DAC Burns Out Upon Touching Ground

[rabbani]

Hi Everyone,

I am using B+ Raspberry Pis for my master's thesis to build 70 devices and deploy them across campus. For the project, I have designed my own custom circuit and I have ordered test PCBs. The RPi is connected to my circuit through the 40pin connector (~20cm ribbon cable). I am supplying power to the RPi through my circuit using 5v pins (2,4) and not the micro-usb. All the GND pins on the RPi are also connected my circuit's GND. This means that the 5v pins on the RPi are connected to my power supply's V+ output (5V, 3A clean power supply. ~$20), and it's GNDs are connected to my power supply's V- and mains ground. The Pi boots up and performs perfectly fine.

On the circuit, I have a DAC that talks to the RPi using SDA and SCL pins. In the beginning, I was using MCP4725 (https://www.sparkfun.com/datasheets/Bre ... CP4725.pdf). In addition to SDA and SCL pins, the A0 pin was also connected to one of the RPi's GPIO pins. Everything worked perfectly fine (I could control the output voltage on the DAC) until I started touching the ground with a hanging wire (it started when I was trying to connect something to the RPi's USB). Sometimes, when I touched the ground with a wire or a piece of metal, it could burn out the DAC.

To be able to experiment more, I replaced the MCP4725 with a MAX5805 (http://datasheets.maximintegrated.com/e ... AX5805.pdf). In this one, in addition to SDA and SCL, AUX and LDAC inputs are connected to the RPi's GPIO pins. With this DAC, I tried doing the same thing, but instead of burning out, the DAC was reset. I tried supplying power to the RPi through the USB, but it didn't work.

Can anyone help me figure out the problem?

[mahjongg]

power supplies for the PI are normally of the double isolated type, and GND of the PI is NEVER connected to mains, (except perhaps via the HDMI cable) it could be your "mains ground" actually contains a large voltage contrary to EARTH, and that large voltage ends up on the PI.

[rabbani]

power supplies for the PI are normally of the double isolated type, and GND of the PI is NEVER connected to mains, (except perhaps via the HDMI cable) it could be your "mains ground" actually contains a large voltage contrary to EARTH, and that large voltage ends up on the PI.

I actually had the problem when the GND for my circuit was not connected to the mains ground. I tried connecting the GND to mains ground in an attempt to fix the problem.

[mahjongg]

use a multimeter and measure if there is an AC voltage between your "mains ground", and a non electric grounded conductor, for example a water mains. mains grounds can be faulty.

[davidcoton]

What was the other end of your hanging wire connected to?

use a multimeter and measure if there is an AC voltage between your "mains ground", and a non electric grounded conductor, for example a water mains. mains grounds can be faulty.

Get an electrician to check your mains ground connection if this seems to be the problem. A supply with a faulty ground is potentially dangerous, as electrical safety often depends on a good ground connection. Measuring voltage on ground to another earth connection is not an adequate test; there are several reasons for this including that water pipes may be bonded to mains ground, or may be plastic and non-conducting, or because of the characteristics of your multimeter. A more sophisticated "earth loop impedance" test is required, together with the knowledge and experience to interpret the result.

In general circuits powered from double insulated supplied should not be grounded. The "0V" connections of each part of the project should be connected to each other, while 5V or 3V3 lines from separate supplies should be kept separate. Note that connections like USB which deliberately carry 5V power may make this difficult. It may not cause a problem, but if it does, diagnosis and correction could be hard.

I've used the OP's terminology of "ground" in this post -- UK readers may translate this to "earth".

[rabbani]

[quote="davidcoton"]What was the other end of your hanging wire connected to?

It initially happened when I wanted to connect a USB powered LCD screen to the Pi, but now that I am testing different scenarios, it happens even when the other end of the wire is not connected to anything.

My guess is that there is a spike in my GPIO outputs (and maybe even SDA, SCL pins), that causes the DAC to reset. I don't know how to verify or fix this though.

[davidcoton]

Are you working in a static-safe environment? It sounds to me like you are building up a static charge on your body (nylon clothing or carpets?) and discharging it into static-sensitive electronics.

[BMS Doug]

I tried supplying power to the RPi through the USB, but it didn't work.

The B+ is designed to prevent backpowering through the USB, if that is what you were trying to do.
If you are saying that you cannot get the pi to turn on even when you connect a micro USB to the normal power supply inlet then I would suggest trying a different Pi for awhile, you may have damaged that one (or maybe just broken it's SD card).
Don't try to power the pi from two sources simultaneously.

I expect you used the bypass capacitor as specified in the DAC datasheet, did you also add a high frequency noise attenuation capacitor?

3.2 Supply Voltage (VDD, VSS)<snip>
This pin requires an appropriate bypass capacitor of about 0.1 μF (ceramic) to ground. An additional 10 μF capacitor (tantalum) in parallel is also recommended to further attenuate high frequency noise present in application boards. The supply voltage (VDD) must be maintained in the 2.7V to 5.5V range for specified operation.

[rabbani]

Are you working in a static-safe environment? It sounds to me like you are building up a static charge on your body (nylon clothing or carpets?) and discharging it into static-sensitive electronics.

Nope, I am not, and I think I shouldn't, because I am deploying 70 instances in the same environment. However, I noticed today, that during a long enough period of running (3-4 hours), the DAC resets without anybody touching the circuit. Therefore, I definitely need the hardware problem I am having.

The B+ is designed to prevent backpowering through the USB, if that is what you were trying to do.
If you are saying that you cannot get the pi to turn on even when you connect a micro USB to the normal power supply inlet then I would suggest trying a different Pi for awhile, you may have damaged that one (or maybe just broken it's SD card).
Don't try to power the pi from two sources simultaneously.

This is not what I was trying to do. Since I was powering the RPi through the 40-pin connector, I suspected that this might be causing the problem. Therefore, I cut the power from the 40-pin connector, and instead, used the micro-USB to power the RPi. Of course, the RPi worked perfectly fine, but it did not fix my burning DAC problem.

I expect you used the bypass capacitor as specified in the DAC datasheet, did you also add a high frequency noise attenuation capacitor?

Thanks for noticing this. I do have the bypass capacitor, but not the high frequency noise attenuation one. I am going to try this and see if it fixes the problem.