Simple'ish Transistor Current Limiting Circuit Help Needed

[beaker353]

I am in the process of replacing a number of switching power supplies in some industrial communications gear that is been in continuous operation for over 14 years.  Within the past three months, two of the 12 power supplies simply stopped dead without warning.  Considering all the unit have sequential serial number, been on the same length of time, equal loading and environment, I decided best to just replace all the supplies as the other 10 are probably not too far behind.  Being that the  power supply originally designed is a rather odd size I am limited greatly what I can replace it with physically.  I purchased three different contenders that fit the electrical spec and physical spec.  The supplies all work beautifully with the exception of how they deal with the initial high current draw when the units are powered up. 

The old supplies current limited just like a typical bench power supply would, dropping voltage across a resistor until the current levels dropped below the max.  But the new supplies all do this “hiccup” mode thing where they pulse on for a brief period, shut off, wait a second or so, then try again.  I guess it does this to prevent damage from a dead short, but it pisses the living hell out of my equipment.  My thought was to build a small current limiting circuit that I can place in unit between the power supply and the primary circuits.

I’ve found a number of rather simple but reliable sounding circuits on various websites that use one or two transistors, some diode, and some passives to do exactly what I need.  I have an OK understanding I think of the basic concept of the circuit.  As long as the voltage across the sense resistor is small enough the transistor gate is left open and current flows as normal.  When excess current on the sense resistor causes  enough voltage, the transistor starts to close and current has to flow through a resistor that drops voltage to limit the current. 

http://www.radio-electronics.com/info/circuits/diode_current_limiter/power_supply_current_limiter.php
http://www.physics.unlv.edu/~bill/PHYS483/current_lim.pdf

However, I am at a loss to determine what specific transistor specs I should be looking for, or the values and ratings of the diodes or passives.  I’m still taking EET classes a local college when I can find holes in my work schedule, and being able to analyze this circuit is beyond what I know at this point.  (I’ll be taking AC and Digital Logic I in the spring.)  Not to say I’m afraid of experimenting and blowing stuff up till I find the right specs.  Nothing has caught fire exactly, but I’ve definitely smoked the room up pretty good a handful of times.  I would just rather love to learn the formulas and math needed to proper engineer this verse rather than just hack it up.

Here are the specs I’m aiming for. DC out of the power supplies is 30Vdc.  I need to limit the supply at 2.4A.  I suspect this limit would not be in effect for more than 10 seconds during boot up.  It should reach normal current levels of 1A within 60 seconds as all the caps across the system charge up.  I would like to limit the voltage drop across the limiter to under 1V if possible while in non current limit state, the lower the better.  I don’t need to go cheap, I have a $15 budget per limiter for 12 of them plus a couple spares. I have a decent amount of physical space I can use (about two deck of cards stacked) and I have space to bolt to the aluminum case to heatsink if needed.  Anyone up to dishing out some education???

-EM

[Kremmen]

Ok, putting other issues to one side, here are answers to your first current limiter:

- the pass transistor need to handle the max incoming voltage for safety (it is not under the full voltage stress during normal operation, but will be during cold start)
- the reference zener diode defines the maximum output voltage. It is the zener breakdown voltage minus pass transistor Vbe (~0.6-0.7 V). Dimension the resistor so that sufficient base current can flow, minding the transisfor hFE. Check zener power loss at that current because it will dissipate all of the power when the psu is idle.
- current limiting is based on the voltage losses over the emitter resistor and the 2 diodes, respectively. When current is drawn, the I^2R loss over the resistor increases until it mathces the voltage loss over the 2 diodes (say 1.2 volts). At that time the diodes start conducting stealin some of the base current and limiting the collector current to that particular maximum.

For the 2nd circuit you already have the explanation in the linked document. The mechanism is the same in both cases, only augmented by an extra transistor in the 2nd case.

[beaker353]

Thanks for the information.  Other than driving a P2N2222AG through a 2.2kohm resistor to turn on a relay from a uC, I have very little background with transistors and knowing how to read specs.  Using the P2N2222AG as a reference (I pretty sure it doesn't have the needed capacity for this application, but just for example) I see specs for max C-E voltage of 40V, C-B of 75V, E-B of 6V.  I'm almost guessing here, but would it be the base-collector voltage that would need to be spec’d higher than my supple voltage? Is it also safe to assume that the max continuous collector current would have to be more than the 2.4A I want to limit at?  Keep in mind I am far away from a solid state course and really don't have the background yet to really understand half of what you laid out.  Specific part numbers and values could help me piece this problem together.

-EM

[Zero999]

The transistor needs to be able to pass the current safely but power dissipation is also a factor. When shorted the transistor will have to burn 30*2.4 = 72W of power for 10 seconds so you need a suitable transistor and heat sink.

How often will it be subject to the overload condition?

T%he good news is, the transistor only dissipates 72W for a short period of time so it doesn't need to be continuously rated.

[Kremmen]

OK, here are some basics then:

Specs:
If the datasheet says voltage between base and emitter is 6 volts, then it must be speaking about absolute maximum ratings. These ratings are the no-not-exceed numbers or else. During normal operation transistor Vbe is very close to a diode forward bias voltage of ~0.6 volts maybe a bit more. This voltage is our target here as well, you can forget the breakdown limit of 6 V. The other significant max rating is the VCEmax that must not be approached too closely. In this kind of low voltage circuit where the decision is not costly, i would leave a considerable margin and select something tolerating about double the needed voltage. That gives plenty of headroom for sporadic small spikes or similar disturbances. So, pick one with VCEmax about 60 volts or so.
The next important spec is max continuous collector current ICmax. Again, this datasheet figure should be comfortably larger that your expected maximum current. Most power devices will tolerate short surges well above the continuous rating, but you as the designer need to be aware if such is the case and design accordingly. Here the temporary max current is 2.4 amps. The continuous current will be lower, but as you don't know exactly how long you need to operate on the limit, best specify for continuous 2.4 amp current. Again, not a costly decision for this particular device.
The next item to check is what is called the safe operating area. This diagram outlines the borders for Vce vs Ic, inside of which the transistor operating point must stay in order to avoid the destructive phenomenon of second breakdown. For this kind of application you will want to use the DC curve, i.e. continuous state.
The final item for a quick selection is to check the large signal current gain of the potential device. This is noted in the datasheet as the hFE. for small amp transistors it can run into hundreds but for a large power device it could be as low as 10. This is the ratio between base current and corresponding collector current. Usually hFE decreases with increasing Ic so you should check the datasheet current for the given hFE.


On to the current limiter:
This is an archetypical linear regulator that is reliable and performs OK for you if it is constructed correctly. The main drawback of this kind of circuit is that the regulation takes place by dissipating all of the excess power as heat. You can use the emitter circuit of either of the linked circuits, calculations are the same for both. The 2 transistor circuit will give a more precise limit, otherwise they are more or less equal.

Basic equations:

Vin    Supply DC voltage into the limiter circuit (RMS value if there is significant ripple)
Vout  Output voltage exiting the limiter circuit. If you use average values, you get average dissipation. Instantaneous values give respective instant values of course
Iout   Output current exiting the limiter. Logic same as above
Re      Resistance of the pass transistor emitter resistor
Vre   Voltage across the emitter resistor. We want 0.6 - 0.7 volts at the current limit to match the transistor Vbe

Vre = Re * Iout ; Vre --> 0.7V @ 2.4 A ; Re = 0.7 / 2.4 = 0.291 ohms

An important design item to verify is the power loss in the pass transistor. The formula for that is:

Pd = (Vin - Vout) * Iout - Re * Iout^2

You don't say what the supply voltage to the limiter ´circuit is, but assuming say 40 V, the numbers become:

Pd = (40 - 30) * 2.4 - 0.291 * 2.4 * 2.4 = 22,3 W

This is a considerable power loss and the pass transistor cannot survive without proper heat sink. Now this loss is not continuous and you can re-calculate for the continuous current, and you will find out that the power loss with 1 A current is just under 10 W. Assuming your heat sink has good thermal capacity, you can dimension it using the latter figure, provided you don't do repetitive starts too much.

The power transistor can be a jellybean TO-220 packaged type that is easy to heatsink. Subject to the hFE of the selected component, the auxiliary transistor can be a 2N2222 or similar.

Power dissipation in the emitter resistor is

Pdre = Re * Iout^2 = 0.291 * 2.4 * 2.4 = 1,67 W, so select a 3W or 5W resistor and it will not be burning hot.


Looking at Digikey component selector, we get several candidates for the pass transistor. in the 60-100V range there are the trusty old TIPxx devices, many of which will do what you need but they have a rather low hFE of 10-15 at higher currents. They can be used in the 2 transistor circuit though, since the aux transistor will mitigate the lack of hFE by its own amplification. The TIP41 has a SOA (safe operating area) DC value of 6 amps @ Vce=10V. So it will not suffer second breakdown at these power levels.
Another alternative would be the TO3 packaged BUX10 power device targeted for industrial and military applications. It will dissipate max 150 watts which is of course gross overkill, but the package favors more efficient cooling. It will also maintain a hFE of min 20 at 10 amps. But there are others and based on the above you should be able to make your choice.

[beaker353]

Thanks Kremmen!  I'm going to spend the better part of this evening, or even longer, chewing over your post.  Everything is starting to make more and more sense so I just need time to take it all in.  Feels like I'm trying to take a sip of water from a firehose.  Sounds like this is very doable and not too complicated or expensive.  I'll post a full schematic in a day or two with my calculations and see if I got everything dialed in.  AcHmed99, I can't count the many pieces of equipment I have brought back to life by replacing electos.  Standard procedure is if I suspect a bad PS, every single electo on the PS board get replaced with prime-spec low-ESR Panasonic.  Funny how sometimes replacing cheap caps with prime-spec actually improves the characteristics of the PS beyond what the original manufacturer engineered.  I guess that's what you get with "value engineering" these days.  However, the original power supplies in this case are pretty much encased in a single block of potting resin with the exception of some extruded aluminum heatsinks.  Zero chance of getting in there and replacing.  My ultimate backup plan is to just put a switch on the front that would bypass a 12.5ohm resistor in series with the power supply.  The resistor would limit the current to 2.4V even under a dead short and then we would manually bypass it once we feel the high current draw is done.  I would prefer something automatic, but can't beat the simplicity of a single resistor and a single switch. Stay tuned...

- EM

[Jeff1946]

Use a power P-MOS.  I think a IRF-5305 would work.  Connect source to postive of power supply and drain to load.  Put 100k resistor between gate and ground and another 100k between gate and source and parallel it with about 100 ufd cap.  Transistor will take a second or two to turn on.   It should be mounted to a small heat sink.  Once it is fully conducting, it will only be dropping a few tenths of a volt so heat will be less than a watt.  Be sure that the gate to source voltage stays below 20V.

[T4P]

If you still want to go the BJT route use the NJW0281G from ONSEMI

[beaker353]

Thanks everyone for your input.  I think I have a solution that at least from my simulations seems to meet all my specifications.  It appears that I am staying well within component operating ranges as well.  The attached PDF has the schematic and E, I, & P measurements across all the components with simulated load from a dead short to 50ohms.  I was able to find pspice data for the ONSEMI NJW0281G and imported the values into CircuitLab, so in theory at least this should be fairly accurate. My only thing I'm not totally thrilled about is the current I have to burn to keep the pass transistor open, but I know there isn't a free-lunch in these types of circuits.  Anything I've overlooked that I need to take into account?  Thanks!!

-EM

[beaker353]

Thanks for the ideas AcHmed99.  Could you be more specific with some part numbers or example circuits?  Keep in mind that at this point in my formal electronics education semiconductors haven’t even been talked about.  I'm working entirely off what I can figure out from my Grob's book and from the 'net.  The application note assumes you have a much higher understanding of electronics than I do, which is why this is in the beginners forum in the first place.

- EM