Hello, I have been designing a power factor corrector circuit. For the control implementation I have decided to use ti's tms320f2812 dsp. I have a Spectrum digital F2812 eZdsp board revision C. I have designed a signal conditioning circuit for ADC inputs using TL084CN opamps (bipolar, short cct protected, JFET input). I have powered the signal conditioner board with external isolated power supply and tested independently. But when I have connected my board to ADC inputs of the DSP board, ADC has failed to work. Every other functions of the DSP chip is working but the ADC is not. I have guessed that the MUX in the ADC input has failed and tested the B port. B port of the ADC is working but there is unacceptable noisy signal. Now I have a new DSP board. Do you have any idea how can I solve my signal conditioning problem? Do i need to buy signal conditioning boards (TI has SLAU105)?
F2812 ADC Signal Conditioning
Started by ●February 13, 2005
Reply by ●February 15, 20052005-02-15
Just wondering if your TL084 opamps were able to drive the ADC i/p beyond
the ADC's own supply rails (due to lack of required hardware clipper cct.).
If so, you have probably fritzed the ADC's input protection diodes, which
can only cope with so much.
You may need non-proportional spacing font to see the following:
___ +ve ADC supply
|
_
^ DIODE (1N4001 or sim)
_
1K | |
FROM TL084 ---->-----\/\/\/\--+-------| ADC i/p
| |
_
^
_ DIODE
|
--- -ve ADC supply
The above cct should be simple to incorporate.
The resistor should be as low as possible to preserve ADC accuracy, but high
enough to limit the max TL084 o/p current. If the ADC -ve supply is GND,
then connect the lower diode to GND.
Jim Adamthwaite
Reply by ●February 16, 20052005-02-16
Thank you very much for your reply. I have used fast switching protection diodes (1N4148) to protect ADC input. I have used a serial resistor to limit the current drawn by the ADC input. Although these I have burned the ADC input. ADC operation is not correct because of the failure of interior analog multiplexer. You may wonder why I have used a separate power suply and not used the DSP card's own supply. This is because the opamps does not have a rail-to-rail operation feature and does not have linear characteristics near the supply voltage. So I decided to use a negative power supply to operate opamps linearly near GND and Vcc. Is the problem using a second power supply? But I must remind that my power supply which is powering the opamps is isolated and the GND is tightly connected to DSP card's analog GND. Thank you for your time. Engin Ciftci
Reply by ●February 16, 20052005-02-16
engcif wrote:> Thank you very much for your reply. > > I have used fast switching protection diodes (1N4148) to protect ADC > input. I have used a serial resistor to limit the current drawn by the > ADC input. Although these I have burned the ADC input. ADC operation is > not correct because of the failure of interior analog multiplexer. > > You may wonder why I have used a separate power suply and not used the > DSP card's own supply. This is because the opamps does not have a > rail-to-rail operation feature and does not have linear characteristics > near the supply voltage. So I decided to use a negative power supply to > operate opamps linearly near GND and Vcc. > > Is the problem using a second power supply? But I must remind that my > power supply which is powering the opamps is isolated and the GND is > tightly connected to DSP card's analog GND. > > Thank you for your time. > Engin CiftciWith thee arrangement as I understand it, you can drive the converter's input well below it's negative rail. That can blow it. Regardless of the power supply arrangement feeding the op-amp, the clamp diodes should be referenced to the converter's supplies. Fast switching diodes are not usually needed, but low-drop diodes (such as Shottkys) are a good idea. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●February 17, 20052005-02-17
To make it clear I have drawn a simplified version of my cct. below. I
have checked the signal level in the ADC input by using an isolated
scope and it is not exceeding or descending the limits of ADC input.
In my opinion the problem can be caused by the connection between DSP
card's Analog GND and my power supply's GND. A current flow between
these two can make this failure. But I am not sure. Does anybody have
an idea?
Thank you again.
.----------.
| |
~ ---- ---| Rectifier|-->--o +3.1V
)||( | & |
AC )||(---| Linear |-->--o GND ---.
)||( | Regulator| |
~ ---- ---| |-->--o -3.1V |
| | |
'----------' |
|
o +3.1V .-----------. |
| | | |
+---------. | VrefLO o----+
| | | | |
Signal | - | An. GND o----.
Input |\| ^ | |
o---|+\ R | | |
| >-/\/\/--+------o ADC IN |
o---|-/ | | |
|/| - | |
o ^ | F2812 |
-3.1V | | DSP Board |
GND o | |
'-----------'
Engin Ciftci
Reply by ●February 17, 20052005-02-17
I forgot to put a link for the cct. image. If you have problems in seeing the above cct. (drawn with proportional space), here is a link that you can follow. http://www.baskent.edu.tr/~engcif/cct1.gif
Reply by ●February 17, 20052005-02-17
engcif wrote:> To make it clear I have drawn a simplified version of my cct. below. I > have checked the signal level in the ADC input by using an isolated > scope and it is not exceeding or descending the limits of ADC input. > > In my opinion the problem can be caused by the connection between DSP > card's Analog GND and my power supply's GND. A current flow between > these two can make this failure. But I am not sure. Does anybody have > an idea? > Thank you again. > > .----------. > | | > ~ ---- ---| Rectifier|-->--o +3.1V > )||( | & | > AC )||(---| Linear |-->--o GND ---. > )||( | Regulator| | > ~ ---- ---| |-->--o -3.1V | > | | | > '----------' | > | > o +3.1V .-----------. | > | | | | > +---------. | VrefLO o----+ > | | | | | > Signal | - | An. GND o----. > Input |\| ^ | | > o---|+\ R | | | > | >-/\/\/--+------o ADC IN | > o---|-/ | | | > |/| - | | > o ^ | F2812 | > -3.1V | | DSP Board | > GND o | | > '-----------' > > Engin CiftciWhat I wrote before: Does the ADC have a negative supply? If not, you can drive its input beyond the negative rail. Depending on the value of R, that could blow it. I think you're missing something. What does the spec sheet say about the converter's allowable input range? Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●February 17, 20052005-02-17
The specs says: (sprs174k.pdf - page 135) Analog input voltage (ADCINx to ADCLO) 0 to 3 V Additional note says: Voltages above VDDA + 0.3 V or below VSS - 0.3 V applied to an analog input pin may temporarily affect the conversion of another pin. To avoid this, the analog inputs should be kept within these limits. Which is 3.3 to -0.3. I am sure that I haven't applied a negative or a possitive voltage above or below its limits. If I assume that there is a voltage difference between the two grounds (mine supply and DSP supply), input can go below or above the limits. You may be right but it also says that this will temporarily affect the conversion. But I am sure that the input signal is in the limits. I have observed the input signal many times using a 500Mhz scope to make sure there are no sparks or jumps in the input signal. Are you using the same DSP card or have you used it before or a similar one? Should I use a signal conditioner card like TI's DAP Signal Conditioning Board (Unipolar) DAPSIGCNDBRDUNPEVM (link below) ? http://focus.ti.com/docs/toolsw/folders/print/dapsigcndbrdunpevm.html Thanks. Engin
Reply by ●February 17, 20052005-02-17
engcif wrote:> The specs says: (sprs174k.pdf - page 135) > Analog input voltage (ADCINx to ADCLO) 0 to 3 V > Additional note says: > Voltages above VDDA + 0.3 V or below VSS - 0.3 V applied to an analog > input pin may temporarily affect the conversion of another pin. To > avoid this, the analog inputs should be kept within these limits. > > Which is 3.3 to -0.3. I am sure that I haven't applied a negative or a > possitive voltage above or below its limits.What makes you sure? The schematic shows the input from an op-amp with a -3.1 supply. Even a poor op-amp supplied like that can put out -2.5 volts. That's far out of spec.> If I assume that there is > a voltage difference between the two grounds (mine supply and DSP > supply), input can go below or above the limits. You may be right but > it also says that this will temporarily affect the conversion. But I am > sure that the input signal is in the limits. I have observed the input > signal many times using a 500Mhz scope to make sure there are no sparks > or jumps in the input signal.Are you sure the input signal never goes below -.3V despite the op-amp that drives it being capable of going to -3V?> Are you using the same DSP card or have you used it before or a similar > one?No.> Should I use a signal conditioner card like TI's DAP Signal > Conditioning Board (Unipolar) DAPSIGCNDBRDUNPEVM (link below) ?I don't think you need it. In your diagram you show a protective diode to ground. A few millivolts difference between that ground and the converter's won't blow anything, but it will create a reading error of that amount. However, if through a wiring error you connected the diode to the symmetrical location, -3.1V, that would explain much. As it stands, the drop across the "protective" diode exceeds the allowed. .----------. � � ~ ---- ---� Rectifier�-->--o +3.1V )��( � & � AC )��(---� Linear �-->--o GND ---. )��( � Regulator� � ~ ---- ---� �-->--o -3.1V � � � � '----------' � � o +3.1V .-----------. � � � � � +---------. � VrefLO o----+ � � � � � Signal � - � An. GND o----+ Input �\� ^ � � � o---�+\ R � � � � � >-/\/\/--+------o ADC IN � � o---�-/ � � � � �/� - � � � o ^ � F2812 � � -3.1V � � DSP Board � � � � � � � '-----------' � � � +-------------------+ Reconnecting the diode as shown makes more sense to me. Shottky diodes will keep the drops low. The drop across a high-speed switching diode will exceed half a volt. (We usually figure 0.7.) A ground clamp useful with high-drop diodes can be built like this: +V ------/\/\/\------+---->|------> Clamp point; won't go below ground � +---->|------o Tie to ground. The diode drops roughly cancel. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●February 17, 20052005-02-17
Thank you for reply.
I was uncomfortable with the 1N4148 as you have mentioned. I thought
they will be sufficient to protect my ADC. Using a Shottky diode can
solve my problem. I will check my resources to find some. If you have
in mind, can you offer me a diode which has low Vdon voltage?
I will redesign my signal conditioning cct by replacing my TL084 opamps
with the TLC2274 or OPA4342 which are single supply and having
rail-to-rail operating feature. By this way I don't need a negative
power supply and I will use DSP card's own supply. I hope this will
remove all doubts.
One additional question. Can a transformer have a leakage current
between the isolated sides? (see below drawing) If this occurs, a
current will flow between DSP board's GND and my cct. board's GND. This
current will badly effect the ADC. The measured voltage in both GND
will be equal but the current sunk from the ADC GND will defect the
ADC. Am I right?
~ o---- ---o ~
)||(
AC In )||( Ac Out
)||(
)||(
~ .---- ---. ~
| |
| ---->>--- |
o Leakage? o
GND 1 GND 2
Image file: http://www.baskent.edu.tr/~engcif/cct2.gif
Engin
