experimental apparatus ( mass spectrometer ) and not in a classic radio tube. Heat sink is large and always stays almost cold. When voltage (minimum one = 1.2 V ) is connected to filament, resistance increases and maximum current is in the order of 350 mA. I don't use any cap in parallel with filament and so I did'nt install any protection inversed diode ... I don't undestand why it works several weeks without any trouble. According to my calculation, even during switching on, transient current couldn't exceed 800 mA !
 

SgtWookie

Joined Jul 17, 2007
22,230
What is the input voltage to the LM317? Do you have a cap across the input?

Realize that the maximum power dissipation in the LM317 will occur when it's output is at the lowest voltage, which (coincidentally) is where your filament will pass its highest current due to its low (cold) resistance. As the filaments' temperature increases, so does its resistance.
 

beenthere

Joined Apr 20, 2004
15,819
The problem is almost certainly related to the current draw of the cold filament. One possible kludge to get around that is to use a switch to start up the filament from the DC supply through a resistor to limit current. When the filament has had a minute to so to warm up, flip the switch to the LM317 for regulating the filament current more accurately.
 

Audioguru

Joined Dec 20, 2007
11,248
Years ago I had a project that used incandescent light bulbs but they needs to flash on and off with DC. They frequently burned out.
I added resistors to keep them warm (but very dim) and then they lasted nearly forever.

Try adding a resistor to keep the filament warm.
 

kubeek

Joined Sep 20, 2005
5,793
You could also use the LM317 as a current source, sourcing the current that is rated when the device is hot. That way you prevent high current peaks during warmup.
 

Thread Starter

Ravaner

Joined Apr 7, 2008
12
Many thanks for your answers and tricks. I've inserted a small value resistor in my filament circuit, and it works.
 
Top