Fubarino SDZ - PIC32MZ?!

Created Tue, 21 Mar 2017 22:20:47 +0000 by MAX232

MAX232

Tue, 21 Mar 2017 22:20:47 +0000

Hey, I've found that board while looking for some open source PIC32MZ design. It's a strange version of Fubarino SD with Z chip, and with weird single-legged crystal. Do anyone here knows anything about that board? Is this something that will be released soon or is it abandoned project? Would it be functional (at least for use with PICKIT3) if I manufacture and solder the board according to Eagle sources from github?

majenko

Tue, 21 Mar 2017 22:25:48 +0000

I don't think it was actually properly made. I made a similar board - the SDZL - which I need to re-work with a new power system.

The "crystal" to which you refer I think is probably a MEMS oscillator: an external clock module. Used to get around a bug in the first run of MZ chips that didn't like some crystals.

One day I will get around to re-working the SDZL. I have the perfect power system on new boards I have been working with recently that should fit the bill.

MAX232

Wed, 22 Mar 2017 01:22:56 +0000

Could you please explain what is exacly wrong with that board? The power supply?

I have used PIC32MZ once on the "tqfp breakout" board from Aliexpress and the TC1264 3.3LDO and it worked good enough with internal oscillator.

majenko

Wed, 22 Mar 2017 09:53:57 +0000

There are two main issues with the smaller MZ boards at the moment: The chip is quite power hungry, so the small low-power linear regulators overheat. Small boards don't have the room for a larger regulator. The second problem is the lack of a USB-based stk500v2 bootloader for them, and boards like the Fubarino don't have the space for a USB UART chip.

The first problem I am tacking with the TPS82130SILT from TI - a minuscule integrated 3A switching regulator, a really gorgeous bit of kit. Just needs a couple of small resistors and capacitors with it.

The second problem I need to get round to tackling. I am contemplating seeing if there's an AN1388 bootloader in the current version of Harmony that could be compiled for it, since pic32prog can work with that bootloader instead of the traditional stk500v2.

A third issue, related to the bootloader issue, is that the normal chipKIT core has no USB support for the MZ chips. I have a separate branch in UECIDE that has some support for it by rather roughly integrating parts of Harmony, but it's not really a very nice system at all - Harmony itself is grossly complex and bloated and I (and many others) hate it.

MAX232

Wed, 22 Mar 2017 11:02:29 +0000

The chip is quite power hungry, so the small low-power linear regulators overheat. Small boards don't have the room for a larger regulator.

This is indeed a helpful information. What's the power consumption at the highest clock speed? Would a TC1264 (800mA) in TO220 package be enough?

Do you know any good but simple open-hardware boards for PIC32MZ? I have Eagle files for one PIC32MZ board from Olimex, but it's very complicated, it has ethernet, LCD, sound output, etc, and I just want a basic breakout.

I don't need a bootloader, I could go with PICKIT3.

majenko

Wed, 22 Mar 2017 11:46:33 +0000

The WiFire is our normal MZ board, but that is also quite complex, with it having a WiFi module on it. It's more Arduino-y though.

At 200MHz you can have up to about 130mA needed by the chip alone. Yes, a TO220 packaged 800mA regulator would cope with it. That is considerably bigger than the ones I used on the SDZL though which were just 300mA SOT-23 regulators. I had one for the main chip and one for the peripherals including the SD card. The chip one got incredibly hot. It worked, but the heat really was too much to be usable.

MAX232

Wed, 22 Mar 2017 17:56:12 +0000

I have found the Eagle sources of that board and inspected them a bit.

It's a nice board but it has 3 issues that are problematic to me:

1. the traces are very thin, and the DRC check for Eagle provided by my board manufacturer is givin me "Width" and "Clearance" warning, even tough it didn't give me those warning for Fubarino
2. Could be MRF24WG0MA replaced with RN131 or RN171?
3. Why it's using FT232 and not (for example) MPC2221? MCP2221 has RX, TX and GPIO pins, one of GPIOs could be used to put board into reset state before uploading the bootloader

majenko

Thu, 23 Mar 2017 11:19:48 +0000

The Fubarino and the WiFire are made by different designers with different manufacturing facilities. My guess is that Digilent can work to finer tolerances than Schmalzhaus (or choose to), and that is finer tolerances than your Eagle installation is configured for.

You could replace the WiFi with whatever you wanted if you have the design files.

The FT232R is a historical thing. Originally the programming software was avrdude, and all that could do was toggle DTR. The MCP222x doesn't have a dedicated DTR pin, but it does have an RTS pin. The situation is somewhat different now, and we have pic32prog, and that can toggle RTS. The MPC222x have a dedicated RTS pin, and I know it has been used successfully. Certainly when using UECIDE, since UECIDE itself can manually control both the DTR and RTS lines of whatever chip it's connected with to reset the board.

Personally I don't use either - I use a smaller PIC32, usually a PIC32MX2xx in a 28-pin QFN package. That way you can program the interface chip to do other things besides just CDC/ACM should you want to, like people sometimes do with the ATMega16U2 on the Arduino Uno R3.

For instance, this is the "programming" board for an set of embedded FX boards I am working on. It's a PIC32MX270F128B acting as a CDC/ACM interface to the PIC32MZ2048EFGxx on the main FX processing board. To keep the main board as small as possible I decided the programming interface should be separate, since it's only needed during development. Plus I didn't want to just use an ICD (although that board does give you the ISP header to the main chip as well) since I wanted UART available over the same channel (yes, I know the PICkit2 has serial emulation, but it sucks).

[attachment=1]IMG_20170323_110612.jpg[/attachment]

The other unpopulated USB is a direct USB connection to the main MZ's USB interface.

One of the FX boards, the more complex, looks like this - it has a TFT screen on the other side, 16-bit stereo audio (I2S DAC), 2.8W class D amplifier, and most importantly, the fancy new power systems I am now working with that I will be incorporating into the SDZL when I have got around to reworking it. They're the three blocks in the upper right. This board has 3 voltages, 2.8V, 3.3V and 5V.

[attachment=0]IMG_20170323_111338.jpg[/attachment]

Each of those power blocks can provide 3A from a supply up to 17V with almost no heat dissipation at all. They're amazing, and so small. And the great thing is, you don't need to faff around finding the right inductor or worry much about layout (though still keep traces short and avoid sharp bends) since most of the high frequency switching stuff is internal to the module. And it's high frequency switching, too, so fine for audio work (you don't get a squeal like you do with low frequency switching).

EmbeddedMan

Thu, 23 Mar 2017 12:01:35 +0000

Wow, those TPS82130SILT are really sweet. But at over $5 each in singles, you're paying a bit of a premium for small space and not having to pick inductors. I sure can see why you like them - the simplicity!

Thanks for the tip.

BTW, I do have about 10 of the SDZ boards built up and functional (with the LDO, which appears to work OK in the limited testing I've done), but I haven't touched them for years, since we don't have a way to turn the USB port on from within chipKIT-core yet.

Hopefully someday.

*Brian

majenko

Thu, 23 Mar 2017 13:45:03 +0000

Much of what I create has size as the most critical factor. Plus most of my customers have more money than sense, so they don't mind paying slightly more for something that a) works well, and b) is smaller than anything else out there.

majenko

Mon, 10 Apr 2017 19:07:55 +0000

I finally got a decent dummy load today. Just tested one of these tiny switching regulators.

It's rated for 3A. I can get about 2.4A out of it before it starts to go into thermal shutdown. That's pretty good for a steady state output. Still got 600mA of peak headroom for coping with transients.

I could compress the components together in my reference design but it would be harder to solder (well, harder to clean up after soldering - getting the excess paste balls out after reflow), and it would be smaller than an LM1117.

You could say I was in lust with them ;)