Breadboard maple « LeafLabs Garden

LeafLabs Garden » Support » Maple Hardware Support

Breadboard maple

(22 posts) (5 voices)

Started 3 years ago by JoshSanders
Latest reply from ala42

12 Next »

JoshSanders
Member

Hi guys!

I'm trying to make a bread-boarded maple using through-hole components.
Using a RET6 chip I mounted on a SMT to through-hole adapter, I built the circuit from the Eagle schematic for Maple R5 (sans the LI-battery charger and barrel jack).

I'm wondering if anyone has tried this before and run in to trouble. Here's my problem:

Using a FTDI adapter (Adafruit, modified for 3.3v), I am able to flash the bootloader binary to Maple r5 following the instructions on the wiki. When I connect the FTDI adapter to my breadboard and try the same thing (of course using the RET-6 bootloader binary), I get the following:

C:\Python27>python.exe stm32loader.py -p COM9 -evw maple_RET6_boot.bin
Reading data from maple_RET6_boot.bin
Can't init. Ensure Boot0=1, Boot1=0, and reset device
[followed by traceback]

It does the same thing when the FTDI adapter is connected to nothing.
Is there anything obvious I'm missing? I triple-checked my connections including Boot0 tied high and Boot1 tied low...

Here are the major through-hole parts I'm using instead of SMT:
> 3.3v regulators: MCP1702-3302E/TO (Maple uses MCP1703T)
> 1uf ceramic Cap on regulator input and output as per datasheet (Maple uses electrolytic 22uf for these)
> Digikey part 887-1233-ND - an 8mhz through-hole crystal spec-matched to the Maple part
> Generic through-hole NPN transistors
> Golden-max series ceramic caps
> Standard through-hole 1/4W resistors

I have a photo of the setup:
https://plus.google.com/photos/110977450351392292564/albums/5835070319517275345/5835070329052827074?banner=pwa

(The huge red button is the reset button - it's the only one I had on hand)

and a photo of the reflow job (I used a reflow-modded toaster and checked under a scope afterwards... all connections look clean.

https://plus.google.com/photos/110977450351392292564/albums/5835070319517275345/5835070327637721682?banner=pwa

I verified that the regulators are delivering 3.3v to the VDD pins.

So I'm stumped.
Are there any fuse bits that have to be set before flashing the bootloader for the first time? Or lines that should be tied high/low that aren't explicitly specified? Or is this whole project hopeless due to stray capacitance or some other breadboard related issue?

I'll post more as I work through this, and a quick guide if I get it working. Thanks for sharing wisdom!

Cheers,
Josh

Posted 3 years ago #
JoshSanders
Member

Got a better pic of the setup (the original link won't work anymore)

https://picasaweb.google.com/lh/photo/JT0LEG6FGIJo740KJ6tmVdMTjNZETYmyPJy0liipFm0?feat=directlink

The red USB breakout isn't connected - I moved it aside for bootloading so I could connect the FTDI adapter to power.

The point of the photo is to help screen whether there's anything obvious I'm doing wrong (i.e. use of long jumper wires, placement of power filtering caps between power rows, etc)

Posted 3 years ago #
gbulmer
Moderator

JoshSanders - I have only flashed the bootloader into 'proper' PCBs, so I might not be able to help you enough. My board was built using a toaster oven, and the soldering on your breakout looks excellent.

There are no fuses which need to be set before flashing.
The process at http://leaflabs.com/docs/bootloader.html#flashing-a-custom-bootloader worked for me on several boards. The major problem was pressing the buttons at the right time.

I am a little concerned by the distance between STM32 pins and the decoupling capacitors. However when the STM32 comes up first time, it should only be running at 8MHz, so it shouldn't behave like a little FM radio station. This might become a bit more of an issue once the bootloader is installed and it is internally bouncing along at 72MHz.

I would try any of:
0. Simplify; remove parts that aren't involved in flashing the bootloader. Remove the crystal (the STM32 will be running off the internal RC oscillator), remove all the Maple-derived USB interface components, remove the user button and LED. You need the power supply, decoupling capacitors, reset button and BOOT pull-up and pull-down. My hope is you will be able "to see the wood for the trees". It has a reasonable chance of working, so the less there is to camouflage the problem, the better. This does have the problem that you'll need it all once the bootloader is working, but you need to get over that hurdle first.
1. If you have access to an oscilloscope, have a look at everything that you can, and try to compare with your working Maple board. Especially try to look at the serial data to see if anything is happening at all.
2. Have you got a ST-LINK/V2, e.g. an STM32F0Discovery, which could be used to check the chip is working? The newer STM32FxDiscoveries all have the two wire 'JTAG/SWD' which supports 'proper' debugging and upload. Texane stlink at github has an upload https://github.com/texane/stlink
3. Wire the FTDI into USART2 of your Maple and write a dumb "read char from serial, write char to USB" program to ensure something is coming up the wire.

Dumb questions:
- are you confident the FTDI Rx & Tx are going to the correct pins?
- are you confident the FTDI is running at 9600 baud (to make it easy for the manufactured in bootloader to synch)?

(Full disclosure: I am not a member of LeafLabs staff.)

Posted 3 years ago #
JoshSanders
Member

Thanks, gbulmer! I'll definitely give those a try.
At first pass, cleaning up the board, removing the crystal and setting transmission to 9600 with "-b 9600" didn't help - but I'll work on it more when I get a chance. The FTDI is definitely transmitting - I'm able to program maple boards and I observed it with the o-scope.

I'll post more soon!

Posted 3 years ago #
JoshSanders
Member

Ok, here's what I tried:

I reduced the schematic to this:

https://picasaweb.google.com/lh/photo/6ug6nTKt3otob0y5HFUqvdMTjNZETYmyPJy0liipFm0?feat=directlink

and set up the breadboard like this:

https://picasaweb.google.com/lh/photo/cpajH0vk0qAHi9y-mtf-1tMTjNZETYmyPJy0liipFm0?feat=directlink

The green wire is the reset button (I ground it briefly in the ground lane).
I attached VDDA to the same power supply as VDD since analog reads aren't part of bootloading so far as I gather.

I checked with an oscilloscope to ensure that all power lines and pins tied high are 3.3v and all grounds are at ground. I checked to ensure that the reset pin goes to ground when I push the reset button. I tried replacing the input filtering cap with a 22uf electrolytic. I tried transmission at 9600bps and default (115200) with and without the crystal. Same error. I'm beginning to think the chip may be at fault... how static sensitive are these? I'd also guess the power scheme is sensitive to noise, so there may be ways to help fix that...

I should also mention that I had the same result as rosendo in this post:

http://forums.leaflabs.com/topic.php?id=1140#post-7024

That the stm32loader.py in maple-bootloader/blob/master/flash (from the /bootloader directory on github) didn't work even for Maple r5, but the one in libmaple/master/support/ worked fine and that's the one I'm using. I should mention I'm on Win7 64bit and python2_7.

In the same post, siy recommends using stm32flash: http://code.google.com/p/stm32flash
I'll try this too.

I'll make a few more breakouts with the RBT6 chip instead of RET6, taking extra precautions against static... (I ordered some of those just in case)

I don't have an ST-link, but I can order one if it turns out to be necessary.
I'll try the read char/write char option later this week. (I'll be busy for the next few days preparing a talk)

gbulmer, when you say your biggest hurdle was pushing the buttons at the right time - could you be more specific about what worked? I've been under the impression that if boot0 is held high and boot1 is held low from power-up, all you need to do is reset by briefly grounding the reset pin... it seems to work on Maple when I hold the BUT down from before power-up, briefly press and release reset while still holding BUT down, start the bootloader, and continue holding BUT until the bootloader finishes... (i.e. BUT can be held high all along)

Thanks again for your advice, gbulmer!! You're the best & don't let anyone tell you otherwise!

~J

Posted 3 years ago #
gbulmer
Moderator

JoshSanders - "..., when you say your biggest hurdle was pushing the buttons at the right time - could you be more specific about what worked"

I'm very sorry, I was trying to suggest that it was straightforward to flash the bootloader by saying the hardest part was pushing two buttons. I apologise, I was not clear.

The STM32F0Discovery, STM32F3Discovery, STM32F4Discovery all have the two wire JTAG/SWD ST-LINK/V2 built in, and are cheaper than an ST-LINK/V2, and you get a free development board.

Posted 3 years ago #
ala42
Member

Try the serial STM flash tool
http://www.st.com/internet/com/SOFTWARE_RESOURCES/SW_COMPONENT/SW_DEMO/stm32-stm8_flash_loader_demo.zip

Posted 3 years ago #
JoshSanders
Member

Ok, I tried the STM flash tool. I was able to flash the bootloader on to a maple board, but not on to the breadboard STM32F103RET6. It just gave me a "No response from the target" message. I tried with and without the crystal, and at different baud rates.
I also tried using Maple's onboard voltage regulator instead of the through-hole version I had been using (i.e. powering a maple board by the same FTDI that's talking to USART1 on the breadboard STM32F103, and bringing the 3v3 and gnd across to the breadboard. Still no dice.

I'll post more when I bake my next round of chips.
Thanks for all your help!

Posted 2 years ago #
gbulmer
Moderator

JoshSanders - if you want to prove the viability of the approach in a smaller step, you might try using the low-cost STM32F051:
http://www.futureelectronics.com/en/Search.aspx?dsNav=Ntk:PartNumberSearch%7cstm32f051%7c1%7c,Ny:True,Ro:0,Ns:P_Price1%7c101%7c1%7c,Nea:True,N:587-4294880381
or
http://uk.rs-online.com/web/c/?sra=oss&r=t&searchTerm=stm32f051&x=0&y=0

They come in 48pin packages, so a smaller SMT.
I do vaguely remember seeing an SMT adapter which had space for decoupling capacitors on board (sadly I'm not sure where). That would reduce the test to the SMT adapter, a smoothed power supply, and a USB-to-serial.
It would need a custom test program though. It wouldn't run the Maple bootloader, or Maple IDE generated programs. On the other hand, you'd have a low-cost Cortex-M0 which is more 'oomph' than an Arduino UNO or Nano.

Posted 2 years ago #
JoshSanders
Member

My original intent was to design a minimal Maple schematic I could "drop" in to my designs to sidestep the need for a separate Maple board (given the uncertain availability of the board in the future). I figured if I could breadboard the circuit, I could more easily determine which parts of the circuit I could change or eliminate.

I've decided the best way to do this is to design a reduced 2-layer version of the core Maple mini circuit, have a PCB fabbed and test it. I'll start another thread once I have made progress on that front.

If anyone decides to pick this up, it would be great to hear what ended up working!

Posted 2 years ago #
gbulmer
Moderator

JoshSanders - "My original intent was to design a minimal Maple schematic I could "drop" in to my designs to sidestep the need for a separate Maple board"

I believe several of us have had the same thought. IMHO a schematic isn't sufficient; the PCB layout is needed too to make it easy to embed in larger designs.

"I've decided the best way to do this is to design a reduced 2-layer version of the core Maple mini circuit, have a PCB fabbed and test it. I'll start another thread once I have made progress on that front."

I recommend looking at Siy's work on the mini48 at https://github.com/siy/openstm32hw though I haven't used this design myself.

Siy's mini48 is a single-sided PCB design for a 48-pin LQFP part, so it is a good start for embedding in other designs. The Eagle files are on github. It improves on the Maple USB interface, being both fewer components, and giving some ESD protection. It also includes the 32KiHz crystal for the STM32F103's on-board battery backed Real-Time-Clock.

Edit:
For prototype PCBs you might try: http://oshpark.com/ for prototype PCB's.
It is $5/square inch of design (to the nearest $0.01), and returns 3 PCBs. They are very good PCBs, with gold finish for SMD soldering. Siy's design would be about $8.50 for three PCBs.

Posted 2 years ago #
JoshSanders
Member

gbulmer, thanks for sharing Oshpark!
I've been using SeeedStudio Fusion PCB and I'm happy with their work - but I'm restricted to 2 layers. This looks great!

I looked over Siy's prototypes. Very elegant!
I may use the mini64 as a starting point - I ultimately want to use the STM32F103RFT6 for its 96k of on-board sRAM.

I had noticed the same USB interface improvement earlier, in Suspect Devices prototype, Maple Bacon
https://github.com/soycamo/maplebacon

I'll post my own designs once I have something I'm happy with

Posted 2 years ago #
feurig
Member

Our bad in not giving credit where credit is due I believe that slys prototype is where we got that from.
That plus 1117 series regulator = much better circuit all around.

I am going to try to do the same with a ret6 based board and run elua on it....

Posted 2 years ago #
ala42
Member

If you go for a 100 pin chip, use a F4 to have 192KB ram, or at least make your board F4 compatible by adding the few needed caps/resistors, which is described in the F4 manual.

Posted 2 years ago #
gbulmer
Moderator

JoshSanders - I certainly agree with ala42, STM32F4 gives a lot more RAM than STM32F103. STM32F4 has many other performance benefits and improved peripherals.

Note: there is an STM32F4 which comes in a 64pin package. Friends use it on micromouse robots.

It would require a bit of effort to make the board compatible with STM32F103 and STM32F4; the power pins are slightly different. But I think it might be worthwhile given the range of chips you'd have access to.

The signal pins are compatible, though IIRC the STM32F4 has more 5V tolerant pins.
The STM32F4 peripherals are much better than some of the STM32F103.

The STM32F103 does have the advantage of being available in low-cost versions, which might be very useful. So I wouldn't suggest making it STM32F4 only.

Posted 2 years ago #

RSS feed for this topic

12 Next »

Reply »

You must log in to post.