I built a brick

[MAkke]

I designed a board based on an STM32F303CC and I am unable to upload any code to it. I'm using STM32CubeIDE with an ST-Link/V2 programmer. I've connected Pin 7 to DIO and Pin 9 to CLK. The board uses USB for it's 5V power supply which is transformed down to VDD = VBAT = 3.312 V. NRST uses the internal pull-up resistor and measures 3.305V. Boot0 is strapped to GND.

I get error message: (4) "No device found on target".

I connected the GND of the board to one of the GND pins on the JTAG connector of the ST-Link/V2, still error (4). I connected RESET (Pin 15) of the ST-Link to NJTRST on the board connector, still error (4).

I connected NRST of the microcontroller to 3.3V (Pin1) of the ST-Link through a jumper wire, and I get error message: (255) "Unknown, please check power and cabling to target".

I connected 3.3V (Pin1) of the ST_link to VDD/VBAT on the board through a jumper wire, again error (255).

PFA my schematic, all the resistors and resistor arrays are 820 Ohm, except for the BOOT0 Pin which uses 56k. What am I missing?

[Danish1]

One thing I don't see in your schematic is power-supply-decoupling capacitors. You need 0.1 uF ceramic placed as close as possible to each Vdd / Vss pair of pins.

Microcontrollers are very fast devices, and the inductance of even a centimetre of pcb track is enough to stop things working reliably. And Tantalum / Aluminium Electrolytic are too slow to work.

I don't use ST-Link V2 and I don't know its pin numbers. But you must wire up all the necessary connections with fairly short wires or you'll get reflections and need to clock things very slowly to make it work (assuming you have the choice). I always use ribbon with every other line grounded.

How confident are you that there are no short-circuits on the pcb? I have now given up trying to hand-solder lqfp packages with 0.5 mm pin pitch!

Hope this helps,

Danish

[MAkke]

I used multiple capacitors in my design, all ceramic.

[MAkke]

The closest is C1 at about 0.7 cm, but it's connected through a ferrite bead.

[Danish1]

Ferrite beads block the kind of high-frequency that we're trying to decouple, so any capacitors connected via ferrite beads won't help at all.

If that's all you have, it might be worth shorting out the ferrite beads to see if that helps.

As to any pins shorted together, most-likely your pcb-manufacturer didn't build-in any shorts*. It's only when you come to solder in your lqfp that you might accidentally have a solder bridge between adjacent pins. So that vastly reduces the number of tests you have to make. And (at the moment) you only need to worry about those pins essential for the microcontroller to be contacted over SWD (or JTAG).

*I have seen a short in one pcb from a well-known manufacturer. They now claim to test 100% of all pcbs and I haven't seen the problem since.

[MAkke]

Thank you very much for your reply, your comments and suggestions are really helpful.

First I will try to solder a capacitor from the corner of VCC/ VBAT to GND and see what that does.

For the oscillator I chose a 24 MHz SMD crystal, it comes with an integrated capacitor (9pF or 18 pF).

I don't think there's a short between the pins of the microcontroller. Solder bridges occurred during assembly, but they are not hard to spot and I made sure to remove all of them before powering the board.

If the capacitor doesn't help I will try shorting the ferrite bead.

[MAkke]

As for shorts, visual inspection reveals nothing and I cleaned all the flux off the board using solvent and an antistatic brush. What worries me is the continuity testing: sometimes I hear a short beep between two points that are not supposed to be connected, but when I keep the probes in the same position, I don't hear a continuous beep.

The strangest thing about his phenomenon is that it already occurred during assembly: the board was built from small to big, starting with all the passive components, MOSFETs and diodes. Before the first IC's and MCU's were placed, I already noticed this issue somewhere around the 24V to 5V step-down converter. I think it's a capacitor discharge or some hysteresis effect in the inductor or diode, but the board was never powered at that point since it wasn't finished yet, so it doesn't make sense.

Furthermore: how confident can one be that there are no shorts on a board that contains both a 48 and 64 LQFP among other components? Lots of probe combinations possible there.

I've broken it down into all the pins involved here: there are no shorts between any combination of GND, 3.3V, BOOT0, NRST, SWDIO, SWCLK and NJRST the voltages, of GND, 3.3V, BOOT0 and NRST are also correct (see above).

[Mike_ST]

1) Make sure you've updated the ST-LINK firmware.

2) SWDIO connection only require 4 cable at minimum, it is quite straightforward.

Wiring pin 1 to target VCC is mandatory.

As you said Pin 7 to SWDIO, and PIN 9 to SWCLK, and ground.

I don't use CubeIde, but maybe there is a setting that let you choose the SWD Interface to connect (not JTAG, it requires more wires).

[MAkke]

1) There's no way to not update the ST-Link software. Whenever you try to debug a microcontroller through the ST-Link V2, CubeIDE will prompt to update the ST-Link first if it wasn't already using the latest firmware version.

2) Interface settings are set to SWD, not JTAG, since I read somewhere that JTAG is not supported by the ST-Link V2. I don't have a dedicated 3V3 pin on the board, but using a jumper wire from the 3V3 to the ST-Link Pin 1 gives me the aforementioned error (255).

[Mike_ST]

>> 1) There's no way to not update the ST-Link software.

Ok I didn't know about that.

>> 2) Interface settings are set to SWD, not JTAG, since I read somewhere that JTAG is not supported by the ST-Link V2.

As far as I know that is not true.

I'm surprised that there is no capacitors on PF0-PF1 for the crystal.

[MAkke]

Soldering an extra capacitor between VBAT and GND resulted in error (13) "Unknown, please check power and cabling to target". Shorting FL1 yields the same result.

[MAkke]

I had a theory that the 820 Ohm resistors used for SWDIO and SCLK limited the current too much. I shorted both resistors with a piece of copper wire and I was able to upload my first code.

So the first time I was able to upload code I had the following configuration:

• Boot0 was still strapped to ground with a resistor
• NRST was still left floating, using the internal pull-up resistor
• A 3V3 Pin was added to connect the ST-Link to the the board
• The 24 MHz SMD crystal with integrated capacitor was replaced by a THT 16 MHz crystal with two separate THT capacitors to GND
• The resistors used for SWDIO and SWCLK were shorted.
• FL1 was still shorted

This raises another question:

The reason I chose 820 Ohm resistors was to limit the current going through the MCU in order to reduce the temperature. In the datasheet it says that the peripheral that consumes the most current is TIM2 at 49.1 uA/MHz. At 72 MHZ that results in a current of 3.535 mA. I chose to go with 820 Ohm, because 3.3V / 820 = 4.024 mA, a little above the required current, but considering I have 31 peripheral pins, 31 * 4.024 = 124.744 mA, which is still below the maximum source/sinking current of 150 mA. Did I miss something? Did I use the wrong values per pin?