Sign in to follow this  
vainajala

Switching power supply

Recommended Posts

Ante,
All the big ordinary transistors have a guaranteed gain of only 5 at 10A collector current. Expensive ones (they select them) have only a little more gain, 10 if you're lucky.
Darlingtons don't fully saturate and therefore will need a big heatsink.
So I believe that the best thing to do is put 3 or 4 ordinary (MJ2955) transistors in parallel and benefit from their much improved gain (maybe 40) at only a few amps of collector current each. They will saturate well and won't need much heatsink. Their base resistor will be reasonable because of their high gain. ;D ;D

Share this post


Link to post
Share on other sites

First of all, thanks for your help, it's surely needed ;)

Audioguru,
What do you think about using three MJE2955's in parallel, with 0,1R emitter resistor in each? By the way, are the "equalizing resistors" needed in switching applications?

As far as I can understand in this transistor's datasheet, the HFE=~40 at IC of 3A. So with three transistors, max. 9A could be drawn with HFE ~40.

So base resistors: 45V / (3A / 40 * 1,3) =460 ohms, and wattage: P = 0,075A2 * 460 ohms = 2,58W.
Am I correct?

9A would be actually more than enough, the circuit's current limit start at ~7,5A...

MOSFET(s) might be good idea. Could P-channel fets be easily used in this circuit? IRF9540 or IRF5305 maybe? ???


MJE2955: http://www.onsemi.com/site/products/summary/0,4450,MJE2955T,00.html

post-2509-14279141649203_thumb.png

Share this post


Link to post
Share on other sites

audioguru vainajala,

The collector outputs in the 494 can sink 200mA maximum, will this suffice as base current for three transistors? What makes you think that a Darlington can

Share this post


Link to post
Share on other sites

Vainajala,
I just now noticed that your very high supply voltage will be putting more than 50V on the TL496's output transistors (42V absolute max. rated) when your SMPS has no load!
When paralleling the output transistors, connect their bases together, connect the TL494's output transistors together and use a single base resistor. Also use a single 15 ohm resistor from the bases to the + supply. The emitter resistors may not be important if you don't match the transistors, read on.
Your calculation for the base resistor is close. The voltage across the base resistor is 45V - 1.5V (max. Vbe of MJE2955) - 1.3V (max. saturation V of TL494 outputs = 42.2V.
Therefore the base resistor is 42.2V divided by [9A divided by 25 (min. hFE of MJE2955 at Ic 3A)] = 117 ohms. I don't use the 1.3 correction term because TI also doesn't. Using 120 ohms, its 352mA is within the max. recommended 400mA of the TL494's paralleled outputs. Its power dissipation will be 42.2V X 352mA = 14.9W full-time, or less than 7W using the SMPS's PWM. Note that many power resistors get extremely hot at their power rating (what temp. does nichrome melt?).
If the output current is reduced to 5A and a 4th MJE2955 is used, then the base resistor can be about 270 ohms, with much less power dissipation.

Ante,
A darlington transistor's output transistor cannot have a saturation voltage that is less than its Vbe, which for an MJ2955 could be as high as 2.5V. By itself, its saturation voltage is half that. So a darlington dissipates twice the heat.
The argument about emitter resistors with paralleled transistors continues. In this circuit, they are not necessary since the transistor with the most gain can hog the whole current without damage. Also, as its collector current increases, its gain decreases down to the gain of the least-gain transistor. Also, the least-gain transistor will conduct less current, so its gain will be more. Therefore they will share the current, but not equally.

Share this post


Link to post
Share on other sites

Ante,
The darlington transistor that you posted has a max. REVERSE voltage rating of 8V, when its eb junctions have avalanche breakdown like a zener, with damaging results if the current is not limited. Its forward Vbe for BOTH transistors is 2.5V max. at 10A Ic.
I was talking about the Vbe for just the output transistor of a darlington, since when the input transistor conducts, it shorts the output transistor's base to its collector. So the output transistor's saturation voltage cannot be less than its Vbe.
In this darlington, its max. output saturation voltage is 2.0V at 10A Ic, with a whopping 0.5A of base drive. A single high-current transistor can saturate with only 1.0V or less (but not a 2N3055, it is lousy at high currents, I just looked at its 3.0V max. saturation voltage at 10A Ic). Also, look at this darlington's saturation voltage at low currents, it is always more than a base-emitter voltage drop of about 0.75V. A single transistor saturates with less than 0.1V at low currents.

Transistor manufacturers have a wide spread for hFE values, so that they can still sell the lousy ones. If they could make them all the same, then they would advertise it on their data sheet.
Look at the 40 to 400 hFE spread on your posted darlington, and even a 2N3055 has a spread of hFE from 20 to 70. The spreads could be much more because they don't spec. a max. hFE. If they can't sell the higher hFE ones at a higher price, then they throw them in with these "ordinary" ones. Since you don't know how much hFE that you are getting, then design the base drive for the worst one and hope that it conducts at least some of the load current. Or use emitter resistors to equallize the gains of paralleled transistors.

Share this post


Link to post
Share on other sites

audioguru,

I still think the Darlington is the better choice. We are talking about pulses that are of different length (PWM) and between the pulses the transistor

Share this post


Link to post
Share on other sites

About darlingtons...

I tried a darlington transistor configuration with dual MJ15004 transistors. 55W halogen served as a load, again. Unfortunately I didn't measure the performance of that circuit (Iwas in bit of a hurry...), but the output-transistor got very hot (i almost burn my finger when touched it ::))
I experimented some different base resistors from 330R to 10K with little or no impact on "heating-effect".

I see you're discussing about darlingtons, well how about Sziklai configuration (also known as the complementary Darlington)? I suppose that the Sziklai-type has lower saturation voltage when compared with ordinary Darlington... I found this information from some finnish electronics website. Text was in finnish, so I put a translation in a form of a picture :)

Some more Sziklai info here:
http://www.ampslab.com/c200cfp.htm

post-2509-14279141650086_thumb.png

Share this post


Link to post
Share on other sites

Vainajala, Ante,
In order to keep the base resistor cool, try a PNP darlington transistor, like a TIP145T, which is rated at 60V. But the 100V rated TIP147T is cheaper, with the same specs. Their data sheet is here:
http://www.fairchildsemi.com/ds/TI/TIP147T.pdf

With a 10A collector current, and only 40mA of base drive, its max. saturation voltage is 3.0V. Its base resistor voltage is calculated to be 45V (supply) - 3.5V (max. Vbe saturated) - 0.2V (TL494 saturated) = 41.3V. The base resistor is calculated to be 41.3V divided by 40mA = 1033 ohms. Using 1000 ohms, its power dissipation is 1.7W continuous, or less than 0.85W with SMPS PWM.

If you are unlucky and get a TIP147T that doesn't saturate well, then it will dissipate about 15W, so will need a medium-sized heatsink.

Share this post


Link to post
Share on other sites

I decided to test N-channel MOSFET as a switch transistor, and it worked well! Typical efffiency was 77% with 21W and ~74% with 55W.
The heatsink was so small that the FET heated pretty quick with 55W and higher loads, so effiency in these loads is more "approximate".

I had to make modification to circuit to drive the FET: I tied the emitters to the FET's gate (instead of collectors) and collectors to Vcc.

This is very promising result, and at least using FETs, the base resistors can be eliminated. So I'm considering to use two or three N-ch FETs in parallel instead of BJTs...

NTP45N06L N-ch FET
http://www.onsemi.com/site/products/summary/0,4450,NTP45N06L,00.html

Share this post


Link to post
Share on other sites

Vainajala,
A darlington transistor does NOT saturate with 2 Vbe voltage drops, when used in a common-emitter configuration that your project needs. Your article refers to the darlington being used in a common-collector (emitter-follower) configuration, where it does have 2 Vbe voltage drops. The Sziklai connection has a single Vbe voltage drop in either configuration, like a darlington does as a common-emitter.

The output transistor of your home-made darlington heated too much because maybe it didn't have a base-emitter resistor to turn it off quickly.

That's good that you got a MOSFET to work well. It sounds like you are using the TL494's transistors as emitter-followers which drive the MOSFET as a source-follower. That configuration has an efficiency loss that is due to the MOSFET's gate-source turned-on voltage.
A P-channel MOSFET will be much more efficient when used in a common-source configuration, with its gate driven though a voltage-divider (so that its max. gate-source voltage rating is not exceeded) from the collectors of the TL494 grounded-emitters transistors. If its Rds is low enough, then maybe it won't even need a heatsink. The voltage-divider could have a zener in series with the collectors to protect them from over-voltage. The voltage-divider resistors must have a fairly low resistance to charge/discharge the high gate capacitance quickly.

Share this post


Link to post
Share on other sites

Vainajala,

To get a better efficiency try to use a MOSFET driver as I suggested earlier the MAX 626/627 or similar. This circuit will assure that the MOSFET:s switches on and of properly.

Ante ::)

Share this post


Link to post
Share on other sites

Audioguru, thanks for information! I must try this when I have more time for this project...

Ante, maybe the MAX626/627 driver could be used to drive the dual P-channel FETs then? And, should I get inverting (MAX626) or non-inverting (MAX627) driver?
Also the max. 18VDC supply must be arranged for the MAX62*, that's one concern too.

I'll maybe use 2-3 of IRF9540N FETs...

post-2509-14279141651443_thumb.png

Share this post


Link to post
Share on other sites

vainajala,

Yes you have to arrange the supply and the TTL level for the input to use these drivers. The IRF9540N is a P-channel and it will work whit the MAX627. There are other MOSFET drivers out there just search for

Share this post


Link to post
Share on other sites

omiela, please post with a new subject. We would like to hear about your ideas, but only those who are interested in the switching power supply will read this thread. Feedback will be limited.

MP

Share this post


Link to post
Share on other sites

Hello every body
I have studied the 0-30 VDC STABILIZED POWER SUPPLY WITH CURRENT CONTROL CIRCUIT and discussed it with audioguru. I think that is a very good circuit.
However, if it can be realized by SMPS (switch mode power supply), it may be much lighter and the project may be challenging.
As a result, I search some circuits(P29-P54). But I don't know how to make these circuits to be variable voltage output with current limitation.
If u have any idea, please tell me, I am going to build this very soon.

Thanks!!!!!!!!
http:
%7Boption%7D

SMPS_POWER.pdf

Share this post


Link to post
Share on other sites

Hi Ante
I mean P29-P54 of the attached pdf file. This file provide some circuit, but I don't know weather they can be adapted to be variable output.
I think it is not necessary to variable from 0V to 30V. Instead, it can vary from 3V-25V. Since I know it is hard for a SMPS to provide 0V.
And u can read page 53. This circuit uses the R5, R6 and TL431B to provide reference. Can I just simply change these three items to provide variable output?
Also can it be current limited.
Thanks very much

Share this post


Link to post
Share on other sites

Hi Ante
Good.
I am going to build a power supply without transformer.
But I still have some problems
1. Like page 53, TL431B is used to control the PWM. Its voltage is 2.5V. What is the variable range of R5 and R6? How to calculate that? Also, how can I determine the output range of a circuit like that?
2. Also can u give me some ideas about adding the current limiter? I have little experience about that. What kind of current limiter? How to calculate?

Thanks very much

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this