Ive been having heat problem with a circuit board which ive traced down to the LM340 5V regulator im using running too hot because its dropping 12V down to 5V, excessive power dissapation.

Does anyone know of any switching regulators that dont require the external inductor/ diode arrangement such and this one from dimension engineering? http://www.dimensionengineering.com/DE-SW050.htm I cant get this one locally in time, just wondering if anyone knows of a similar part or I might have to bite the bullet and make an ugly mess on my board with a DIP package and inductor & diode as well

 

I think you will find that All switching regulators of that size and type will require external components, as they are too large to go onto the chip.....Dissipation will always be a problem using linear regulators, that 7 volts at X current will have to go somewhere.Have a look at some of the Linear technology switching regulators, they only require very few external components.....Daniel.

 

You didn't say how much current you require. Are you using a heatsink?

The thermal resistance of the TO-220 case is around 60DegC/W. With a heatsink
you probably drop that to the 10-20DegC/W range. With moving air you could get
below 10DegC/W.

As was mentioned switching regulators require a diode and either an inductor
or capacitor. One of the switching regulator manufacturers was packaging
a complete switching regulator onto a TO-220 style PCB. Unfortunately I can't
seem to find the information in my component notes. You could try National Semiconductor, Texas Instruments or Micrel.

(* jcl *)

 

If you place several diodes, such as 1N4001's in series with the LM340's input, each will drop the voltage by about .7 volts. Reduce the input voltage and a lot of your dissipation goes away - and 1N4xxx diodes are only $.05 a pop.

 

I require around 300mA maximum current. At the moment the regulator is not heat sinked, I have tested with a heat sink and yes infinitely better. Unfortunately There is not room on the PCB for heat sinking easily, so Im thinking it may be just as easy to go for something like a LT1376 from linear technology.

I am assuming of course that these switching regulators will generate a lot less heat than a linear regulator? Inductor and capacitor choice is the only confusing issue.
Judging by the LT1376 datasheet http://www.farnell.com/datasheets/27602.pdf I figure about 10uH will do for the inductor

 

although.... I really dont want a noisy switching device on my board, because I have SSI clock lines and data lines on the board. They are not in the direct vicinity of the regulators though. I might try something like an 8V linear regulator in front of the 5V one to see if the heat is tolerable. This way the efficiency of both should be around 62 - 66%, rather than 41% when stepping 12V down to 5V

 

SMPS regulators are not that noisy...digital lines have lots more noise margin compared to analog signals. You can design the switcher to minimize ripple

 

I require around 300mA maximum current. At the moment the regulator is not heat sinked, I have tested with a heat sink and yes infinitely better. Unfortunately There is not room on the PCB for heat sinking easily, so Im thinking it may be just as easy to go for something like a LT1376 from linear technology.

I am assuming of course that these switching regulators will generate a lot less heat than a linear regulator? Inductor and capacitor choice is the only confusing issue.
Judging by the LT1376 datasheet http://www.farnell.com/datasheets/27602.pdf I figure about 10uH will do for the inductor

The temperature rise is apx. 60DegC/W * (7V * 300mA) = 126 DegC which
puts the junction temperature at 23DegC + 126DegC = 149DegC.
I believe the absolute maximum junction temperature for that device is 150DegC!!!

Can you do SMD parts? If so I would look at some of the LTC converters that run
in the 1-2MHz range. The higher frequency devices will use much smaller L's and C's.

If you solder QFN parts take at look at the TI TPS series of parts. You could easily
make a small TO-220 size converter on a separate PCB with the TPS parts.

The layout can be tricky for the high frequency switching converters. It is best to
find a reference design that is close to what you need and make minor modifications. When you do the PCB layout follow the recommended layout *exactly*.

I have a layout that uses the LT1616 at

http://www.luciani.org/works-in-progress/works-in-progress-index.html

scroll down to the wall transformer power supply. Above the wall transformer jack
is the LT1616 (U71).

(* jcl *)

 

Three terminal 1A switching DC-DC convertor 12V to 5V pin compatible with standard three terminal regulators PT5041A. They're $15 each http://focus.ti.com/lit/ds/symlink/pt5041.pdf

I've seen *somewhere* where you can use three terminal regulator in a switching circuit, I think it only works as a voltage reference but I can't find it just now.

 

The temperature rise is apx. 60DegC/W * (7V * 300mA) = 126 DegC which
puts the junction temperature at 23DegC + 126DegC = 149DegC.
I believe the absolute maximum junction temperature for that device is 150DegC!!!

(* jcl *)

I just had a look at the data sheet, I calculate the junction temperature to be a lot less, because the power dissapation is calculated using their given formulas for Psw ( switch loss) , Pboost (boost loss), and Pq (quiescent loss) . This gives me a dissapation of only 0.093W when using 12Vin, 5Vout, and 300mA current.

 

I just had a look at the data sheet, I calculate the junction temperature to be a lot less, because the power dissapation is calculated using their given formulas for Psw ( switch loss) , Pboost (boost loss), and Pq (quiescent loss) . This gives me a dissapation of only 0.093W when using 12Vin, 5Vout, and 300mA current.

My calculation was for the LM340 5V linear regulator that you said was running
hot.

(* jcl *)

 

ah ok, quite correct

 

well i like the circuit, works a treat... except... with a LM1117 3.3 regulator on the output ( I need this to drive a Lantronix XPORT device ), I get too much ripple ( 1.4Vp-p on the 5V rail). Ive tried bigger caps etc around the 3.3V regulator, with no real difference in ripple. Any ideas as to why this would be? Currently there is a 10uF tantalum on the input and 10uF low esr on the output of the 3V regulator. The 5V switching regulator has a 10uH inductor and 100uF low esr cap on the output. Is it just the type of load im wondering...

 

Is the LM1117 the adjustable one or is the fixed output?

 

fixed output. on further investigation its not the LM1117 causing the ripple, its the extra load. as i increase load my ripple increases. Im using a 10uH inductor, maybe ill try both smaller and larger inductor. The switching regulator is wired exactly as the diagram on page 1 http://www.farnell.com/datasheets/27602.pdf, using a 12V input. and approx 130-170mA output.

 

Inductors can not source current instantaneously so a larger cap can help also.

Larger inductors can reduce ripple current but the trade-off is slower response to load changes. When adding more capacitance be careful you stay within the stable ESR range-- this has bitten me before.

 

When adding more capacitance be careful you stay within the stable ESR range-- this has bitten me before.

Check your layout and the ESR of the output capacitor. For the output capacitor ESR information check pages 12 and 13 of the datasheet. What does the input voltage
ripple look like?

For that much ripple that changes with load current I would look at the ESR
of the output capacitor first.

(* jcl *)

 

The input capacitor of the 3.3V regulator is also part of the output capacitance of the switcher.

 

the input voltage is atrocious when the switching regulator is hooked up to it. otherwise it is fine. It is 12V coming from a linear reg. I havent got any low esr caps of large value here, so ill get some in and see what happens. how much larger than 100uF would I need to go though, this already seems big enough to me?

 

You can parallel caps for lower ESR