AVR-Chat

I am going to try a project using the SMT version of the Atmega16.  This
will be my first SMT project.

I must admit, I’m a bit nervous at the thought of not being able to pull the
processor out of a socket to replace it – not that I’ve ever had to in the
past, but at least I knew the option was there  :-)

I note that, in the datasheet - near the top where it shows the pin
assignments, it shows the TQFP (gull wing) package along with "NOTE: Bottom
pad should be soldered to ground." with an arrow to a large pad on the
bottom of the package.  This note does not appear near the end of the
datasheet where they show the package dimensions.

What is the reason for this pad to Gnd - thermal stability?

In a home environment how does one solder this pad - provide a hole under
the part and solder from the back side?  Would thermal/conducting grease
work rather than soldering?

My application does not use a lot of output drive so heat dissipation should
not be a consideration ...


Another issue:  In the past I have only used commercial prototyping boards
with the processor already in place.  I had never noticed that, in the
through-hole part, there are two "GND" pins.  The TQFP package has three VCC
and three GND pins.

I assume that good design requires that all of these pins be connected
(i.e.: connecting one won't work)?  Do all the VCC pins require their own
decoupling or will one do (and if so, which one?) ?

I'm looking forward to designing a board with 603 size resistors that are
WAY smaller than my usual 3/8" long beasts.  I'll be able to get a whole lot
more in the board area provided by the free version of Eagle PCB  :-)

Thanks in advance for the help ...
 
Cheers,

Chuck Hackett
"Good judgment comes from experience, experience comes from bad judgment"
7.5" gauge Union Pacific Northern (4-8-4) 844
http://www.whitetrout.net/Chuck

Chuck Hackett wrote:
> I am going to try a project using the SMT version of the Atmega16.  This
> will be my first SMT project.

Welcome to the world of SMT.  I think I'm a little older than you and 
enjoy using surface mount parts -- although I would hesitate using 0603 
size as my current limit is around 0805.

> I must admit, I’m a bit nervous at the thought of not being able to pull the
> processor out of a socket to replace it – not that I’ve ever had to in the
> past, but at least I knew the option was there  :-)

If you suspect a bad part with this type of mounting, it is often that 
it is best to sacrifice the mounted part by cutting the leads at the 
package edge and desolder the remnants in order to avoid ruining the 
PCB.  I've never had to do it.

> I note that, in the datasheet - near the top where it shows the pin
> assignments, it shows the TQFP (gull wing) package along with "NOTE: Bottom
> pad should be soldered to ground." with an arrow to a large pad on the
> bottom of the package.  This note does not appear near the end of the
> datasheet where they show the package dimensions.
> 
> What is the reason for this pad to Gnd - thermal stability?

Yes.

> In a home environment how does one solder this pad - provide a hole under
> the part and solder from the back side?  Would thermal/conducting grease
> work rather than soldering?

Yes to hole, no to grease.

> Another issue:  In the past I have only used commercial prototyping boards
> with the processor already in place.  I had never noticed that, in the
> through-hole part, there are two "GND" pins.  The TQFP package has three VCC
> and three GND pins.

Correct.

> I assume that good design requires that all of these pins be connected
> (i.e.: connecting one won't work)?  Do all the VCC pins require their own
> decoupling or will one do (and if so, which one?) ?

YES! Definitely connect ALL Vcc and Gnd pins!!!  And yes, each Vcc pin 
should have its own decoupling.  Side note: I recall a customer's 
engineer who ran Vcc to one of two package pins and then connected a lot 
of the other board Vcc circuitry to the other package pin.  Not only is 
this horrendous practice, but depending on bonding wire size and current 
draw, it could act as a very expensive fuse.

> I'm looking forward to designing a board with 603 size resistors that are

As I said, that size is uncomfortable for me although I could probably 
do it if necessary -- it would not not be enjoyable.  One more thing 
I've found is in laying out boards and etching them myself (double 
sided), I can etch them fine, but when it comes to drilling the small 
holes with precision, it is beyond my equipment and age/eye/steadiness 
capability, so I've gone to services like SparkFun and their PCB for 
single boards.

Regards,
Don Jackson
http://ae5k.us http://hamavr.ae5k.us http://4smicrowave.ae5k.us

Hey Chuck,

 

If you are going to get into SMD work more than just this once, I would
recommend getting a hot air rework station for do any repair work. I bought
one of these from a Chinese company,

 

http://www.aoyue.com/en/ArticleShow.asp?ArticleID=329

 

It was about $240 at the time. I had a rather expensive PCB made up and I
could not get the AT90CAN128 to programme. All connections where checked and
the other PCB was working so I suspected a damaged device. Using the hot air
pencil I was able to safely remove the device without damage to the board.
Some cleaning with solder wick and re-soldering a new device onto the board
and I was back working so for me this purchase easily paid for itself. I
have since used to 2 or 3 times more. The hot air pencil is also great for
heat shrink installation.!!

 

Another purchase I made from China was a re-flow oven and a solder paste
dispenser. All in the 2 off them cost me around $700 with duty and VAT etc.
Doing SMD PCB's is now much quicker and with some PCB shops offering free or
cheap stencils I am considering this method for past application.

 

As Don quite rightly pointed out, make sure you have a decoupling cap next
to each GND and VCC pin. My friend used to make boards without these and
always wondered why the thing kept failing at odd times. Once he started
using decoupling caps his designs where rock solid.

 

Good luck with the 0603 sizes. They are doable by hand and if you have a
magnifying lamp they are even easier. You'll need a really fine tip
soldering iron and see if you can find the finest solder you can get as this
makes it a lot easier, otherwise you end up with too much solder and it
looks poor (although it will work).

 

 

Good luck,

Dave.

PS. I like the UP484 on your website.! Nice toy.

From: AVR-Chat@yahoogroups.com [mailto:AVR-Chat@yahoogroups.com] On Behalf
Of Chuck Hackett
Sent: 24 January 2010 09:59
To: AVR-Chat
Subject: [AVR-Chat] Soldering SMT Atmega16

 

  

I am going to try a project using the SMT version of the Atmega16. This
will be my first SMT project.

I must admit, I'm a bit nervous at the thought of not being able to pull the
processor out of a socket to replace it - not that I've ever had to in the
past, but at least I knew the option was there :-)

I note that, in the datasheet - near the top where it shows the pin
assignments, it shows the TQFP (gull wing) package along with "NOTE: Bottom
pad should be soldered to ground." with an arrow to a large pad on the
bottom of the package. This note does not appear near the end of the
datasheet where they show the package dimensions.

What is the reason for this pad to Gnd - thermal stability?

In a home environment how does one solder this pad - provide a hole under
the part and solder from the back side? Would thermal/conducting grease
work rather than soldering?

My application does not use a lot of output drive so heat dissipation should
not be a consideration ...

Another issue: In the past I have only used commercial prototyping boards
with the processor already in place. I had never noticed that, in the
through-hole part, there are two "GND" pins. The TQFP package has three VCC
and three GND pins.

I assume that good design requires that all of these pins be connected
(i.e.: connecting one won't work)? Do all the VCC pins require their own
decoupling or will one do (and if so, which one?) ?

I'm looking forward to designing a board with 603 size resistors that are
WAY smaller than my usual 3/8" long beasts. I'll be able to get a whole lot
more in the board area provided by the free version of Eagle PCB :-)

Thanks in advance for the help ...
 
Cheers,

Chuck Hackett
"Good judgment comes from experience, experience comes from bad judgment"
7.5" gauge Union Pacific Northern (4-8-4) 844
http://www.whitetrout.net/Chuck



[Non-text portions of this message have been removed]

At 01:59 PM 24/01/2010, you wrote:
>SMT version of the Atmega16.
>.
>.
>I note that, in the datasheet - near the top where it shows the pin
>assignments, it shows the TQFP (gull wing) package along with "NOTE: Bottom
>pad should be soldered to ground." with an arrow to a large pad on the
>bottom of the package.

Huh? The TQFP is just plain old PLASTIC. What version of the data sheet
do you have? I have used well over 1,000 TQFPs AVRS and never seen
the pad.

Are you using the QFN package? That's the one with the pad IIRC, not nice.

Anyway it would be nicer to switch to the M164 instead if you could.

Regards

John Samperi

********************************************************
Ampertronics Pty. Ltd.
11 Brokenwood Place Baulkham Hills, NSW 2153 AUSTRALIA
Tel. (02) 9674-6495       Fax (02) 9674-8745
Website  http://www.ampertronics.com.au
*Electronic Design * Custom Products * Contract Assembly
********************************************************

Chuck,

The TQFP (and any other TQFP as far as I know) version of that chip does not
have a bottom pad. That note is there because the same pinout is used for
both the TQFP and MLF/QFN version of the chip, and the latter does have and
require the bottom pad. That is why it is not mentioned at the bottom in the
package description, but only at the top pinout diagram.

One thing to note for future reference: If you use Eagle, many of the
standard MLF/QFN packages do not include the bottom pad. That may work fine
for hobbyist designs, but it will definitely be a potential trouble spot. I
usually avoid QFNs, but if you have to use one, be sure the package includes
the center pad. If not, make your own package.

On Sat, Jan 23, 2010 at 6:59 PM, Chuck Hackett <egroupscdh@up844.us> wrote:

> I am going to try a project using the SMT version of the Atmega16.  This
> will be my first SMT project.
>
> I must admit, I�m a bit nervous at the thought of not being able to pull
> the
> processor out of a socket to replace it � not that I�ve ever had to in the
> past, but at least I knew the option was there  :-)
>
> I note that, in the datasheet - near the top where it shows the pin
> assignments, it shows the TQFP (gull wing) package along with "NOTE: Bottom
> pad should be soldered to ground." with an arrow to a large pad on the
> bottom of the package.  This note does not appear near the end of the
> datasheet where they show the package dimensions.
>
> What is the reason for this pad to Gnd - thermal stability?
>
> In a home environment how does one solder this pad - provide a hole under
> the part and solder from the back side?  Would thermal/conducting grease
> work rather than soldering?
>
> My application does not use a lot of output drive so heat dissipation
> should
> not be a consideration ...
>
>
> Another issue:  In the past I have only used commercial prototyping boards
> with the processor already in place.  I had never noticed that, in the
> through-hole part, there are two "GND" pins.  The TQFP package has three
> VCC
> and three GND pins.
>
> I assume that good design requires that all of these pins be connected
> (i.e.: connecting one won't work)?  Do all the VCC pins require their own
> decoupling or will one do (and if so, which one?) ?
>
> I'm looking forward to designing a board with 603 size resistors that are
> WAY smaller than my usual 3/8" long beasts.  I'll be able to get a whole
> lot
> more in the board area provided by the free version of Eagle PCB  :-)
>
> Thanks in advance for the help ...
>
> Cheers,
>
> Chuck Hackett
> "Good judgment comes from experience, experience comes from bad judgment"
> 7.5" gauge Union Pacific Northern (4-8-4) 844
> http://www.whitetrout.net/Chuck
>
>
>
>
>
> ------------------------------------
>
> Yahoo! Groups Links
>
>
>
>


[Non-text portions of this message have been removed]

Hello Chuck,

I agree with John the m16 TQFP shown in the Atmel specification doesn't have ground pad ... I believe you may have been looking at the MLF version of the m16. But it should not be feared as with the right (and inexpensive) equipment with practice/equipment you can put on or remove MLFs or DFNs or whatever with ease. 0602 are also easy and no problem.

But if you are going to begin to do SMT soldering you have some equipment requirements and some learning required. Neither are expensive or difficult, but you will have to spend around $400+ US to get basic equipment dependent on how far you want to go in making boards. If you only are going to make a few, you will need couple hundred $ for a hot air rework unit, <$100 for a temperature controlled esd safe soldering iron, assorted proper tips, proper solder, proper flux, good tweezers to place the part, a hold down device to secure the part before soldering and a headband magnifier. A small "hoof tip" soldering tip is definitely required. It takes a little practice but will make it easy to put on fine pitch TQFP, SO and the like. The right solder and flux are also necessary for good results.

If you are going to make many like 2 or more at a time, another few hundred $ for a proper fry pan and convection oven in which you can do small batches of boards dependent primarily on the size of the boards. To do many, you will need the artwork and get stencils and solder paste. And the cost rises a little more but still significantly less expensive that hiring the board assembly out. And it is sort of fun to. But you will soon tire of doing one at a time, resistor by resistor and long to do the entire board in one cooking session. A shortfall of the convection oven and fry pan is knowing how to set them properly, sounds easy - isn't. I can give you my reflow data and settings as your starting point but "your mileage may vary". I've had very good results.

I also suggest that "YouTube" will help you learn what is involved. Search "smt fine pitch soldering" or similar and watch the videos on how it is done. Very instructive and will give you a strong background.

I've been brief here as this is a complex and wide subject. Feel free to contact me direct to spare the forum (unless there is great interest in this topic) if you wish more details. Good luck and I know you can do SMT soldering successfully.

Jim

  ----- Original Message ----- 
  From: Chuck Hackett 
  To: AVR-Chat 
  Sent: Saturday, January 23, 2010 6:59 PM
  Subject: [AVR-Chat] Soldering SMT Atmega16


    
  I am going to try a project using the SMT version of the Atmega16. This
  will be my first SMT project.

  I must admit, I'm a bit nervous at the thought of not being able to pull the
  <<< snip >>>


[Non-text portions of this message have been removed]

>good tweezers to place the part
I like to use a manual vacuum pickup tool for placing ICs with pins on four sides.  They're relatively inexpensive and work well.  The link below is but one example.
http://www.tecratools.com/product973.html

>a hold down device to secure the part before soldering
For hand placement/soldering, I've found solder paste an invaluable aid.  With it you can put a small dot of paste on two or four corner pads of an IC and then place the part.  After adjusting the position, a quick touch with the soldering iron on each of the pads with solder paste will tack the IC in place.  I've begun using soldering paste to aid in soldering some through-hole parts, too, because it eliminates the need for a third hand to hold the solder.

The solder paste that I've used is Digi-Key p/n KE1512-ND.  I see that they now carry a smaller tube from a different manufacturer, p/n SMD4300AX10-ND.  I use the water soluble but you may prefer no-clean.

>and a headband magnifier.
If you can afford it, a stereo zoom boom microscope is a great aid.  I find it much more convenient to use than the headband type magnifier but the latter is much less expensive.

>A shortfall of the convection oven and fry pan is knowing
>how to set them properly, sounds easy - isn't.
I've had good results on small batches/single boards using a home-made reflow oven built using a small convection oven and a microcontroller to implement the temperature profile.  You can find several articles on the subject on the Internet.  A brief description of mine can be found here:
http://www.zbasic.net/forum/about368.html

Don Kinzer
ZBasic Microcontrollers
http://www.zbasic.net

On Sat, Jan 23, 2010 at 9:59 PM, Chuck Hackett <egroupscdh@up844.us> wrote:

>
>   I note that, in the datasheet - near the top where it shows the pin
> assignments, it shows the TQFP (gull wing) package along with "NOTE: Bottom
> pad should be soldered to ground." with an arrow to a large pad on the
> bottom of the package. This note does not appear near the end of the
> datasheet where they show the package dimensions.
>
> What is the reason for this pad to Gnd - thermal stability?
>

The TQFP has no pad.  The MLF/QFN has pad, and it is for mechanical
stability in this case.
The solder on the pins alone will not hold the QFN to the board if the board
is dropped or vibrated.



>
> In a home environment how does one solder this pad - provide a hole under
> the part and solder from the back side? Would thermal/conducting grease
> work rather than soldering?
>

For a prototype or one-off project a hole under the part is fine for hand
soldering.
In a production environment you do not want to have that as it can theave
solder
into the hole away from the chip.

No grease, but you do want a *lot* of good flux.  A flux pen is a good first
step.



>
> My application does not use a lot of output drive so heat dissipation
> should
> not be a consideration ...
>
> Another issue: In the past I have only used commercial prototyping boards
> with the processor already in place. I had never noticed that, in the
> through-hole part, there are two "GND" pins. The TQFP package has three VCC
> and three GND pins.
>
> I assume that good design requires that all of these pins be connected
> (i.e.: connecting one won't work)? Do all the VCC pins require their own
> decoupling or will one do (and if so, which one?) ?
>

Unless the pin is marked "do not connect" or "not connected" always connect
the pin to something appropriate.
Just because you don't use the ADC does not mean you can not leave AVcc
unconnected.
 Unused I/O pins do not need to be connected, but they do need to have
pull-ups enabled, or set them to outputs.



>
> I'm looking forward to designing a board with 603 size resistors that are
> WAY smaller than my usual 3/8" long beasts.
>

The consensus on the gEDA/PCB list is that 0402's can are the smallest that
can be done by hand with practice, and good magnification.
01005 parts are hard to even see with the naked-eye at all. :-)


> I'll be able to get a whole lot
> more in the board area provided by the free version of Eagle PCB :-)
>
If you are not stuck on Windows give gEDA/PCB a shot, it has no such
arbitrary limits.
http://www.gpleda.org/  The Windows version is still a work in process...


-- 
http://www.wearablesmartsensors.com/
http://www.softwaresafety.net/
http://www.designer-iii.com/
http://www.unusualresearch.com/


[Non-text portions of this message have been removed]

For hand assembly, you'll probably find that there is already enough solder 
on the board (especially if its a standard tin/lead board) that all you'll 
need is flux.  Douse the area with solder flux (we like the flux pens); 
position the part and gently run the iron's tip over the pins.  Just be sure 
to wash the board if you're using water soluble solder!

Good luck,

Tim Gilbert
JEM Innovation Inc.
303-926-9053 (office)
303-437-4342 (cell)
720-890-8582 (fax)
www.jeminnovation.com
www.pdksolutions.com


----- Original Message ----- 

From: "Don AE5K" <don@ae5k.us>
To: <AVR-Chat@yahoogroups.com>
Sent: Saturday, January 23, 2010 9:25 PM
Subject: Re: [AVR-Chat] Soldering SMT Atmega16


> Chuck Hackett wrote:
>> I am going to try a project using the SMT version of the Atmega16.  This
>> will be my first SMT project.
>
> Welcome to the world of SMT.  I think I'm a little older than you and
> enjoy using surface mount parts -- although I would hesitate using 0603
> size as my current limit is around 0805.
>
>> I must admit, I’m a bit nervous at the thought of not being able to pull 
>> the
>> processor out of a socket to replace it – not that I’ve ever had to in 
>> the
>> past, but at least I knew the option was there  :-)
>
> If you suspect a bad part with this type of mounting, it is often that
> it is best to sacrifice the mounted part by cutting the leads at the
> package edge and desolder the remnants in order to avoid ruining the
> PCB.  I've never had to do it.
>
>> I note that, in the datasheet - near the top where it shows the pin
>> assignments, it shows the TQFP (gull wing) package along with "NOTE: 
>> Bottom
>> pad should be soldered to ground." with an arrow to a large pad on the
>> bottom of the package.  This note does not appear near the end of the
>> datasheet where they show the package dimensions.
>>
>> What is the reason for this pad to Gnd - thermal stability?
>
> Yes.
>
>> In a home environment how does one solder this pad - provide a hole under
>> the part and solder from the back side?  Would thermal/conducting grease
>> work rather than soldering?
>
> Yes to hole, no to grease.
>
>> Another issue:  In the past I have only used commercial prototyping 
>> boards
>> with the processor already in place.  I had never noticed that, in the
>> through-hole part, there are two "GND" pins.  The TQFP package has three 
>> VCC
>> and three GND pins.
>
> Correct.
>
>> I assume that good design requires that all of these pins be connected
>> (i.e.: connecting one won't work)?  Do all the VCC pins require their own
>> decoupling or will one do (and if so, which one?) ?
>
> YES! Definitely connect ALL Vcc and Gnd pins!!!  And yes, each Vcc pin
> should have its own decoupling.  Side note: I recall a customer's
> engineer who ran Vcc to one of two package pins and then connected a lot
> of the other board Vcc circuitry to the other package pin.  Not only is
> this horrendous practice, but depending on bonding wire size and current
> draw, it could act as a very expensive fuse.
>
>> I'm looking forward to designing a board with 603 size resistors that are
>
> As I said, that size is uncomfortable for me although I could probably
> do it if necessary -- it would not not be enjoyable.  One more thing
> I've found is in laying out boards and etching them myself (double
> sided), I can etch them fine, but when it comes to drilling the small
> holes with precision, it is beyond my equipment and age/eye/steadiness
> capability, so I've gone to services like SparkFun and their PCB for
> single boards.
>
> Regards,
> Don Jackson
> http://ae5k.us http://hamavr.ae5k.us http://4smicrowave.ae5k.us
>
>
> ------------------------------------
>
> Yahoo! Groups Links
>
>
>
>

On 24/01/2010 20:58, tim gilbert wrote:
> For hand assembly, you'll probably find that there is already enough solder
> on the board (especially if its a standard tin/lead board) that all you'll
> need is flux.  Douse the area with solder flux (we like the flux pens);
> position the part and gently run the iron's tip over the pins.  Just be sure
> to wash the board if you're using water soluble solder!

I use a mini-hoof cartridge on my Metcal system for drag-soldering. It 
holds just the right amount of solder.

Leon