Question regarding pass transistors

[Resqueline]

The first schematic on page 18 now makes sense to me, but I am completely dumb founded how to mathematically analyse the circuit, since my knowledge of opamps is very limited. But I do get the jist of it, the op-amp changes the voltage applied to the ADJ. It is essentially running in differential mode, and will adjust it's output to try and equalize the two inputs I believe? How close am I?

Spot on in fact.
Let's call +Vout "common". The op-amp non-inverting input is connected to this.
R3 is the sense resistor and drops 1V at 5A.
The op-amp inverting input is connected to the positive side of R3 via R2, and to "minus 1.24V" via R5.
As the op-amp inverting input is pulled above "common" (by a rising voltage across R3) the op-amp output will swing negative and pull ADJ down via the diodes (LED lights).
At the 5A limiting point there's 1.24V across R5 (= 3.757uA) and R2 has to be set at 1V / 3.757uA = 266k Ohms. At 1A limiting it has to be at 0.2V / 3.757uA = 53k Ohms.
The circuit is not ideal, since the (somewhat significant) current that the LM contributes with is not measured by R3.

But Dave has some very sensible points. If you don't feel the need for having an adjustable current limit etc. now then just make the simple version (and fuse the mains input).
There's no point in biting over more than you can chew (at the moment). Making an ideal lab supply is also better/easier done without using these simple 3-terminal regulators.

 

[jackorocko]

One more question. Can this type of circuit be built on prototype board or are the currents to much for the traces in prototyping board.

One day I would like to try my hand at etching, but that would require me to learn how to do pcb prototyping software. Only one man, can't do everything at once.

There's no point in biting over more than you can chew (at the moment). Making an ideal lab supply is also better/easier done without using these simple 3-terminal regulators.

Yeah I did some more googling and found this chip. http://cache.national.com/ds/LM/LM2679.pdf $33 dollars though, but that is still a cheap power supply I think.

 

[davelectronic]

Pass transistor

You can build any permanent circuit on copper strip board, its no different to fully clad board used in pcb production. There is another type, or quality of matrix and strip board, there used as interfacing and are known over here in the uk as euro cards, all of these posses the same dialectric strength. I have a psu built about 4 years ago on copper strip board, and its never caused a minutes bother, prototype boards are fine for permanent builds, or until you try pcb production. Dave.

 

[davelectronic]

Pass transistor

The humble voltage regulator can be found in loads of electronic circuits, domestic, retail, industrial. I think there great versitial components. Yes you can build a psu from other descreates, transitors, op amps zeners to name a few, and both the humble VR and the descreates circuit will serve equally well, its the design that matters, components are only as good as the circuits there built in, a psu can be any thing you want it to be, simple, or bells and wistles, both as good as each other. Dave.

 

[Resqueline]

If it weren't for the holes the track would rise 20 deg C at 4.5A but with the holes narrowing the tracks as much as they do I'd guess at a 1.5A capacity.
So either put the current in & out at neighboring holes, distribute the current over at least 3 tracks, or solder an extra copper wire along the track.

 

[davelectronic]

Pass transistor

The most ive put through copper strip is 6 amps in a single track, it was fine, but currents exceeding that i probably would join two tracks, or use an additional link to be on the safe side, but permanent circuits i fine on matrix and copper strip, i have never had a problem. Dave.

 

[jackorocko]

Can someone here help me come up with a proper heat sink size for this circuit. The one like resqueline made?

I finally got my mill working again and am ready to make my heat sink. I have a couple 5x5 inch chunks of extruded aluminum about 1/4 in. thick or so. I know this should be big enough, but I really want to cut it down to the proper size.

 

[davelectronic]

Pass transistor

The forum has a thread on working out excact size of heat sinks for semiconductors, cant remember where i found it, but it was a sticky thread. Degrees C per watt etc, typical flat finned heat sinks are mounted externally, i mount a lot internally, but have good air flow cooling, only because i like to keep things small on some psu's. but an apperture in the case rear to allow cabling to get through the case to the heat sink, passive or air cooled. Dave.

 

[davelectronic]

Pass transistor

This looks handy, it also has other calculators for other app's Dave.

http://www.daycounter.com/Calculators/Heat-Sink-Temperature-Calculator.phtml

 

[(*steve*)]

The problem with a random chunk of metal is that you have no idea what thermal properties it has.

In most simple thermal calculations, we model the interface between the function and the atmosphere as a simple set of "resistors" in series. Knowing the value of those "resistors" is critical. In the case of a large chunk of metal, another factor may need to be taken into account. This is the thermal mass or inertia of the system.

A typical heatsink is designed to radiate heat as fast as possible. The surface area is large compared to its mass (but not too large...). Heatsinks like this will reach thermal equilibrium quickly.

If you have a solid block, the surface area is small compared to the mass and this will heat up slowly before reaching an equilibrium. A system like this is slow to react, and thus any temperature measurements need to be taken with some care (essentially you need to allow the system to reach equilibrium).

This is effectively modelled as some sort of "capacitance" in parallel with the "resistors".

The modelling of such things is more than this bear of very little brain wants to do.

If you want to model the heatsink in a very simple way, you will need to determine a value for its thermal resistance in degC/W. The simple way id to attach a device and have it dissipate a known amount of power. When the heatsink reaches an equilibrium temperature, measure the difference between the device's case temperature and the ambient temperature and divide by the power.

Once you've done that you can use this in the simple calculations and the results will normally be very close.

 

[(*steve*)]

Oooh, Dave, that site you mention has 5 degC/Watt as the thermal resistance junction to case for a TO-3 device.

I thought this was a bit high and some random datasheets confirm that values between 1.5 and 2 degC/W are far more typical (which is pretty much as I recalled)

 

[davelectronic]

pass transistor

Yes thats a bit high in thermal resistance i suppose, the site looked interesting, i guess all that glimmers is not gold, i should have looked closer. Dave.

 

[jackorocko]

If you have a solid block, the surface area is small compared to the mass and this will heat up slowly before reaching an equilibrium. A system like this is slow to react, and thus any temperature measurements need to be taken with some care (essentially you need to allow the system to reach equilibrium).

Why does everything have to be so hard? It is not a block of aluminum, it is a beefy extruded heat sink(that means it has fins with maximized surface area). I bought a foot section and had it cut up into equal pieces to be used on a amplifier kit I bought and built. Since I only needed two, I have one extra now which I would like to use with this project.

I will post a pick soon, but I don't know how I am gonna find out it's thermal properties since I bought it off ebay.

 

[(*steve*)]

OK, your description made it sound like a large block of metal.

I have a couple 5x5 inch chunks of extruded aluminum about 1/4 in. thick or so

If you can find a similar heatsink, you may be able to take a guess at the thermal rating.

However, there is a law of diminishing returs. This means a heatsink twice as large is not twice as good. Conversely, cutting a heatsink in half doesn't reduce it's efficiency by a factor of two either.

The best thing to do is to do the calculations to determine what the minimum heatsink is that you require, then compare what you have to heatsinks of that rating. Maybe you can then cut it down a little.

All other things aside, I'd leave it as large as practicable because the life of a semiconductor is significantly increased if you can keep it cooler. Also you can make a heatsink smaller, but it's a lot harder to make it larger.

If you can place a device with a known dissipation on the heatsink you can arrive at a good estimate.

 

[davelectronic]

Pass transistor

Hi all.
Ive put a couple of heat sinks together, and what i know they can handle.
As you probably guessed iam a fan of air cooling using axial fans, small 20mm large 120mm.

I have added the misses cigarette papers to give you an idea of scale, this next bit might sound a bit clinical, we dont need to enter the lab for every electronic procedure, my way of going is looking at production items to get an idea of power and heat dissipation and thats amps psu's etc any other components that are heat sinked.

The smallest in the picture will easily handle between 0.5 amps to 1 amp, the one down from the largest will handle 2.5 to 3 amps, and the biggest will cope easily with 5 to 6 amps, convert to watts if you want, these are guesstimates, but on experience of other supply's and heat measurements ive taken with a thermal couple connected to one of the meters i have, i normally measure the semiconductor tab and the middle of the heat sink, and for about 30 minutes at full rated load usually halogen lighting and i check simultaneously checking voltage drop and current before during and after loading the supply , none of it has to be rocket science.

The heat sinks in the pic's are based on axial fan cooling, starting at a rise to 40 or 50 degrees C. I use some supply's made and built this way daily, and have one in the electric cupboard that runs 24/7. It does not have to be difficult, with all due respect. Dave.

As simple as these two supply's are both can cope easily with 5 amps. One makes use of an AMD CPU heat sink, the other is an LM338K.

 

[davelectronic]

Pass transistor

Hi jackorocko.
Just to give an idea on size for power, these units are simple, one contains the transformer, the other next to it is a 2 amp unit, and even the tiny one on top is capable of 2 amps, its driving 2 CB radios i use. A bench supply can be any configuration you want it to be. Dave.

Yes there simple supply's but it gives an idea on heat sinking.
The two machine screws in the second pic down, the one with the small heat sink on bolt up the thermal switch, and the smallest is a chip set cooler from an old pc.