AC ripple on DC power supply

Thread Starter

crobertsbmw

Joined Sep 7, 2011
28
I have built a 3.3 volt power supply, to run my microcontroller and a microphone/op amp, out of a step down transformer, a full wave bridge rectifier, a 100microFarad capacitor, and a 3.3voltage regulator. This is tapped off of a standard 110 wall supply. In series with the transformer, I have a triac switch which turns on/off a load (in my case, a light bulb). I can't figure out why but when I turn on the power, the 3.3v dc line has a pretty significant ripple on it. When I turn the lightbulb off (using the triac switch) the dc voltage has no ripple. Also when I take the amplifier and microcontroller off the 3.3volt line then my dc voltage looks fine. Any ideas why this may be happening? I put in like 12 100microfarad decoupling capacitors in parallel and I was able to smooth out the dc line pretty well. But there has got to be a better way..

I initially thought that maybe my voltage regulator wasn't big enough. So I bought one rated up to 1.5 Amps and this did not solve the problem.

I will try to post up a schematic so that this is a little easier to understand.
 

praondevou

Joined Jul 9, 2011
2,942
In series with the transformer, I have a triac switch which turns on/off a load (in my case, a light bulb).
Yes, a schematic would be nice to explain this better.
The voltage ripple will depend on the smoothing cap capacitance (as you already discovered) and the load current.
How much is the load current?

You can also use a bigger cap, instead of 12 x 100uF, use 1000uf.
 

praondevou

Joined Jul 9, 2011
2,942
Ok, it's not in series, it's in parallel. The only thing I can imagine is that you are slightly distorting the input voltage and this reflects on the output . What's the model number of the triac switch? Is it possible that it's not fully turned on, i.e. it turns on too far away from zero?
Is this triac just being switched on and stays on or is it being switched with pulses?

What's the ripple frequency? 60Hz?
 

praondevou

Joined Jul 9, 2011
2,942
Thinking again, where does the voltage come from to switch the triac switch? Also from the 3.3V? Then maybe you are drawing too much current when turning the switch ON? Did you try to power the triac switch from another source?
 
Last edited:

Adjuster

Joined Dec 26, 2010
2,148
I hope you are not driving the LED in your opto-coupler with excessive current (no resistor in series?) so that it loads down the supply.
 

Thread Starter

crobertsbmw

Joined Sep 7, 2011
28
I hope you are not driving the LED in your opto-coupler with excessive current (no resistor in series?) so that it loads down the supply.
I do have a resistor in series with the opto-coupler LED.

Is this triac just being switched on and stays on or is it being switched with pulses?
No pulsing. It is on or off.

Thinking again, where does the voltage come from to switch the triac switch? Also from the 3.3V? Then maybe you are too much current at when turning the switch ON? Did you try to power the triac switch from another source?
Yes I am using the 3.3v to flip the LED in the optocoupler. I bet that's it. So if I am pulling too much current then what is the solution? I guess I can mess around with the resistor that is in series and ensure that it is pulling the least amount of current possible while still being able to run the LED. If that doesn't work then am I going to have to have to separate power supplies?
 

studiot

Joined Nov 9, 2007
4,998
Sorry but 6volts alternating from a transformer does not become 6 volts direct out of a rectifier.

How are you viewing the ripple?

Do you know what its frequency is?
 

Adjuster

Joined Dec 26, 2010
2,148
I would not recommend turning down the LED current too far. These things can age and vary a bit with temperature, so a moderate margin of current above the minimum is a good thing.

It may seem puzzling that you are getting ripple on your regulator output, but 100μF is not very big. Perhaps the ripple on the rectified 6V is pretty big anyhow even with the LED off, and the extra current is just enough to drop the minimum too low for the regulator drop-out. We might have a go at estimating how much current is possible without too much ripple.

If you assume that the transformer is exactly 6V RMS, then the peak would be 6√2 or about 8.5V, maybe 7.3V at most after the bridge drop. Your 3.3V regulator may require a 2V excess or "drop-out" voltage to give full output, so maybe 5.3V minimum input is needed. That implies that there is only 2V of headroom for ripple before the output will start to wobble.

We can estimate the amount of current available with a given size of reservoir capacitor from the relation Q=CV, or in the case of a constant current drain IT=CV, I=CV/T. Here C=100μF, V=2V, and T is approximately half a mains period, for 60Hz 1/120 s or 8.33ms. That comes to only 24mA for the micro-controller, microphone plus op-amp, and the LED.

I think you should go look for a bigger capacitor, say 470μf or 1000μF. Make sure it is rated for at least 10V, a 6.3V rating is too low.

Edit: That said, note the last posters comment about the "ripple" frequency. Is it twice mains, or is something else going on? I still think it will be the small reservoir though.
 
Last edited:

praondevou

Joined Jul 9, 2011
2,942
I guess I can mess around with the resistor that is in series and ensure that it is pulling the least amount of current possible while still being able to run the LED. If that doesn't work then am I going to have to have to separate power supplies?
Good idea, try to adjust the resistor. You could also provide the datasheet of the triac, so we can have a look at it.

If that doesn't work, use a bigger capacitor. 10V voltage rating would be ok, so it's going to be small, even at 1000uF.

If you don't want to do that then use one more rectifier to rectify the AC out of the transformer and use the obtained DC to supply current to the Opto-Triac switch. (that way you don't discharge your 100uF to much)
You can do that with a transistor being driven by your original drive signal. If you want us to help you with that tell us if you are sinking or sourcing current through the LED.
 

MrChips

Joined Oct 2, 2009
30,711
Three things.
1) When you increased the "decoupling capacitor" to 12 x 100uF, which capacitor are you referring to? You should be increasing the reservoir capacitor. The cap on the output of the regulator should be 10uF.
2) If you have a voltmeter or scope, measure the voltage across the input (reservoir) capacitor.
3) What is the part number of the voltage regulator?

I suspect that the input voltage to the regulator is not high enough.
 

ErnieM

Joined Apr 24, 2011
8,377
Try taking the light bulb out of the socket and see if the ripple goes away. If it doesn't, your supply isn't up to par.

If it does get better, turn it off and don't touch it again. It's either hexed or has a short to the AC line that could kill you.
 

Thread Starter

crobertsbmw

Joined Sep 7, 2011
28
Try taking the light bulb out of the socket and see if the ripple goes away. If it doesn't, your supply isn't up to par.

If it does get better, turn it off and don't touch it again. It's either hexed or has a short to the AC line that could kill you.
I took the lightbulb out and I have the same problem. I noticed that just running my microphone/amplifier off the 3.3v line it is taking too much power out of the circuit. putting 15mv across the LED in the optocoupler is just making it worse.

Three things.
1) When you increased the "decoupling capacitor" to 12 x 100uF, which capacitor are you referring to? You should be increasing the reservoir capacitor. The cap on the output of the regulator should be 10uF.
2) If you have a voltmeter or scope, measure the voltage across the input (reservoir) capacitor.
3) What is the part number of the voltage regulator?

I suspect that the input voltage to the regulator is not high enough.
I tried increasing the reservoir capacitor. I guess I just need to throw down 1000-2000micro farads on this thing. Although, I do think that ideally I may need a larger voltage coming out of the rectifier.
 
Last edited:

Thread Starter

crobertsbmw

Joined Sep 7, 2011
28
Ok so I got a transformer that goes from 115 to 12. It works wonderfully.

The odd thing is that when I measure the voltage coming out of the transformer it isn't 12v at all. Its like 44v. I then took my previous transformer that is supposedly 230V to 6v and measured it. Putting in 110 from the wall the transformer is outputting 12v. Why are the output voltages so different from what they say in the spec sheet??
 

Thread Starter

crobertsbmw

Joined Sep 7, 2011
28
Very good to know...

I am not powering anything huge with the transformer. Just a micro-controller, op amp, and a few LED's. So I guess I should take that into consideration when buying these things.
 

bountyhunter

Joined Sep 7, 2009
2,512
If you assume that the transformer is exactly 6V RMS, then the peak would be 6√2 or about 8.5V, maybe 7.3V at most after the bridge drop. Your 3.3V regulator may require a 2V excess or "drop-out" voltage to give full output, so maybe 5.3V minimum input is needed. That implies that there is only 2V of headroom for ripple before the output will start to wobble.
Bullseye.

Use an O scope to measure the input going to the regulator. If it is a standard NPN regulator it needs 2 VBE's + 1 sat voltage to stay in regulation which is about 2V above the regulated voltage.

Don't forget there is AC ripple on that input due to the capacitor.
 

bountyhunter

Joined Sep 7, 2009
2,512
Ok so I got a transformer that goes from 115 to 12. It works wonderfully.

The odd thing is that when I measure the voltage coming out of the transformer it isn't 12v at all. Its like 44v.
You are getting 44V at the input to the reg? Check the data sheet, that might be over the spec.

Putting in 110 from the wall the transformer is outputting 12v. Why are the output voltages so different from what they say in the spec sheet??
Because the output voltages are specified at RATED LOAD CURRENT on the transformer secondary. At no load, the voltage has to be much higher. All transformers have a loading effect.
 
Top