core/s on SMPS need some info

[djQUAN]

I also got a EE core with gap and used a small grinder to remove some ferrite material from the sides (the gap is in the tongue) so that the tongues meet.

be careful though to prevent breaking the ferrite.

[RX5]

thanks for that tip.. but if I grind the sides, It might not fit the bobbin anymore..

[RX5]

Eva,

what MODE (current mode?) is it when I use a TL494 push/pull and with NO output regulation....

NORMAL mode perhaps

[RX5]

anyways...... w/c is better(or is the right way)........ filter coil in between the fast recovery diodes and filter cap/s ?? or placed AFTER the filter cap/s???

[djQUAN]

I'm not sure but I prefer to use (and have used) between the rectifier and filter caps.

but from what I know, if you're using a regulated supply, you need a higher inductance and placed between rectifier and caps. if you just want plain old filtering, you need to put it after the caps but can use smaller inductance value.

[poobah]

TL494, as I recall, is a "voltage mode" device. Current mode control earns its name from the fact that the primary current ramp is compared against an error signal/reference to determine the duty cycle of the switches. Voltage-mode control uses some fraction of the output voltage error compared to a ramp waveform for the determination of duty cycle.

Both methods have pros and cons. Voltage mode requires more complexity in feedback compensation. Current mode control may require ramp compensation depending on the duty cycles involved.

Current mode control is thought to be superior in most regards because of its ability to compensate for line variations, and, also because its solves (for the most part) saturation problems (DC flux/imbalance) in push-pull and bridge converters. You also pick up current limiting "for free". Current mode control also can yield higher better response to load steps with less grief (compensation design).

The catch is, current mode control requires a resistor to sense the primary current... and a decent level of voltage to measure... a half Volt or so. So, efficiency can be hurt by the power loss across such a resistor.

New voltage-mode techniques are being used that feature feed-forward ramp compensation. These designs can rival current-mode control in all respects except for the flux imbalance issue. So for SMPS that do not have the flux imbalance problem... buck, boost, forward, flyback etc... you see the voltage mode coming back in popularity because it offers higher efficiency when primary/input voltages are low.



Let's see if EVA tears me a new one...

[RX5]

djQUAN,

ok.... 50uH ferrite coil between diode//filter cap... if I could find those in junks shops.. hehe anyways, finally found my long time friend who HAD -lots- of electronic junk!! finally, some MORE junks.... got LOTs of big U-U cores (from flyback transformer).. nice selection for bigger SMPS.... and found a 47000uF 5.5V memory cap.. man!! have I been looking for this a long time now(maybe 3 years? haha) from a fax machine..i dunno what make/model..for my BEAM robotics projects.. and some cute ALPS pots..and some IR demodulator modules(+5V) .....




poobah,

nice one, man!!! great explaination....
anyway, for my SMPS I think I dont need regulation since it(SMPS) is run with a car battery 12V.. w/c hardly? gets off voltage 12V...... maybe a little higher if the battery charger is charging it.. so It has maximum effeciency




Cheers,
Raff

[poobah]

Rx5,

You can run SMPS without regulation. When you do so you will find a large time constant based on your inductor and cap value. Your inductor must "spin" to a new level of current in order to handle new values of output power. Feedback helps this... GREATLY.

Using feedback is not simple, but, it is not that hard either... the circuits are drawn for you. You do not have to understand the math to build a good circuit. Most engineers don't use the math anyway... you have to build a real circuit and play "capacitor roulette" in order to mkae a good SMPS.

Go to Texas Instruments and search the literature... the "Unitrode" app notes are very good. Also Linear Technologies App Note 19 ia very good.

[djQUAN]

Originally posted by RX5
djQUAN,

ok.... 50uH ferrite coil between diode//filter cap... if I could find those in junks shops..

Cheers,
Raff

same value I used. note that you would need to use powdered iron here to prevent saturation at the large DC currents involved.

Originally posted by poobah
Rx5,

You can run SMPS without regulation. When you do so you will find a large time constant based on your inductor and cap value. Your inductor must "spin" to a new level of current in order to handle new values of output power. Feedback helps this... GREATLY.

"spin"? does that have something to do with the large current changes imposed by the amplifier? so does that mean using an LC filter at the output on an unregulated supply is not a good idea? I originally used one in mine since I was planning it to have a dual output voltage mode, a higher rail for high power and lower voltage rail for "power save" mode. the idea was to run at 100% duty cycle at high power and reduce the duty cycle (without a feedback loop) by a fixed amount to lower the rail voltage.

[RX5]

djQUAN,

I use ferrite RODs for coils

I used iron powder cores before, as coils, but seems SMPS is noisy, was radiating EMI/RFI to my diy ucd... when I changed to ferrite rods(coil), it helped to lower(if not eliminate) the EMI/RFI....


poobah,

thanks for the tip..