Heating operational amplifier OP284FSZ-REEL7

Thread Starter

Glebiys

Joined Mar 11, 2019
16
Hello,

Task: to make a square wave with a frequency of 1 kHz in the voltage range (-12V - +12V), using a PWM signal with voltage (0V-5V).

Circuit:


Originally LM258WYDT (ST) was used as an operational amplifier, it did not heat up, but since it was not from the Rail-to-Rail series, there was a 1.5V drop at the output, which did not allow us to give out a voltage close to 12V. So I decided to switch to Rail-to-Rail, for these purposes I found OP284FSZ-REEL7 (AnalogDevices). After replacing with this amplifier, the value of the output voltage became close to 12V (+-11.85 V), but the operational amplifier began to heat up (after 5 minutes its temperature was about 45-60 degrees).
I did not find a short circuit, but noticed that at the output of the voltage divider (R1, R2), instead of 2.5V, the voltage became 4.5V. On the LM258WYDT, this voltage was 2.5V. Perhaps there are elements on this input of the operational amplifier that should lift it?
Unfortunately, at the moment I do not have an oscilloscope and I can not see the output signal.

My suspicions:
  1. The amplifier is not designed for such a load or, conversely, it is not enough for it, and it begins to oscillate.
  2. Is this an accurate amplifier, perhaps additional elements are needed for its operation?
  3. The amplifier is damaged.
Parameters of currents from datasheets:
LM258WYDT
  1. Input current: 5mA (DC)
  2. Source output current (Isource): 20-60mA
  3. Output sink current (Isink): 10-20mA
OP284FSZ-REEL7
  1. Supply current: 2.25mA
  2. Output current: 10mA

Question: What could be the reason?
 
Last edited:

ci139

Joined Jul 11, 2016
1,179
maybe
Untitled-1.gif

it does not end to there - you likely need to compensate overshoots and provide some feed forward to sharper edges . . .
the visual examination of the op amp simplified circuit diagram may hint some peculiarities (i'm not into analysing it right now)
 
Last edited:

Thread Starter

Glebiys

Joined Mar 11, 2019
16
@ci139 Thank you for the answer!

The voltage difference between the inverted input (2.4 V) and the non-inverted input (5.1 V) was more than 0.6 V. But the divider 100k-100k can not give a current of more than 5mA, what is the disconnected operational amplifier?
The non-inverted input is directly connected to the microcontroller, which could give out 20mA.

The divider should have produced 2.4V, but most likely inside the op-amp this voltage rises to 4.5V to correspond to a difference of 0.6V. And is this causing high heat?
 
Last edited:

ci139

Joined Jul 11, 2016
1,179
( i actually never used any AD's opAmp-s ) . . . the product page says "Military temperature range (−55°C to +125°C)" so i speculate 80°C would be acceptable . . . maybe -- the d/s and app notes must be read carefully to figure this out . . .

. . . "they" had a ground loops cancellation shown as cap to (in this case) from Vee of opAmp to GND of your MCU and another across the opAmp-s supply pins . . . i donno if this changes anything here . . .

. . . will be reading those ↑ documents . . .
there are internal feedback capacitors that seem to ? slow down the inverting input ? (why ... or i am guessing the effect wrong ...) ? thus flash-boosting the positive feedback ?
AND ? an INP protection ? QL1 QL2 that may make the op amp logarithmic for "high" difference input signals . . .

? what if you configure your op amp as this https://hackaday.com/2018/11/13/whats-the-difference-ask-an-op-amp/
 
Last edited:

ci139

Joined Jul 11, 2016
1,179
it looks the internal positive feedback does exist + that your ±12V is best fed from ±5:N (V)
(but the Spice model behaves somewhat stubbornly ... might be a good sign of a ? realistic model ?)
OP248F - TEST - E01A.pngOP248F - TEST - E01B.png
 
Last edited:

ci139

Joined Jul 11, 2016
1,179
measuring the differential input impedance of the Spice Model
+ attempting to make sense of this opAmp . . . and failing (the "differential opAmp" wiring + input symmetrizing in prev. post works simulates . . . better )
but the AD820 seems to have no "pre-rise delay" issue
Op-Amp - dRinp - TEST - OP248F.png


OP248F - TEST - E01C.png!!! fixed the biasing ↑bug↓ !!!'

OP248F - TEST - E01CC.png
INSANE = Back to ® device

OP248F - TEST - E01CX.png
 
Last edited:

ebeowulf17

Joined Aug 12, 2014
3,153
Maybe this is just wishful thinking, but the footnote says to limit differential input current to less than 5mA, so couldn't you just put a 5-10k resistor in series between mcu output and op amp input and see if that prevents heating while maintaining functionally?

This problem is very interesting to me. I'm still a newbie when it comes to op amps, and I'd never seen a spec like this before. I knew there were some limitations when using op amps as comparators, but it never occurred to me that it would be this easy to violate input specs!

Anyway, this brings up another question/possible solution. Since you don't need linear output, why not use a comparator. It's specifically designed for larger differential inputs and is likely optimized for fast switching and good saturation characteristics, so it probably delivers a more square output too. Although people always assume that comparator means open collector or open drain output, there are plenty of push-pull output models out there which would be a drop in replacement.
 
Top