Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Reducing ripple in the power source

Status
Not open for further replies.
So here's the scenario:
I have a circuit that requires +12VDC and +9VDC. Although I could use +12V for everything there are two catches. First, it powers an Arduino Uno and +12V is at the upper input voltage limit, secondly the ripple on the +12V supply is reflected in the attached sensor readings. I can't use straight +9 because of the +12V components (mainly relays) elsewhere in the circuit. The plan from the start was to add a separate 7809 based power supply (just a couple of caps and the chip). The ripple from the power supply is about +/- 0.5V at 7ms (~142Hz). How do I size the caps for the 7809 to reduce the ripple as much as practically possible?
 
More likely, the ripple is getting into your Arduino measurements via a ground-loop in the external wiring. Regulating the 12V down to 9V and then feeding that to the Arduino will do nothing to solve your root problem...
 
More likely, the ripple is getting into your Arduino measurements via a ground-loop in the external wiring. Regulating the 12V down to 9V and then feeding that to the Arduino will do nothing to solve your root problem...

Before jumping to conclusions, you should really try to determine if a ground loop is even possible.

(a) There is no "external" wiring.
(b) There is no place for a ground loop.
(c) The sensor output "ripples" in direct proportion and phase with the power source, with and/or without the Arduino.

The plan is to use the 2+ volts of overhead (difference between 9V and low point of 12V source) to supply the necessary power to smooth the ripple. I'd like to select the capacitors based on an engineering approach rather than just selecting some big ones and adding them to the circuit. It may be possible to just add the capacitors to the existing circuit, but I would like to shift some of the voltage drop for the regulators from the tiny regultor on the Arduino board to the nice fat TO-220 packaged 7809. Kind of a 3-for-1 deal if I can find the right caps.
 
Last edited:
So you are feeding the sensor an unregulated voltage? You should feed the sensor the same voltage that the Arduino ADC uses for its reference voltage, which AFAICR is the 5V Vdd created by the Arduino's built-in 5V regulator. That makes the sensor ratiometric to the ADC reference.

What kind of sensor is it? How much current/voltage does it need?

You seem to imply that the sensor wants 9V, so use an opamp to amplify the ADC reference by 9/5, supplying the opamp with the same unregulated 12V. The 100db of PSRR of the opamp will suppress the ripple. It will make the 9V ratiometric to the 5V in the process.
 
Supposedly the power supply is regulated. Supposedly. I have tried several (same manufacturer) they all have nearly the same ripple. The amplifier board for the circuit cannot reach the power rail (have no idea why they choose this board), so it must be run at a higher voltage (9V) than the arduino needs/supplies (5v). I think it must have been originally designed to be used this way because I cannot figure out why any would design the board for use in a 5V only system when you lose 1V of the upper output range. The sensor is a linear position sensor. The sensor has a slight proportional response to the input voltage (about 10:1), so for the 1V ripple in, I see 0.1V ripple out. I'm trying to convince them to get a better power supply but they are preferring this route. To make it work all I need are (I think) the right size caps.
 
So as far as I can determine what you are doing ( a schematic would have helped)...

You have an improperly-filtered 12Vdc supply that sags to 11.5V at a 120Hz ripple frequency. The Arduino on-board regulator can deal with that, except you are worried about dissipation in that regulator. Have you ever touched it to see how hot it gets?

You have a poorly-designed sensor which now operates from the bad 12V, but this sensor has terrible power-supply rejection ratio, and its 0-5V signal output is contaminated with the ripple from the 12V supply. (What Idiot designed that sensor?)

You want to put an intermediate 7809 regulator ahead of the sensor to make clean 9V for the sensor. You might also power the Arduino from the 7809? (I have been powering several Arduino projects from 13.8V, and they do just fine)

You want to know what value of capacitor(s) to put ahead and behind the 7809, but you haven't specified the sensor current and probably have never measured the Arduino operating current.

The answer to your 7809 capacitor question is right on Fig 8. of the data sheet. Only the two capacitors shown (short leads, non-inductive) are required to suppress possible oscillation of the regulator. Making them bigger does nothing...

Of bigger concern is the Vdrop spec for the 7809. If the input (output from the 12V supply sags below 11V, you are in trouble. You could go with a 7808.

The only other possibility to go fix the 12V supply by decreasing its ripple.
 
Last edited:
So as far as I can determine what you are doing ( a schematic would have helped)...

There is no real schematic, only a block diagram. The items are all premade, with the exception of this 9V power supply

You have an improperly-filtered 12Vdc supply that sags to 11.5V at a 120Hz ripple frequency. The Arduino on-board regulator can deal with that, except you are worried about dissipation in that regulator. Have you ever touched it to see how hot it gets?

The frequency is noticeably faster than 120Hz, but perhaps it is based on it. I have not tested the regulator for temperature but thinking if I have to add a 9V regulator I might as well buy some margin. Additionally the sensor is powered from a board that mounts on the arduino and uses the incoming power directly, hence this direction.

You have a poorly-designed sensor which now operates from the bad 12V, but this sensor has terrible power-supply rejection ratio, and its 0-5V signal output is contaminated with the ripple from the 12V supply. (What Idiot designed that sensor?)

Yep and its not the only one I have found.

You want to put an intermediate 7809 regulator ahead of the sensor to make clean 9V for the sensor. You might also power the Arduino from the 7809? (I have been powering several Arduino projects from 13.8V, and they do just fine)

Yes, might as well. The 7809 in a TO-220 package should have plenty of power to spare.

You want to know what value of capacitor(s) to put ahead and behind the 7809, but you haven't specified the sensor current and probably have never measured the Arduino operating current.

The answer to your 7809 capacitor question is right on Fig 8. of the data sheet. Only the two capacitors shown (short leads, non-inductive) are required to suppress possible oscillation of the regulator. Making them bigger does nothing...

Of bigger concern is the Vdrop spec for the 7809. If the input (output from the 12V supply sags below 11V, you are in trouble. You could go with a 7808.

The only other possibility to go fix the 12V supply by decreasing its ripple.

The 12VDC PS is OTC. Fixing the source is not an option. Getting a different one might be, if I can convince them.

So the 0.33uF on the input and the 0.1uF on the output of the 7809 will be fine to 'filter' the ripple.
 
...So the 0.33uF on the input and the 0.1uF on the output of the 7809 will be fine to 'filter' the ripple.
The regulator gets rid of the ripple by "regulating"; the capacitors have nothing to do with filtering low frequency ripple. They are there to keep the from oscillating (taking off), and the output capacitor improves the regulators transient load response. It only has to be what is specified on the data sheet; bigger is not better.

Here is the way to make the 9V sensor voltage ratiometric with respect to the Arduino's 5V AD reference voltage. This is the way I would solve your problem... This way you do not have two unrelated voltage regulators drifting in opposite directions.

I am using an opamp to produce the 9V for the sensor. (Need to check the sensor current draw). The 9V is slaved to the 5V ref. using two 1% resistors. The ripple on the 12V supply is eliminated by the > 100db PSRR of the opamp.


225.gif
 
Thank you for the idea. I will have to see if it will be possible (physically in the enclosure) to get this in between the arduino and the sensor. I know the amplifier board sits on the arduino directly,. I would have to verify the sensor connection. I think it plugs directly into a connection on the board which is bulkhead mounted leaving no way to 'inject' a voltage reference between the arduino and the sensor.
 
...I think it plugs directly into a connection on the board which is bulkhead mounted leaving no way to 'inject' a voltage reference between the arduino and the sensor.

So how where you going to do it with the 9V regulator?
 
The arduino power plug does not show through the enclosure. The PS and the Arduino are inside. The plan is to "intercept" the dirty 12V with the regulator board, and then plug that board into the arudino. Same as would be the plan for the sensor, only on the front end instead of the back end.
 
Another example of a engineer being directed to "fix it but don't change it".
 
Yeah. Pretty close. They don't want to go through the work of spec'ing, sourcing, and qualifying a new power supply.

Sounds like it wasn't properly spec'd the first time. Or at least, the real needs weren't known at the time.

Another way of reducing noise voltage to a low current sensor, is a one or two stage RC filter.
 
Sounds like it wasn't properly spec'd the first time. Or at least, the real needs weren't known at the time.

Another way of reducing noise voltage to a low current sensor, is a one or two stage RC filter.

From what I can gather they changed the sensor to one that can provide more range plus a little more accuracy. They were happy with the results but realized they were not reaching the full potential of the new parts. Even the old parts had similar issues (I always baseline new vs. old to make sure the project is headed up). In short, I think the new sensor has highlighted and amplified a old problem. They are looking at adding an off the shelf DC-DC converter. I'm trying to convince them to combine the cost of the old PS with the new converter and just buy a better 12V power supply. I'm an engineer not a businessman, in my corner of the building it makes sense to me.
 
Ha! Win one for engineering. I had the intermediate 9V PS built and now they say the design looks bandaided. Duh, it is a bandaid. They are now letting me spec a new PS, but I have a few opamps I need to swap out to be single voltage supply rail-to-rail. They are handcuffing me a little as they would like to use an opamp they are using elsewhere. I have narrowed the options down to the NE5532, LMC6482, and the LMC6032.

On the NE5532 datasheet, I do not see (can't determine) what V out low is for a 0-5V supply is.

Looks like either the LMC chips would work. The LMC6482 looks better to me (closer to the rail output) than the LMC6032.

Am I on the right track? Comments? Suggestions? What did I miss?
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top