Why 2.56V references?

I'm pretty sure that beyond 8 bits (so it should be 2.55V) any other argument is facile......

Unless someone else corrects me I think this might be my first attempt at a technical troll...... but you didn't read that, serious minds need to know.

DNA

Ignore this one, I'm pissed again.

DNA

munchie time is it?

martin

I am seriously disanointed.

Not only was I forced to consider buying various breakfast cereals and chocolate to melt and mix amongst them with some of those paper things but I was also forced to consider the fact that I would have to share my tasty things..... Or they would miraculously disappear according to the rule of 'Their own violition'.

Ah well. I suppose that means I'm not going to enjoy Chocolate Rice Crispies like what Mum used to make.... or Chocolate Cornflakes.... or Chocolate Fruit and Fibre.... or.....

It's a sad life when you have to restrict other peoples enjoyment because they take the piss.

DNA

Learn how to cancel your messages

--
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

Last time I tried to cancel the one stating I was a screaming alien animalist homosexual it didn't work..... Then the paranoia started.

Why does your address sound like something that would be easy for an extremely drunk person would say?

Castle in the garden of the street.

DNA

--
Why? 

At the very least Genome is entertaining.  In a nice way.

Consider: He writes: "Ignore this one, I\'m pissed again."

Having read it, however, one (maybe) becomes aware that it cannot be
ignored since it already burned its way into one\'s brain.  Not quite
where two ships tie up, but close...

Actually, I think it should. With a 2.56V ref, you'd get 2.56 V at 0xFF, which is actually 255.

So it should be 2.55V for exact 10 mV steps - my previous reply wasn't thought out properly.

Thanks, Rich

Are you thinking ADC or DAC ?

Hmmm.. turns out it may not matter. I looked up the AD5300 8-bit DAC and it also has the extra resistor. Total of 256 resistors, 257 reference voltages, the chip selects one of the first 256, ignoring the 257th (=Vref) one, giving output voltages of 0 through

255/256*Vref. For a Vref of 2.56, it would produce 0.01V steps.

Of course, both DAC and ADC chips can do whatever they want, just by putting in whatever resistor values they want in the divider ladder. For example, they could arrange the reference points semi-logarithmically, or sinusoidal, for special purposes. A semi-logarithmic ADC would be useful for audio; it would be more precise for quieter signals.

Genome a écrit :

Look at how R-2R networks are done:

R R R R Vref ___ ___ ___ ___ >-|___|-+-|___|-+-|___|-+.......+-|___|-+----. | | | | | | .-. .-. .-. .-. .-. .-. | | | | | | | | | | | |

In message , Genome writes

For an 8-bit DAC, data bits D7 to D0, input range 0 to 255.

Vout= Vref*(D7/2 +D6/4 +D5/8 +D4/16 +D3/32 +D2/64 +D1/128 +D0/256).

A single step of D0 changes Vout by Vref/256.

When D7-D0 are all 1's........

Vout = Vref*(128 + 64 + 32 + 16 + 8 + 4 + 2 + 1)/256 = Vref*255/256.

So to get a Vout(max) of 2.55V, Vref has to be 2.56V.

--
Tony Williams

Hint. An 8 bit R2R DAC, connected to 2.56v, will _not_ give 2.56v at 0xFF. It will give 2.55v. A R2R DAC, always gives as it's maximum output, one step _less_ than the reference voltage. Have a look at:

Which explains this. The same is also true of most designs of ADC. Hence the 2.56v reference.

Best Wishes

On a sunny day (Tue, 21 Nov 2006 11:29:49 GMT) it happened "Roger Hamlett" wrote in : .

How about this one, only 8 resistors, if you can get the values:

5V 8bit port D7-- 128R ----- D6--- 64R -----| D5--- 32R -----| D4--- 16R -----|_________ out, 0V at '0', 5 V at '255'. D3---- 8R -----| D2---- 4R -----| D1---- 2R -----| D0---- R -----

I have tried this with an up counter, nice ramp. Also used it for 6 bit video DAC.

Well, that was fun? Actually I wasn't so much bothered about whether it should be 2.56 or 2.55V, which is why I suggested there wasn't much reason for it other than the numbers work out......

The real question is is it really necessary to scale things like that? Most, all(?), times the voltage to be measured/produced is not going to be related to 2.56(2.55) volts and will need scaling externally to the AD/DA so, other than the bits matching up, the number chosen is kind of arbitrary......

I suppose it gives the digital folks a warm cuddly feeling though.

Unless of course someone knows different.

DNA

Well, that was fun? Actually I wasn't so much bothered about whether it should be 2.56 or 2.55V, which is why I suggested there wasn't much reason for it other than the numbers work out......

The real question is is it really necessary to scale things like that? Most, all(?), times the voltage to be measured/produced is not going to be related to 2.56(2.55) volts and will need scaling externally to the AD/DA so, other than the bits matching up, the number chosen is kind of arbitrary......

I suppose it gives the digital folks a warm cuddly feeling though.

Unless of course someone knows different.

Correction: Of course D0 should be D7 etc... LSB has the highest resistor.

Bitter experience: The lsb accuracy, resolution and linearity of a DAC is precious, easily lost by being casual.

Be nit-picky about doing the correct sum to calculate the number to send to the DAC, even to the extent of doing the calcs to more bits than required and then a round_to_nearest for the width of the final number to send to the DAC. And when you have done that do a test to make sure that the calculated number hasn't overflowed, otherwise the DAC will suddenly get set to a low value.

--
Tony Williams.

[Voltage reference]...

The forward voltage across a diode is a crude regulated voltage value. With amplification, and the use of mismatched currents, one can make a semiconductor reference that has much better regulation, and IF one sets the operating point at the voltage which is silicon's bandgap value (about 1.22V at room temperature) the reference has zero temperature coefficient.

These are called bandgap references, and all are multiples of 1.22V for best performance; the TL431 adjustable zener and LM185 reference and LM317 adjustable regulator all use bandgap references. Best Zener diode technology (lowest temperature coefficient) is at 5 or 6.8V, and if you h ave a +5-only power supply, the bandgap reference always wins.