Stack ADC Differential Pi

Hello,

I'm trying to stack two boards together (eventually up to 4 in the future) but only one board is detected from them. Connected individually I can see them with i2cdetect -y 1 no problem.

One board has i2c address 68 and 69. The other board 6a and 6b. Both are working perfectly in isolation but cannot get the Pi to recognize both simultaneously.

I'm using a standard raspberry pi DC power source.

Many thanks in advance for any hints.

Felipe.

Hello Felipe

When you have the boards stacked together is the one that is recognised on the top or the bottom of the stack? If it is on the bottom then it could be a bad connection between the two boards and I would recommend cleaning the pins to make sure there isn't any flux or other contaminants that could cause a bad connection.

It may also be worth checking the solder joints on the address select headers to make sure there are no dry joints or bridges and try some other address combinations to see if that fixes the issue.

Hi Many thanks for the answer. The one being preferentially recognized is the one on top. I changed the addresses but no effect.

I think solder joints on address select headers are fine.

Can it be a power issue?

I have Raspbian 2018-03-13 lite

It shouldn't be an issue with the power supply as the ADC Differential Pi only uses a few milliamps so it would have almost no impact on the power consumption of the Raspberry Pi. If the top board is being recognised then power should be getting to the one below as well but it may be worth checking the 5V and GND pads on each board with a multimeter just to be sure.

If you swap the two boards around so the one on the bottom is on top does it still only recognise the top board or is the bottom one recognised instead?

Hi,

Still the one recognized is the top board. Once, both boards were recognized but not longer.

Do you have a spare SD card you could use to install a fresh copy of Raspbian Linux just in case it is a software problem?

OK, now one of the boards is not recognized anymore...

I can measure 5V on the board and the solded connections look fine. Very strange indeed.

I will try a fresh installation of Raspbian and see what happens.

No improvement. I want to try the following according to your site:

"You may need to short the 5V and ground with a resistor to discharge the capacitors in order for the new addresses to be recognised"

What kind of resistor do I need to use?

Thanks.

It looks the only option would be to try another card... :-(

Holding a 100R or similar value resistor onto the power pins for a few seconds when the board is disconnected will completely discharge the capacitors which should reset the ADC. What sometimes happens is on the I2C addresses where one of the address pins is floating instead of being pulled high or low the ADC can get confused when you change the address as any internal capacitance on the address line could make it read as still being high or low.

You could try selecting addresses where the pins are pulled either high or low, so 0x68, 0x6A, 0x6C or 0x6E.

Do you have a second Raspberry Pi you could test the boards on just to make sure it isn't a fault with the I2C bus on the GPIO header?

It is unusual for both ADCs to stop responding, normally when a board stops working only one of the chips fails. When both fail to respond it usually means there is a problem with the I2C communication. You could try reflowing the solder on the small 6-pin MOSFET U6 as a dry joint on that or the multiway resistor R1 would stop it from appearing on the I2C bus. Heat each pin on the MOSFET in turn with a clean soldering iron tip and make sure you don't accidentally bridge the pins.

Have the ADC Differential Pis been used to measure voltages or is this the first time you have tried using the boards? If there was a voltage over 2.048V on any of the ADC inputs it could have damaged the ADCs which would cause them to act erratically or stop working completely.

Hi Andrew,

Many thanks for the input.

I tried a different Raspberry Pi but same story. I will try the resoldering you suggest in the next following days and will revert soon. I'm getting in any case more boards since I'm working in a project where I need to measure around 30 voltages more or less simultaneously. These are electrochemical potentials which will never overcome +- 1.5 V so I thought these boards were ideal to stack in a singel Pi.

Until which resolution do you think I could trust the readings? I see vriations of a few mV when testing a reference voltage.

Thanks in advance.

Felipe.

Question:

Do you offer pre-soldered boards even at an extra cost?

You should be able to trust the readings. The variations you are seeing are probably noise being picked up on the wires going to the ADC inputs or from the power supply. The MCP3424 ADC should have an internal noise level of about 5µV and we use tantalum capacitors next to each chip to reduce the power supply noise as much as possible but some switchmode power supplies can be fairly noisy. Using a separate low-noise linear power supply for any sensors or other devices you have connected to the ADC inputs should reduce the noise level.

We don't normally offer pre-soldered boards but as you are having trouble with the boards you already have I can solder the headers on and test the boards before sending them to you. If you could add a note with the order I will know which boards I need to solder.

Many thanks for your support. It is great you can solder the headers and test the boards. I'm located in the Netherlads and get the boards from Kiwi electronics. How should I make the order in this case then? Directly with you?

Thanks a lot in advance.

If you could place the order directly with us on this website then I can solder and test the boards before sending them to you.