Current Limiter with MOSFETs

Go To Last Post
19 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi, I am trying to add a current limiter to my little circuit at about 80mA.

Theres 5V comming in, then I drop it down to 3.3V with a LDO (NCP1117) regulator.

The 3.3V regulator needs to be 1.2V above 3.3V, so then I have (5V - 3.3V - 1.2V =) 0.5V spare to make a current limiter with. Those few milli volts are precious!

I have designed a standard two PNP transistor current limiter, but the transistor along with the series resistor drops too much voltage.

:idea:I was wondering why does no-one seem to use MOSFETs in their current limiters? Are they too slow?

I feel I could save a few tenths of a volt with MOSFETS, but I am sure I am not smart enough to realise why people dont do this.

I am open to suggestions, however I prefer to use standard off the shelve components because its hard for me to source anything other than a 1K resistor or 100nF cap.

Thank you for any advice/help!

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In fact, Linear Technology makes a whole raft of devices in this general category. Some are "active diodes" for power switching, and there is also a current limiter. They are not cheap, as they take both an IC and a FET. But, they work nicely and are relatively simple.

One (of several) is here:

http://www.linear.com/pc/product...

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
In fact, Linear Technology makes a whole raft of devices in this general category.
http://www.linear.com/pc/product...
Jim

Thanks for the recommendation, they are neat little devices. I tried my local suppliers for any one of them in that series, and they dont have.

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I tried the MOSFET route, can you guys please critize it. Fire away, dont hold back :P. I will buy the components on the way home from work, and play with it this weekend.
Hopefully you can make recommendations before I hit the shops!

There are two parts to my circuit. The first is the power in part (I will post the current limiter part next). It takes iether 220VAC, 12VDC, or 5V USB as the power source. I am trying to make it so that if two or more sources are plugged in, they wont fight each other.

I would of used a diodes to protect the USB power (instead of a MOSFET), but I cant afford the 0.7V drop. Even a 0.2V drop is too much.

Attachment(s): 

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here is the current limiter part. Basically the +5V from the previous part is regulated to +3.3V for the AVR, and is also passed through a current limiter then regulated to 3.3V for an external device (Vtest).
The current limiter is set to 100mA.

VUSB = 4.9V (what my computer measures at)
Normal working current = 50mA
V drop across MOSFET Q2 = 50mA*0.12E = 6mA
V drop across MOSFET Q3 = 50mA*0.12E = 6mA
V drop across Rsense = 50mA*6E = 0.3V
V drop across 3.3V regulator = 1.2V
Output = 4.9 - 0.012 - 0.012 - 0.3 -1.2 = 3.388

Voltage spare = 3.388 - 3.3 = 88mV!! Cutting it close! So if you can see any way of improving it, or any recommendation to do things better I would love to hear your recommendations.

I would prefer to set the current limiter to about 60mA, but then I have to increase the sense resistance, which will result in more volts wasted.

Thanks for any advice!

Attachment(s): 

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Can the LM358 really handle inputs that close to its supply rails?

There are LDOs that have an extremely low dropout voltage below 100mV. You could choose an LDO that has an enable input. Would save the MOSFET if the polarity happens to be right?

Wouldn't a circuit like this oscillate like crazy? It does not latch, so it detects overcurrents, shuts down, overcurrent situation goes away, starts up again, repeat and rinse...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Linear Technology also allows you to order parts directly from them, online.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Those are some good points thank you jay.

jayjay1974 wrote:
Can the LM358 really handle inputs that close to its supply rails?
The datasheet states that it can go from 0V to Vcc- 1.5V. I am operating it off 5V, and asking it to put out 3.3V max.

Quote:
There are LDOs that have an extremely low dropout voltage below 100mV. You could choose an LDO that has an enable input. Would save the MOSFET if the polarity happens to be right?

Brillant idea. I can get a SMPS 3.3V regulator (LM2574M-3.3) that has an enable, and the drop out Voltage will be about 0.8V (but I am not keen on all the addedcaps and inductors). Or I could use a (LP2951ACDG with a drop out 0.38V, but it can only supply 100mA, so I am worried about it dying).

Quote:
Wouldn't a circuit like this oscillate like crazy? It does not latch, so it detects overcurrents, shuts down, overcurrent situation goes away, starts up again, repeat and rinse...

I thought the Q4 would be biased more and more to a certain point where the maximum current flows, and then no more. Why would it oscillate? I will look into it further though.

Thanks once again for your insight!

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jayjay1974 wrote:
Linear Technology also allows you to order parts directly from them, online.

Thanks for the recommendation.

Found a nice regulator (LT1763CS8-3.3), decided to buy 10 of them so I have them for lots of future projects. Turns out US$67 shipping for economy, and US$75 for priority.

Uncoolness. :(

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Do you not have Farnell or RS in South Africa?

You may pay a bit more per item, but you will get free next day delivery. YMMV.

David.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I spent a bit of time simulating this little circuit in LTSpice. It seems to work, but not as a foldback. Above the trippoint the MOSFET shuts off quite abruptly, and the loads get a bit of current via the PNP emitter, base and R6 which keeps the PNP transistor conducting, hence the MOSFET remains non-conducting until the load fall below 4mA or so.

[edit] I attached the LTSpice file

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jayjay1974 wrote:
I spent a bit of time simulating this little circuit in LTSpice. It seems to work, but not as a foldback. Above the trippoint the MOSFET shuts off quite abruptly, and the loads get a bit of current via the PNP emitter, base and R6 which keeps the PNP transistor conducting, hence the MOSFET remains non-conducting until the load fall below 4mA or so.

[edit] I attached the LTSpice file


Wow Jay, I am amazed you would go to all that trouble. Thank you so much!!! I have read your post three times, and will continue reading it until I fully understand it.

So if I understand you correctly, I should increase R6 so that the transistor turns off easier.

I have downloaded LTSpice, and am currently installing it. I always wished for a simulator, I didnt know it was free!
Awesome! :D

Note, I just saw you attached your model, its like you can read my mind :D
Thank you, you are a star!!

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

david.prentice wrote:
Do you not have Farnell or RS in South Africa?

You may pay a bit more per item, but you will get free next day delivery. YMMV.

David.


Thanks for the suggestion.

We do, Fernell takes about 1 week, and costs me about 2x the quoted Fernell price.
And RS takes 5 working days, and the shipping is reasonable (equivalent to US$5), but one switcher is US$7.00 (as opposed to US$2 on linears website).

But I like to use generic components incase it breaks, then its easy to repair.

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Farnell has sometimes massive discounts on LT components, sometimes the 50% discount starts at just 5 pieces :shock: That means you spend less money if you buy 5 then when you buy 3 ;)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Jay I have success without even breadboarding it or multiple trips to the electronics store! All thanks to you!!! :D

For any one else who would like a MOSFET current limiter please see the attached file. It is very independant of all the resistors except R1. R1 sets the Current limiting, and be calculated as follows:

Ilimit = R1 / 0.66V

I found like you said the base emitter diode of the transistor was conducting all the time. So I moved it before the MOSFET, so the current from the base emitter must flow through the MOSFET. It made things much more stable.

Once again thanks for your help.

I am loving this simulating thing!
You can get LTSpice here:
http://www.linear.com/designtool...

Note: In the attached simulation the load (R3) goes from 100Ohm to 0.001Ohm in 1s econd.

Attachment(s): 

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jayjay1974 wrote:
Farnell has sometimes massive discounts on LT components, sometimes the 50% discount starts at just 5 pieces :shock: That means you spend less money if you buy 5 then when you buy 3 ;)

Heheheh :lol:

Reminds me of the last set of circuit boards I got made. If I got 8 made, they cost R156 each, it I got 10, then cost R104 each. So I went with 10! Score!!

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Remember that simulations do not guarantee your circuits work exactly the same in real life; GIGO applies :)

But they can be very useful if used properly and with care :) especially the possibility to measure current anywhere without any influence on the circuit itself is very nice.

A neat little trick in LTSpice is when you alt-click a component it will plot a power dissipation trace :)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jayjay1974 wrote:
Remember that simulations do not guarantee your circuits work exactly the same in real life; GIGO applies :)

But they can be very useful if used properly and with care :) especially the possibility to measure current anywhere without any influence on the circuit itself is very nice.

A neat little trick in LTSpice is when you alt-click a component it will plot a power dissipation trace :)

Wisewords!

That is indeed a cool trick to quickly see the power, thanks!!! =)
Measuring current in spice sure beats multimeter cable swapping.

I also noticed R4 is redundant in my circuit and can be chucked.

Heres a common 0.5V Drop out @ 500mA ADJ regulator if anyone is interested: LM2941.

Just a noob in this crazy world trying to get some electrons to obey me.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Simulations are usually very close to reality. (depending on complexity). I have been astound several times how exact they are.