Full wave rectifier filter

[jnnewton]

I have a full wave rectifier (input 24VACrms), has an LM2675 switcher circuit as the only load. Based on the webench design, I'm going to have about 0.5 to 0.6A input. I"m wondering if i need capacitance betwen the output of the bridge and the input to the switcher circuit (which has a 1uF and 100nF input capacitors) My standard electrolytics are too tall (limit is 0.287" due to mechanical concerns.

If so, I'd like PN recommendations for capactors.

Thanks,
Josh

 

[ChrisP58]

I have a full wave rectifier (input 24VACrms), has an LM2675 switcher circuit as the only load. Based on the webench design, I'm going to have about 0.5 to 0.6A input. I"m wondering if i need capacitance betwen the output of the bridge and the input to the switcher circuit (which has a 1uF and 100nF input capacitors) My standard electrolytics are too tall (limit is 0.287" due to mechanical concerns.

If so, I'd like PN recommendations for capactors.

Thanks,
Josh

You need to have enough input capacitance to keep the rectified DC above the minimum needed for the LM2675 to maintain regulation. Otherwise the output voltage will drop out 120 times per second.

Also, if you want the cap to survive for very long, you need to factor in it's ripple current capacity.

 

[ronv]

Your not going to have much fun.
You don't state the voltage you want out of the 2675, but lets say it is 20 volts..... You would need about 600Ufd if you have a full wave bridge. A quick look I found 100Ufd. @50 volts so you would need 6 to meet your height requirements.

 

[jnnewton]

ChrisP58,
I was / am aware of these issues. I am not sure what the minimum level is, as it is stated as DC. Ripple current for each DC link will split depending on how many there are and their values.

Ronv,
I should have put in there that it's 12V out of the 2675.

Some other info:

I have my webench sim for the switcher set to 20-40V input. I've used LTSpice to determine that i'll need 470uF to keep the voltage input between 20 and 40VDC. Now, that leaves a lot to be desired, because the webench sim is for DC, and I've got a spikey 100-120Hz 20-40V input out of the rectifier. The transient analysis output from the webench has a timescale too short for what I'm looking at, so it's not too helpful.

Anyway, I've also got the option to cut a 1" x 1" section out of the pcb, and lay an axial cap or two across the gap, thereby doubling my available height. When doing that, or doing what you are talking about, I run into area issues. My working area for thd or smt without the cutout and thereby putting me back into the original height requirement is about 1" x 1.5"

The best i've come up with is 2x330uF Axials, and I end up calculating the ripple @ 0.645A through each. I've found the TVX1J331MCD cap at
https://www.electro-tech-online.com/custompdfs/2013/07/e-vx-30194.pdf
which has a ripple rating of 650mA @ 120Hz, but no derating for 100Hz. (this thing needs to run on both 50 and 60Hz)
Plus, that gives me about 5mA of margin, which is too close for my tastes.

 

[ChrisP58]

The data sheet lists 15V as the minimum input for 12V out, so technically you get by with the input dropping that far. But, remember that as the input voltage droops, the current into the LM2675 increases. You need to think of a buck converter with a constant output current as a constant power load. So when the input voltage sags to 20V, the current needed to sustain your 12V output will be ~twice what it is at 40V.

 

[ronv]

If you lay down a radial cap in the slot you can find one easy.

 

[The Electrician]

The best i've come up with is 2x330uF Axials, and I end up calculating the ripple @ 0.645A through each. I've found the TVX1J331MCD cap at
https://www.electro-tech-online.com/custompdfs/2013/07/e-vx-30194.pdf
which has a ripple rating of 650mA @ 120Hz, but no derating for 100Hz. (this thing needs to run on both 50 and 60Hz)
Plus, that gives me about 5mA of margin, which is too close for my tastes.

The ripple rating is for an ambient temperature of 85°C. If this power supply isn't going to get that hot, the capacitor ripple rating will be substantially greater.

 

[ChrisP58]

The ripple rating is for an ambient temperature of 85°C. If this power supply isn't going to get that hot, the capacitor ripple rating will be substantially greater.

85°C is not the ambient temperature, it is the operating temperature of the cap itself. It will only be at ambient at the moment you turn the unit on, then it will increase. How far and how fast is dependent on many factors, including self heating due to the ripple current flowing in and out of the cap each half cycle.

And there is a lifetime rating attached to that 85°C value. For the TVX1J331MCD, that number is 2000 hours. Generally, the lifetime doubles for every 10°C drop in operating temp. All else being equal, a 105°C rated part will last ~4 times longer than an 85°C rated part.

So again, how long do you want the product to survive? Electrolytic caps are usually the first parts to fail. The cooler you run them, the longer the product will last.

 

[The Electrician]

85°C is not the ambient temperature, it is the operating temperature of the cap itself.

Have a look at this manufacturer's document:

https://www.electro-tech-online.com/custompdfs/2013/07/AEappGUIDE.pdf

There, on page 6 under the general heading "Operating termperature range", you will find this:

"The Operating Temperature Range is the temperature range over which the part will function, when electrified, within the limits given in the specification. It is the range of ambient temperatures for which the capacitor has been designed to operate continuously. "

The internal hot spot temperature will be higher than ambient due to the losses occasioned by the ripple current. However, the manufacturer has taken that into account with the relevant maximum ambient temperature rating.

 

[jnnewton]

I built up the circuit, and it works. 68V 220uF caps, and I was actually able to squeeze 4 of them across the back of the pcb. I learned that I am horrible at estimating current draw, as a measurement on the output of my buck regulator yeilded a 12V draw of only 22mA. While I may be able to reduce the capacitance or number of caps at this point, I'm going to choose not to. I would like some further help with specifiying components for the regulator portion as I now plan to try an emitter follower voltage regulator to save space / cost and complexity associated with that switcher i'm using now. attached is my schematic an proposed part numbers. I would welcome criticism and / or reinforcement before i try this and let the magic smoke out because of something i missed.

 

[ronv]

The problem with a linear regulator is that the extra power gets dissipated in the transistor. So in your case 21 volts X .03 amps = .6 watts so it will want to be a TO220 package (temperature rise of 62C/watt) or something with a heatsink. You might look at an LM317HV as a solution or maybe a 12 volt transformer instead of the 24 volt one, then a little TL750L would probably just work and be better regulated and protected.

 

[jnnewton]

I'm not providing the transformer. My input is 24VAC. I'm going to test an IFX25001TC V10 and see how loud the piezo is at 10V. Thanks for catching that power dissipation problem. Saved me a meltdown.