These figures are below the recommended minimum reference voltage of 
1V4. In practice it would be difficult to obtain such a low reference 
voltage with equivalently low noise. It is much easier to obtain good 
SNR at slightly higher voltages. The the programmable amp solution is a 
better option.

Al

Rolf F. wrote:
> Hi,
> 
> ok, but which resolution can be achieved with a reference voltage of 15 
> mV or 150 mV?
> Is it 1/10 and 1/100 mV?
> 
> Rolf
> 
> 
> onestone schrieb:
> 
> 
>>You have to differentiate betwen resolution and precision.
>>
>>The best way to approach this might be through the use of an external 
>>instrumentation amplifier, such as the AD62x series. Low noise, 
>>programmable gain. there are also some excellent interface devices from 
>>Maxim for this purpose, such as the MAX1452, which has digitally 
>>programmable gain. This device worked superbly in a weighbridge system 
>>for me. This method allows you to sample the signal, then adjust the 
>>gain of the amp for maximum dynamic range. Nominally the finest 
>>resolution of the MSP aa2 bit processors is VREF/4096 (being a 12 bit 
>>converter). However this precision requires trmendous attention to 
>>detail, especially with regard to calibration and temperature 
>>compensation. If care is taken you should be able to get close to 10.5 
>>bits of true precision. Thus with a 1.5V reference the ACTUAL PRECISION 
>>would most likely be 1.5V/1448, or 1.0358mV. However using a 
>>programmable gain amplifier that has also been calibrated will increase 
>>this resolution by typically up to 1000 times, although there would 
>>likely be some slight loss in absolute precision.
>>
>>Al
>>
>>Rolf F. wrote:
>> 
>>
>>
>>>Hi,
>>>
>>>what's the minimal resolution of the 10 Bit ADC in micro Volt?
>>>I want to measure little votages (some mV) with 10 bit resolution
and 
>>>therefore need to know the lower limit of the reference voltage (to
0 V 
>>>Vref-)  for a resolution of  10 bit.
>>>
>>>Rolf F.
>>>
>>>
>>>
>>>
>>>.
>>>
>>>
>>>Yahoo! Groups Links
>>>
>>>
>>>
>>>
>>>
>>>
>>>   
>>>
>>
>>
>>
>>
>>.
>>
>>
>>Yahoo! Groups Links
>>
>>
>>
>>
>>
>>
>> 
>>
> 
> 
> 
> 
> 
> .
> 
>  
> Yahoo! Groups Links
> 
> 
> 
>  
> 
>

Beginning Microcontrollers with the MSP430

Hi,

> The minimum recommended reference voltage is 1.4V,
so that 
> would be a resolution of 1/4096 * 1.4 ~= 342uV. In practice 
> it might be better to use a higher reference and amplify the 
> signal as Al says. The data sheet says"Lower reference 
> voltage levels may be applied with reduced 
> accuracy requirements.", but there
> is no guarantee of performance with lower reference voltages 
> than 1.4V. It 
> certainly will not damage
> the chip by applying lower reference voltages, but..
> 
> In some situations you can increase the resolution by 
> oversampling and 
> dithering, but IME,
> especially for load cells and thermocouples, you're far better off 
> amplifying the signal or using a
> micro with a PGA/delta-sigma ADC in it such as TI's 8051-based series.

The F42x has 16-bit Sigma-Delta ADCs.  It's preview and is for use in
energy meters; I do have information on it, but am not absolutely
familiar with the device.

Regards,

--
Paul Curtis, Rowley Associates Ltd http://www.rowley.co.uk
CrossWorks for MSP430, ARM, and (soon) Atmel AVR processors

At 11:09 AM 3/30/2004 +0200, you wrote:
>Hi,
>
>ok, but which resolution can be achieved with a reference voltage of 15
>mV or 150 mV?
>Is it 1/10 and 1/100 mV?
>
>Rolf

The minimum recommended reference voltage is 1.4V, so that would be
a resolution of 1/4096 * 1.4 ~= 342uV. In practice it might be better
to use a higher reference and amplify the signal as Al says. The data
sheet says"Lower reference voltage levels may be applied with reduced 
accuracy requirements.", but there
is no guarantee of performance with lower reference voltages than 1.4V. It 
certainly will not damage
the chip by applying lower reference voltages, but..

In some situations you can increase the resolution by oversampling and 
dithering, but IME,
especially for load cells and thermocouples, you're far better off 
amplifying the signal or using a
micro with a PGA/delta-sigma ADC in it such as TI's 8051-based series.

Best regards,

Spehro Pefhany --"it's the network..."            "The
Journey is the reward"
speff@spef...             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

I think Rolf means what the noise floor is of the overall converter system.
Do you mean, Rolf, what minimum reference voltage is needed to still have
the converter resolving xyz bits properly ?
I agree with Al though that it'd be a lot easier to give proper span with a
frontend
amp.
Still good question though, what's the lowest referece usable at which # of
bits ?

-- Kris

> Hi,
> 
> ok, but which resolution can be achieved with a reference voltage of 15 
> mV or 150 mV?
> Is it 1/10 and 1/100 mV?
> 
> Rolf
> 
> 
> onestone schrieb:
> 
> >You have to differentiate betwen resolution and precision.

Hi,

ok, but which resolution can be achieved with a reference voltage of 15 
mV or 150 mV?
Is it 1/10 and 1/100 mV?

Rolf


onestone schrieb:

>You have to differentiate betwen resolution and
precision.
>
>The best way to approach this might be through the use of an external 
>instrumentation amplifier, such as the AD62x series. Low noise, 
>programmable gain. there are also some excellent interface devices from 
>Maxim for this purpose, such as the MAX1452, which has digitally 
>programmable gain. This device worked superbly in a weighbridge system 
>for me. This method allows you to sample the signal, then adjust the 
>gain of the amp for maximum dynamic range. Nominally the finest 
>resolution of the MSP aa2 bit processors is VREF/4096 (being a 12 bit 
>converter). However this precision requires trmendous attention to 
>detail, especially with regard to calibration and temperature 
>compensation. If care is taken you should be able to get close to 10.5 
>bits of true precision. Thus with a 1.5V reference the ACTUAL PRECISION 
>would most likely be 1.5V/1448, or 1.0358mV. However using a 
>programmable gain amplifier that has also been calibrated will increase 
>this resolution by typically up to 1000 times, although there would 
>likely be some slight loss in absolute precision.
>
>Al
>
>Rolf F. wrote:
>  
>
>>Hi,
>>
>>what's the minimal resolution of the 10 Bit ADC in micro Volt?
>>I want to measure little votages (some mV) with 10 bit resolution and 
>>therefore need to know the lower limit of the reference voltage (to 0 V 
>>Vref-)  for a resolution of  10 bit.
>>
>>Rolf F.
>>
>>
>>
>>
>>.
>>
>> 
>>Yahoo! Groups Links
>>
>>
>>
>> 
>>
>>
>>    
>>
>
>
>
>
>.
>
> 
>Yahoo! Groups Links
>
>
>
> 
>
>
>  
>

You have to differentiate betwen resolution and precision.

The best way to approach this might be through the use of an external 
instrumentation amplifier, such as the AD62x series. Low noise, 
programmable gain. there are also some excellent interface devices from 
Maxim for this purpose, such as the MAX1452, which has digitally 
programmable gain. This device worked superbly in a weighbridge system 
for me. This method allows you to sample the signal, then adjust the 
gain of the amp for maximum dynamic range. Nominally the finest 
resolution of the MSP aa2 bit processors is VREF/4096 (being a 12 bit 
converter). However this precision requires trmendous attention to 
detail, especially with regard to calibration and temperature 
compensation. If care is taken you should be able to get close to 10.5 
bits of true precision. Thus with a 1.5V reference the ACTUAL PRECISION 
would most likely be 1.5V/1448, or 1.0358mV. However using a 
programmable gain amplifier that has also been calibrated will increase 
this resolution by typically up to 1000 times, although there would 
likely be some slight loss in absolute precision.

Al

Rolf F. wrote:
> Hi,
> 
> what's the minimal resolution of the 10 Bit ADC in micro Volt?
> I want to measure little votages (some mV) with 10 bit resolution and 
> therefore need to know the lower limit of the reference voltage (to 0 V 
> Vref-)  for a resolution of  10 bit.
> 
> Rolf F.
> 
> 
> 
> 
> .
> 
>  
> Yahoo! Groups Links
> 
> 
> 
>  
> 
>