Mega8 MLF

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

I was searching for a physically small AVR (few I/O) with lots of SRAM and EEPROM, and the perfect match is of course the Mega8 in a MLF package. That thing looks sweet!
It looks to me like it can be hand soldered quite easy, I've done quite a few 0.5mm QFP packages with no problem.
MLF Should be no problem, right?

However, I stumbled across one problem... I can't find a distributor that has it!
Any suggestions?

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

Geez, this 32pin MLF package is sweet!
I compared the size with a small 2313 AVR 20pin SO package, the MLF is one fourth the size! :)

I wish I could find a place that can sell me a few of them! :)

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

Whats the spacing and size of the pins? not sweet.....

admin's test signature
 

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

Check the datasheet, the solderpads are underneth the package, (like a BGA) but along the edge of it, so it should be quite possible to solder it by hand.

The pitch (c-c distance between two pads) is 0.5mm. I've soldered 144pin 0.5mm pitch QFP pacakges, this shouldn't be a problem.

Do you know where I can find a few of theese sweet chips?

16MHz AVR with 1kByte SRAM, 512Byte EEPROM, lots of flash, 32 pin package, and much more. All in a package smaller than my tiny finger nail! :)

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

I've been looking for these for 2 months and nobody has them (in the States). If you call up All-American or Pioneer they may have new information. They didn't know a couple of weeks ago.

The DIP version is finally back in stock at most places now.

admin's test signature
 

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

I have no use for large DIL packages...
Do Atmel send samples on request? (To a electronics company in sweden)

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

atmel is supporting three micro lead frame packages ( MLF) , the little 5x5 and also a 7x7 and a 9x9 .. the mega128 is available in the 9x9 ... that means the mega128 package area is only about 1/3 that of the tqfp ... mega sweet ! .. the 128 is being shipped like that and I have been promised samples are in transit but i have not been able to find any of these in the distribution chain ... only from the factory ...

admin's test signature
 

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

AVR MLF == Wapourware?
The Mega series in MLF packages was "released" a year ago... And there still is no trace of them on the market.

Has anyone actually seen a AVR in a MLF package??

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

As far as I can tell, the socket for a 32 pin MLF costs more than an entire PCB with parts all built up an assembled. Even if you could find a socket, it would be as large as using a 28pin DIP anyway.
Thus there isn't any demand for MLF's.
Everybody is just soldering down the TQFP devices onto the boards and going with it

Atmel probably has production facilities to make MLF's but nobody ever ordered any. Plus you'd have to order thousands before they'd make them anyway. Then you'd have to get someone to make thousands of sockets for them. But the sockets are too expensive. Nowadays you have to make everything really inexpensive so every penny you save is important.
When I looked at Digikey and some other places, they didn't have any Atmel chips of any type in MLF formfactor that I could see.

Granted we may not need sockets per se, but the intention for MLF's was to use a socket.

It is doubtfull they'll ever make them, as if you need a socketed chip you could use the simple DIP-28 package as those sockets are pretty cheap in comparison.

The TQFP 32 pin package works well for me, I just got done making up some under 1 square inch CPU boards with the AtMega 8's and they worked very well.
I could have made it smaller, but I had to allow for a Crystal oscillator circuit and a little voltage regulator too, along with a couple of tiny smt 603 size LED's.

Although the MLF is really sweet at being even smaller, trying to solder those pins under the chip down to the board may be a real problem as the pins are recessed slightly under the chip. I do agree that it would be nice to try some though.

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

Where do you get the idea that MLFs were intended for sockets? Very few SMT parts are intended to be socketed, and most sockets manufactured for SMT components are for prototyping. The one possible exception are PLCC PAL/GALs and FLASH parts. Those are the most commonly socketted SMT parts. It's my contention that the socket for MLF is nothing more than a debugging tool, and never was intended for mass production. It just makes no sense.

--John

admin's test signature
 

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

Earl: I don't want no sockets! I want a slim MCU. And as John says, the MLF was never intended for sockets, I highly doubpt that MLF sockets even exists!
I need a _small_ MCU with lots of RAM & EEPROM, the Mega8 in a MLF package is perfect!
You say there is no demand for MLF packages.. Right... Today everyone want everything to be as small as possible. It saves cots, and results in slimmer products.
I'd say a more reasonable explanation is that the big boys on the market buys every MLF package Atmel can produce.

About the TQFP32 package... it's like four times larger than the MLF package! Totally out of the question! ;)

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

I agree with what you guys say.
But in looking at some old 80286 and 80386 chips I had laying around in a junk box, they sure as heck look like MLF packages. And they are quite small in comparison to the "other" package sizes.
And the boards that they are on have these weird little/big sockets, where you put the CPU down on it and then balance a square metal clip and either put in a couple of screws or press down a couple of clips to hold the metal square down on the chip, clamping it down on the socket. It was a production socket too. I forget what they used to call that package, but it sure as heck looks like the MLF's as they call them today. The raised square on the bottom of the chip allowed it to line up in the socket before you clamp it down.

Here is an example of the socket as I know it:
http://micromouse.cannock.ac.uk/...
http://micromouse.cannock.ac.uk/...
It looks like he didn't want anyone to know what cpu he was using. But it does look like a Intel chip.

I haven't figured out how to get rid of this old junk, nowadays it's considered low level toxic waste, so it sits in the corner of the shed in a old box.

I apologize for showing my age.
And if I'm wrong I apologize for that too. But the chips sure as heck look like the same package format to me.

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

I think the reason that particular part is socketted is because it may be an EPROM part, instead of FLASH. As such, to be put in the UV eraser, it has to be removable. It may be they're using the tape to be secretive about the processor used, since I can't find any indication of what it is after a quick glance over the pages. But I really suspect it's because it's an EPROM type device.

In the old days, the 80286 and 80386 were available in a ceramic package with no pins, known as a LCC, or Leadless Chip Carrier. That packaging option only applies to ceramic packages, to the best of my knowledge. In comparison to the MLF package, which I believe means Micro Leadless Form (if I believe who I talked to, who is right as often as wrong), the LLC pin spacing is huge. It's less than 100 mils, but not by much. Whereas the MLF is much much finer.

I guess you could say they have some common conceptual ancestry, but the LCC isn't considered an MLF package.

Oh, the reason the ceramic was usually used is better heat dissipation, which was a problem for the NMOS parts. It's damn expensive to put pins on a ceramic package, since they're usually brazed, and it's an extra manufacturing step. Back then, SMT was a lot more expensive process, and fallout was higher, so a socket was almost a requirement.

--John

admin's test signature
 

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

I've got ONE sample on my desk, nice 5*5 mm and as my distributor says, it should be available in autumn in production quantities. My part has got a marking ES on it, seems to be engineering sample. There is a large copper-field in the center of the footprint, I don't know if it should be connected or just for "heat dissipation". We should be able to put it our IR-soldering like the tqfp's.
I thought of vapour-ware also until I got this example and three tqfp's last week with real date-code on it.
Now we'll start changing our 4433-design to the mlf-chip.

Andreas

admin's test signature
 

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

kemper: The Mega8 MLF was "released" A year ago... and still they have only released a few engineering samples? Let's hope your distributor is correct.. Who is it btw?

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

Thanks for the info, I guess my brain's full, as I learn new things the old stuff is getting erased.

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

At the Atmel seminar they talked about the MLF packages, and that big pad in the middle is supposed to be secured to the PCB somehow, they said glue or (preferably) soldered... The reason is that because it's such a small package, small expansions in the PCB caused by heat tends to make the MLF's pop off.. So they said unless your application is guaranteed not to be put under ANY kind of stress from heat, you must secure that big pad.

admin's test signature
 

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

pepsi: Oh, sounds reasonable... An idea... How about putting a groundplane with a few vias in the middle, and then when you mount the MLF you put solderpaste under the package and heat the vias from the back of the PCB in order to melt the solderpaste... Should work, don't you think?

Did they say when they'll ship some MLF packages?

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

Hi hello,

THE MLFs are nice to work with, we have hand soldered a few and it was not a problem. Basicaly you put some solder on the PCB and put it in an owen (or heat stream). Works very nice. Atmel produces them in large quanteties, we have had mega8 in MLF from the first wafer they produced, begining of March. They have worked without any larger suprices so I think they will be very successfull.

/Jonas

admin's test signature
 

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

it wasn't clear what chips will be offered in MLF form, but they said that because they are beginning to shrink all of their new designs down, they can start to use MLF, so it sounded like most new parts will be offered in MLF form. They didn't give the availability of any MLF parts, so I'm not sure when you can expect to be able to get them in low volume.

admin's test signature
 

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

i take that back (about it being unclear which chips will be offered in MLF)

there were 2 slides that showed each device and the packages it will be offered in...

all of these will be (are?) offered in MLF form...
tiny26, tiny28, mega8, mega8515, mega8535, mega162, mega16, mega169, mega323, mega32, mega64, mega103, mega128

it wasn't a comprehensive list of devices, some werent in those slides, and most of the tiny/at90s series didn't specify MLF

they stressed that this is a new technology for them, so I wouldn't expect to find low volume MLFs for a little while still

admin's test signature
 

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

pepsi: I think 32pins is the smallest practically possible MLF package, so tiny's in MLF packages sounds kinda strange (?)
It would be nice if the could get them out on the market... The 0.8mm pitch TQFP packages are _huge_ compared to most other circuits today. 0.5mm pitch QFP packages would be okay (too bad no AVR's are offered as 0.5mm QFP's), but MLF looks even better.

Jonas: I think it should be quite easy to solder them using just a iron and some solderpaste, don't you think?

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

AmPz,
you're right... a tiny part in MLF form would be kind of strange, but from what i understand from the seminar is that the process used to make the parts is what determines eligibility for MLF... because there is less surface area to dissipate heat, they need to use more advanced production processes, and they didnt sound like they had plans to make older parts with the new process

you're also right about the 32 pin minimum.. even the 20 pin tiny26 uses 32 pins when 24 would have worked

smaller MLF's would be EXTREMELY sweet, but they are definately very nice even with the 32 pin minimum

admin's test signature
 

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

AmPz, the MLF goes much smaller than 32 pins but just not for AVR at this time .. how about an 11 pin 3mmx3mm processor, checkout www.cygnal.com to see pictures, in fact I have one of these little guys here on my desk unless it has gotten lost again ! althought I hate the 51 architecture I have one application where I may use it just for the size ... what i really want is a low power attiny in one of these 3x3 packages ...

admin's test signature
 

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

I have used the micro-BGA parts, but we use Cygnals in one of our products. Except for being expensive, I can only say good things about it. Personally, I like the 8051 architecture. Beats the heck out of a PIC, but it doesn't compare to some others (the MSP430 is really growing on me).

I will say that Cygnals have one of the best A/D sections you can ask for. Cygnal was starting by a handful of analog guys from Maxim, and it shows. The core was something they subcontracted, then bought the company. It's well thought out, with the only flaw being the switching matrix means you *really* have to plan ahead on your pin allocation. That's the only bugaboo, is that if you want to go back later and turn something on, you might find the pins have already been "cascaded". But it's still a nice part.

--John

admin's test signature
 

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

Hi,

The package manufacturer's recommendations for soldering MLF packages can be found at
http://www.amkor.com/products/no...

Doing a search for "MLF" at http://www.amkor.com/products/
will show links to more documents. Here you also find a list of all the other possible MLF packages, too.

The 5x5-32pin, 7x7-44pin and 9x9-64pin versions are the ones we are using for AVR products.

The same package can be named LPCC or LLP from other manufacturers, but we will keep "MLF" as the name for the AVRs to avoid too much confusion.

The complete JEDEC spec is found at
http://www.jedec.org/download/se....
Here the package is called "HP-VFQFP-N"....

The soldering process for this package is closer to the process for BGA than for TQFP. There are no flexible leads on this package, so reliability will be low if the pad in the middle is not soldered properly.

(This was the great disadvantage of the LCC packages mentioned earlier in this thread: If they were soldered to ordinary PCBs, they had a tendency to jump off the board whenever there was a change in temperature. Just dropping the board on the table could sometimes make the packages jump off. The LCC packages had to be mounted in sockets or on special boards with the same temperature coefficient as the package.
The large pad under the MLF package takes all the stress and removes this problem.)

As for all new products, you will probably not find them in the distributors' inventory. Counting all voltage, speed and package options, there are more than 100 different versions of the AVR available now, and the distis will usually not stock them all. But if you order the device from a distributor, they will order it from Atmel if they do not already have orders in the system.

If you are able to hand solder 0.5mm pitch TQFPs, you should be able to handle the MLFs, at least on boards with solder masks. The pad may be tricky to solder, though, but you can use glue instead.

Regards,

Bård M. Pedersen

Atmel AVR Technical Marketing

admin's test signature