9

I have a digital output, driven by the high-side driver with nominal voltage of 24V DC. The load current normally is below 100 mA. The output is monitored, so I can switch it off quickly if I detect a short circuit at the load side. The problem is that the driver itself is not protected and short-circuit makes it generating a lot of smoke. So what I need is a simple circuit at the output of the driver which:

  • has low resistance of below 10 Ω if output current is under 100 mA
  • rapidly increases its resistance to limit the driver current at 500 mA level or lower
  • withstand capability at short-circuit current shall be at least 20 ms for short circuit to be detected and driver switched off
  • has a working voltage of 50V or higher
  • has minimum components and cheap (0,20$ per channel max)
  • is not single-source supplier

I tried PTC resettable polyfuses, but they are too slow. Microchip's FP0100 should be good but it's expensive (I need at least 60 channels on my PCB). Bourns TBU series are also OK, but also expensive.

Any other options?

UPD1. My current output circuit is MIC2981/82 driven by 74HC594 shift register. On each output I have Littelfuse 1206L012 PTC. On my board I need 64 channels like this, and this is small series board so total price per channel and footprint are important.

syoma
  • 191
  • 1
  • 6
  • How much voltage are you willing to lose? – Trevor_G Jan 28 '18 at 19:46
  • Is this for a single job? or high volume? – Tony Stewart EE75 Jan 28 '18 at 21:00
  • 1
    This looks like a typical high side switching application found in ECU and PLC units. Only your 50V rating excludes almost all switches on the market (except BTS4140N), can you elaborate on this? – Jeroen3 Jan 28 '18 at 21:00
  • 1
    Can you show us a schematic of your high-side driver circuit? – Bruce Abbott Jan 28 '18 at 22:34
  • What is your short circuit current? Some 140mA PTC's trip in 8ms with 8A. What actually smoked ? The driver? If so that's the solution. – Tony Stewart EE75 Jan 29 '18 at 01:02
  • You can make the PTC trip very fast if you attach an SCR crowbar to the output. Monitor current with a shunt resistor and current sense amplifier. If the current gets above the theshold, fire the SCR. If shunt voltage drop is unacceptable, use a hall-based current detector (ACS 711 or similar). – user57037 Jan 29 '18 at 03:55
  • 1
    Yeah, I think this is an XY problem. You should really post your whole driver circuit. There may be a better way to get what you need. – user57037 Jan 29 '18 at 04:00
  • **It is definitely an XY problem** with the method of measuring RTD's not having proper active current limiting and using a voltage switch instead. Probably using a 2 wire method too , instead of 3 or 4 wire. Question needs full disclosure on RTD requirements with missing *must-have* and *nice to have* specs **= Y** rather than how do I short circuit protect 60 smoking voltage drivers? **= X** – Tony Stewart EE75 Jan 29 '18 at 15:43

6 Answers6

13

Your typical double transistor current limiter may be your best bet. Shown below is the top-side and bottom side versions.

schematic

simulate this circuit – Schematic created using CircuitLab

Note there is a penalty of about a volt drop with this circuit.

Buy dual transistors in a single 6 pin package.

The small resistor will cause the current to fold back when it reaches Vbe. The other resistor sets the base current and needs to be calculated to produce sufficient collector current taking into account Hfe.

HOWEVER: Be aware that transistor needs to handle a few watts for the duration of the short since it only limits the current to your threshold value.

Trevor_G
  • 46,364
  • 8
  • 68
  • 151
  • 1
    2nd this. I am just incorporating such a design to provide short-ciruit protection to an external RTD circuit. The pass BJT being a SOT223 and the sense BJT being a SOT23 –  Jan 28 '18 at 20:35
  • The other thing to consider with this is the min/max current your expected load is to draw & equally the min/max you can afford to sink during a short/fault. The variation of beta makes this topology quite susceptible to device characteristics BUT as long as you know the load and the parts its all good. –  Jan 28 '18 at 20:42
  • Very classical. OP: Put some attention to the voltage rating! – Marcus Müller Jan 28 '18 at 20:57
  • 1
    I've posted an answer that builds upon yours and focuses on the multi-channel usage. I'd like to ask you to proof-read, please! – Marcus Müller Jan 28 '18 at 22:19
  • My load can be various - from just some LED taking as low as 5ma up to relays, taking 20-30ma which maybe paralleled with appropriate transients. Your circuit looks interesting, thanks, but has 3 components, making footprint a bit large – syoma Jan 29 '18 at 09:40
  • @syoma you will not find much smaller than this though and is only one more resistor than the microchip part. The NPN circuit can also be modified to drive the biasing resistor from your control circuit so you do not need an extra switch. – Trevor_G Jan 29 '18 at 11:32
  • Yes. I understand. I was hoping that there is some cheap integrated solution available which I'm missing. – syoma Jan 29 '18 at 16:16
  • @syoma unfortunately small and cheap are sort of conflicting requirements. This solution is likely the best you can do. But getting rid of the MIC2981s may be a trade-off you can handle. – Trevor_G Jan 29 '18 at 16:22
7

Take a look at the ProFET high-side driver ICs. These devices give you a switchable high-side drive with protection from all sorts of things, including output over-current.

You can find and select ProFETs readily enough from distributors.

Have a look at the BSP752T, which is cheap, small and can be driven directly from 3.3 V or 5 V logic.

enter image description here

TonyM
  • 21,742
  • 4
  • 39
  • 62
  • Thank you. BSP752T costs 0.9€ per channel. This is a bit too expensive. It's current is 1.2A, so a bit too big too. Are there any lower cost/lower rating alternatives? – syoma Jan 29 '18 at 09:44
  • @syoma, you're welcome :-) As per the answer, you can search distributor websites (e.g. Digikey, Farnell) for ProFET and look at the characteristics vs price local to you. Only you know your costs and business constraints. – TonyM Jan 29 '18 at 09:52
  • Yeah will do. What wonders me is that for high-side switching there are almost no switching arrays available, and the ones available are very pricey. – syoma Jan 29 '18 at 09:57
  • @syoma, the pricing usually explains the demand to us. It's asking a lot of one package to carry lots of power dissipation at the voltages and short-circuit power you're looking at, and you want small and very cheap. If one channel is taken beyond its protection and goes bang, it'll take fewer or none with it, which may be very important in other applications. You say expensive a lot but give no budget...? Otherwise, spend the 90c/chan and enjoy the protection benefits. – TonyM Jan 29 '18 at 10:13
  • I found cheaper alternative from Infineon - ITS41k0S for 0,5$. This sounds reasonable price for it. This would also eliminate need for PTC, so I save couple of cents. To drive it I will need to change to tpic6c595 or similar. – syoma Mar 12 '18 at 19:43
  • @syoma, excellent, glad it led to a solution. There's a broad range of these parts, under the ProFET banner and other trade names. – TonyM Mar 12 '18 at 21:30
5

To build upon Trevor's excellent answer:

There's semiconductor devices that are constant current sources (or sinks); many of these will internally look exactly like Trevor's circuit (maybe adding a few temperature-compensating elements).

One very simplistic device (constant current sink with exactly two pins, designed for voltages <= 50 V and a max/constant current of 350 mA) is the NSI50350AD. I don't know what it internally does, but the datasheet calls it "self-biased transistor", so chances are it might be a combination of some bipolar transistors, a JFET and a couple resistors internally.

Now, your 50 V limit really hurts – it's hard to find integrated current sources that will work at that voltage. For smaller currents, a self-biased JFET might work, but at 100 mA that's going to be expensive.

So, I'd really roll with Trevor's solution, albeit I might recommend a few things:

  • Check whether you can't simply increase the speed of your fault detection. That would solve the problem.
  • Because (as far as I know – correct me if I'm wrong) it's hard to come by transistor arrays (which you'd prefer if you need reduce effort and board space), you might want to spend a little more on the component than just a NPN for Q4, but save on pick&place cost by using a device with multiple comparators in one case. Luckily, 4x comparators and 4x opamps cost around 13 ct when bought in hundreds, so that's ca 3ct in opamp per channel; use the opamp/comparator to compare the voltage over R2 to a constant reference voltage (here, a simple zener might do) and to drive Q3. Notice that you don't need an R3 for every single channel anymore. (the same applies for the high-side approach with Q5/Q6)
  • Use resistor arrays instead of individual resistors, thermal design permitting.

Another relatively crazy approach would be using a high-side 8.2Ω resistor before your load. After that, insert a current divider between your load and the LED side of a transistor-out optocoupler, with an appropriate series resistance. Design that series resistance so that for 100 mA \$I_\text{Load}\$, the transistor is in saturation, but for 500 mA, you significantly pinch of. Put the C-E of the optocoupler output in low-side series with your load:

schematic

simulate this circuit – Schematic created using CircuitLab

A cheap candidate for the optocoupler would be Lite-On CNY17.

Marcus Müller
  • 88,280
  • 5
  • 131
  • 237
  • Looks interesting... – Trevor_G Jan 28 '18 at 22:38
  • Basically my idea was not to protect the driver on it's own but just keep it alive until I detect the SC and switch the output off. The limitation of current driver is that it does not saturate at high currents and breaks. So my other stupid idea is to take something like tpic6c595 and PNP transistor for high-side(for example PBSS9110T). It can survive with up to 3amps shortly, long enough to trigger the protection. – syoma Jan 29 '18 at 10:05
2

This one works out to $0.2/port x16 https://ca.mouser.com/ProductDetail/NXP-Freescale/MCZ33996EKR2?qs=sGAEpiMZZMuCmTIBzycWfKe9ppy40BrEybgj5eCsa3I%3d enter image description here

Tony Stewart EE75
  • 1
  • 3
  • 54
  • 182
1

Here is the basic idea for the SCR circuit. Might have to add a resistor in series with PTC1 to get the right value of resistance. The total resistance in parallel with the base emitter junction of Q1 will set the trip current. Once Q1 starts to conduct, the SCR will fire, and then the load will be protected until the PTC trips. Q1 can be a SOT-23. R3 and R4 are just guesses. They are just there to prevent over-current damge to Q1. Most SCR's are kind of large. I will let you look to see if you can find one small enough to suit your needs.

Note: Once the SCR fires, you will probably have to de-energize the power supply before it will stop pulling down the rail.

schematic

simulate this circuit – Schematic created using CircuitLab

user57037
  • 28,915
  • 1
  • 28
  • 81
-1

I was going to suggest the series double transistor circuit but Trevor_G has already done an excellent job of that.

Instead I thought it worth revisiting the PTC fuse option. You say they were too slow, but that suggests you might have marginal power supply design instead.

Consider the Littelfuse RXEF017. While it might take 8s to trip at 500mA, surely that's a low enough current for your short-circuit protection to have time to kick in? At 2A trip time is <0.2s, which is not a great deal of energy in a 24V system. In fact the point of a fuse is to be the most susceptible component in the circuit to current, so it's a bit of a worry that something else can give up its smoke before the fuse.

I just fear that you'll go to the trouble to limit current to a narrow window below 500mA, and then find other things become marginal because they can't draw enough in-rush current to charge caps or drive a pulse or something.

Heath Raftery
  • 3,101
  • 1
  • 11
  • 16
  • 1
    No. As I wrote in my question I was using Littelfuse 1206L012 PTC. When I was watching with Oscilloscope what happens on my driver output when short-circuit happens, I have seen, that there is a voltage drop of about 10V on PTC, suggesting that I have about 3-4 Amps of current during that time. Unfortunately rest of my 24 Volts are falling over driver, which heats it too much. – syoma Jan 29 '18 at 09:51
  • Okay. That's an even more conservative fuse than the one I assessed, so if it's still not enough to protect your driver than you will have to take more drastic action. Maybe the driver is weak, but maybe you were unlucky and the PTC put the driver in a partial short circuit that was high enough to damage the driver but too low to blow the fuse quickly. – Heath Raftery Jan 29 '18 at 11:46
  • Such PTC fuses are simply too slow to protect from short circuits. I've checked that during first milliseconds it doesn't increase it's resistance significantly. In my design the PTC is acting primary as overload protection - the driver has a constant current limit of 100ma per channel. – syoma Jan 29 '18 at 12:41
  • @syoma **then your choice of CC design is at fault not the short circuit.** what is it? It can me made with integral SCP – Tony Stewart EE75 Jan 29 '18 at 15:31