Atmega32U4 Availability Problems?

[JohnSL]

I have a project I'm working on that needs an Atmel Atmega32U4. When I looked a couple of weeks ago, it seems like just about everywhere had zero stock and were projecting to get more this past week.

Checking today I see that Digikey has some of the QFN parts, but none available in other packages. Mouser doesn't have any, and they're projecting more in January.

Does anyone have any insight into why this particular microcontroller is in such short supply? I'm wondering if it's realistic to create a product based on this chip given that getting a reliable supply looks like it might be a problem.

Thanks,
  -- John

[V_King]

because everybody wants to be arduino compatible with their little gadget and atmega32u4 cuts the bom. go with xmega improved chips over atmega for lower price 

[dannyf]

The tight supply could be due to small demand; or over demand; or poor scheduling; etc.

Sourcing has been an issue for atmel parts over the last few years.

[LukeW]

Or just use the QFN one?

[Jeroen3]

If I recall correctly Atmel itself doesn't own a fab. That means that when they produce an amount, it is the stock they have to deal with for the next quarter/year.
If demands suddenly rise, you've got problems.
This also is one of the reasons Atmel parts are relatively expensive compared to their similar competitors.

[Jianan Li]

Just use the QFN package, and get yourself a stencil at OSH Stencils. It worked perfectly for my last project.

[JohnSL]

Thanks for all the info. My current design uses the QFN part. I don't want to order chips until I know what the demand will be, and I am concerned that they will be out of stock for some months once we are ready to order parts-- this will be for a Kickstarter campaign.

Now that I know more about how Atmel manufacturing works, I think we can work this into the schedule.

Using an Atxmega is also really intriguing, and I'll have to look into this when I get home on Thursday. I brought my brand new laptop with me, a Surface Pro 3, but left the charger at home. Woops. So I'm replying from my phone.

[dannyf]

Using an Atxmega is also really intriguing

I would think hard before investing in the Xmegas - I don't think there is a future for those chips.

[JohnSL]

I would think hard before investing in the Xmegas - I don't think there is a future for those chips.

Could you elaborate more on this? Also, where are they focusing if not on the Atxmega?

[dannyf]

Xmega doesn't offer much beyond what a typical 8-bitter does, with the potential exception of its encryption engine.

I would be surprised if you find a lot of solutions utilizing it. As a matter of fact, you are likey to have more problems sourcing it reliably than you do with the megas - I suspect that has a lot to do with low demand for the Xmegas.

If you want to move up within the Atmel family, you may want to check out their very nice ARM offerings.

[TerminalJack505]

If you watch at about 25:00 in this Atmel video...



They say that Atmel is going to be pushing the SAM D20s (an ARM Cortex M0+ MCU) in the space formally occupied by the XMEGAs.  At least the D series and E series XMEGAs.

So, yes, you have to wonder about the long-term viability of the XMEGAs.

[sporadic]

The XmegaE5 is an amazing series for the price. Can do some pretty cool stuff mixing the event system, XCL and DMA system together. Haven't played with the fault module yet but looks handy for motor control. Their internal oscillators are very stable compared to Megas and they're much cheaper as well.  I'm testing one project with a 460,800 rs485 link using the internal 32MHz oscillator and dfll without any issues yet.

[Jeroen3]

I would label Xmega as NRND, but if you insist in not using Cortex M0's, its the chip for you!

[dannyf]

I view xmega as vaporware: I suspect its purpose is to provide a perceived upward mobility for avr users so they don't migrate to arm chips.

with atmels own arm offerings now, the xmega has served its goal

[sporadic]

Perceived upward mobility, vapor ware and NRND? Have you all even looked at a Xmega*E5 datasheet or even used any of the Xmegas? Their current M0+ offerings seem to work ok as a Xmega D replacement. Give me a D20 with a full Xmega E feature set. XCL and DMA mainly in a 32pin package. ARM isn't the be all end all....