EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: taste_tester on August 21, 2022, 09:42:00 am
-
i have attached the schematic of a very basic linear power supply, but there is not much need to inspect the circuit because I will describe it.
Transformer has 2 windings; each going to a separate diode bridge. One goes to a L78MR05 which has an integrated reset signal generator. Between pin 2 and 3 of L78MR05 is a 1uf / 50V electro.
The other diode bridge goes to a pair of 7815 and 7915 to generate +15V / -15V. At each point before the diode bridge is a 0.01uf ceramic disc. Every other small capacitor is 0.1uf / 50V ceramic disc.
Another circuit with a 1mh inductor, 100uf electro, 6.8uf / 10v tantalum, and a BJT transistor generate high voltage AC for the EL panel backlight.
If I replaced the ceramic disc capacitors with X5R MLCC, L78MR05 circuit only puts out 4.0V.
If I put back the ceramic disc, or replaced with film, it works again, +5V output is stable 5.1V and everything works.
Also, EL backlight circuit still works but not as well with film instead of ceramic disc.
I don't understand why the different types of capacitors behave so differently, until now I thought with few exceptions, ceramic disc and MLCC and film were interchangeable.
-
To answer my own post, it seems like maybe the X7R dielectric which exhibits a change of capacitance due to frequency which the film and ceramic disc types should not be affected by, since the datasheets for these two only mention temperature derating and not frequency derating.
I might have to simulate the circuit to find out the frequency the PSU runs at.
Sources:
https://www.alliedelec.com/m/d/a24bb5190bea413b2ff164816f55e87c.pdf (https://www.alliedelec.com/m/d/a24bb5190bea413b2ff164816f55e87c.pdf)
https://www.mouser.com/datasheet/2/212/1/KEM_F3127_R41B_Y2_350-2932676.pdf (https://www.mouser.com/datasheet/2/212/1/KEM_F3127_R41B_Y2_350-2932676.pdf)
-
Aside from being frequency-dependent, don't forget that most (all?) the MLCC dielectrics except for C0G (NP0) have a DC bias voltage derating as well. For example, a 1206 X7R might lose 10-20% of its rated capacitance at 5V, but in an 0402 of the same capacitance, it might lose 80%!! Despite being quite important information, this is basically never found on the datasheets, and requires you to go to the website and look up the specific part number to find the voltage/capacitance loss curve. The higher the volumetric efficiency (i.e. the "density" of the capacitance), the worse the effect.
See the attached PDF.
-
What size MLCCs?
Frequency dependency isn't relevant; that occurs because C is traded for ESR, and the amount is fairly small because ceramics have low ESR.
The inductance of THT vs. SMT parts (or are these leaded MLCCs you're testing with?) shouldn't make any difference in a circuit like this, and it should be down simply to value.
If the original discs are Z5U, they can have quite a voltage dependency as well, but it may be skewed to high enough voltages (-30% value at ca. 20V?) that it doesn't matter. Not that 0.1uF should have any problem at the same voltages, for MLCCs of 0603 or larger (even 0402 isn't bad). Note that resoldering these types also anneals out the aging effect, so you will not get identical results to the original state(!).
Tim
-
Thanks so much for the informative replies. I learned a lot about MLCC from asking, I'm very appreciative.
The original discs, I don't know what material they are made of. They are from 1985 and each rated to 50V. I would not be awfully surprised if the low ESR of the MLCC were too low for the circuit to function properly.
They were leaded MLCC I was working with as the discs I replaced were 5mm lead spacing. I found the MLCC I bought are based on the 1206 MLCC, for the dissipation factor of this size, voltage and the capacitance I chose the dissipation factor was at 10%!!! So no wonder that with these, the capacitance was not sufficient.
I think that clears things up for me.
-
You could add 1 Ohm of ESR to the MLCC, to see if low ESR is the explanation.
-
You could add 1 Ohm of ESR to the MLCC, to see if low ESR is the explanation.
Wouldn't the problem be low ESR relative to ESL and dissipation factor? I thought the film capacitor was working because the insulation resistance was vastly higher but also dissipation factor of the film capacitor is under 1% whereas on the MLCC it is 10% or even higher.
regardless, how would one do this in practice...?
-
ESR is easily added. Just put a 1 Ohm resistor in series with the capacitor.
The 7805 is not an LDO so I would not expect it to be sensitive to ESR.
Check your electrolytic caps carefully. Is the input voltage stable without excessive ripple? If the reservoir cap is faulty, voltage dependent loss of capacitance in a small MLCC may not give sufficient capacitance for correct operation.