Practical home method to solder SMD with ground pad below?
photomankc
Posts: 943
I'm wanting to make a PCB that uses two L6470 Stepper driver chips in the HTSSOP28 package on a single PCB. The signal routing is easy enough as would be getting the pins soldered but I'm at a loss as to how I would reliably connect the bottom ground pad without burning up the chip in the process. I have not found another dual board that meets my requirements for 3.3V supply and connection for the external OSC input. I could make my own fairly inexpensively but that ground pad has me stumped on how to get it soldered down without some kind of oven.
Anyone here had any experience in doing this? I do have a toaster oven I have used to cook on powder coat for parts before. Not sure that it's temperature stable enough to do PCBs though.
Anyone here had any experience in doing this? I do have a toaster oven I have used to cook on powder coat for parts before. Not sure that it's temperature stable enough to do PCBs though.
Comments
It seems like a a soldered connection would provide better heat transfer so I'd favor Leon's idea over SparkFun's instructions.
Edit: Others who know more about this say one should always solder the center pad. I'd suggest going with what they say.
I think you're gonna have to connect it somehow to keep the chips from overheating.
Use 0.3mm thermal via and if you are not using solder paste (why not?) then use flux and get aggressive with a hot iron with the board upside down and let the solder flow through the vias. A bit of flux on the bottom of the chip before soldering would help too.
You can use my boards if you want, I would just solder a wire to pin 7 and bring it out if you really want the oscillator. I just stack two of these together side by side normally.
In some instances I cut out a strip of brass and fold it up and solder it to bottom of the board for extra heatsinking, but that's really only because I'm running it hard as otherwise it will trip the thermal shutdown when you are constantly accelerating/decelerating fast.
I keep meaning to revive that site, my bad. The thing is I don't really have a need for it either, unless i work out how best to sell and ship components and boards etc. We'll see then.....
I found the source of my "don't solder the center pad" quote. As I mentioned, it was from a SparkFun tutorial. Here's the quote.
Here's a link to the tutorial.
I am not implying the tutorial is correct in "don't worry about it". I'm not sure if any of the parts Nate wasn't worried about produced a lot of heat or not.
Also this is dangerous talk: "Almost all center pads are ground connections in addition to ground connections on exterior pads (so you don't even need the center pad connection)." Bad things can and do happen if ground and power pads are not connected. And it's not all about heat. The Propeller has two power and ground pads, for example, better connect them all else problems with blown PLLs ensue.
It may well be rarely required to solder the centre pad and perhaps it can be ignored for many combinations of chip and application. But that does not mean you should never worry about it.
For example: That tutorial is about the CP2102. If we check the data sheet we see: "Thermal Resistance 32C/W" with the note that "Thermal resistance assumes a multi-layer PCB with any exposed pad soldered to a PCB pad".
Is that a problem? No idea but you can check the expected power consumption of the thing in your application and decide if it's going to cook or not.
But there is a problem if that pad is not soldered down. The thermal resistance is then not known at all!
Yeah, hugely bad advice, he forgot to RTM on this one but then again he probably only used baby motors and never saw a problem so he figured it didn't need it. So in his own words then "this is a common misconception and problem", but "don't worry about it" cuz the magic smoke will find it's way out if it needs too
Still bad advice to make general statements like that. For example I imagine a stepper motor driver could really need all the thermal help it can get.
The L6470 manual says the thermal resistance of the chip is 22C/W with this note:
"HTSSOP28 mounted on the EVAL6470H rev 1.0 board: a four-layer FR4 PCB with a dissipating copper surface of about 40 cm2 on each layer and 15 via holes below the IC."
But hey, that chip will protect itself by shutting down if it gets too hot, nothing to worry about
I use 2oz copper on an otherwise standard DS FR4 and keep the bottom layer solder mask free and mostly just ground plane. Although it certainly does shut itself down, that doesn't help the production line much!
Peter, Thanks for the offer but I'd like to stick to the connectors I have designed as part of my robot. Bonus for me on my own board is that I can match the 10 pin pin-out from my mainboard SPI headers and simplify the connection even more. The ability to slightly tweak OSCIN is kind of central to calibrating my robots wheels to dead-straight when running together.
A little off-topic but this is what is prompting me to try a new design since I always thought these things were too big anyway:
I just learned the hard way, you can thermally kill the chip before the shutdown can react. By using settings for a much smaller motor on a bigger one I induced smoking failure almost immediately. Was driving WAY too much current to the windings on the low-resistance large motor using the settings for that much higher-resistance coil on the small motor. I think I even saw a flash of light momentarily when the holding torque was applied.
Funny you ask. I tried to remove just the chip by cutting off the legs and I was almost successful but two pads ripped up as I cut through the leg and ruined the PCB. The chip was still firmly mounted when all legs were clipped. I had to use hot-air to knock the chip loose from the center pad which was clearly soldered down on this PCB.
And, just for the record, ROHS solder is an abomination.
Now the tricky part.
1: use via's and they drain all the solder paste to the other side, not good.
2: cap the via on the other side with solder mask openings-disable on them, now you get gas-out problem.
3: cap the via's on top, actually the preferred way but having solder mask over a large square pad is not what most pcb design software allows.
but with some manual epad gerber design it should be possible.
P.S use the oven, 4 minute pre-heat and 1minute toast is better than any hand soldering you can do, get stencil from oshstencils
Thermal vias are normally used and absolutely no solder mask at all. My vias are 0.3mm hole size and in actual fact the solder paste has the center pad to adhere to so it's not that inclined to go wandering off.
As for removing these chips you could of course cut the legs with a nice sharp knife while you brace you fingers on the board to prevent punch through on the tracks. The sharp blade helps as you don't need to apply too much pressure. But once you have cut it and it still soldered underneath then I just heat up my iron as hot as it will get and sit it on top of the package while feeding solder onto the tip to create a heated mass. If you have access under the board through the thermal vias then it will work from that side too. This way you are only heating up the bit that you want heated up.
BTW, The loose pins after being cut can be removed using the hot tip and fresh flux solder blob method, they just cling to the solder blob as you run down the row and with the board vertical or slightly tilted downward it becomes an easy matter to lift the tip away when you get to the end pins and the blob goes with it (hopefully onto somewhere safe, and not your leg!). Also, I just use the syringe and run a line of paste down the rows of pins rather than use a stencil which is mainly for big volume runs.
Duane, I wasn't implying you were wrong or promoting that comment, but just putting in my opinion regarding the center bottom pads. As someone who deals with these types of parts on a daily basis, I am quite familiar with the proper ways to handle them. Granted, I deal with them in a high volume manufacturing environment, with processes and equipment not readily available to hobbyists, but many of the practices can be adapted to toaster ovens and kapton film stencils.
Yes, that can be a problem if the vias are oversized or paste is applied over uncapped/unplugged vias. It is common to design a custom aperture for the stencil for the center pad areas. They are much too large to use a single opening, so they are typically "windowed" (reduced in size and split up like panes in an old style window), but when vias are present, the windowing needs to compensate for them. Odd-form shapes can be used to eliminate applying paste over the vias. There will still be flow to the vias in some cases, but starting out clear of the vias makes a very big difference with undesired via flow as well as reducing voiding.
There are lots of ways to reduce via issues with plugging/capping as well, but as you mentioned, outgassing can be an issue depending on the quality of your fabs from your supplier. I have seen some prototype PCBs supplied by customers that were very inexpensive, but in turn very low quality. We saw the proof in the process with various problems associated with vias. On one occasion, I even bet a customer that if I ordered their fabs, their fab quality related design issues would go away without them having to change their design. My fab house came through and we built a batch of boards with no issues using the same process settings from the run with their fabs.
The easiest way I've found by far of removing a bad chip without damaging the board it to use Chipquik. A few years ago Leon suggested Chipquik and when SparkFun started carrying it, I gave it a try. It's amazing stuff and the only surefire way I know of to remove a chip without damaging the PCB.
Thanks for the suggestion Leon.
just my 0000 0010 cents.... (oh, that was lame...)
Rick
Not following..... how would doing that connect the ground/heatsink pad under the chip?
Yeah, it's an old trick, ....... and he fell for it