I bought an LTC3780 step-up/down converter:

Photo of LTC3780 PCB


It is working fine but the LM358 on it is getting pretty hot. It seems no matter what I do with the LTC3780 the LM358 is always pretty hot. Like I have to remove my finger after 3-4 seconds hot.

I have put (non-permanently) a small heatsink on it (held by gravity) and when I remove it and touch the chip again I am able to touch it for pretty long without having to remove my finger again.

Is it normal?

I have applied a 2.5A current @ 5V during 4 hours and everything remained stable. The LM358 was not particularly hotter.

EDIT: Input voltage 12v from a switching power supply.

Thank you!

  • 1
    \$\begingroup\$ Can you measure the temperature? Op amps usually should not get too hot, but it depends how they're used in the circuit. Without a schematic we can't know how the LM358 is being used or if high temp is expected. It could be the result of a poor design. \$\endgroup\$ – vofa Jan 18 '18 at 20:17
  • 1
    \$\begingroup\$ @vofa - 60 C is generally accepted as the "don't touch" threshold, so a case temp in the 50s seems a reasonable estimate. \$\endgroup\$ – WhatRoughBeast Jan 18 '18 at 20:34
  • 3
    \$\begingroup\$ We need a schematic as we are guessing right now. Include details like the Vcc volts, component values, etc. Is it a breadboard design or a dedicated PCB? \$\endgroup\$ – user105652 Jan 18 '18 at 21:05
  • 1
    \$\begingroup\$ @MikeGleasonjrCouturier FYI - that probe type sounds like a thermocouple \$\endgroup\$ – user253751 Jan 18 '18 at 22:04
  • 1
    \$\begingroup\$ Fyi, here’s a blog where the board is discussed- specifically regarding how its LM358 caused a failure: wiki.beyondlogic.org/… \$\endgroup\$ – Blair Fonville Jan 19 '18 at 4:12

From the image I assume that the LM358 is the SOIC-8 device. I also assume that it is a real LM358. In your case it could be counterfeit, which is unfortunately common in these kind of modules.

Referring to the TI LM358 datasheet:

LM358 Thermal Metrics

The parameter of interest here is the junction-to-case thermal resistance, $$R_{\Theta JC(top)}\ = 72.2 °C/W$$

This means that for every Watt dissipated by the IC, the case temperature will rise by 72.2°C above ambient. You've measured 55-57°C, which is roughly 30°C above ambient. This means the chip is dissipating about $$P_{LM358} = \frac{30°C}{72.2°C/W} = 400mW$$

Is that too much power dissipation? The datasheet gives a maximum junction temperature of 150°C. How hot is the junction? If the case is 57°C and you're dissipating 400mW, it must be roughly $$T_{junction}=57°C + 72.2°C/W * 0.4W = 86°C$$ The chip is operating well below its thermal limits.

Is it normal? That depends on how the circuit is designed. Since your unit functions properly, I would say that it is not an issue in this case.

| improve this answer | |

Is it normal?

We can't know whether it is normal for that board, without seeing (and then analysing) its true schematic. (I specifically say its true schematic, as I've occasionally seen schematics provided for boards by vendors, after many requests, which didn't match the actual board!)

Unfortunately you have now seen the trap of these pre-built cheap boards, which are usually supplied without data, schematic, or even much confidence that the main components or specifications are genuine :-(

That op-amp being hot might be typical for that specific board, due to a valid design decision. Or it might be that the device marked LM358 is really a copy of the original device, which responds badly to that specific design and will have limited life, although an original LM358 would be fine. Or any of several other possible reasons...

Without having the schematic, some confidence in the origins of the components, and then perhaps needing to perform further measurements e.g. with an oscilloscope, we can't know the expected behaviour for that board. As I said, that's the trap with such boards - lots of unknowns...

You could start by buying another one of the boards, taking photos of the locations of the components, and then desolder every component until you are left with a bare PCB. Then reverse-engineer the schematic from the tracks on the bare PCB and the devices which you removed (you might need to measure unmarked components e.g. some L & C, using a suitable meter). Then you can start to evaluate why that op-amp is hot. As you can see, this isn't something which we can do for you "remotely". :-(

I see your options as including either:

  • get the true schematic from the vendor; or
  • reverse-engineer the schematic yourself; or
  • trust that the designer knew what they were doing; or
  • accept that you'll never know whether the board is truly working as designed.
| improve this answer | |
  • \$\begingroup\$ I don't want you to debug everything for me :) The answer could have been "yes they are always in that range"! I am a software engineer I don't know much electronics. I bought a second board from another supplier which seems of better quality according to some other reviewers. BTW I like to solder kits. I don't mind building my own step-up down converter if I can find a good tutorial. It is for a home made variable lab bench power supply (how original!). Thank you for your detailed answer! \$\endgroup\$ – Mike Gleason jr Couturier Jan 18 '18 at 21:16
  • 1
    \$\begingroup\$ @MikeGleasonjrCouturier - Hi, 'The answer could have been "yes they are always in that range"!' In which case, I can say that LM385 op-amps do NOT always operate in that temperature range. Very often, they run cool, due to having little power dissipation. It all depends on the circuit design (which is why the answer depends on seeing the circuit design i.e. the schematic). As shown by the Arrhenius equation, devices which run hotter generally have a shorter life than those which run cooler. \$\endgroup\$ – SamGibson Jan 18 '18 at 21:25
  • 1
    \$\begingroup\$ I will add the heatsink permanently and wait for the second unit. If the second unit is better I will swap it! Thank you! \$\endgroup\$ – Mike Gleason jr Couturier Jan 18 '18 at 21:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.