Hi!
i'm building a 28V/2.5A DC power supply and i'm using a 24V 100VA transformer. I have used a simple capacitor filer after the full wave bridge rectifier with 4x3300uF capacitor bank and i was aiming for about 2V p-p ripple. But at full load the voltage after the capacitor filter was dropping to about 29.7 volts while the transformer output voltage was normal - about 24.4V AC.
I assumed that there was a big power loss in the bridge rectifier (KBPC2504) because it was heating up anyway (finger test - about 70C), and i did some simulations. I notice big current spikes at the input of the bridge rectifier (about 28A) and that will explain where the missing voltage drop was going (yes i know simulations can be a bit wrong but i pulled out the rectifier and passed 2.5A of dc current thru 2 of its diodes for some time and it didn't get hot at all, not even warm). Then i did some internet researching and it turned out these current spikes are absolutely with the simple capacitor filter.
So my question is the simplest solution just to mount the rectifier to a heat sink and call it a day? And i know that the voltage after the rectifier is 1.6-2 volts lower than the input voltage but i am getting quite a bit more average voltage drop. I was reading about a inductor in series before the capacitors and its purpose is to reduce exactly these current spikes but i'm not sure how to choose a value and what are (and are there any) downsides of using an inductor (effectively a LC filer).

This is the current circuit:

 

You could decrease the size of the filter cap because ripple won't be much of a problem. Or you could use an NTC thermistor to limit inrush current.

Your post would be easier to read if you used paragraphs.

 

Hi!
i'm building a 28V/2.5A DC power supply and i'm using a 24V 100VA transformer. I have used a simple capacitor filer after the full wave bridge rectifier with 4x3300uF capacitor bank and i was aiming for about 2V p-p ripple. But at full load the voltage after the capacitor filter was dropping to about 29.7 volts while the transformer output voltage was normal - about 24.4V AC.
I assumed that there was a big power loss in the bridge rectifier (KBPC2504) because it was heating up anyway (finger test - about 70C), and i did some simulations. I notice big current spikes at the input of the bridge rectifier (about 28A) and that will explain where the missing voltage drop was going (yes i know simulations can be a bit wrong but i pulled out the rectifier and passed 2.5A of dc current thru 2 of its diodes for some time and it didn't get hot at all, not even warm). Then i did some internet researching and it turned out these current spikes are absolutely with the simple capacitor filter.
So my question is the simplest solution just to mount the rectifier to a heat sink and call it a day? And i know that the voltage after the rectifier is 1.6-2 volts lower than the input voltage but i am getting quite a bit more average voltage drop. I was reading about a inductor in series before the capacitors and its purpose is to reduce exactly these current spikes but i'm not sure how to choose a value and what are (and are there any) downsides of using an inductor (effectively a LC filer).

This is the current circuit:
View attachment 98717

I think your biggest problem is trying to get 28 volts out of a 24 volt transformer. This won't leave you much head room no matter what you do.
Since your transformer is holding up well you could make a bridge out of some schottky diodes to reduce their voltage drop.
That might pick you up a volt or so. Try it in your simulation.
http://www.mouser.com/ds/2/427/88721-103280.pdf
The inductor probably won't do you much good.

 

The large value of capacitance is causing the high peak currents and dissipation in the diodes.
If you can tolerate more ripple (why 2V?) then reducing the capacitance would reduce the peak current.

You could add an inductor but that can cause overshoot and ringing from the LC resonant circuit formed.
A large inductor will cause the output voltage to approach the average value of the sinewave rather than the peak.
With an inductor you could significantly reduce the size of the filter capacitor for a given value of ripple.

The easiest way to determine the inductor and capacitor size is with simulation.
For example an LTspice simulation (at 60Hz) with a 1mH inductor and a 5000μF capacitor gave a peak current of about 8A and a ripple of 2.5V.
Be sure and put in some realistic values for the transformer and inductor resistances as they have a significant effect on the initial voltage overshoot, and the ripple current and voltage.

 

So my question is the simplest solution just to mount the rectifier to a heat sink and call it a day?

Yes.

The 29.7 V reading is your meter trying to make sense of a small, slow, very distorted sine wave sitting on a large DC offset. It is consistent with a couple of volts of diode voltage drop and a couple of volts of ripple.

ak

 

I measured that voltage with a peak detector circuit (diode and a capacitor). It was showing about 29.1 volts + 0.6 for the diode ~ 29.7V.

ronv, thanks for the suggestion, this is the kind of diode i was searching to do some testing. But in real world situation that DO-201 package will it be enough to handle the power dissipation ?

 

I added a 100uH / 0.06ohm inductor to see what would happen and the current spikes went down a lot (from 25A to 16A). There is a bit of voltage overshoot (1-3volts) at the beginning but i'm thinking of adding a soft start circuit anyway so that will take care of the overshoot. With the smaller current spikes the rectifier should not heat as much.

The first 3 simulations are without an inductor and the last 3 are with one.

 

My first contact to electronics was with this and in this forum. Theses guys are really experts.

First thing that i learned, aim for a good voltage regulator, just stabilize with a single cap( you can try to limit the inrush current with a NTC), not a large one, and aim for a regulator that can regulate with like 75dB ripple rejection, i'm not a specialist, but for simple purposes, LM350 can give 3A, if you use a very nice heatsink and a full wave rectifier the datasheet says that you can have a 80dB ripple rejection.

This means that your ripple will be reduced by a factor of 10000. Let's say you could get that 2vpp just from the caps, using a factor of 10000, this would reduce to 0.0002vpp, but this is very idealistic case, running in 28v you won't get that at all. Nevertheless is worth a try i think.

Also some protection circuitry for this psu would be vital. Bleeder resistors, clamps or crowbars, fuses(If the fuse is a fast one, put it after the capacitor).

Things to note: Minimum load current(stated in the datasheet) to maintain regulation, dropout voltage of the lm350, also stated in the datasheet( I believe the input needs to be at least 1.25v bigger). If you wish a variable one and want to reach a 0v with this LM350 it's possible to override the reference voltage by offsetting it with -1.25v on the adj pin, but that's a subject for another topic.

EDIT: If you liked this idea you should definitely read the datasheet of the component. I'm picking these values from my head.

 

I was aiming for a switcher regulator like LM2596. I cant find the ripple rejection in the datasheet for it tho
As for the circuitry, this is not the final version with everything in it. I just want to optimise the basic functions of it.

 

If you look at the the Line Regulation, Figure 4 in this data sheet, that should give you an idea of the ripple rejection.
Thus a 10V change in the input gives about a 0.05% change in the output voltage or a 0.005% change per Vpp of ripple.

 

I measured that voltage with a peak detector circuit (diode and a capacitor). It was showing about 29.1 volts + 0.6 for the diode ~ 29.7V.

ronv, thanks for the suggestion, this is the kind of diode i was searching to do some testing. But in real world situation that DO-201 package will it be enough to handle the power dissipation ?

Yes, I think they will be ok. Only about 1 watt each.
Your current will probably be quite a bit lower than your simulation because the transformer has some resistance. It's still pretty high, but not quite as bad as your simulations.

 

I have done some further measurements on my current circuit and i got some pretty odd results.
Since i don't have an oscilloscope i'm trying to measure the ripple with a simple diode-capacitor peak detector to hold the upper edge of the waveform. I have changed the rectifier with standalone schottky diodes and the voltage (full load/2.5A) at the filer caps was still about the same - 30V DC. The filter was designed to produce 2Vp-p ripple at full load, so 30V - 2V gives about 28V for the regulator to work with. This explains why when i set the regulator output voltage to 28 volts and connect the load the voltage drops to 26.5 volts.
At the same time i measured the AC voltage of the transformer winding and it was about 24.4V AC. Then 24.4AC x 1.414 gives 34.5 volts peak. That means the i am loosing about 4.5V at the rectifier So either the rectifier is taxing me 4.5 volts or the transformer voltage is dropping more and my multimeter is lying to me.
I decided to connect the peak detector at the transformer winding (input of the bridge rectifier) and try to measure the voltage that way. I measured 31.1V DC. That divided bt 1.414 gives about 22V AC effective voltage.If this measurement is more accurate than when i was trying to measure the AC with my multimeter set to AC voltage then that would explain where are the missing volts - the transformer doesn't drop to 24.4 but to 22 volts.
And now the question is why a 24V 100VA transformer is dropping to 22V AC when im pulling about 2.5A of the power supply. Are those current spikes cousin the problem ? Is all of this just normal?

This is my setup. The multimeter is connected to the cathode of the 1n4148 diode.
Measuring the filter voltage:

Measuring the transformer voltage:


Edit: Another interesting thing. This is my transformer:
INDEL TST100/004
The datasheet says that the diameter of that transformer is 97mm and the height is 40mm. I'm measuring mine to be about 87mm by 32mm.
On the label it clearly says "TST100/004 100VA 2x12 2x4.16A" .... i don't know what to make of that.

 

Just a side question.
Is it practical to make bridge rectifier with a schottky diodes? Does the bigger reverse current do any harm? Should i put some small capacitors across the diodes?

 

get yourself an oscilloscope. There is a kit for $35 at least from Amazon that goes to 200 Khz , which would be fine for Power supply building, plus it would not be grounded whit a battery supply. and thus safer.

 

Just a side question.
Is it practical to make bridge rectifier with a schottky diodes?

Yes. Why would you think it's not?

Does the bigger reverse current do any harm?

Not in your application.

Should i put some small capacitors across the diodes?

Not unless you are feeding a load sensitive to EMI.

My answers above in blue.