• State Resolved
  • Locked Locked
  • Replies 11 replies
  • Answers 1 answer
  • Subscribers 81 subscribers
  • Views 7971 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Connection between AVcc and DVcc

seb
Genius 4170 points
Other Parts Discussed in Thread: TPS63031, MSP430F5659, REF3333

Hello all,

 

i just have found out about the issues concerning the AVcc connection and i cannot find a real answer for myself by reading all the older forum posts.

 

So the problem in my Layout is, I supply the AVcc-Voltage directly from one GPIO Pin, which gets put as an Output with logic "high".

 

So generally there should be around 3,3 V output from the GPIO-Pin to the AVcc Pin, but whenever i meassure it it says about 2,5 V.

 

Now strange things first:

1. I read a lot of things about possible oscillation, when AVcc and DVcc are not about euqal on startup, which it is in my case.

 

My code is working properly, i do not have any high currents in my device, my LPM3 is working just fine, so am I just a lucky guy, or why is it working for me.

 

2. Now i am also wondering if it is intelligent to supply the AVcc voltage form within the MSP GPIO, or should I as everywhere recommended, put the AVCC directly to my battery Vcc, just seperated with a LC-filter from my DVcc?

 

3. Another question came up. I do use an external reference vor ADC12, i do not have a buffer network attached to the Veref Pins, do I have to put one in there, or is it dependend on how much current i use of my reference?

 

Thanks for your time.

Best wishes.

Seb

  • 0
    TI__Expert 7125 points

    Seb,

    You should not be sourcing AVcc from a GPIO. AVcc will sink more current than the GPIO can source, and that is why the voltage drops. There are graphs in the datasheet showing this phenomenon. This is also why it works in LPM3, because almost no current is being consumed, so the GPIO is capable of sourcing it. That doesn't mean that it is a good idea however. Sourcing AVcc through the GPIO is a bad practice.

    It is typical to simply connect AVcc and DVcc. On the devices without a specific AVcc pin, the AVcc and DVcc pins are simply shorted on the inside. In most use cases connecting AVcc to DVcc directly will work fine. My guess is that it will be fine for you if you were originally trying to source AVcc from a GPIO. If for some reason you want AVcc and DVcc to be separate, you can do that, but don't use MSP GPIO.

  • 0
    Guru 227245 points

    Sorry, Mike, that's not exaclty right.

    the GPIO pins will definitely be able to source enough current. AVcc only draws one or two milliamperes even with all analog circuitry is active. Any GPIO pin will be able to source that.

    The reason why thi sis not allows is a different one:

    AVCC must not differ from DVCC by more than 0.2V. So it may not be switched off at all, and sicne the IO pins have a voltage drop, they won't provide enough voltage.
    If external AVcc is not sourced by a sufficiently high volage, a cross-current will flow internally from DVCC to AVCC, stressing the internal ciscuitry.

    The reason for separating AVCC and DVCC is not to provide two different voltages. Is is meant to decouple the two so that small changes in DVCC due to digital current ripple can be filtered-out and do not affect the analog circuitry.

    seb said:
    whenever i meassure it it says about 2,5 V.

    How do you measure? By a scope or by a multimieter?  A multimeter will show you the average voltage and you won't see the occurring high-frequency oscillation. This oscillation, if it happens, won't cause high currents. At least not what one would typically call 'high'. If power consumption increases by 0.1mA, this is extremely high for a low-current system that tries to reach <10µA total consumption. For a normal system that already consumes a few mA, it is negligible or not noticeable.

    2) you should do the RC filter. LC is usually not necessary.

    3) indeed, it depends on the impedance of your reference. The MSP doesn't draw much (see the datasheet). However, a capacitor couple (10µF tantalum and 100nF ceramic) ensure that the reference will be stable and withstand crosstalk as well as the small current changes during conversion (the MSP only draws from the reference during the 13 conversion cycles, not during the sampling time or when not converting).

    It depends on how high precision you need.

  • 0 in reply to Jens-Michael Gross
    TI__Expert 7125 points

    Jens-Michael,

    You are right, thanks for clarifying.

  • 0 in reply to MikeS
    Prodigy 120 points

    Hello: 

    I have a MSP430F5659 powered by a TPS63031 3.3V Buck-Boost Regulator and I have concerns about the reference voltage on the ADCa12. 

    Is it better to source AVcc pin from a 3.3 Reference voltage generator (such as REF3333)  rather than DVcc (Switching 3.3V) ? Is there any advantage on using external reference or the capacitors between DVcc and AVcc are good enough ?  Should be the REF3333 connected to the external Vref pin or to AVcc ? 

    thanks 

    Santiago 

  • 0 in reply to Santiago Barros
    Genius 4170 points

    Wow long lost post, did also forget to answer back then...

    If you want some precise ADC results you have to use some precise Voltage references, your ADC result will be dependent on those references.

    So you will have to decide for yourself if its important or not to measure precisly.

    Yout TPS volatge regulator will have a ripple for sure and that ripple will be in your ADC results as well, although the ripple will be different for different load scenarios, for example how much current will be drawn from your TPS, wheather the MSP430 is running at full speed or slowly, all this matters then.

  • 0 in reply to seb
    Prodigy 120 points

    That ripple would only affect the ADC results if  AVcc is selected as Vr+, but not if the Internal reference generator is active (@1.5 or 2.5V). Am I right? 

  • 0 in reply to Santiago Barros
    Genius 4170 points

    Thats of course right, if you use the internal 2,5V or 1,5 V you have to look into the Datasheet for further specifications and compare with your demands, perhaps its accuracy and long time stability, temperature etc will fit if not consider using an external reference which does.

    If you do not know yet, you can of course plan for both, install an external Vref. By c-code you can change the settings in the MSP430 for either external or internal Vref, as described in the FamilyUserGuide, so you could easily compare both or even more settings.

  • 0 in reply to Santiago Barros
    Guru 227245 points

    Santiago Barros said:
    That ripple would only affect the ADC results if  AVcc is selected as Vr+, but not if the Internal reference generator is active (@1.5 or 2.5V). Am I right? 

    Partly. The internal reference is also affected by changes on AVcc. Yet the effect is by magnitudes smaller.
    Another thing is the current ripple on the GND line. It causes GND to also change its voltage relative to Vcc - and to the reference. This is why there is a separate AVss. IT doesn't carry any current and  should be routed separately to the supply GND point. Remember that not only the upper voltage of a reference is important, but also the lower end.

  • 0 in reply to Jens-Michael Gross
    Prodigy 120 points

    Thanks Jens-Michael, you're right about considering the ripple on Vr- as well. 

    Jens-Michael Gross said:
     The internal reference is also affected by changes on AVcc. Yet the effect is by magnitudes smaller.

    Regarding the resulting ripple on the internal reference output, would be better then, to source the AVcc pin from the 3.3V external reference ? therefore, que 1.5, 2 and 2.5V internally generated voltages would be "cleaner" ? Is this worth it in terms of measurement improvement ?or the ripple in the internal reference is just insignificant even with a switching 3.3V AVcc? .  

    The other  problem i have doubts about is that there would be a small  varying voltage differences between AVcc and DVcc (while the microcontroller is in operation )  since one DVcc would be sourced from the TPS 3.3V switching regulator  and AVcc from a reference voltage. In the datasheet says that the difference between these two voltages should not be greater than 0.2V. Is this difference relevant ? 

    best regards 

    Santiago 

  • 0 in reply to Santiago Barros
    Genius 4170 points

    AVcc and Reference inputs are 2 completly different voltages supplied by you on completly different pins of the MSP430

    I you should supply Vcc with 3V3 and AVcc with 2V5 you can expect some excess current flowing and perhaps the dead of your MSP430.

    I did this once in my design nothing did break or overheat, but i drew way too much current, since I didnt expect it, so I already did this failure some years ago, learning by doing :)

    Another question comes into my mind with hte GND ripple, I have some batterie run device, I did couple AVss and DVss ( analog and digital GND) right beside the Batterie not right beside the swichting supply, but I always wondered what solution is best, any ideas to that? I did argue with current will be drawn from the batterie directly, so alongside the batteriepath there will be some voltage ripple or at least some voltage offset on the GND line, so I decided to couple them both at the very source.

  • 0 in reply to Santiago Barros
    Guru 227245 points

    Santiago Barros said:
    Regarding the resulting ripple on the internal reference output, would be better then, to source the AVcc pin from the 3.3V external reference ?

    You can do it, and indeed, it might give a cleaner result. However, you must ensure that AVcc and DVcc never drift apart for more than ~0.2V (diode voltage), even during power-up or power-down.

    Santiago Barros said:
    therefore, que 1.5, 2 and 2.5V internally generated voltages would be "cleaner" ? Is this worth it in terms of measurement improvement ?

    They would, but I doubt it will be worth the effort.

    BTW, I use an external reference of 2.495V+-0.5% (TS431BCX). It is a shunt regulator that draws a constant current from the supply, no matter how much the reference is loaded. It is more precise as the internal reference and can 'drive' up to 100mA (for external circuitry). of course, being a shunt regulator,  it  isn't exactly 'low power' :)

    Santiago Barros said:
    or the ripple in the internal reference is just insignificant even with a switching 3.3V AVcc? .  

    Can't say yes or no here. Depends. It is measurable and may affect the result. But depending on the other circumstances it often is insignificant.

    Santiago Barros said:
    In the datasheet says that the difference between these two voltages should not be greater than 0.2V. Is this difference relevant ? 

    Between the two, there are clamping diodes. As soon as the voltage difference is too large, current will flow from one to the other to equalize the two. Just like the voltage on any input pin must not exceed Vcc or Vss by more than 0.2V

**Attention** This is a public forum