LoRa Mote DM164139在电池供电时偶尔会被重置

我试图找出为什么我的DM164139上的PIC偶尔会重置，同时指示RN2903发送数据包，并最终将其缩小到根本原因。一双新的AAA)，然后突然大量抽取电源（例如，开始LoRa传输），然后电源可能下降到足以使PIC复位。通常的症状是LCD背光闪烁和固件重启。我能够可靠地重现这个场景，方法是：使用设置在>=3.1V的直流电源代替电池；将分机引线连接到J3天线端口（奇怪的是，除了连接天线之外，还有其他任何东西）到此端口会有所不同，包括各种直角SMA连接器、各种150mm延伸引线或它们的组合；以及发送“mac tx uncnf 100 DEADBEEF”到RN2903。在3.1V时，尘粒是稳定的。这正好在一对新的AAA碳锌电池的电压附近，可能也是碱性的。我想也许在3V时，MCP16252升压调节器不会在关键时刻触发，以触发另一个升压循环，因为输入电压太接近输出电压。这允许PIC电源下降到足以进入复位的程度。下面是C5正端电压（PIC附近的+3.3V轨）的克罗轨迹，在复位事件之前、期间和之后：[参见下面的5篇文章获取图像]锯齿电压范围为3.06V到3.52V。平线是2.88V。我认为最好的解决办法是降低fr。ESH电池在使用它们之前有点小！在这种设计中，升压变换器似乎没有很好的规定，因为它大部分都保持在具有巨大输出纹波的PFM模式中，但是对于效率来说，这可能是一个很好的折衷。



      以下为原文

    I was trying to figure out why the PIC on my DM164139's would occasionally reset while instructing the RN2903 to send a packet and have finally narrowed it down to the root cause.

If:

• running from the battery supply (V+ as opposed to the USB supply +Vu***), and
• the battery supply is more than 3.1V (eg. a fresh pair of AAAs), and
• there is a sudden big draw on the supply (eg. a LoRa transmission is commenced)
  

Then the supply may dip enough to put the PIC into reset. Usual symptoms are the LCD backlight flickering and the firmware rebooting.

I'm able to reliably reproduce this scenario by: using a DC power supply set at >=3.1V in place of the batteries; connecting an extension lead to the J3 antenna port (oddly, having anything other than an antenna connected to this port makes a difference, including various right-angle SMA connectors, various 150mm extension leads or combinations of these); and sending "mac tx uncnf 100 DEADBEEF" to the RN2903.

The supply being >3.1V was crucial. At <3.1V the mote is stable. This happens to be right near the voltage of a fresh pair of AAA carbon zinc batteries and probably alkalines as well.

I think perhaps at 3V the MCP16252 boost regulator doesn't get triggered to fire another boost cycle at a crucial time because the input voltage is too close to the output voltage. This allows the PIC supply to drop far enough to go into reset. Here are the cro traces of the voltage at the positive terminal of C5 (the +3.3V rail near the PIC), leading up to, during and after the reset event:

[See 5 posts below for images]
  
The sawtooth voltage ranges from 3.06V to 3.52V. The flat line is at 2.88V.

I think the best workaround is to run down fresh batteries a little before using them in the Mote! The boost convertor seems generally poorly specified in this design, given it mostly stays in PFM mode with the huge output ripple, but maybe that's a good tradeoff for efficiency.