250W Adjustable SMPS Design / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

250W Adjustable SMPS Design

Move Thread

LAN_403

cjk2

Fri Jan 08 2010, 08:59AM

Registered Member #51 Joined: Thu Feb 09 2006, 04:17AM
Location:
Posts: 263

This is my attempt at a 250W current and voltage regulated switching supply. I am planning on using primary side current feedback because I have read that it is better in response to changing load conditions. The only aspects of this design that worry me are the compensation section (I am using Type II) and the current transformer section.

Does this supply look like it will be stable and regulate current/voltage correctly?

GeordieBoy

Fri Jan 08 2010, 12:59PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

Basic design looks okay. Only a few suggestions:

1. Disconnect the voltage-doubler C8/C10 mid-point from the mid-point of C11/C12 half-bridge capacitive divider. Also C22 is un-necessary, as C11/C12 provide DC blocking to prevent flux walking.

2. Move the CT from the "hot" end of the primary winding to the "cold" end that goes to the capacitive divider. It will pick up less interference from capacitive coupling if it is fitted here, rather than on the wire with high dv/dt leaving the inverter.

3. Provide space on PCB for secondary side snubbers across D9 and D10. (Also provide space for additional post-filtering L & C outside the voltage-loop if you want the lowest possible residual switching noise at the output.)

4. Replace C15 with multiple 150uF capacitors in parallel. Gives lower total ESR, so reduced switching freq ripple at output.

5. Be careful if you ground the negative output terminal of the supply as it will short out your I-sense resistor R17. Better to ground the right end of R17 where it leaves the supply and sense a negative voltage at the left end of it with respect to ground.

6. Add 10uF and 100nF decoupling from +12V to AGND as close as possible to UC3825 to support gate-drive demands from supply.

7. Decouple I and V reference signals from pots going into the two error amplifiers. Removes any switching noise picked up here that would be fed into the EAs.

8. You might want to include some output over-voltage protection. This will prevent the output voltage overshooting to dangerous levels if the load current suddenly drops and the energy stored in L1 gets dumped into C15. This can happen when a s/c is removed from the output.

I can't comment on the stability of the design and loop compensation etc, without seeing open loop bode-plots or knowing a whole lot more parameters like output capacitance ESR, choke resistance, choke inductance at various DC loads etc. All I can say about this design is that it is possible to stabilise a PWM controlled voltage-mode supply like this using a type II compensator provided that the open-loop phase margin is high enough.

If the damping of the LC filter at the output is particularly low, then you might find achieving a satisfactory response with type II compensation a challenge. Type III is a better bet as it puts a phase boost right where the double pole of the output filter causes the most phase lag.

Three important things to remember to take into account are:

1. The ESR of the output capacitors at minimum and maximum operating temperature. ESR multiplied by the choke's ripple current is what determines output voltage ripple, and good knowledge of ESR is essential for stability calculations.

2. The inductance of the output choke under varying DC bias. Gapped ferrite changes least with DC bias until it abruptly saturates, whereas iron-powder permeability drops as soon as you start dropping load current. A swinging choke made using iron-powder can be cheap, compact and lower the discontinuous current boundary, however the range of L must be taken into account when doing stability calculations!

3. Make sure that you've worked out the turn's ratio on the transformer properly. You need to make sure you can still maintain the maximum output voltage under full-load, given how low the DC bus ripple will dip when the AC input voltage is at it's minimum spec. Remember to take into account winding resistances, and diode voltage drops if they are significant, and any dead-time that limits the maximum achievable duty-ratio.

Re: Current limiting, I would always include primary side CT sensing to limit the peak switch current under fault conditions, even if you plan to have separate current sensing on the secondary side with closed-loop feedback. Cycle-by-cycle peak current limiting rarely fails to catch damaging faults so is valuable at the development stage!

I hope this info helps?

-Richie,

cjk2

Sun Jan 10 2010, 09:34PM

Registered Member #51 Joined: Thu Feb 09 2006, 04:17AM
Location:
Posts: 263

Thanks for the reply!

I have made the changes you suggested. My main worry in this design was its stability and its ability to regulate to 0v and 0a output. I am now feeling better about the stability after reading more about current mode control. It seems that many current mode designs can be stabilized with a simple type 2 or even type 1 compensation circuit because they make the output inductor act like a current source. I still don't know how well this design will regulate at very low voltages and currents, but most of the fun in this project will be learning and finding out!

Here is the revised schematic.

Mads Barnkob

Mon Jan 11 2010, 05:23AM

Registered Member #1403 Joined: Tue Mar 18 2008, 06:05PM
Location: Denmark, Odense C
Posts: 1968

Hey cjk2

I am quite interested in knowing what resources or method you used to design your output transformer, there are not many examples on adjustable SMPS transformer design in the books I have looked in

cjk2

Tue Jan 12 2010, 09:13PM

Registered Member #51 Joined: Thu Feb 09 2006, 04:17AM
Location:
Posts: 263

There were a few resources I used to design my transformer:

http://schmidt-walter.eit.h-da.de/smps_e/smps_e.html Has a good design tool that can help you visulize what is going on and predict the core size you will need. I did find that this site calculated the primary turns my transformer would need to be higher than what I actually ended up needing.

http://www.powerstream.com/Wire_Size.htm Proved to be very useful in deciding what wire gauge and what type of litz to use.

http://www.bcae1.com/trnsfrmr.htm Has a calculator that allows you to predict primary turns to achieve a given flux density. I aimed for less that 1500 gauss (.15 Tesla) in my design. You need to look at the data sheet for your given core and decide how much power loss you are willing to tolerate, this will determine your maximum flux density. If you don't know, it is probably safe to guess and stay under 1500 as long as your frequency is under 100khz. If you pick too small a core, then to achieve a reasonable flux density you will need a lot of primary turns which means your winding will either not fit on your core, or your wire will be very fine and you will have lots of copper loss and a very inefficient transformer. If this is the case, just pick the next size larger core and try your design again.

I had decided from the beginning that I wanted to use an ETD core because i have many of them sitting around from power supplies I have taken apart. My cores came from a company called adamsmagnetic who is a distributor for ferroxcube, I highly recommend them.

I have not actually wound my transformer yet but it looks like 50 to 60 primary turns will work well and my primary should not dissipate more than 1W of heat.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob