Smps

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I would recommend a Push-Pull converter in Current Mode..
With a 70V secondary with Center Tap... This way you get you + and - 35V ..... 6 Amps is straight forward.....
What is the mains voltage in your country????
240V ???
The toplogy and basic circuitry has been around for-ever...All you need to do is design the values of the components...I can help with that..
Look at the LTC3721-1 controller.....

Chris
 
A push pull converter is not the best solution for what you want.
STK4192 is a 100W IC so that falls right in the range of the flyback converter.

A simple flyback converter using UC3842or(3) can be made with very few components.

UC3842 ON semi's datasheet has some good examples and info if your intrested.
 
A 100W chip will certainly need more than 100W that a flyback will output. I would recommend NOT going with a center-tap push-pull because at 240VAC, your main switching transistors or MOSFETs would have to be able to stand off close 1kV. 240V x 1.414 x 2.6 = 883V.

Let's say the 100W chip needs 200W to run properly, and we assume you're doing two channels for stereo. In that case, you would need a power supply capable of outputting over 400W comfortably, and 600W without too much difficulty. At power levels over 150W, the peak drain and rectifier currents of the flyback topology become way too high to.

At this point, I would recommend a half-bridge, very similar to what is inside almost every AT & ATX pc power supply, or possibly a full-bridge topology for peak power levels over 500W. For your PWM controller IC, I would consider any of the five most common PWM ICs suited to run push-pull, half-bridge, or full-bridge layouts: TL494, SG3524, SG3525, MC-33025, or the SG1846. The '494 and '3524 are voltage-mode only (the '3524 does have a current-sense amplifier, but it is current-limiting, not current-mode), the '3525 is a voltage-mode, but pin access at the chip's comparator enables pseudo-current-mode operation. The MC33025 can be run either voltage- or current-mode, and the SG1846 is strictly current-mode. The '33025 is a newer generation of, and close cousin to the '3525, but with faster outputs and higher-speed oscillator (1MHz).

There are MANY threads on deisgning SMPSs in the Power Supply Design forum, and I encourage you to do a search function to see them all. A good threat, like this one, will usually attract the PSU Gurus to throw their two cents in, like EVA, jackinnj, poobah, richey00boy, and others. EVA always has some nuggets of golden advice she showers us every so often, and has designed many AC Mains powered SMPSs for audio and other uses.

I did a SG3525 controlled half-bridge a while back, and it actually worked! Since then, I have cannibilized its parts from it for other projects. For your needs, +/- 35V @ 6A is not unreasonable, nor uncommon. 70V x 6A = 420W output. Taking a worst-case scenario, if your psu is 80% efficient, you will need at least 525W to get the full desired outout from the your psu to the amp.

Looking foprward to your reply. Anyway, I hope these comments give you some insight as to how to best proceed.

Regards,

Steve
 
Practicality of Flyback falls away over 50 Watts... unless there other motivating factors. Your inductor will be HUGE.

You are definately in P-P/ half bridge land.

I wouldn't consider a custom switcher for a one-off DIY design... unless learning switchers is an ambition. The reason is that there are really no "off-the -shelf" magnetics elements for you. If you wish to learn switcher design AND design and make your own magnetics, they go for it!

Don't expect it to be a whole lot cheaper than a linear supply... probably the opposite. Cannibalizing parts is an option... but that is easier AFTER you know what you are doing.

Methinks N-channel knows a wee more than he lets on about SMPS!

:)
 
Because you have modified a SMPS without knowing exactly what you are doing. You have a (design) fault either in the transformer or in the output filter.

Also, you are quite lucky because the switching transistors are just overheating instead of exploding.
 
a2005r2003 said:
tanx all of u.
i finaly convert atx 250w psu to give 2*35v. but its driver Transistors became very hot (with heatsink & fan). it is very abnormal. why did the tr. become very hot?!?!?!?!?!?!?

I tried this too, and have same problem.
Number of primary turns ( on the output ransformer) is the same as on original - 40 turns, onlu the secondary is modified ( I put 2x15 turns ). Also increased output inductor, but problem is still here.


P.S.
My English is bad, so I hope you understand me.
 
Jack,

TNX for the comps. :) I may know a thing or two, but from what I have read from EVA's posts, it appears she probably knows more than all of us combined. :worship:

a2005r2003:

I would recommend Coupled inductors for the output as EVA recommends, but you need to keep in mind that the BIG yellow toroid immediately after the output rectifiers is an energy-storing coil, and not a filtering coil. Its function in an AT/ATX PSU is much the same as in a srtaightforward buck converter: It integrates (averages) the output voltage(s) over time. Together with the filter cap, it forms an LC filter that will affect your supply's response to chaing line voltage and output load. This coil is best wound bi-filar, with one winding for each output. This will provide excellent cross-regulation between the equal, but opposite outputs (+/-35V), and help to keep things symmetrical about the zero-volt line. The filter inductors after the first caps can be wound on individual cores, as they do not perform cross-regulation. Then you are free to place additional caps after the filter inductors, effectively resulting in an L-C-L-C output filter.

Choosing the L-C values is critical, because as previously mentioned, it affects the PSU's response to changing conditions, but also affects things on the primary side, because the inductance in the energy-storing coil is "reflected" back to the primary side. This can have adverse affects on the switching conditions of your main switching transistors, as EVA also mentions.

Did you, by chance, change anything in the driver transformer? I hope not, as this type of drive (proportional-base drive) is specifically suited to the transistors and inductance(s) in the original PSU. It is NOT suited for driving MOSFETs, nor is MOSFET drive suited for driving bi-polar NPNs.

The driver transformer's design is critical, as its core must saturate much more quickly than the main transformer. Indeed, part of the primary current path is wound on this core. This helps the supply start up before the PWM control chip kicks in. In the start-up mode, the primary side essentially runs like a free oscillator, but only for a second or two, as the PWM action of the TL494 has not yet taken control. Because the PWM chip is on the secondary side, there is no start-up circuit for it like there would otherwise be if it were on the primary side, but that is another issue. As the secondary side voltage begins to build up (usually within the first few cycles) voltage is applied to the PWM and things go from there.

The reason the values of the inductance of both the driver- and main- transforemer are critical is that they determine the free-running frequency in start-up mode. Changing the secondary windings of the main power transformer definitely will afftect things, and changes will need to be made elsewhere in the whole circuit to bring things "back into balance". Start by checking out the feedback compensation at the PWM's error amp input. Following Marty Brown's chapter on Feedback compensation, you can choose compensation components based on your new configuration. This would be a good start. Anyway,. I'm beginning to babble here, so I will stop for now and catch up with my thoughts. As EVA said, you're lucky your main transistors merely got hot, and didn't let the magic smoke out (voilently)!

Hope these provide some insight.

Steve (N-Channel)
 
The main transformer will be OK as soon as the primary windings remain exactly the same, wire insulation is not damaged, and the core is not damaged during disassembly. Any air gaps in the ferrite due to a cracked core or due to bad assembling will cause trouble (of an overheating or exploding flavour).

The turn ratio has to be carefully chosen so that the PSU is capable of producing the required output voltage with the minimum desired mains input voltage.

For a 160V AC minimum input and +-35V output a quick calculation tells us to use...

176*1.4142~=250V (mains peak value)
250V/2=125V (it's a half bridge)
125*.8=100V (accounting for max. duty cycle and rectifier losses)
n=100/35=2.86 (turn ratio)
40/2.86=14 (turns in each secondary)

So with 15 turns you did the right choice.

Calculating an output filter is not so simple, though.
 
Output cap

M-tech

If you have Pressman's "Switching Power Supply Design" refer
to pg #448 if not see below:

Quote:
Using a constant of 65 x 10E-6 (This assumes that over a large range of aluminum electrolytic capacitor magnitudes and voltage ratings, that ResrCo is constant and equal to 65 x10E-6) End Quote:

Co = di/Vor where di is twice the minimum output current and Vor is your required output ripple voltage.

His example is based on 1 amp min output current and .05 Vor


Co = 65 x 10E-6 X (2 X 1 amp) / .05 volts = 2600 ufd

chas1
 
After Several Years :
I Can Modify PC PSU and It is Very Important than U Use Original Driver Section( don`t modify it) and use GOOD Feedback

Hi, I got lot of PC ps from old dell p3 servers and sun and have pretty good specs...can u help me out how to modify it for higher voltage output like 24 to 35 volts. or if possible convert the +/- 12vdc output to at least +/-24v with equal amperage?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.