EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: hgg on March 04, 2014, 07:26:19 pm
-
Hi,
A quick question.
I am using the MAX6325 2.5V precision reference and I cannot understand something.
Its voltage output is a steady and noise free 2.497V regardless of the power supply
voltage, from 8V to 15V. When I place a load of 10mA on the power supply, then the
reference output will increase by 2mV. I am not placing any load on its output.
Can someone explain that behaviour? Is this something that is expected?
Thank you.
-
line regulation. Measure the input and output voltage with 10mA on the input. Wouldn't have thought it would have effected it too much, though.
What capacitance do you have on the input and output?
-
Is this something that is expected?
Definitely not. You are doing something wrong.
Either GND-Line or R42 are oscillating.
With best regards
Andreas
-
line regulation.
WITH NO LOAD WITH LOAD
Vcc = 9.008V Vcc = 8.981V 27mV drop
Vout = 2.497V Vout = 2.499V +2mV
I am connecting a 1K resistor with an LED from VCC to Ground.
The MAX6325 output is floating.
The thing is that if I drop the power supply down to 8V, there is no difference to the output.
Why is there with only a 27mV drop? Only if I connect a load to Vcc.
Voltage regulation of the IC version I am using is 30ppm/V.
I am using the following setup:
(http://s5.postimg.org/ewiztllev/MAX3625.jpg) (http://postimage.org/)
(http://postimage.org/)
Either GND-Line or R42 are oscillating.
I see no oscillation on the oscilloscope.
What is R42 ??
-
Are you probing gnd close to the ref or at the V source ? ,have you got a poor gnd and your losing 2mV in the gnd trace when you connect load .
-
Kevin you are right!
If I probe the ground from the power supply terminal, then I get a 2mV rise when I connect a load.
If I probe the ground on the breadboard then I get a 1mV voltage drop when I connect a load.
Only if I measure the voltage just next to the reference pins I get a steady reading regardless
the load I connect to the power supply.
The breadboard ground rail resistance has to play a role here as well. It has a 3.5R resistance at
best and a steady 3.5M ohms from positive rail to negative.
-
I am trying to build a precision reference for multimeter calibration.
Things look good now down to 3 significant digits.
It will have a voltage output of 2.5V, a Vishay S102K precision resistor and a 2.5mA current source.
I have placed an order for a chopper stabilized op amp LTC1049 for the final setup.
Of course it will need to be calibrated at the end by a high accuracy meter which I don't have...
One question.
Do you think that if I use IC sockets for the chips they will affect its accuracy or drift?
(I have some good quality gold plated ones.)
-
What is R42 ??
"42" is the answer to "all questions" (Douglas Adams)
Meaning: Too few information that one can answer the questin (no schematic, no picture of the wiring).
Now with the schematics:
Although the MAX6325 has a minimum supply voltage of 8V you should supply a well stabilized voltage of around 10V.
Below 10V the line regulation is much worse than above 10V. Not much more than 10V because of self-heating of the chip.
Well stabilized: so that the self-heating of the chip will be constant.
Star ground wiring is essential for precision cirquits to avoid GND shifting.
And you should never use sockets: In my ageing box I use sockets. It went a long time good. But since a few weeks some of the references (sockets) are drifting away with up to 120 ppm where formerly the drift was below 1ppm/month.
With best regards
Andreas
-
Andreas thank you for your advice.
I will be using a 9V battery with the voltage reference circuit because I want it to be portable.
It looks pretty stable and noise free down to 8.0V and with a 2.5mA load on the reference
output.
Star ground wiring is essential for precision cirquits to avoid GND shifting.
What about if I use a ground plane in the final PCB design? All the return signals will
follow the path of least resistance. Do you suggest that a star ground wiring will be a
better approach in this case?
George.
-
What about if I use a ground plane in the final PCB design? All the return signals will
follow the path of least resistance. Do you suggest that a star ground wiring will be a
better approach in this case?
Hello,
if you know what you are doing -> with a star ground design you will get better results than with a ground plane for low frequency analog. Additionally the thermocouple effects will be more symmetrically without ground plane.
But its a question to which precision (stability) level you want to go. (0.1%, 0.01% (thermally controlled), or below)
With best regards
Andreas
-
with a star ground design you will get better results than with a ground plane for low frequency analog. Additionally the thermocouple effects will be more symmetrically without ground plane.
Ok, I see. Thank you Andreas.
The MAX6325 uses a low power temperature compensation scheme to achieve temperature
stability. It does not require to be heated. I would be more than happy if I can achieve a 0.01%
precision.
That is why I've also bought the LTC1049 chopper and zero drift op amp to use it as a voltage follower
from the 2.5V reference through a 0.01% precision Vishay 1K resistor, in order to have a precision current
source of 2.5mA as well.
http://cds.linear.com/docs/en/datasheet/1049fb.pdf (http://cds.linear.com/docs/en/datasheet/1049fb.pdf)
Previously in my test circuit I was using the LM358N op amp and it was working fine. The 2.500V from
the reference was very stable and remained that way when I connected the 1K load at its output.
But when I replaced it with the LTC1049 I could not make it to work. I doubled checked all the
connections.
I removed the LTC1049 to test it on a separate board. I used voltage divider to feed its non inverting
input (pin 3), and connected pin 2 to 6, for a voltage follower setup. I get strange results. :palm:
Vcc+ = 9.013V
Input from Reference = 7.785V
Op Amp Output Pin 6 = 8.012V
I've connected its output to the scope and I get horrible oscillation:
(http://s5.postimg.org/dyunl4xbb/LTC1049_Oscillation.png) (http://postimage.org/)
(http://postimage.org/app.php)
In the datasheet it is mentioned that the op amp does not need external capacitors because they
are inside. But worst than that, when I connect the 1K ohm load to its output and ground, the output
drops to 0.8V ! So either I got a faulty op amp or I am doing something terribly wrong.
How can I test the LTC1049 or any other op amp to make sure that its not faulty?
Any ideas of what might be happening?
Thank you.
p.s. this reminds me of an ancient Greek saying: "The enemy of the good is the better"...
-
Hello,
The LT1049 is not specified for a 1K load. (only above 10K)
The maximum output current is in the range of 1mA.
The precision version of the LM358 is the LT1013 (or OPA2234)
You could also try a LTC2057 which has more output current than the 1049
with best regards
Andreas
-
The LT1049 is not specified for a 1K load. (only above 10K)
The maximum output current is in the range of 1mA.
Hello Andreas.
I am beginner in electronics. I may have chosen the wrong op-amp for a reference buffer...
I am looking at the datasheet and I cannot explicitly find its current output capability. How
did you find that its maximum current is 1mA? The 10K is mentioned in the Maximum Output
voltage swing but this value is just a test condition. One of its applications mentioned in the
data sheet is 4mA to 20mA Current Loops. (If 1ma is the maximum maybe I can use a BJT
at its output.)
But that is not my problem to start with.
Before I even connect a load at its output, I get a value slightly higher than the input and
a terrible oscillation shown in the graph above.
Is this normal?
Thank you.
-
Andreas you are right.
I don't know where you found the 1mA specification but this is indeed the maximum it can source
with a 15V Vcc. 0.70mA at 8V Vcc. I connected a variable resistor as a load and its output started
to drop at 1mA/15V.
I managed to find one of the few op amps that cannot source more than a 1mA... :)
-
The oscillation is probably your crappy breadboard striking again (parasitic capacitance) . If you look at the phase/gain graph plot for that opamp in the data sheet it's barely stable (only 15 deg phase margin and thats with no load capacitance), so it's not really suited to be used at a gain of 1 without some compensation, it will overshoot even without any load cap ,and it would only take a small output capacitance to eat up a 15 deg phase margin and cause it to oscillate .It will be stable though at higher gains ,try at a gain of 10 and see if it's stable then .
-
The oscillation is probably your crappy breadboard striking again (parasitic capacitance)
LOL! :) Yes probably, because it decreased today. A 0.1uF capacitor at its output made it a lot more
stable with 5mVpp ripple. Very touchy op amp. I managed to create a steady 2.5mA current source with
a BJT at its output and the feedback sensing its emitter, but the precision resistor has to be tied to ground
at all times. I need the current source for multimeter calibration, so I need the op amp output to be floating
until you short to ground the load resistor with your multimeter, in order to take the reading. So I think
I will have to choose a different op amp.
(only 15 deg phase margin and thats with no load capacitance)
I need more reading on that because I am not sure I understand it well.
Thanks. :)
-
Ok,
I am between three op-amp candidates in the same price range:
1) ICL7650S by Intersil.
http://docs-europe.electrocomponents.com/webdocs/0e27/0900766b80e27842.pdf (http://docs-europe.electrocomponents.com/webdocs/0e27/0900766b80e27842.pdf)
2) TLC2654A by Texas Instruments.
http://docs-europe.electrocomponents.com/webdocs/0c90/0900766b80c90e79.pdf (http://docs-europe.electrocomponents.com/webdocs/0c90/0900766b80c90e79.pdf)
3) LTC1250 by Linear Technology
http://docs-europe.electrocomponents.com/webdocs/10ed/0900766b810ed836.pdf (http://docs-europe.electrocomponents.com/webdocs/10ed/0900766b810ed836.pdf)
Ok, I already don't like Linear Technology...
(Please state source current explicitly guys. Its not difficult...)
Output source current, Io of TLC2654A is +/-50mA max and of ICL7650S 4.5mA typical.
I would be sourcing 2.5mA.
(http://s5.postimg.org/w1sbq3ijr/opamp_comparison.jpg) (http://postimage.org/)
(http://postimage.org/)
The TLC2654A looks like a better option for a precision current source. What do you think?
To avoid further mistakes any suggestions are appreciated.
Thank you.
George.
-
Output source current, Io of TLC2654A is +/-50mA max
You are looking at the wrong data. The maximum ratings are those where the chip may be destroyed.
Take a LT1013. Your resistors will be much worse than this chip.
With best regards
Andreas
-
You are looking at the wrong data. The maximum ratings are those where the chip may be destroyed
Intersil is more honest and lists the typical value (4.5mA) of the output source current.
There is only a maximum value of +/-50mA in the TI datasheet so I think its safe
to assume that it can easily source 2.5mA.
Take a LT1013. Your resistors will be much worse than this chip.
The 1K resistor is the Vishay metal foil S102K series with 1ppm/C and tolerance of ± 0.005 %
http://www.vishaypg.com/docs/63001/63001.pdf (http://www.vishaypg.com/docs/63001/63001.pdf)
In what respect they are much worse?
Thank you.
-
Hi,
Still haven't decided on which chopper op amp to use for the 2.5mA current reference... :-//
I might buy at the end the LTC1250 which explicitly says that it can drive 4.3V into 1K load from a single 5V supply.
http://docs-europe.electrocomponents.com/webdocs/10ed/0900766b810ed836.pdf (http://docs-europe.electrocomponents.com/webdocs/10ed/0900766b810ed836.pdf)
but,
The TLC2654A by Texas Instruments has a 0.003uV/mo offset long term drift and a much lower Idd.
http://docs-europe.electrocomponents.com/webdocs/0c90/0900766b80c90e79.pdf (http://docs-europe.electrocomponents.com/webdocs/0c90/0900766b80c90e79.pdf)
I am still not sure if it can drive a 1K load @ 2.5mA though.
Can somebody help me by having a look at the datasheet above and see if he can datamine... that info?
Thanks. :)