rf52840 saadc vreference

[ziv123]

hello Nordic

i work with nrf52840, SDK 16.0

this issue was discussed in the forum but i wish to get a clearer answer

to my understanding (relaying on the saadc example in the SDK), it is possible to have only 2 Vref, one is the internal which is 0.6v and another is the VDD/4

both force me to use gain of 1_6, or 1_4 respectively. 

but, using gain of less then 1, i loos information, like loosing resolution on camera. this is not good for my application, what solution can Nordic offer to measuring 0-3.3v with gain of 1 ?

(i use 14bit resolution, and my sample rate is 32KHz)

(p.s. i added audio and music to the tags of this topic cause i am actually measuring sound)

best regards

Ziv

[Jared]

Hi,

You're correct in that the internal reference should be either VDD/4 or the 0.6V but the gain is not dependent on the internal reference. You can chose any gain from 1/6 to 4, see the config register. 

regards

Jared 

[ziv123]

can you explain what's the values you expect to read when v ref is VDD/4 and gain is 1 ? because to my understanding it will not be in the range equivalent of 0-3.3 v

[Jared]

Hi,

With a gain of 1 and using VDD/4 as internal reference you'll get an input range +- VDD/4:

Input range = (+-VDD/4)/Gain    ------------>  Input range = +- VDD/4*1  = +- VDD/4

A lower gain will enable you to have a higher input without going over the internal reference. If you want an input range up to VDD = 3.3 then you have to lower the gain to 1/4 if the SAADC is using VDD/4 as it's internal reference:

Input range = (+-VDD/4)/Gain    ------------>  Input range = +- (VDD*4)/(4*1)  = +- VDD

regards

Jared 

[ziv123]

hi 

sorry to say but this discussion does not advance me in any way, i already know that in order to get 0-3.3v range i have to use gain 1/4 if my V ref is VDD/4, the point was that in that way i loos some accuracy of input, the input was reduced to VDD/4 (~3.3/4 ~0.825v) and bringing it back to 3.3v is done by some calculation, like a camera chip that compensate on less accurate lens or sensor, so missed pixels are calculated in and were not captured by the input itself (hope i made my self more clear with this example) 

so my wonder was that with the nrf52 there is no way to get the full original range of input and my question was am i missing something ?

[Jared]

Hi,

I understand. The main issue here is that you would want the internal reference to be at 3.3 V. This would mean that VDD would have to be 13.2 V if you chose VDD/4 as the internal reference. This is obviously way over the maximum ratings for VDD at 3.6 V. In short we don't support this. The internal reference can either be VDD/4 or 0.6 V. While VDD can be up 3.6V

best regards

Jared