what are the symptoms of a ADC (MCP3202, MCP3204 or MCP3208) going bad?
courtens
Posts: 101
I have a control box that uses 4 MCP3208 chips. Some time ago I switched out my power supply and did not check how clean the 3.3VDC looked on the scope. Now, 4 months later things have started to act strangely. My control unit started to act up. and sure enough; the power supply had a extremely strong ripple (well over 0.7VDC), but a capacitor fixed that. Could that ripple have caused the ADC to go bad?
After doing some troubleshooting I discovered that one of the ADC is bleeding the CLK signal from pin 7 right back out through Vin on pin 2. (Note, for reference, image of ADC is a MCP3202, not a MCP3208). This bleed through did not look right. Is this a symptom of a ADV gone bad? Vin is sourced by a 100K pot. (I should have used a 10K, but it used to work with a 100K pot, just fine.)
In addition, I noticed a very strange time related phenomena. Over time, Vin slowly creeps down to 0VDC. When I set the pot to somewhere in the middle (1.6DC) and turn the unit on, at first it's all good; but about 2-3 minutes later, Vin has dropped significantly, and keeps on dropping all the way to 0VDC. It stays put at 0VDC and now to get back to 1.6VDC, I need to turn the pot to about 3/4 of a turn towards 3.3VDC. I can repeat this, by turning the unit off. I can't understand what is happening.
After doing some troubleshooting I discovered that one of the ADC is bleeding the CLK signal from pin 7 right back out through Vin on pin 2. (Note, for reference, image of ADC is a MCP3202, not a MCP3208). This bleed through did not look right. Is this a symptom of a ADV gone bad? Vin is sourced by a 100K pot. (I should have used a 10K, but it used to work with a 100K pot, just fine.)
In addition, I noticed a very strange time related phenomena. Over time, Vin slowly creeps down to 0VDC. When I set the pot to somewhere in the middle (1.6DC) and turn the unit on, at first it's all good; but about 2-3 minutes later, Vin has dropped significantly, and keeps on dropping all the way to 0VDC. It stays put at 0VDC and now to get back to 1.6VDC, I need to turn the pot to about 3/4 of a turn towards 3.3VDC. I can repeat this, by turning the unit off. I can't understand what is happening.
Comments
I'm guessing, but if Vin was ever over Vref it may cause some internal damage that could cause /CS to show up where it shouldn't. It might be time to replace it, if the wiring is good.
How are your grounds setup, are digital and analog on the ADC tied together? If so it might be too noisy, especially if you have more than one ADC.
Check this article - http://www.analog.com/en/analog-dialogue/articles/grounding-again.html
All the grounds are tied together. Thanks for the helpful link.
Per Microchip, any input or output that exceeds Vdd + 0.6V may cause permanent damage.
I am guessing from the image and description that Vdd = Vref, and that each Vin source was also powered by Vdd (albeit with some time-shifting capacitance). A power supply ripple "well over 0.7Vdc" could certainly create a damaging scenario.
Do all 4 chips behave/fail the same way ?
What you describe is roughly consistent with a bad or damaged Vdd connection.
With no proper power, CLK ends up 'powering' the device, but rather poorly.
Time related issues could also be something now floating.
Maybe I should have sourced all the pots with lower voltage to be on the safe side with Vin not maxing out!
I replaced one - which fixed the problem for that part of the unit. Finding out which other IC's to replace should not be too difficult...
My longs line is 76mm (not overly long)
I have been having some ground issues. My unit (control-box) is hocked up to multiple devices. The motor and strobe "network" are isolated -- they have their own power supply and do not share ground. But a TV (via Video) and a camera (via HD analog Y) have the ground tied together. The camera and the control-box share the same power supply. The camera is also hocked up to a PC (via HD-SDI), and the PC sometimes has a problem locking up to the camera's HD-SDI signal. It can take multiple very frustrating attempts (turning thins on and off) to get it to find the signal. I found that disconnecting the camera from the control-box and completely unplugging the TV's power cord can help the PC fined the camera -- sometimes. But whatever is going on is frustrating.
For now I will replace the faulty ADC's and lower the Vin's voltage, in order to create a safety margin.
How many amps at 3.3 V do you need?
There are a lot on eBay.
As for the long traces all they need is some kind of series resistance or termination. If you always allow for extra parts on your PCB design then it doesn't mean that you have to place them, but it does give you the option of doing so later. Once you insert series components though it's only a matter then of the value that you use.
Btw, I can't help but ask why so many SMD PCB designs bother to try and display the component designators on the silk screen? This was the standard back in the old DIP days when components were far larger but they are hardly legible and not at all useful on a final PCB although I do have designators on an inactive mechanical layer for reference but normally I print the component values along with the silk screen and pad layer plus outline as a zoomed print to fit a printed page. This is then used as a clear to read guide for manual pcb assembly and the pcb itself stays nice and uncluttered with more room for useful information about signals and labels etc.
I have a mcp3208 spin object that I run using an lcd screen and Labview either at the same time or independent. It displays 4095 total bit for max. Voltage. So if I have an over voltage or cable glitch I can see it on the bit display. If the bits freeze or all channels start reading max bits at the same time when only testing one channel I know there is a problem. It Could be cable noise, over voltage, bad IC etc. I like dip so I can just swap it out.
In your case if you have another board and bench power supply to test it.