12V in switchmode power supplies

One of the major problems with my big system is coming up with the money for it. In order to help with the funds, I need to start producing high quality modular 150W/channel car amps in fairly large quantity. To do this, I must have a design for a SMPS with +/-45V out capable of at least 250W. Has anyone built ESP project89? PCB layout for it anyone? I noticed he does not have PCBs for this project for sale. Are there any other switchers I could make or maybe even complete units or kits for sale?
If you are going to produce them then you will either have to buy someones design or make one yourself.

It won't blow up. Not more than the amp board. It's the same thing; high current - wider traces, high voltage - wider spacings. Since it is somewhat higher frequencies involved you will have to think twice about decoupling and noise reduction. But it is NOT an impossible task even for the DIYer.

I say go for it!

Unlikely, I have a fairly specific design in mind to keep the modules as compact as possible and so they can be connected to form a unit almost undecernable from a single amp. I'm using TO-3 output devices, and they must be completely hidden because of their live cases. I need a cross section something like this. There will be a 92 cm fan at one end, and a panel with input, volume, gain, Xover controls and power connections on the other. There will be holes in the sides for air from the fan that goes inside the amp to escape from. The - power terminal will be part of the case, and the + will be a strip inside the area where the top of the TO-3's come out which will come out the end to be accessed. This strip will only be in the dominant units, the slave units will be connected via lateral cables within that corresponding area. There will be a plug at each end to connect dominant units at the the amp's input level (after the preamp, so that only adjustment of the controlls on the end unit will be valid. Because of this, each preamp will need to be capable of driving lots of units). There will be a plug on the side to connect slave units. The speaker outputs will be on the sides. The units will be connected end to end by long bolts (or threaded rods with nuts) and side by side with metal plates that screw on. It looks like this might require welding on the fins, but I don't have a TIG welder, nor do I have access to one, but I will when school starts. Are there any aluminum extrusions that look like what I need? I think not. It's OK because I need to keep cost down anyway.


  • crossec.jpg
    35.6 KB · Views: 908
Make sure that current-mode control is utilized or the output FETs will fail due to core saturation. This does not happen immediately but slowly over time by not subjecting the core to equal and opposite Volt-second products or a DC current. This is the trickiest part of switchers especially with dynamic power demands such as audio. Also keep in mind the skin depth at whatever frequency you are running. It may be necessary to use parallel smaller guage wire for the windings in order to minimize the copper losses. Good Luck!!!



2001-02-01 2:43 am

Look for my old replies about smps or car amps. There are some references about SMPS power supplies there, inclunding a complete project, in Audio Amateur, for a car amp using +/- 45 VDC SMPS.


Using current-mode control in SMPS power suplies in this kind of SMPS using MOSFETS is not necessary to avoid flux imbalances, and, if I remember well, the design is more complex. You can see some information about the matter in "Switching Mode Power Supply Design", second edition, chapter 2.25, by Abraham I. Pressman - ISBN 0-07-052236-7. Actually, I have some designs using push-pull SMPS running since 1997 with no problems. Your tips about skin effects are very usefull and I just add the design and construction of the power transformer (where the perfect simmetry is necessary) have to be very well made.

Without current-mode, there is no way to prevent flux imbalance and there is no way to prevent it from happening without it. At least use a DC blocking capacitor to avoid DC currents from flowing through the primary. The Pressman book is also highly recommended by myself. I own a copy of the second edition. Also the "Switching Power Supply Handbook, 2nd ed" by Keith Billings is also highly recommended. Also for reference, "The Inductor and Transformer Design Handbook" by Col. Wm McLymann. However, for a true beginner, the first two books cover transformer and inductor design fundamentals and basics rather well.



2001-02-01 2:43 am

As Beanz recommend, the Billing's book is a very good start. See the chapter about self oscillating supplies. It's an easier way to start learning to work with high frequency transformers.


Since you have the book, I think you might read the chapter to see what I'm saiyng about flux imbalance, specially the last two paragraphs (2.2.5, page 45).

An electrolytic is not the proper choise for this application because of it polarization. A film cap should be fine in this application. It is placeed in series with the primary winding of the transformer. Dual primaries may also be desireble too. Make sure that your output filter caps are rated for the ripple current specified for the output filter inductor.

As for flux imbalance and saturation of the core: It is almost guaranteed using MOSFETs and no combatant for their different Rds(on) values. A slight difference in the Rds(on) of the transistors will lead to catastrophic failure. When the core gets into the non-linear saturation region, the core will no longer support flux and the transistors will be driving a low impedance (short circuit) load. The transistors will fail. The process is much slower however with MOSFETs which have an increasing "on" voltage with increasing temperature. This helps to minimize flux imbalance on a short-term pulse by pulse basis but over time the core will eventually walk up the hysteresis curve and make it to the non-linear region. A small primary resistance is often added in series with the power transformer primary in car audio SMPS. This will slow the process even more. But, a few billion cycles down the road...the MOSFETs will go. Unitrode (TI) has some current-mode PWM controllers. They also have PWM drivers which have internal high-current totem-pole outputs for driving the high capacitance gate charge of a MOSFET.

"A small primary resistance is often added in series with the power transformer primary in car audio SMPS."

And a DC blocking capacitor would provide just that.

Anyway, I need these to be very reliable and last practically forever, so I need some way to combat this non-linear thing. I assume almost perfectly matched parts are a must here (I wonder if I could design a current mirror into the thing:) ). This is a push-pull SMPS, for those who don't know. Rod suggested using 6x0.4mm wires (what AWG is that?) for the primary, I think I'll use 7. Please describe how I can keep it from saturating.

10AWG is much too big of a wire to use for the primary at 50kHz. This is a number to indicate the equivalent amount of copper cross sectional area needed to carry the primary current. Due to the skin effect at 50kHz, several smaller guage wires are used to achieve a total copper cross sectional area about that of a ten guage solid wire. There are plenty of sites on the web that will give you this information, but here is some:

10AWG=11740 circular mils or 58.13cm^2 x 10^-3
resistance=.9987 ohms/1000'
current capacity = 13840 [email protected] Circ.Mil /A
= 20768 [email protected] Circ. Mil/A

20AWG=1246 circular mils or 6.315 cm^2 x 10^-3
resistance=10.13 ohms /1000'
current capacity = 1365 [email protected] 750 Circ. Mil/A
= 2050 [email protected] 500 Circ. Mil/A