View previous topic :: View next topic |
Author |
Message |
ac34856
Joined: 14 Dec 2009 Posts: 33 Location: Wales
|
AD7705 Conversion time (sigma-delta ADC) |
Posted: Thu Mar 11, 2010 12:24 pm |
|
|
I've been testing a design with an AD7705 ADC using the standard bit banged CCS driver code.
I noticed that as the DC voltage to the ADC goes up the conversion speed
drops dramatically by a factor of four. These devices have some on-chip buffer but this is disabled in software.
Anyone know what the issue is and a reasonable work around. The input
to the ADC is a diode clamped circuit with 200R to the ADC and a further 600R prior to the diodes. The diodes are fixed to the 2.5V reference from an AD780. |
|
|
FvM
Joined: 27 Aug 2008 Posts: 2337 Location: Germany
|
|
Posted: Thu Mar 11, 2010 12:38 pm |
|
|
Quote: | I noticed that as the DC voltage to the ADC goes up the conversion speed drops dramatically by a factor of four. | Unfortunately, these words makes no sense at all. What goes up and what drops down? |
|
|
ac34856
Joined: 14 Dec 2009 Posts: 33 Location: Wales
|
|
Posted: Fri Mar 12, 2010 5:32 am |
|
|
It gets slower as the input voltage increases ... very annoying.
It also locks up and the frequency with which it locks up depends
on the level of the input voltage.
At present my code forces a reset when there is a lock up but
thats slow and if something is wrong I'd like a fix.
The CCS drivers for this work "OK" but it sounds like some hardware
issue - in other words the input impedance has an effect on some internal
capacitors and the higher the voltage the slower this internal capacitance
changes .... thats my hypothesis. |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19513
|
Ttelmah |
Posted: Fri Mar 12, 2010 11:13 am |
|
|
I think it would help if you actually told us what you are doing.
There is nothing in the CCS code, that will change in performance with voltage. It clocks the same number of bits whatever the reading, at the same rate.
The rate _is_ changed, if you alter the gain setting, so if you are performing some degree of auto-ranging, and reducing the gain as the voltage rises, this would give the effect you are seeing. However it really does sound as if there may be an issue with the circuit.
So, post a simplified code, just doing the minimum of taking a reading the way you are doing it, using the standard library. Post the connections to the chip, how supplies are generated, how the PIC is clocked, what smoothing is present, where the protection actually connects, what voltage is involved, etc. etc..
Best Wishes |
|
|
ac34856
Joined: 14 Dec 2009 Posts: 33 Location: Wales
|
|
Posted: Fri Mar 12, 2010 11:24 am |
|
|
Yes but I was wondering whether there is some buffering setting or combination that would improve things. As per usual the gremlins have gone away and its working (mostly) correctly now. |
|
|
|