Replies: 6 comments 8 replies
-
|
Drawing too much current from an individual pin can result in:
I've never observed flakiness in behavior due to this as you describe it. There is also a limitation on total current used among all pins, (i think it was 200mA), did you check that? You could also check that the input voltage is not too close to the brown out voltage, and if you are powering it via usb remember there's a protection diode that lowers the voltage half a volt or so. What flakiness do you observe? Resets or fluctuating gpio voltage? In any case, 30mA per pin is probably too high if you want reliability. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks. The flakiness I observe:
I'm tried different power sources, including a bench power supply, USB and 3.7V battery. The MCU is running at 3.3V. I've monitored Vcc with the oscilloscope and it looks stable. I've also monitored GPIO I/O and that looks consistent. Sample size is less than 10 units. |
Beta Was this translation helpful? Give feedback.
-
|
I'm working on a documentation of the differences bewtween LGT8Fx and their Atmel / Microchip counterparts for some time now. Maybe the following citation helps here. "All high-current capable GPIOs are accessible on all three packages. However, when using the The following table shows the internal resistance in Ohm of the LGT8F328P ports at Vcc= 5V in comparison to the Atmega168/328. Taking these values into account, the limit of 12 resp. 80 mA cited in the datasheet seem very conservative. However, while the Atmega will withstand a momentary external short-circuit of one GPIO at 5V , since the peak current does not exceed 200 mA, a short-circuit of one of the LGT8F328P outputs, specially in the high current mode, has the potential to destroy the chip, if the Vcc supply doesn‘t limit the current to safe values of i.e. 500 mA. Besides the danger of thermally overloading the chip, high sink /source currents can distort the voltage distribution between the functional units of the MCU, leading for instance to a shift oft the H/L threshold for digital inputs and the offset of analog functions like ADC, DAC and comparators. This is aggravated by the fact, that the QFP32 and the SSOP20 packages have just one Vcc and one GND connection for all functions . (QFP48 has two of each for the digital functions and an additional AVcc and AGND for the analog part. ). A conservative upper limit for the total sink/source current per package seems to be 200 mA. |
Beta Was this translation helpful? Give feedback.
-
|
This is this very helpful. Thanks! My project uses the QFP48 and at any one time a maximum of 8 pins are sharing a load totalling 80ma. (each set of 2 pins provides the sink and the source for 1 device and there are 4 devices total. Each device is 160 ohms.) It sounds like this should not be a problem based on what you are sharing? Also I am only using 1 Vcc and one GND connection currently and it sounds like it would be unnecessary but still prudent for me to also connect to the other Vcc and GND connection on the QFP48. Does that all sound correct to you? |
Beta Was this translation helpful? Give feedback.
-
|
This looks like you are using the GPIO pins as H bridge or similar , correct ? At 5 V, you would have 30 mA per load, which is well in the acceptable range. Is the load purely resistive ? If there is an inductive component by i.e. a relay coil or similar, the switching induces voltage spikes, which could explain spurious behaviour of the MCU. If you use the QFP48 package, it is essential to connect all 3 GND pins to ground and the 2 Vcc and the AVcc to the supply voltage. The different pins are connected to different regions of the chip and must be on the same potential. In the QFP32 and SOP20, these are connected internally . I'm not sure whether this solves your problem, but it's possible |
Beta Was this translation helpful? Give feedback.
-
|
Yes I am using GPIO pins as H bridge. There is an inductive component as the pins are powering coils. I've been worried about flyback voltage and suspected that first when I started to see problems. My understanding is there will be no flyback voltage from the coils
We have both Vccs and all 3 GNDs connected to each other already. AVcc is currently connected to a .1uF cap that connects to GND. I could try connecting that to Vcc as well. How did you learn that the Vccs weren't connected internally? Is that something I missed in the datasheet or did you figure that out emperically? |
Beta Was this translation helpful? Give feedback.
-
|
The chinese datasheet is only marginally specific about the electric properties of the chip, in contrast to most western manufacturers, which have example application schematics and lots of diagrams to show power consumption and the like at various operating conditions. Having to connect all GND and Vcc pins of the QFP48 is not explicitly mentioned anywhere . In an inofficial statement in a chinese electronics forum, a representative of LGT stated that the voltage difference between Vcc and AVcc should be below 0.1 V. A separate cap for the AVcc pin is o.k. , as long as it's connected to Vcc via a resistor of less than 50 Ohms or a choke of i.e. 500 uH. If preventing digital noise to interfere with A/D conversions is not necessary, AVcc is best connected to Vcc directly. Switching inductances is always tricky. If you keep both sides of the coils connected to GPIOs in either H or L state, there should be no extreme voltage spikes. You can improve the situation by having an additional resistor of i.e. 500 Ohms in parallel to each coil. If switching speed is not of the essence, using an RC snubber on each coil ( cap of 0.1 uF and resistor of 22 Ohms in series ) could also help. I have used the LGT8F328P to directly drive the 30 Ohms coils of a stepper motor ( in high current mode of course ) without any problems. By the way, have you seen any improvement of your problems already ? |
Beta Was this translation helpful? Give feedback.
-
|
Ummm, no. Turning off the high and low side switches at the same time will
not prevent EMF from inducing a voltage across the coil. The voltage will
be present across the coil and will both overvoltage the high side and
undervoltage the low side. If you're not seeing this on the scope, it may
be that your scope is just loading the circuit enough to supress it--or
your scope has insufficient bandwidth to see the peaks. If you really want
to drive an inductive load directly with a uC, you may want to look into
adding some diodes to all of the driving pins to protect them from >Vcc and
<GND. Two diodes/GPIO will do it. There are small diodes on the uC die,
but they're not meant for taking this much energy and will fail and that
can lead to other problems with your chips.
…On Sun, Nov 19, 2023 at 12:23 PM John Harrison ***@***.***> wrote:
Yes I am using GPIO as H bridge. There is an inductive component as the
pins are powering coils. I've been worried about flyback voltage and
suspected that first when I started to see problems. My understanding is
there will be no flyback voltage from the coils if both ends of the coil
are floating after being charged so I always make sure to do that in the
software and not leave one side at GND or Vcc. From my oscilloscope
readings it appears this trick is preventing flyback but it's worth keeping
this on the radar as a potential issue for sure.
If you use the QFP48 package, it is essential to connect all 3 GND pins to
ground and the 2 Vcc and the AVcc to the supply voltage.
Yikes that didn't even occur to me to do. I have one Vcc pin and 1 GND pin
connected. AVcc is currently connected to a .1uF cap that connects to GND.
I feel silly that I missed that the pins are not connected internal. Did I
miss this in the datasheet or did you figure this out through emperical
evidence?
—
Reply to this email directly, view it on GitHub
<#305 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACPEX7EM7FBQZRH6RTOAKHDYFI6B7AVCNFSM6AAAAAA7QWDUVSVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TMMJSGI4DI>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
Oops I misspoke. I don't turn both sides off. I power both sides of the coil to GND. I'll edit my earlier comment to reflect that. |
Beta Was this translation helpful? Give feedback.
-
|
Protecting a H bridge is not a simple task. Look at this complex protection: https://i.stack.imgur.com/aVFr3.jpg
It is recommended that these be fast schottky diodes. |
Beta Was this translation helpful? Give feedback.
-
|
And let's not forget the mechanical energy stored by the rotating mass. When braking, the elecric motor produces energy. |
Beta Was this translation helpful? Give feedback.
-
|
In case you can still modify the design, I had success with DRV8833, they are very cheap low voltage mosfet double h-bridges (2.7-10.8v, around half an ohm rds-on) |
Beta Was this translation helpful? Give feedback.
-
|
no question at this pont I'm modifying the design! :) |
Beta Was this translation helpful? Give feedback.
-
|
Or a BDR6122TC if you like'em cheap and tiny.
…On Sun, Nov 19, 2023 at 3:17 PM David Buezas ***@***.***> wrote:
In case you can still modify the design, I had success with DRV8833, they
are very cheap low voltage mosfet double h-bridges (2.7-10.8v, around half
an ohm rds-on)
—
Reply to this email directly, view it on GitHub
<#305 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACPEX7B4ZMGAD2A7YEQZMGTYFJSNJAVCNFSM6AAAAAA7QWDUVSVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TMMJTGI2TI>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
There seems to be a lot of seemingly conflicting information online and in the datasheet about what current the pins of the MCU can and should be able to handle. Referring to this datasheet:
I'm deep into trying to understand flakey behavior on a project which currently requires many of the pins to deliver 20ma (160 ohm load at 3.3V) and I'm trying to determine if the load on the pins could be a factor here.
Beta Was this translation helpful? Give feedback.
All reactions