Shop OBEX P1 Docs P2 Docs Learn Events
Pin bending question — Parallax Forums

Pin bending question

max72max72 Posts: 1,155
edited 2014-12-23 09:16 in General Discussion
I have a small circuit using a BTS555 ad a switch and the prop to control it (via an optocoupler and a mos).
The manufacturer website is this one:
http://www.infineon.com/cms/en/product/power/smart-low-side-and-high-side-switches/smart-high-side-switch-profet-tm/BTS555+E3146/productType.html?productType=db3a30442ba3f989012bcf898eee271e

I assume the chip is to be mounted "standing" on the board, but I would like to bend the pin and have the chip tab laying on the board.

I'm wondering if there are best practices or suggested steups to mount this or a similar chip laying on the board

Thanks in advance,
Massimo

Comments

  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-12-20 06:26
    You might try practicing on something that is trash. This kind of modification is all trial and error. The chip manufacturers are really providing product that is intended for machine mass fabrication and not for human modification.

    My own experience is with similar packages that are linear voltage regulators. So far, they seem to tolerate quite a bit of abuse. But I would try to just bend them once. You might get away with a later straightening. But sooner or later, bending back and forth will cause a failure.
  • Heater.Heater. Posts: 21,230
    edited 2014-12-20 06:30
    Best practice would be to use a through hole board layout. "standing" on the board as you say.

    However I have done what you want recently. I put kinks in the legs with a pair of long node pliers such that I could solder it to surface mount pads on the board and have the thing lying down. I snipped the legs a bit shorter as well. It's a bit fiddly to do but worked out OK.

    Then you can solder the big tab to a large copper area on the board for a little heat dissipation.

    WARNING: First time I did this the tab on the device ended up touching some traces on the board that it should not. There was much heat and smoke....
  • ElliottOiElliottOi Posts: 1
    edited 2014-12-20 07:09
    I must say I'm impressed. Really not often I come across such sites as forums.parallax.com
  • max72max72 Posts: 1,155
    edited 2014-12-20 07:14
    Thanks for the suggestions.
    I plan to print/carve a small tool to bend the pins in a controlled way and with a radius not too small (and only once).
    The in current must pass trough the pad, and I'll solder the wires directly on it. So I can either provide a generous exposed copper area or leave the solder mask.. I think I'll expose the area and solder the component on the board.
    Massimo
  • RDL2004RDL2004 Posts: 2,554
    edited 2014-12-20 07:18
    With that type package you don't want it free standing. It will tend to wiggle back and forth and eventually the solder joints can fail. They're meant to be attached to a heatsink which is in turn attached to the board. This will give it more support.

    You can also mount it flat. Just grip each pin right where it comes out of the plastic with a pair of needle nose pliers to keep it from moving and breaking something inside, then bend the pin down just past where the pliers are gripping it. Like the regulator in this picture.


    attachment.php?attachmentid=96442&d=1350854263
  • Heater.Heater. Posts: 21,230
    edited 2014-12-20 07:43
    ElliottOi ,

    Welcome to the forum.

    Yes, we are awesome. And I'm only one of the least awesome ones around here.

    Do join in and be awesome:)
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2014-12-20 09:08
    Fro the datasheet, referring to pin 3:
    Positive power supply voltage, the tab is electrically connected to this pin. In high current applications the tab should be used for the Vbb connection instead of this pin.

    So, yes, you do want to bend the pins so the device lies flat in such a way that you can make a soldered connection to the tab. The device may also require a heatsink, depending upon your current draw.

    When you bend the pins, hold them with thin needle-nosed pliers between the bend and the package. This will prevent strain at the package interface.

    -Phil
  • max72max72 Posts: 1,155
    edited 2014-12-20 14:07
    Thanks for the suggestions.
    I'm more confident now on the approach.

    Massimo
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2014-12-20 16:30
    max72 wrote: »
    Thanks for the suggestions.
    I'm more confident now on the approach.

    Massimo

    As RDL mentioned, make sure you take the tension off the internal lead-frame by gripping the pin between the package and where you are bending it. With this particular type of device and such a low Rds(on) it is unlikely that it will ever need a heatsink because it is not designed for regular PWM, maybe very low frequency PWM, but with a max turn on time of 600us this component is really an on/off high-side switch.
  • Cluso99Cluso99 Posts: 18,069
    edited 2014-12-20 16:39
    As others have said, this is done all the time, even on professional boards.
  • kwinnkwinn Posts: 8,697
    edited 2014-12-20 21:25
    Cluso99 wrote: »
    As others have said, this is done all the time, even on professional boards.

    Done so often that I used a grinder to modify a pair of needle nose pliers so I could bend all the leads at the same time and the same distance from the package.
  • Cluso99Cluso99 Posts: 18,069
    edited 2014-12-21 16:05
    Also done often (when the short tab versions are not available) is to cut off most of the tab.
  • WBA ConsultingWBA Consulting Posts: 2,934
    edited 2014-12-22 09:06
    Yep, that's a common prep, but usually we buy SMT versions of the part that will be prepped that way by the manufacturer. TO-220 parts are also commonly prepped into SMT versions.
  • max72max72 Posts: 1,155
    edited 2014-12-22 13:32
    Thanks.
    The chip is available only with this package. I tested it mounted upright and it works well with no extra heat dissipation.
    It will be used as an on/off switch.
    The option to mount it laying flat is appealing for the final dimension. I was worried to break some rules bending the pins...
    Massimo
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2014-12-22 13:43
    max72 wrote:
    I tested it mounted upright and it works well with no extra heat dissipation.
    Okay, but do remember that if you're switching large currents, the V+ connection needs to be made to the tab, not just to the center pin.

    -Phil
  • max72max72 Posts: 1,155
    edited 2014-12-22 14:03
    Thanks Phil.
    I agree, the key point is to follow datasheet and directly connect Vin to the tab.
    For Vout there are two pins available to distribute current flow. I also plan to use 70 oz copper on both sides.
    Massimo
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2014-12-22 15:12
    Phil Pilgrim (PhiPi) wrote: »
    Okay, but do remember that if you're switching large currents, the V+ connection needs to be made to the tab, not just to the center pin.

    -Phil


    The center pin and tab are one piece of the lead frame and the package limit is dictated by the weakest link and since the current must flow through another pin there is no advantage to directly connecting to the tab itself. Not saying we wouldn't connect there though. For many to220 packs that I lay flat on the board I have a large exposed (no mask) pad for the tab and I find it beneficial to sweat the tab down but in this case there is no advantage here either as the device runs cools anyway.

    Remember that there are many times when devices like these have their tabs mounted to a common heatsink or the chassis via silicon pads so there can be no direct connection to the tab anyway.

    EDIT: having a detailed look at the datasheet the things it mentions about extra resistance are normal part of the design since you can have up to 0.5mohm "more" resistance if you don't connect to the tab. Now pcb traces and wires add a lot more than that and this kind of information is not necessary for "low" (not 165A) current use :) Mind you even 10 or 20A would still be consider large currents for this forum. The datasheet also reminds the designer to connect the two output pins together as well, so this is where they lower the pin resistance for the circuit then. Basically there is no real need in this application to connect to the tab and even though max hasn't mentioned what he is switching I extremely doubt it will be anything close to 165A!

    EDIT: that's an expensive part considering (~$10) but it sounds like max wanted to be extremely safe and never sorry :)
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2014-12-22 18:49
    Peter Jakacki wrote:
    The center pin and tab are one piece of the lead frame and the package limit is dictated by the weakest link and since the current must flow through another pin there is no advantage to directly connecting to the tab itself.

    Peter, did you even read the datasheet? It specifically says to connect V+ to the tab in high-current situations. Having had a bad experience with a similarly-limited dual H-bridge driver (UDN2998W), when I ignored the datasheet and connected only to the pin, I know that this is not a specious requirement. If the datasheet says connect to the tab, you connect to the tab, or suffer the consequences.

    -Phil

    Oops, missed your edit. Sorry.
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2014-12-22 19:25
    Phil Pilgrim (PhiPi) wrote: »
    Peter, did you even read the datasheet? It specifically says to connect V+ to the tab in high-current situations. Having had a bad experience with a similarly-limited dual H-bridge driver (UDN2998W), when I ignored the datasheet and connected only to the pin, I know that this is not a specious requirement. If the datasheet says connect to the tab, you connect to the tab, or suffer the consequences.

    -Phil

    Oops, missed your edit. Sorry.


    :) While we have these "discussions" it seems par for the course that we butt heads. Though we may butt heads it's all friendly like, then we sit down and have a nice cold beer :)

    BTW, with the 2998 the ground pin is not central but on devices where the tab is connected to the centre pin then the silicon substrate usually makes contact right there. So I trust the centre pin at least.
  • RDL2004RDL2004 Posts: 2,554
    edited 2014-12-22 20:52
    Just to reiterate, those in-line pin TO-220 style packages are not meant to be mounted free standing on a circuit board. Doing so is not considered "good practice". The part will wiggle back and forth, the solder joints will eventually fail. Since no heat sink is needed here, it would be best to bend the pins and mount it flat to the board, either with a nut and bolt or soldered. If there is no room for this, an alternative might be to bend the center pin out of line to create a tripod arrangement.
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2014-12-22 22:09
    RDL2004 wrote: »
    Just to reiterate, those in-line pin TO-220 style packages are not meant to be mounted free standing on a circuit board. Doing so is not considered "good practice". The part will wiggle back and forth, the solder joints will eventually fail. Since no heat sink is needed here, it would be best to bend the pins and mount it flat to the board, either with a nut and bolt or soldered. If there is no room for this, an alternative might be to bend the center pin out of line to create a tripod arrangement.

    Au contraire mon ami this mounting of standard TO-220 packages vertically without support is fairly standard on a lot of equipment I've come across especially when they are tucked in amongst electrolytic caps etc. But then again most of that stuff tends to be the mass produced single-sided wire linked boards. Personally I don't like it and I will always seek to lay them flat but sometimes it's a cram and has to be done. I've not known a free standing vertical mount TO220 to wiggle back and forth and fail unless there are some careless mitts mangling them. Mind you too, I have supported a bank of TO220 mosfets by stacking them one behind the other so that they are supporting one another too.

    However the part about "it's best to bend the pins and mount it flat" I am in total agreeance* with you.


    * Funny thing and hopefully not to start a long debate, I note that American English (esp Paul Brian, an American) considers this word an error in grammar since it protested so, but not so as it is commonly used in various parts of the world and it is similar to but different from the word agreement which is normally something done bilaterally if you like (the "ment" is an action or result") vs agreeance ("ance" is state or quality of as in governance vs government) which is to state one's view from which one can then come to an agreement, do you agree? :)

    I notice that there are a lot of mangled words in various English "dialects", what is the difference between resistor and resister? An "-or" indicates a thing vs a person. Suffixes used to mean something but over time people have forgotten their use unless of course you read a lot of history and older books.
  • Heater.Heater. Posts: 21,230
    edited 2014-12-23 07:05
    Yes, "agreeance" is an annoyment :)

    One of those ugly and unnecessary mutations of English that makes English speakers around the world wince.

    Why not just say "I totally agree with you"?

    It's two words shorter, or eight letters. Succinct, precise and avoids making up new redundant words.

    Or how about "We are in agreement" Save another word with no loss of meaning.

    Or how about just "Agreed". Which works the same if the context is known.

    Still, the English word "agree" came to us from the French. Perhaps they roll their eyes at how badly we mangle "their" language.

    I put "their" in quotes because the French language gets the word from Latin and it's Indo-European roots.
  • RDL2004RDL2004 Posts: 2,554
    edited 2014-12-23 08:54
    Yes, mounting TO-220 type packages vertically is something you will see done in commercial products - that doesn't make it right, or good. It's certainly possible get away with it for a home-made DIY type of thing, something that goes from the bench to a shelf and no more. However, when something has to be shipped in trucks and planes, vibration will eventually take its toll. It's basically a pendulum after all.

    On a well manufactured product you will almost never see these devices flopping around in the breeze. Most of the time if a device is in this type package it will require a heat sink and that normally takes care of the problem right there. If no heat sink is needed and it's mounted vertically they will at least be anchored with silicone or hot glue, along with any other large, heavy parts such as electrolytic capacitors.
  • Heater.Heater. Posts: 21,230
    edited 2014-12-23 09:16
    Peter Jakacki,
    ...the package limit is dictated by the weakest link and since the current must flow through another pin there is no advantage to directly connecting to the tab itself.
    That is what I thought initially. I wondered why Phil was being so weird about it. Normally such devices have high current in pin and a high current out pin and the tab is connected to something but not required to be in the circuit.

    But in this case the device has two high current pins in one direction and only one in the other. Therefor one has to connect the tab to carry the current of the missing pin on that side of the circuit. Assuming you want to max out the current that is. Very strange.
Sign In or Register to comment.