Switching power supplies 1

[sparkyyyyy]

Hello,

Does anyone have a recommendation where I can find some good free information on all of the parameters dealing with switching power supplies. I am also interested in any good books on the subject.

Thanks,

sparkyyyyy

 

[logbook]

You could start here …

http://www.national.com/appinfo/power/

The technical papers and app notes will keep you busy for quite a while.

Other manufacturers like Unitrode and Motorola also have good websites.

 

[sreid]

I would always start with "Practical Switching Power Supply Design" By Marty Brown. This is the place to start period in my humble opinon.

"Switching Power Supply Design" by Abraham Pressman is more in depth but less intuitive to the uninitiated. There is also a good book by Keith Billings.

 

[Madcow]

What type of power conversion? AC-DC or DC-DC
How much power? Volts , amps ect.
If low power most sites have software for their chips.
National,Fairchild,Ti,Motorola,lineartech.,ect.

Heres some links I use;

Good software for simulating, but very expensive.

http://www.intusoft.com/psbook.htm

This site has alot of good links

http://www.smps.us/

Ridley is very good training and the software is also
very good.

http://www.ridleyengineering.com/

http://www.switchingpowermagazine.com/

Worked with this guy , very good book on using spice
to simulate power supplies. Has alot of tips on convergence
problems.

http://www10.pcbcafe.com/book/parse_book.php?article=SMPS/SMPS.htm

Some free training from TI/Unitrode.

http://focus.ti.com/general/docs/training/trainingevents.tsp?familyId=2

 

[buzzp]

Sort off subject but can a switching supply be damaged by to small a load? I can't imagine what could be 'damaged' per say but I could believe the regulation may not be within spec. The question pertains to a supply that has 48VDC input with a 5VDC (amongst others - +-12, +-15) output.
A colleague of mine is saying they can be damaged by a light load. Has anyone else ever heard of this? If so, what would fail? I realize more details may be needed but in general, has anyone seen a switcher fail (DC-DC) by too small a load? Thanks in advance.

 

[Mstrvb19]

It depends on what kind of load and what the design of the switcher is. If the switcher is dependant upon an exterior load to establish output inductance, then yes, a load that is "too small" (i.e. not enough inductance) could give it some problems. There may be some other issues involved here that are more design/application specific. Not sure.

 

[buzzp]

This supply is not dependant on inductive loading. It is for powering TTL logic chips. I will have to do some more digging into the application. I just wanted a little hindsight into the possibility of 'lack of load' causing problems.

 

[sdmays]

buzzp,
Yes, a switching power supply will most likely be messed up due to too little load current (this is true for pretty much ALL switching power supplies that do not occasionally employ discontinuous mode). There are various failure modes associated with insufficient loading.

A good example is with a boost DC-DC converter with insufficient loading. The inductor of a boost DC-DC converter acts as a constant current source feeding constant current (with some non-zero ripple current) to a filter capacitor and your load. With insufficient loading the capacitor will increase it's voltage until it breaks down.

Your power supply is most likely either some kind of a forward DC-DC converter or some kind of a fly-back DC-DC converter. Newer off-the-shelf DC power supplies will have controls that will employ discontinuous mode operation to avoid the problem I mentioned in the paragraph above. It is possible your power supply will require a minimum load.

So,...what to do...If your power supply requires a minimum load (quite often 10% of rated output current for a given rail), you will need to determine that from the data sheets and then apply the minimum load. If the data sheets have no mention of a minimum load, then you must have a different problem.

 

[buzzp]

sdmays,
Your explanation makes a little sense to me. I have not looked at switching design to know exactly what is going on with light loads. Our application certainly is lightly loading the supply (150A rated and pulling 20A). I will be getting the part numbers, etc soon to figure out if this is a possible cause of the supply failures. Sounds like it could definately be an issue. Of course, its hard to decipher without a data sheet. Thanks for the info.

 

[Madcow]

Buzzp,
In my experience at light loads the power supply
may not preform to the specs. (load regulation,ripple & noise, and transiant response),but it should not be damaged.

 

[MacGyverS2000]

I'm not an expert in the field, just learning some details myself, but from what I HAVE learned... low loads will significantly lower the efficiency of the supply, but not affect its reliability. YMMV...

 

[buzzp]

The posts are helpful guys, I appreciate it. I should have a part number/data sheet today to see what the manufacturer says. I will post the specs when I get them if there is no clear answer in the data sheet. I intend to call the manufacturer as well. Thanks again.

 

[sdmays]

The failure of an 'insufficiently loaded' switching power supply is largely dependent on how the design of the power supply handles transferring energy to light loads. As I said above, newer supplies handle light loads (or no load) much better, which is great.

I have seen several switching power supplies that spent a good amount with 120VAC applied to the input and the output running open circuit fail early. The failure mode was the output voltage was floating between 16V to 20V instead of the intended 12V. Inexpensive switching power supply IC's one might use as part of a larger design might not survive if the intended load is completely removed (or never there).

Lastly, a number of newer supplies already have a 'dummy load' on the PWB. I have two supplies I purchased in the last 3 months sitting next to me. Looking at the traces associated with the 'main rail', i.e. the 5V rail or, for the sake of the other supply, the 24V rail, I see power resistors across the rail to ground. I would guess these resistors serve several purposes, amongst which would be when the power supply first turns on and begins transferring power. If there is no load on the output and the supply is trying to transfer the rated power, the voltage across the output filter caps will probably continue to increase. The resistors will probably 'bleed' off the extra stored energy, allowing the supply's controls to adapt to the current load (no load, light load, etc...).

The supplies I mentioned at the top of this particular post were supposed to have loads that drew at least 10% of the rated current. I had to dissipate a minimum of 9 watts (12V * 7.5A * 0.10W). This project was 'mobile' (think shipping abuse), so I had to 'batten' everything down which created quite a challenge.

 

[buzzp]

I finally got a mfg and PN. It is HC Power and the model number is DC10-C1032. Of course, this company was bought out by Power-One, Inc so I am still trying to get a data sheet or talk with an engineer (HC powers' website has been taken over by a search engine hcpower.com). I will post their reply after I talk with them. Again, thanks for the posts.

 

[buzzp]

I got the data sheet for this supply and its plastered all over it that "no minimum load is required on any output". I wish some people could read before they call me. Thanks guys for the posts.

 

[agiza001]

I need help desiginig an ac to dc switching power supply. That has atleast a 150W output and multiple V outputs of 5V,+12V, and -12V with each at 5A or more. Where do I begin with the desgning? and any good website on desiging multiple output switching power supplies? Thanks.

 

[betkro]

SMPSs deserve special consideration in circuit engineering. They require a deep knowledge of electromagnetism besides electronic circuits and components. Magnetic materials and inductor and transformers manufacturing and testing is essential when you design an SMPS for commercial purposes. I suggest the following basic itinerary:

- Define the topology. Buck, boost, fly-back, etc. Also, specifications like isolation, output power, voltage input range, power factor, efficiency, true-hole or surface mounting components and applicable standards.
- Choose the controller that fit your needs: Voltage or current mode, built-in or external switcher, seeking info into the Application Notes and Prototype models at the main semiconductor manufacturers, for instance, Unitrode (Texas), ST, Maxim, National, IR, LT, etc.
- Define the operation mode of your controller: In general, discontinuous mode is preferred over continuous mode. However, this is not always the case, in PC power supplies (for example the PS of my PC doesn't start unloaded.) This operation mode is dependant on the controller, the inductor and load conditions.
- Search for suppliers and/or manufactures for the main components, named: low impedance capacitors, switching transistors, fast rectifier diodes, magnetic materials, coil forms and heatsinks.
- Prepare a worksheet for calculations of the several sections of the SMPS: Primary switcher, snubber circuits, inductors and transformers, rectifier section, feedback section. Make what-if scenarios.
- Simulate in PSPICE, if possible, the primary switching for several inductances, turn ratios, capacitances, etc. Do not forget nor disregard your calculations when manufacturing or requesting samples. Fortunately, the behavior of magnetic materials is close to reality.
- Technology and new products appear every day, so check for manufacturer’s advising on new designs.
- Search the Internet on specific studies about snubber calculations, transformer design and PSPICE models. Unless you are designing for hobby, it is difficult to found ready made circuits.
- Printed circuit design is another important issue.
- Built and test your circuit by sections in open loop mode first, then in closed loop, making the necessary adjustments.

Good luck.

 

[BrianG]

Do you actually need each of the 12V outputs to be rated at 5A (i.e. 60W each)?

It may not be worth designing a power supply if you can use a "standard" unit designed for powering a PC. These normally share out the total power with the highest power to the 5V output, but a 150W unit normally has 12V outputs rated at 2A and a standard 250W power unit could run to 3 or 4A on the 12V rails.