TLC548 ADC question

This is the place for any magazine-related discussions that don't fit in any of the column discussion boards below.
Post Reply
Spymat
Posts: 6
Joined: Mon Jan 21, 2002 1:01 am
Contact:

TLC548 ADC question

Post by Spymat » Mon Jan 21, 2002 3:14 pm

When using a serial ADC TLC548 sampling a signal between 1 and 5 volts, my output jitters by 1 and 3 units. I've put a 1uC cap on the input to get rid of high freq noise.<p>Is this normal operation for the ADC or is there a way to improve performance? I mean get a more dc output?<p>Thanks

russlk
Posts: 563
Joined: Thu Dec 06, 2001 1:01 am
Location: New Hampshire
Contact:

Re: TLC548 ADC question

Post by russlk » Mon Jan 21, 2002 7:08 pm

This 8 bit ADC should give 20 mV sensitivity but if the analog and digital grounds are not well seperated, ground noise will cause the effect you see.

Spymat
Posts: 6
Joined: Mon Jan 21, 2002 1:01 am
Contact:

Re: TLC548 ADC question

Post by Spymat » Mon Jan 21, 2002 7:12 pm

Here's my power setup.<p>12Volt DC ---> 5Volt DC/DC converter ---> Logic
|
|
Analog Tranducer<p>So I guess the grounds are the same, how would I go about separating the grounds?<p>Regards,

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: TLC548 ADC question

Post by MrAl » Tue Jan 22, 2002 4:23 am

Hi there,<p>Normally you create a 'star' ground pattern,
bringing all grounds individually to a common
point close to the power supply. In this way
the analog signals never see the small voltage
drops present in the ground lead when the
digital signals switch states.
Another good idea is to bypass the supply
with individual caps across the chip, power supply,
and analog circuits. The best caps to use
are the low ESR caps. Although they are more
expensive, they work much better. You could
also try something like 10 to 100uf in parallel
with 0.1uf disc capacitors which work almost
as good. Sometimes 0.1uf disc caps alone
work ok too.<p>Good luck with it,
Al
LEDs vs Bulbs, LEDs are winning.

Ron H
Posts: 360
Joined: Tue Dec 18, 2001 1:01 am
Location: Boise, ID
Contact:

Re: TLC548 ADC question

Post by Ron H » Tue Jan 22, 2002 11:51 am

Matthew, if I am interpreting your ASCII schematic correctly, you are using a DC-DC converter to power the ADC. The noise from this alone could be the source of your jitter. If you are also using this for the reference, it's almost certain to be the cause. These parts draw 3ma max. I would power the ADC from a linear regulator which runs off the 12v supply. Your logic could still be powered off the converter, as long as you pay attention to grounding considerations, and use good power supply bypassing as previously mentioned.<p>You can use a 78L05 or LM317L for the regulator. If you need better accuracy and/or stability, you can either run the ADC and reference off a precision shunt regulator such as the TL431 or LM336.
Ron H

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: TLC548 ADC question

Post by MrAl » Wed Jan 23, 2002 4:49 am

Hi again,<p>Yeah i would go with a linear regulator too
instead of a switcher for that app.
If you must use the switcher, another idea is
to build a pi filter for the output of the
switching supply. You can probably get away
with a small ferrite bead for the inductor.
Wrap 5 to 10 turns of #32 gauge wire around the
little bead and connect it in series with the +
power supply lead. Connect two 100uf caps
on either sides of the inductor to ground.
This arrangement swamps out most of the noise
caused by the regulators' switching. Use a
low ESR cap on the output if you can, and you
probably wont see any more noise out of that
thing :-)
LEDs vs Bulbs, LEDs are winning.

Spymat
Posts: 6
Joined: Mon Jan 21, 2002 1:01 am
Contact:

Re: TLC548 ADC question

Post by Spymat » Thu Jan 24, 2002 10:29 pm

Using the linear regulator has definetly helped, but I believe a big problem is the sensor itself. It's an ashcroft K1 pressure transducer. Kinda noisy. If I want a nice DC output, I should put a inductor on the output of the sensor (a choke right?) and then a R and C off of that inductor's output for a second order discrete LPF. <p>Also do you think the bread board and all its unintended inductances could be causeing a problem that would lessen if the entire circuit was just built?<p>Any objections?<p>Thanks for all the help.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: TLC548 ADC question

Post by MrAl » Fri Jan 25, 2002 6:52 am

<blockquote><font size="1" face="Verdana, Helvetica, sans-serif">quote:</font><hr>Originally posted by Matthew:
Using the linear regulator has definetly helped, but I believe a big problem is the sensor itself. It's an ashcroft K1 pressure transducer. Kinda noisy. If I want a nice DC output, I should put a inductor on the output of the sensor (a choke right?) and then a R and C off of that inductor's output for a second order discrete LPF. <p>Also do you think the bread board and all its unintended inductances could be causeing a problem that would lessen if the entire circuit was just built?<p>Any objections?<p>Thanks for all the help.<hr></blockquote><p>Hi there again Matthew,<p>I would say you should definitely try breadboarding
the circuit using a physical layout as close
as possible to the final circuit as you can get.
If that doesnt help and the sensor really is
the main source of the unacceptable noise,
then you will have to investigate just exactly
what kind of noise it is putting out.
There are many possibilities here, too numerous
to cover all of them (including physical vibrations)
, but we can divide the
possiblilities into two general catagories:<p>1. known frequencies
2. random<p>You might check the data sheet on your particular
sensor to find out what kind of noise it puts
out to begin with, and what kind of level to
expect, relative to the true signal. If it
happens to be high frequency noise, then your
LP filter will help very much if your app can
stand a loss in high frequency response.<p>Unfortunately, many sensors put out a random
noise (looks like jitter in the signal for
no apparent reason) which is impossible to
filter with linear devices such as caps and
chokes. Adding a LP filter simply filters out
the high freq stuff, while the low freq stuff
still gets though.
The best way to filter such a signal is with
a digital filter. The digital filter makes
educated 'guesses' about the signal using
probability theory and rejects all that it
thinks is not part of the true signal.
Last i read, you can get 20db noise rejection
with this kind of filter. Of course you
wouldnt want to build this up out of discreet
parts though, and im not sure how complex you
can allow your design to become, but there are
chips that will do this made by Analog Devices.
The chips were originally made for audio apps.<p>Alternately, sometimes the same manufacturer
makes several models of almost the same sensor,
so you might find one that does the same thing
with less noise output. It might be more
expensive, but it could help quite a bit.<p>Good luck, and it would be interesting to
hear about what you end up doing to solve the
problem.
LEDs vs Bulbs, LEDs are winning.

Spymat
Posts: 6
Joined: Mon Jan 21, 2002 1:01 am
Contact:

Re: TLC548 ADC question

Post by Spymat » Fri Jan 25, 2002 8:45 am

Its too bad that Microchip hasn't come out with the dsPIC yet. <p>So are there any low cost random noise filters out there? Programming a dsp I think goes beyond the scope of what I'm using it for. <p>Maybe I'm stuck if its white noise. I have one LTC1562 which is a resistor programmed filter, but as you say, the low freq. noise will still cause problems.<p>Thanks again.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: TLC548 ADC question

Post by MrAl » Sun Jan 27, 2002 7:45 am

Hello again,<p>The chip i had in mind was a stand alone
chip if i remember right, and used in
audio systems to take the noise level
down some orders of magnitude. I dont
believe you would have to program anything.
If i get a chance, i'll check the site and
try to find the exact part number. It's
been a few years since i looked into that.<p>Of course the easier way is to get a better sensor,
if one is available.
LEDs vs Bulbs, LEDs are winning.

Post Reply

Who is online

Users browsing this forum: No registered users and 33 guests