No connect pins on MLF packages

Go To Last Post
22 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Does anyone know if it is a problem (or better yet has anyone tried it) to connect to the NC pins on a tiny13 in the MLF package? Does NC indicate that the pin is not connected internally within the package (therefore I could connect them to ground without causing harm) or that the pin is connected to something inside the package and must not be connected externally?

I'm planning a project that will use a tiny13, and it's so simple that I'd like to use a single sided PC board. If I can connect to the NC pins safely, I'd route ground from the central pad through one or more of pins 16-20 so that it would be easy to place a decoupling capacitor next to VCC at pin 15. Otherwise, I'd need a double sided board to get a ground trace next to VCC without effectively cutting off access to one side (pins 11-14) of the chip or the other (pins 1-5).

If I could I'd try to use the 10 pin version of the MLF package that is mentioned in the datasheet because this would make single sided PC board routing easier, but I don't have a local source for the 10 pin version of the chip.

Anyone with experience or better knowledge?

Thanks,

Michael

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The NC pins on the tiny44 are OK to connect to GND. So it is likely that it should be OK on the tiny13.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi

If you look on page 2 of the ATTiny13 data sheet it mentioned "NDC".
Which means "DO NOT CONNECT"

I can only see 20 pins for QFN/MLF & 8 pins for PDIP/SOIC.
No where I see a 10 pins version.

Ken

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think for a definitive answer, you have to ask avr [at] atmel com. If the
datasheet mentions them as "do not connect", it's your risk if you connect
them to something anyway. OTOH, I guess they are simply not connected at
all, and are only there since there's simply no smaller MLF package available.
In that case, it's probably OK to tie them to GND but before doing that in a
real project, I'd ask for Atmel's blessing to do so.

Jörg Wunsch

Please don't send me PMs, use email if you want to approach me personally.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

pykedgew wrote:
Hi

If you look on page 2 of the ATTiny13 data sheet it mentioned "NDC".
Which means "DO NOT CONNECT"

I can only see 20 pins for QFN/MLF & 8 pins for PDIP/SOIC.
No where I see a 10 pins version.

Ken

Interesting - I suspect we have different versions of the datasheet. I'm reading version 2535H-AVR-10/07, which says "Not Connect" instead of "Do Not Connect", and includes a pinout for the 10-pin MLF on page 2. There's two notes in the revision history section of the document, so I guess the 10 pin chip was only introduced some time after the datasheet you have.

Michael

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Michael

Now you have me worried.
The one I was looking at was 2535G–AVR–01/07 on my system at home has 'NDC'.
As you have pointed out the 2535G–AVR–10/07 from Atmel site has 'NC'.
Take your pick.

I agree with Jörg that clarification is warranted from Atmel.

Strange things do happen & today was not my day.
But at least you highlighted the issues.

Ken

Edit:2

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Bonding out to a pin is expensive and typically not done unless there is a test need. You can probably ground an unconnected pin (NC), but not a "Do Not Connect" (DNC) pin. The latter label is typicallyt applied to pins that are used in manufacturing test. You can wind up with "interesting" behavior.

As to MLF parts, I would stongly discourage the use of SS boards. It is difficult to impossible to reliably connect the bottom pad of an MLF-20 package to ground, which is recommended by the data sheet. Even if you use the DNC pins as a heat path to the pad, getting a good solder joint is tough.

What happens if you don't connect the pad? Likely, as the connection is recommended instead of required, the bottom of the die is not electrically connected to the pad. You should expect to see higher noise problems in the A/D area, but the chip should still be expected to function.

Go DS and ground the thing the easy way.

BTW, why MLF instead of the SOIC? The size difference btween an SOIC-8 and a 20p MLF isn't a lot. The MLF cuts out a big patch under the part (both sides) so a lot of routable real estate is lost and the advantage of the MLF can be limited.

"It's easier to ask forgiveness than it is to get permission" - Admiral "Amazing" Grace Hopper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Once I saw a 5.000V reference chip (for ADC) that had NC pins (Not connected). I don't remember what the datasheet really said about them (not connected or do not connect), but if the NC pins were connected to GND the chip heated.

So best to ask Atmel if you wish to go solo. I'd just leave them unconnected, as there is no way for me to know.

- Jani

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think the bottom of the MLF package *is* connected to the die.
The recommendation to connect it primarily results out of
mechanical/stability considerations; the electrical connection
of the normal GND pins would be sufficient for that purpose.

Jörg Wunsch

Please don't send me PMs, use email if you want to approach me personally.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Jorg,

Did you check continuity? Many die are glued to the pad, but not with electrically conductive adheasive. I don't have one of the ATTiny13s around to check myself, Though I would not be at all surprised if the connection was Ohmic.

I always solder the pad to ground anyway. It is good practice and, as you noted, the pads on MLF parts aren't particulaly large and don't give you a lot of adheasion to the board.

I know that sometimes you can upset the bias of a chip by not connecting the pad. The bottom of the chip needs to be at ground potential. Such devices have, in my experience, come with a warning that the pad MUST be connected. In this case, Atmel hasn't chimed in. We can only guess.

"It's easier to ask forgiveness than it is to get permission" - Admiral "Amazing" Grace Hopper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you are going to use MLF then please read the following:

http://www.atmel.com/dyn/resourc...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Atmel speaks with forked tongue.

Do you know what this is?

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Microsuffer wrote:

BTW, why MLF instead of the SOIC? The size difference btween an SOIC-8 and a 20p MLF isn't a lot. The MLF cuts out a big patch under the part (both sides) so a lot of routable real estate is lost and the advantage of the MLF can be limited.

In light of the potential complications, I think you're right - space is not so critical that I couldn't just use the SOIC version instead.

This has been an interesting discussion though, so thanks to everyone for the input. When I do need to use an MLF package, I'll be a lot better equipped to pretend I know what I'm doing :).

Michael

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Just find a good toaster oven and you are set to go.

"It's easier to ask forgiveness than it is to get permission" - Admiral "Amazing" Grace Hopper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

It is difficult to impossible to reliably connect the bottom pad of an MLF-20 package to ground, which is recommended by the data sheet.

I have seen discussions where "home users" of MLFs used some good-sized vias under the "post". This not only facilitated mounting, but dismounting as well if needed. (See the link below to the Amkor document--they spend quite a bit discussing the vias) [Hmmm--maybe a big mother hole nearly the diameter of the post?]

Quote:

The recommendation to connect it primarily results out of mechanical/stability considerations;

I thought it was for heat dissipation. lol

So, is it for:
a) grounding
b) mechanical stability
c) heat dissipation
d) all of the above
e) none of the above

Wikipedia leans toward a):

Quote:

...the leadframe's metal die-attach paddle is exposed on the bottom of the package, enabling the use of down-bonds or conductive die-attach material, which in turn allows for a stable, low-impedance ground.

Amkor's app note leans toward c):
https://www.avrfreaks.net/modules...

Quote:
The MLF package is designed to provide a superior thermal performance. This is partly achieved by incorporating an exposed die paddle on the bottom surface of the package. However, in order to take full advantage of this feature, the PCB must have features to effectively conduct heat away from the package. This can be achieved by incorporating thermal pad and thermal vias on the PCB. While thermal pad provides a solderable surface on the top surface of the PCB (to solder the package die paddle on the board), thermal vias are needed to provide a thermal path to inner and/or bottom layers of the PCB to remove the heat.
A quick scan saw nothing to indicate that the pad >>was<< ground, leading me to guess that it may not be in the general case.

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think the center pad is there to cause trouble for the naive. Sparkfun's soldering tutorial says they don't use it and recommend you don't use it either. Atmel doesn't use it, at least on the Butterfly. A good thing too or I wouldn't be able to remove the mega169.

I don't believe the mechanical stability thing. If all those pads around all four sides won't hold the chip in place, then nothing will. There can be thermal stresses with MLF packages but I don't see how soldering the center pad would help that.

I could believe the center pad gives better grounding and cooling, but if that were needed then the AVRs in other packages wouldn't work, and surely by now, someone would have noticed that. :)

If you try to solder the center pad and there is too much solder, it will squeeze out and cause shorts with the other pads. I follow Sparkfun's advice. Make sure there is little or no solder on the center pad before you attempt to solder the chip.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

> If you try to solder the center pad and there is too much solder,
> it will squeeze out and cause shorts with the other pads.

*Who* would press the chip down so it can squeeze out? Surely, not
by the gravitational forces itself. The chip rather “swims” once
all the solder is melted.

Jörg Wunsch

Please don't send me PMs, use email if you want to approach me personally.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

dl8dtl wrote:

*Who* would press the chip down so it can squeeze out? Surely, not
by the gravitational forces itself. The chip rather “swims” once
all the solder is melted.

Well Jörg, I don't know who would do that. By the way, it isn't easy getting those funny looking dots over the o. Did this cause you problems in school? :)

I can think of 3 ways to solder the chip. I'm not familiar with the oven method.

I use the soldering iron drag method. If there is solder on the center pad, that could hold the chip above the board and the pads along the edge might not touch the pads on the board.

Sparkfun uses a hot air gun. They say this:

What happens if you get solder onto the center pad? If the center pad becomes domed in a big way, and you hot-air the IC onto this domed pad, one of two things will happen. More than likely, you will press the IC down. The domed solder will squish to the exterior pads shorting many things to the center pad. Or if you don't press down, the IC will float a couple hundred microns above the PCB leaving all sorts of open connections. Bad news all around. A little solder is not a big deal but try to leave the center pad as flat as possible.

Last Edited: Wed. Dec 19, 2007 - 01:51 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

> it isn't easy getting those funny looking dots over the o.

In the era of the world-wide web, it's fairly easy even for you guys:
just copy&paste it from my signature. :)

> Did this cause you problems in school?

Of course not. Mind you, this is not the US here, and our language uses
the umlauts regularly, so what kind of problems would you expect arising
out of that in school? We also learned Russian in school, which is an
entirely different writing. With chalk on a blackboard: no problem
whatsoever. That's been all there used to be when I attended school...

Back to the center pad: of course, if your solder forms a dome, than you
likely won't have much fun with it. But at least for an industrial
production line, I would expect them to handle that correctly instead.
At least, the electronics house who do our board designs here don't appear
to have the kind of problems sparkfun is describing when they solder the
AT86RF230 chips, and yes, they *do* solder the center pad.

Jörg Wunsch

Please don't send me PMs, use email if you want to approach me personally.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes Jörg, the copy and paste works. Thanks for the tip. I'm a slow learner. I guess you know I was kidding about the umlaut. My sense of humor is a bit strange.

Blackboards were big when I went to school. English was about the only language we learned. Now the government has taken over the schools and apparently we don't learn even that.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

steve17 wrote:
English was about the only language we learned. Now the government has taken over the schools and apparently we don't learn even that.
"In America, they haven't spoken [English] in years." -- Prof. Henry Higgins, "My Fair Lady"

Stu (Yo!) Bell

Engineering seems to boil down to: Cheap. Fast. Good. Choose two. Sometimes choose only one.

Newbie? Be sure to read the thread Newbie? Start here!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

For experimental boards I use the toaster oven method. Excess solder appears as bumps on one or more pins, requiring touch-up with a fine point iron. With a proper land pattern and solder mask setup there is hardly ever a problem. Rework once is generally doable. Twice seldom works well. Unless, of course, the part shorts real good and welds to the copper pads. But those parts are switching supply devices and failure there is generally entertaining.

"It's easier to ask forgiveness than it is to get permission" - Admiral "Amazing" Grace Hopper.