What Can You Do With the World’s Smallest Arm® Cortex® M4 with FPU Microcontroller?

Hello Randall and other MAX32660 watchers:

I've just posted a blog about my MAX32660 + FPGA board

Ultra cheap and Tiny FPGA Board

MK

This is from Maxim:

The land pattern can be found here:

https://pdfserv.maximintegrated.com/land_patterns/90-100089.PDF

The link is in the data sheet for the ‘660.

The non-secure bootloader can be downloaded here:

https://www.maximintegrated.com/en/design/software-description.html/swpart=SFW0007580A?cq_ck=1539960966781#

I’m downloading it now. I think you’ll need to create a MyMaxim account to make it work, but it seems to be in order.

With regard to the SWD pins, your assessment is correct: you can perform debugging via P0.8 and P0.9, but on every reset the default pin assignments are re-established. That means you have to retain access to P0.0 and P0.1 to load code via SWD. If you’re using the bootloader, you can load your firmware to the ‘660 by accessing the bootloader via I2C instead of SWD, but I don’t think we have IDE support for loading via the bootloader and then debugging via the alternate function pins.

Hoping Randal is listening.

I've got 2 (simple) tech support questions on this part which so far Maxim's tech support havent answered. Perhaps Randal could trigger his hotline to them.

1) I need footprint drawing for the 24 pin QFN package - this does not seem to exist on Maxim's website - the sketch (not dimensioned) on the data sheet shows it to be an unusual footprint so I need to be sure.

2) The link on Maxim's website to download the unsecured boot loader does not work.

There is one less simple question:

The chip has two pairs of pins that can be used for debugging - if you want to use the non-default pair is it necessary to retain access to the default pair to load initial code into the chip ?

I've got 25 of these parts on order  - it would be nice to have a board ready for them.

MK

RE the TQFN package.

Neither Farnell nor Digikey are listing these packages and they have no stock.

Mouser have lots on order and expect to have them by 19th October.

These will be the 24 pin 0.4mm pitch TQFN.

I have no problem hand soldering 0.5mm pitch TQFN I expect 0.4mm to be feasible but a little harder (I think Ive done a couple of 0.4mm but can't remember).

The 0.35mm pitch BGA will be expensive to prototype since it will need filled in vias for the 4 middle pins. I've had no joy hand processing 0.5mm pitch.

To "hand solder" the TQFN, design a footprint with a decent sized centre hole in the pad (1mm is good). Alternatively go for 5 much smaller holes.

Paint on the solder paste (I use a grotty Chinese air pressure driven syringe but with decent tips).  You want paste on the centre pad and on the peripheral pads but NOT in the space in between.

It doesn't need to be that even. Use leaded solder - the lead free stuff needs too much heat and doesn't flow well - it can work in production but we are bodging here and need to stack the cards a bit.

Place the chip and put the board on a preheated hotplate (chip up !!!) - I set mine to 215C. Once the solder has melted (will be very visible) give it another 30 seconds or so to let the flux get out.

Remove from the hotplate, cool, and drench in runny (water like viscosity) flux, remove excess solder with solder wick, solder any pins that didn't 'take' using a fine point temperature controlled iron at 350C (possibly more if very pointy).

I've done this on more than 20 different board designs with up to 5 TQFN chips on them. It's a perfectly viable prototype process but obviously NOT suitable for anything where failure has serious consequences.

(The holes in the pad are needed because the usual problem is excess solder under the chip and the holes give it somewhere to go - they also allow you (especially if you go for one big hole) to inspect the soldering of the pad.)

You will need steady hands and a microscope - I used to use an optical binocular thing which killed my back - I now use a sub £200 Chinese video camera with suitable lens and it's much better.

People will tell you that you need some fancy stuff with special cameras and lenses costing thousands - but you don't (unless you want to make high quality video of what you see.)

I tried an expensive optical thing with an optical screen and built in camera (about £4k worth) and it isn't as good as the cheap set up I use now.

MK

Webinar Questions Answered:

• For prototyping, a BGA is quite dificult to use, so is there a "socketed" alternative?
  Two options: the ‘660 is also available in a TQFN, but that’s pretty hard to use for prototyping purposes. But the MAX32660EVSYS exposes all pins to the 0.1” pin array, including the SWD pins, so you can use the EVSYS for your “prototyping part” if you want.

• Why is hyteresis a necessity ?
  On an I/O pin, having input hysteresis allows you to connect the pin to a slow-moving signal without worrying about metastability. The input will always be valid even if it’s not at a strict logic level.

• How much Wake-up Charge is needed from backup mode?
  The wake-up charge isn’t independently specified, but we can draw an upper bound by noting that the wake-up time from both deep sleep and backup modes is typically 1ms. Let’s assume you’re waking to full speed, full power. That’s a current draw (not counting the current required to charge any external capacitors) of about 8.6mA. You’re drawing that current for 1ms, so that’s a wake-up charge of 8.6uC. That’s just a back of the envelope calculation, and in actual practice, it may be less than this, but if you design toward this value I don’t think you’ll go wrong.

• Once broken, is there any way to re-program the board on the RHS?
  Sure! The SWD pins are exposed on the 0.1” pin array. You’ll need to provide an SWD header or other means to get the signals to P0.0 and P0.1.

• How would you add USB support for data transfer to a Host computer?
  We discussed this a bit: if you’re adding, for example, a serial port (USB CDC class) there are a whole world of devices out there, mostly from FTDI. If you’re trying to add some other USB endpoint, we need to talk…

• What is the Moisture Sensitive Level (MSL) for these different packages?
  Maxim WLP packages are classified MSL1, and require no prebake. I didn’t find any information about TQFN packages, but many of our TQFN package devices are explicitly MSL1.