Power Op Amp (mis)Application

Thread Starter

Afterburner

Joined Mar 25, 2013
4
I just burned up four STMicro L165 power op amps driving with a 0.5V amplitude square wave at ~1380 Hz using the Unity Gain circuit suggested in their datasheet. I was powering with +/- 15VDC and it seemed to work initiially, but saw a current draw of about 800 mA (versus a spec of 40 mA), increasing to 1.6 A after it was toast. As far as I can tell I have set up their unity gain configuration correctly, but the 1k across the inputs seems kinda odd and the 0.22 uF cap seems kinda big to be swinging around. Anyone have any insight on what I've done wrong? I generate the input square wave with a couple of TL082's set up as a relaxation oscillator passed through a variable gain attenuator then summed with a DC offset. I've been starting out with a low amplitude square wave with no DC shift while breadboarding the circuit, though. The square wave generation works fine. Ultimately I'm trying to drive a couple of Peltier modules in series, hence the need for a power op amp.
 

Ron H

Joined Apr 14, 2005
7,063
What was your load?
Did you include the 220nF in series with 1Ω on the output?
Did you have 100nF decouplng caps from each power supply pin to ground?

The 1k to ground on the input is for stability. Otherwise, it would oscillate. The same probably goes for the 1Ω/220nF.
 

Thread Starter

Afterburner

Joined Mar 25, 2013
4
Load was initially a 10k resistor, then a 1k later. I didn't have any caps on the power supply leads to the op amp. I had a small heat sink attached to the TO220 tab. I'm pretty sure I didn't have any wiring errors with respect to the unity gain configuration shown in Figure 2 of the L165 datasheet and I actually got the correct inverted square wave output (for a little while) a couple of times before it got hot, started severely oscillating, and died. I reduced the power supply to +/- 10 VDC after I was surprised by the current draw (the spec says the quiescent drain current is 40 mA and the load shouldn't be more than 30 mA (1k) + 57 mA (0.22 uF + 1 ohm) = 87 mA. I was seeing ten times that. The datasheet isn't a lot of help, either. I didn't expect a problem with such a small AC waveform and no average DC component, so now I'm doing more research/reading. I'm just wondering if there is some corporate knowledge out there with respect to pathologies of the L165.
 

Ron H

Joined Apr 14, 2005
7,063
Power supply decoupling is mandatory. The op amp will probably oscillate without them.

Did you notice that the input pin numbers on the inverting unity gain example (Fig. 2) are swapped?
 

Thread Starter

Afterburner

Joined Mar 25, 2013
4
Yes, I caught the swapped pinouts. I've never tried to use a power op amp before, so I didn't have the PS caps even though they are shown in Fig. 1 of the datasheet. I've never had a PS issue with the small signal op amps I've used so I wasn't keyed in to that. Thanks!
 

Ron H

Joined Apr 14, 2005
7,063
Yes, I caught the swapped pinouts. I've never tried to use a power op amp before, so I didn't have the PS caps even though they are shown in Fig. 1 of the datasheet. I've never had a PS issue with the small signal op amps I've used so I wasn't keyed in to that. Thanks!
Many small signal op amps will also oscillate if they don't have supply decoupling.
A good rule of thumb is to use a 100nF cap to ground from each supply pin on each IC (digital and analog) in your circuit. Keep the leads short. A ground plane is very helpful.
 

SgtWookie

Joined Jul 17, 2007
22,230
The 0.22uF cap & 1 Ohm resistor in series form a Boucherot cell: http://en.wikipedia.org/wiki/Boucherot_cell
Additionally, since you are feeding the input of the L165 a ~1380Hz square wave, you're actually feeding it ALL of the odd harmonics of 1380Hz (that is, up to the bandwidth limitations of your TL082 circuit), which will greatly increase the power dissipation in the power opamp.
 

Ron H

Joined Apr 14, 2005
7,063
The 0.22uF cap & 1 Ohm resistor in series form a Boucherot cell: http://en.wikipedia.org/wiki/Boucherot_cell
Additionally, since you are feeding the input of the L165 a ~1380Hz square wave, you're actually feeding it ALL of the odd harmonics of 1380Hz (that is, up to the bandwidth limitations of your TL082 circuit), which will greatly increase the power dissipation in the power opamp.
But, at 0.5V p-p, the peak current is less than 0.5A, and the average power (in addition to quiescent) is about 50mW. That should be a piece of cake for this amplifier.
 

Ron H

Joined Apr 14, 2005
7,063
If the transition times become long the power dissipation will skyrocket.
Can you explain that?

If high amplitude oscillation exists, e.g. due to missing power supply decoupling, that will cause the power to skyrocket. That 1 ohm resistor will suck a lot of power if the frequency is high enough to make that 220nF cap look like a short circuit (or even a medium-length circuit :D).
Of course, the resistor might be the first thing to fail.
 

ramancini8

Joined Jul 18, 2012
473
I am thinking of the output stage power rising because slow transition times keep the output stages in the linear region too long. Am I off base?
 

Ron H

Joined Apr 14, 2005
7,063
I am thinking of the output stage power rising because slow transition times keep the output stages in the linear region too long. Am I off base?
It wouldn't be of much use as a general purpose op amp if it couldn't handle slow transition times.
It has to be something that defeats the internal protection circuitry.
 

Thread Starter

Afterburner

Joined Mar 25, 2013
4
I didn't notice that the 1 ohm resistor in the Boucherot cell got hot, but the minus Vsup on the L165 sure did - maybe because it is electrically connected to the heat sink tab and it was conducted heat not Joule heating in the lead. So would 0.1 uF mylar film be a good choice for Psup decoupling caps?
 

Ron H

Joined Apr 14, 2005
7,063
I didn't notice that the 1 ohm resistor in the Boucherot cell got hot, but the minus Vsup on the L165 sure did - maybe because it is electrically connected to the heat sink tab and it was conducted heat not Joule heating in the lead. So would 0.1 uF mylar film be a good choice for Psup decoupling caps?
You should use 100nF (0.1uF) multilayer ceramic caps. Leadless chip caps are the best, but you can also use leaded ceramic caps.
Here are some examples:

 
Top