Problems with high-side current sensing with 12V comparator and logic

Question

I am designing a system with circuit protection using high-side current sensing. Expected shut-off current is in the range of 1~3 Amps. The current being sensed is 12V so I have designed a circuit that automatically shuts off a power transistor when excess current is detected. To avoid having to use voltage shifters I am using 4000-series logic gates that run at the same 12V supply which do the detection and shutoff. My problem is that most comparators can't operate with their inputs close to their upper rail (i.e. power) voltage and need to be brought into their linear ranges in order to do compares accurately. Those that do seem to be expensive and my board needs to have 14 of these automatic shut-off circuits on the board. But by their nature, high-side current sensing means that the voltages that will be monitored are small fractions of a volt below the upper rail voltage. Since the comparator needs to run with say 1.5V above measured voltage ~12V then it needs to be supplied power of at least 13.5V. So it looks like this might be a case where some sort of solid state voltage generator is called for. I am aware that RS232 transceivers can generate the higher voltages needed for line output and sensing line input using an on-chip device (called a charge pump). Is anyone aware of such a discrete device for creating a voltage that is higher than supply by some known amount for use by other onboard devices? It would be similar to an LDO regulator except that it produces a voltage above supply.

Another problem I see is that, if the comparator is running at a higher voltage than the logic then the 4000-series logic gates might experience inputs that are higher than their power supplies. A possible solution to this problem is to run all of the logic at the higher voltage as well. I am loath to use a higher valued current sensing resistor to bring the sensing voltages down into a comparator's linear range as it would make them burn a lot more power as well as lowering the voltage that is available to drive the load that is being controlled.

Any advice or insight is appreciated.

Answer 1

If you have a few hundred mV of signal and don't care about an extra mA or so on each output (or 100uA if you use more expensive parts) you could use TL431 parts to level shift the signal into the common mode range. If the accuracy of the TL431 is 1% that's a maximum error of 24.95V mV, so if your current limit is 250mV your maximum error due to the LM431 is +/-10%. Not great but maybe good enough.

Or create a 12V-5V rail with a 7905 regulator or a shunt reference. Then use a RR input comparator such as a MCP6564 (about 25 cents/channel) from a 5V supply (+12 to +7) and use a P-channel MOSFET and pull-down resistor to get CMOS compatible output.

Or give up and use a purpose-made current amplifier/level shifter.

Answer 2

There are a couple of options that you can explore.

1) Use op-amps that include Vcc within their input common-mode range as comparitors. For example, the TL07x / 08x parts will work well with both inputs sitting at the (+) supply rail.

2) Use a high-side voltage-to-current converter chip such as the ZXCT1008. This is a 3-terminal chip that takes a tiny voltage across a current-sense resistor on the (+) rail and converts it to a ground-referenced current. That is: a small voltage dropped across the high-side current sense resistor is converted into a current proportional to that voltage. That current is then converted back to a voltage in a resistor that is connected between that node and ground.

Then do all of your comparisons referenced to the 0V rail.