EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: TinkeringSteve on July 13, 2013, 01:46:36 pm
-
(yeah I have made the same post at Analog Devices EZ board)
http://www.analog.com/static/imported-files/data_sheets/AD5290.pdf (http://www.analog.com/static/imported-files/data_sheets/AD5290.pdf)
I recently got an AD5290 for evaluation, I put it in a test circuit with +/- 15 V analog power supply rails, and hooked the SPI lines up to an MCU devboard.
There is one GND connector, pin-4, which is designated as "digital ground".
However, I only get sensible output, i.e. measured DC resistance via DMM between terminals W and B (pin 10 and 2), if I also connect the digital ground to the 0V center of my analog dual supply.
If that's the case, I can adjust the output resistance to the whole specified range.
If it's not connected, the output was seemingly random the other day, today it seems to remain at about 25 kOhm.
Just in case my PSU is to blame, I also tried it with two 9V batteries, same result. If the zero volt center is not connected, I measure nonsense, if it is connected to "digital ground", it works perfectly.
The test circuit consists only of the IC on an adapter PCB, test leads, power supplies and the microcontroller board.
Now according to the datasheet this shouldn't be the case, and it also seems weird, since some people would probably not like to connect their analog ground to the digital one at that point.
So I take it this simply should not be the case.
What could be wrong here? Could this behavior be a result of damage due to experimenting with the chip? (I remember having grilled one when I first tried them out many months ago, corrected the circuit (a bigger circuit, not this here test circuit), and then put things aside for a while, I'm not sure why that one now should be damaged, but who knows. Unfortunately I only have those 2 specimen of the IC.
-
Now according to the datasheet this shouldn't be the case,
Can you quote the bit of the datasheet that say you can float the digital ground with respect to the analogue ground? There does appear to be some lee way in that digital ground does not have to be at the same potential as the analogue ground, but that's not the same as allowing it to float.
The datasheet does specifically say (page 17):
"The ground pin of the AD5290 is a digital ground reference.
To minimize the digital ground bounce, the AD5290 digital
ground terminal should be joined remotely to the analog
ground (Figure 32)."
-
Now according to the datasheet this shouldn't be the case,
Can you quote the bit of the datasheet that say you can float the digital ground with respect to the analogue ground? There does appear to be some lee way in that digital ground does not have to be at the same potential as the analogue ground, but that's not the same as allowing it to float.
The datasheet does specifically say (page 17):
"The ground pin of the AD5290 is a digital ground reference.
To minimize the digital ground bounce, the AD5290 digital
ground terminal should be joined remotely to the analog
ground (Figure 32)."
Thanks for the quick reply,
Okay, that was an oversight, I was too focused on the "digital ground" in the pin explanation table. I'm not very experienced with those kind of chips, and the number of data sheets I have read so far may still be in the lower two-digit range, and most of them some MCU ;-)
BUT the funny thing is, in my bigger circuit, those analog & digital grounds *are* "remotely" joined, and it doesn't work, but when I give the ground pin of the AD chip a dedicated wire to the analog ground connector on the board, it works.
And
Could you explain to me in what way the pin could float here? Certainly not digitally, since my SPI out is never floating, but then I don't understand how the digital ground could or should have anything to do with the analog part so that they could "float apart" or something - but what do I know, maybe you could shed some light on it?
Can/should I check whether it does that with my oscilloscope?
I guess I might also change my test programm to *not* continuously send the current pot value, to see whether the output value also recovers with the ground on/off thing if it doesn't send constantly...
-
Ground noise.
Read up on the problems with non-perfect ground connections, and self inductance. A 'remote' ground tie doesn't mean a meter away, up on the shelf with your PSUs, it means elsewhere on the same PCB a few cm away.
Having a ground wire running halfway around the work table causes it to act as a pickup coil, which continuously changes the voltage difference between the DGND and Vss pins. Apparently the chip gets confused by this, and interprets the SPI data from the MCU with this noise voltage layered 'on top' of the digital voltages.
Try using your scope, AC coupled, to inspect the noise voltage between the DGND and Vss pins, without connecting them locally, Ie. like when you are having problems. This ought to be a stable DC voltage, so anything you see is picked up noise.
Additionally, by your description, it sounds like you are not bypassing the Vdd and Vss pins of the AD5290. That may also cause serious difficulties, particularly if you are constantly hammering it with data.
-
Oh yeah you're right, I forgot the bypass caps, didn't fit on the tiny adapter PCB, I'll put them directly around the holes where the adapter's pins go on the mother board.
I'll look whether I can see any noise on the scope, thanks
-
I just wish we had digital pots where the 3 pot terminals could go +- (and not just 0V to VCC) for use on split supply opamps.
It's a pain.
That one here can actually do that, as can others which are probably cheaper if they're made for just +/- 5V or so,
the one I'm using here is for +/-15V which I need for my project.