Shop OBEX P1 Docs P2 Docs Learn Events
How hot is too hot — Parallax Forums

How hot is too hot

xanaduxanadu Posts: 3,347
edited 2014-12-27 12:29 in General Discussion
I purchased a small OSD (on screen display) from eBay China. It uses an LM2940C 1A LDO.

The specs say it can handle up to 26v input and output 1 amp at 5v. I am using ~12v source (3S lipo) so there is a large drop. The problem is it gets so hot you can burn yourself touching it for more than a second. It gets so hot the entire board it is soldered to gets hot, and starts to smell.

It is within voltage spec, but doesn't seem right at all. It is secured to a very small section of copper plane, and I think this is why it gets so hot. The copper plane is about three times the surface area of the LDO itself.

According to the datasheet-

The TO-220 can be attached to a typical heatsink, or secured to a copper plane on a PC board.
If a manufactured heatsink is to be selected, the value of heatsink-to-ambient thermal resistance, RθSA, must first
be calculated:
RθSA = RθJA − RθCS − RθJC
where
• RθJC is defined as the thermal resistance from the junction to the surface of the case. A value of 3°C/W can be
assumed for RθJC for this calculation.
• RθCS is defined as the thermal resistance between the case and the surface of the heatsink. The value of RθCS
will vary from about 0.5°C/W to about 2.5°C/W (depending on method of attachment, insulator, etc.). If the
exact value is unknown, 2°C/W should be assumed for RθCS. (3)
When a value for RθSA is found using Equation 3, a heatsink must be selected that has a value that is less than
or equal to this number.
RθSA is specified numerically by the heatsink manufacturer in the catalog, or shown in a curve that plots
temperature rise vs power dissipation for the heatsink.

Without taking it apart I can't verify the numbers above and do any of the math, so I'm guessing. Let's say you really don't want this thing to fail, in general practice would you run an LDO that was so hot you couldn't touch it?

It seems like a red flag, however I am known to be a little too critical sometimes. The little devil on one shoulder is saying don't worry, the angel on the other is saying get a second opinion.

Many thanks.

Photo.jpg

Comments

  • DomanikDomanik Posts: 233
    edited 2014-12-26 12:16
    IMHO, it's too hot. You might consider dropping the voltage to 6.5V or a switcher reg to replace the LM2940C. With 12V input it operates at 50% efficiency, so your battery discharges twice as fast. And the extra heat/watts shortens the life of components in the adjacent area.
    My guideline is if you can't hold your finger against the IC for a minute it's too hot.
  • xanaduxanadu Posts: 3,347
    edited 2014-12-26 13:01
    I forgot to mention I can't drop the supply voltage. If I switched the ~12v supply to a steady 7v it would run cooler but it wouldn't scale with the incoming voltage which I'm displaying on the OSD.

    Thanks Domanik.
  • RDL2004RDL2004 Posts: 2,554
    edited 2014-12-26 13:03
    Power dissipation (watts) = Vin - Vout x current.

    While the 2940 may be able to handle 26 volts in and can output 5 volts at 1 amp, it probably can't do both at the same time. It might help if you can add some kind of real heat sink, but the best thing would be to reduce the Vin as low as possible.

    You might want to watch the EEVblog episode #102. Dave builds a constant current dummy load using a TO-220 MOSFET and goes over the thermal calculations pretty well.
  • xanaduxanadu Posts: 3,347
    edited 2014-12-26 13:34
    Watching that now and thinking heat sink may be the only thing to do.

    I added a pic to the original post. I was thinking about soldering some light copper wire in a U shape to the ground plane tab. Thoughts on heat sink?

    I guess something that is attached in a way heat will conduct, also something that has enough surface area to dissipate a seemingly large amount of heat. I just found my temperature probe it's hitting 220f but it feels hotter, like the head of a match after you put it out.
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2014-12-26 16:08
    Forget about the heatsink, you are way out of spec because you have not RTM properly, just the tiny bit that you were happy with. The package you show and you quote are different because a TO220 is a big through-hole package, not the little SOT89 or perhaps SOT-223 that is shown in the photograph. Anyway it doesn't matter even if it's spec'd for 35V and 3A max the truth is that you could have one but not the other, a bit like reading fuel consumption and top speed, as linear regulators work by dissipating the excess voltage*current as heat and even with a good heatsink there are still limits. A bit of copper will not help and besides needs to be in intimate contact with the area behind the die (silicon chip itself) to be really effective.

    The only practical way you can run off 12V then with your little board which we don't have a link or the current consumption figures for is to use a switching regulator to feed the LDO. Most of these "LDO" boards are designed to run off close to their regulation voltage which might be 3.3 or 5V so that they expect 5 or 6V inputs. That's why they use "Low Drop-Out" regulators and not switch-mode.
  • xanaduxanadu Posts: 3,347
    edited 2014-12-26 16:52
    I'm not sure a link would help because there is not much information but here it is - http://www.ebay.com/itm/OSD-FPV-Remzibi-OSD-OSD-ultra-thin-GPS-module-new-/111312327749?pt=Radio_Control_Parts_Accessories&hash=item19eabb5445

    Everyone is running them on 12v (3S lipo).

    I found a pretty good link with others having the same issue so at least I know it isn't me being overly critical - http://fpvlab.com/forums/archive/index.php/t-366.html?

    It looks like some people were okay with a heat sink and some people changed out the regulator. At this point I'm not sure what I will do. Seems like an easy enough problem to fix, I was just trying to gauge whether or not I should fix it. After finding the forum post above and the advice here I'm definitely not using it as is.

    Thanks.
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2014-12-26 17:48
    xanadu wrote: »
    I'm not sure a link would help because there is not much information but here it is - http://www.ebay.com/itm/OSD-FPV-Remzibi-OSD-OSD-ultra-thin-GPS-module-new-/111312327749?pt=Radio_Control_Parts_Accessories&hash=item19eabb5445

    Everyone is running them on 12v (3S lipo).

    I found a pretty good link with others having the same issue so at least I know it isn't me being overly critical - http://fpvlab.com/forums/archive/index.php/t-366.html?

    It looks like some people were okay with a heat sink and some people changed out the regulator. At this point I'm not sure what I will do. Seems like an easy enough problem to fix, I was just trying to gauge whether or not I should fix it. After finding the forum post above and the advice here I'm definitely not using it as is.

    Thanks.
    That link shows a GPS connected as well, if that is so then there is definitely more current consumption going on there. Clicking on another link for the same product comes up with some info:
    FPV Remzibi OSD System APM with GPS TTL for Multicopter Airplane Helicopter

    Specifications:
    - Power supply: 3S 12V
    - Video voltage measurement range 0 - 15V
    - Power voltage measurement range of 0 - 30V for
    - Universal ADC estimates range.
    - Video signals - standard PAL or NTSC
    - Global Positioning System - send and receive signals with any standard GPS NMEA protocol instant system .5 V power supply

    What! It says power supply 12V and yet the regulator is not designed for this voltage because of its load current. But then again, it's ebay and they just tell you and sell you what you want.

    However searching a bit more shows that there is a new 2013 version with step-down regulator (switcher).
    http://www.goodluckbuy.com/diy-remzibi-osd-open-source-v1-79-5hz-gps-for-quadcopter-fpv-system-2013-new-version.html
    Interesting they seem to allude to the fact that the old version is supposed to have an external step-down:
    The 2013 version not included external BEC step-down circuit, has been integrated into the main board.

    So there you go, your version will work on 12V, provided you use the external BEC step-down circuit (didn't we tell you? Oh, that's right, you didn't ask)
  • xanaduxanadu Posts: 3,347
    edited 2014-12-26 18:20
    Thanks Peter, that is great I must have missed it. Appreciate the hand holding lol... I will order the newer one.

    I didn't even think about the GPS current in addition to the OSD.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-12-26 23:51
    In general, the solution is to just dump all the heat by pre-regulation. (Yes, that would be an external step-down)

    Insert an LM317 with a heat sink between that higher voltage and the LM2940. And then regulate it down to something that runs right.

    And you don't have to use an LM317, there are nice adjustable switcher regulators that are very cheap that will do the same.
  • DomanikDomanik Posts: 233
    edited 2014-12-27 12:29
    xanadu wrote: »
    lol... I will order the newer one. I didn't even think about the GPS current in addition to the OSD.
    If it was me I'd buy a new one too, the reason: 1) My time 2) Waiting on parts 3) Installing a mod that may or may not work. Curiosity, however, would dictate a quick and dirty solution with a high wattage resistor of about 5 ohms to start with or a regulator as Loopy suggested. Or the copper wire coil as a heat sink. Then you'd have something to play with while you're waiting on a replacement.
Sign In or Register to comment.