# DAC zero code = zero volts, best way??

Discussion in 'Electronic Design' started by John, Jul 24, 2006.

1. ### JohnGuest

Instead of manually adjusting a pot (voltage divider) to set the
desired current level through a servo loop (op-amp driving a FET), I'd
like to use a PIC controlling a DAC. My problem is that even
rail-to-rail DACs only go down to a few millivolts, not true zero
volts, when the DAC code is zero. This few millivolts prevents me
from setting my current levels low enough for my use. I have a bunch
of servo loops and it adds up quickly.

I was thinking that I could have the PIC multiply the desired DAC
output code by X and then set up a resistor voltage-divider on the
output of the DAC to divide the voltage by that same X amount (before
going on to the servo loop's op-amp). This would reduce the zero-code
output voltage level by X times and give me a much lower minimum
current level for each servo loop. Using my existing setup, I can set
X = 30 or so, maximum.

Is this a reasonable way to reduce the effect of the DAC output not
being able to reach a true zero volts? Is there a better way? Level
shifting of some kind?

Thanks!
-- remove SPAMMENOT for e-mail responses --

2. ### JohnGuest

Forgot to mention, I have a +5V supply to the PIC and DAC and a +1V
reference to the DAC. No negative voltages available.
-- remove SPAMMENOT for e-mail responses --

3. ### Phil HobbsGuest

Hmm. Why not post a schematic on a.b.s.e?

Cheers,

Phil Hobbs

4. ### Ben JacksonGuest

["Followup-To:" header set to sci.electronics.basics.]

I can't tell if your problem is that you can't hit zero or that
you just can't get *close enough* to zero. In other words, is
the code '1' too high (or too coarse) or is '0' too high for '0'.

If you want to scale the output ('0' still not 0V, but '1' lower
and each step finer) then an output divider is fine. However, mind
the input impedance of your next stage. If you are relying on a
buffer built into the DAC to drive your next stage, the relatively
high output impedance resistor divider is going to need another
buffer. That buffer, even if *IT* is rail-to-rail is going to cause
you more offset/scale problems near 0 again.

5. ### Tim WescottGuest

You mention in your add-on that you have no negative voltages available.
You may find that the easiest and most compact solution you can do is
to supply a negative rail so you _can_ bring your voltages down to some
true zero.

An alternative is to supply yourself with a virtual ground. Normally
I'd choose a voltage that's 1/2 my supply voltage (less if I'm using el
cheapo op amps like the LM358 that go close to ground but only approach
the top rail to within 2V or so). It sounds like you don't need to
servo on the negative side, so you may be able to do OK with a 0.5V, or
some other low value which you know your DACs can always fall below. If
you need a really precise zero you'll have to develop this voltage
carefully.

If you don't need an exact amount of shift you can drive a resistor
through a diode. This will always put you one diode drop below the DAC
output, but you will learn amazing things about how much a diode drop
can vary with current and temperature. This technique also won't work
if your following stages are dumping current into the resistor.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google? See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

6. ### Frank BemelmanGuest

Why not make a small negative voltage? Have one of the PIC outputs
generate a square wave, couple the output with a nice cap and
follow with a rectifier, regulator etc.

7. ### JohnGuest

Yea, I shold have posted the circuit. It's Win Hill's recommended
circuit from a while ago and has worked well when manually adjusting
the pot off the reference.

I'm using a .005-ohm shunt so if the DAC (MCP4921) can only drop to
10mV with a DAC code of zero (instead of 0.000V) then each FET will
set up to pass (0.010mV/0.005 ohms) = 2 amps per FET as the minimum
current level I can set. Too high. I can't increase the resistance
of the shunt anymore (current level is too high) so I was thinking I
could drop the DAC's min. output voltage with a voltage divider.

Using my example of divide-by-30 resistors on the output of the DAC,
the 10mV output for a DAC code of zero would drop to 0.33mV. This
would give me a min. current level of 67mA through each FET. Not
ideal, but I could live with that.

I'll look over the other ideas brought up in other posts. I'm sure
I'll have a couple more questions though.

SHUNT = 0.005 ohm, 1%, 5W
R6 = 1K, 1%
.. Vcc R2, R3 = 49.9K, 1%
.. | R4, R5 = 10K, 1%
.. R1
.. |
.. +------, Vcc ,-----+-----
.. | | | LT1013 | x |
.. 1V REF POT <-+--R2--+-----|+\ D |
.. | | | | | >--+--R6-- G |
.. -+------+-, | | ,--|-/ | S |
.. | | | | | === C1 | |
.. low-level | | | | | | | |
.. signal GND +--|--R3--|--+----|---+---R5----+ |
.. | | | | | |
.. | | | | SHUNT |
.. | | | | | y |
.. | | '--R4---+-------------+-----|-+--- RTN
.. | | | |
.. | | | |
.. | | Vcc ,-----+ |
.. | | | | | |
.. | +--R2--+-----|+\ D | |
.. | | | | >--+--R6-- G | |
.. | | | ,--|-/ | S | |
.. | | | | | === C1 | | |
.. | | | | | | | | |
.. GND +--|--R3--|--+----|---+---R5----+ | |
.. | | | | | | |
.. | | | | SHUNT | |
.. | | | | | | |
.. | | '--R4---+-------------+-----|-+
.. | | | |
.. : : etc etc - N stages : :
.. : : : :
-- remove SPAMMENOT for e-mail responses --

8. ### Rich GriseGuest

I vote with this - that's what I'd do if I needed a "real" zero.
Admittedly, the last time I did this I already had the negative
supply, but as Frank writes, enough negative volts for bias should
only cost a few parts.

Good Luck!
Rich

9. ### Tim WescottGuest

"A rectifier"?

Two caps, two diodes:

||
o----||--o--->|---.
|| | |
| V
|
'---|<----o-----O -Vout
|
===
|
V

With reasonable op amps and not-too-strict requirements you may not need
to regulate, either. Check you power supply rejection ratio, and
consider a simple RC filter to decouple the ripple.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google? See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

10. ### JamieGuest

you need some small amount of - Voltage.

11. ### JamieGuest

you need to use a + - rail supply so that you can balance
the op-amp at zero.

12. ### Frank BemelmanGuest

Oh sure, that must be a language problem. Rectifier and diode are the same
thing for me. I certainly didn't mean a bridge-rectifier. There is a
nice improvement half way down this page:
http://murray.newcastle.edu.au/users/students/1999/c9221792/doubler.htm

13. ### JohnGuest

You may find that the easiest and most compact solution you can do is
I'm beginning to think that may be the case. The neg voltage ripple
might be a problem though but testing will tell.

Yea, might get ugly quick though. A quick&dirty true neg. voltage or
fully regulated one (if needed) seems to be my best bet so far.

Thanks!
-- remove SPAMMENOT for e-mail responses --

14. ### JohnGuest

Why not make a small negative voltage? Have one of the PIC outputs
Thanks for the recommendation and the ckt! I'll check some specs and
whip up a prototype to test.
-- remove SPAMMENOT for e-mail responses --

15. ### PeteSGuest

If you use a charge pump at -5V (you have +5V available) and then
regulate to -3.3V (or some other arbitrary value more suitable in your
view, but at least 1.5V away from the rail) you should be able to keep
the negative rail ripple quite manageable.

I have found the charge pumps work perfectly well from -Vin to -Vin + 1
across the load range of perhaps 100mA. With a followon LDO, that
should work perfectly well.

Cheers

PeteS

16. ### almoGuest

I'd just tie one of the unused I/O pins to the DAC output via a
resistor. I think you can set the I/O pin to high impedence. Then,
when you want to go to 0 volts (ground,) just let the second I/O pin be
a digital output and set it to "0" and draw the voltage to ground (sink
the current.)

Almo

17. ### Frank BemelmanGuest

First of all you are a stupid top poster, and secondly the OP
wants to get rid of the last few mV so your idea is totally
worthless.

PLONK and good riddance.

18. ### almoGuest

Hey...I'm only in 7th grade. Cut me some slack.

Hey...I'm only in 7th grade. Cut me some slack.

Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Continue to site