TineFPGA-GOLD Metal detector

[SaltyDog]

Ok, my first real stumbling block.

The LED driver arrived and promptly blew up when I tried it. I had assumed the FET's in the output of the driver would not have body diodes, and the schematic of
the internals seems to verify this, however, all the outputs conducted momentarily with no current limit and smoked the chip .. (it is a STP08CP05MTR)

If you have a look at the schematic attached, I am driving the LED's through the FET's with reverse currrent i.e -12v compared to ground. I had assumed that would be ok, as once turned on, mosfet's don't care in which direction the current flows .. they act as a resistor ..

I think they only would have had problems if there was a body diode, which would have conducted instead ..

Have a look at how I wired it up and see what you guy's think ...

The whole reason for doing it this way, is because I need to level shift the FPGA 0..3v to the -12v to drive the LED's .. (because GND is the positive of the 12v battery)

Maybe there is a better way?

Cheers

[6666]

How did you prototype it

[SaltyDog]

How did you prototype it

How do you mean? This first rev of the board is my prototype ...

By the way, I have used individual fet's in this "reverse current" mode just fine ...

I guess I could isolate the SPI bus with one of these, then I could operate the driver in "standard" positive supply mode .. means more cost though ..
https://www.analog.com/media/en/technical-documentation/data-sheets/ADuM150N-151N-152N.pdf

[6666]

I thought you said a few weeks ago that you had a fully working prototype on the bench what were you driving the leds with then, I always use a current limit resistor with leds.

all the outputs conducted momentarily with no current limit and smoked the chip

just re read this saves me looking up the cheat sheets

[6666]

Looks like about 300mA if all leds on at once

[SaltyDog]

I thought you said a few weeks ago that you had a fully working prototype on the bench what were you driving the leds with then, I always use a current limit resistor with leds.



just re read this saves me looking up the cheat sheets

That testing was done on the Lattice board that directly drives led's off the FPGA pins ....
a) I am now using the TinyFPGA board
b) .. those LED's on the Lattice board not bright enough for outside use ..

I am designing for real conditions ..

[SaltyDog]

Looks like about 300mA if all leds on at once

No,That driver has a singe resistor that programs current through all LED's ..
With no current control (with reverse voltage) current is only limited by the LED's ... hence the smoke ..

[Skippy]

Never pull an IC input or output below the negative supply rail, disaster is guaranteed. If it's not FET body diodes ( which pretty much every MOSFET has ) it's parasitic diodes or transistors inherent in the manufacture of IC's. There's a well-known phenomena called SCR-latchup that can occur with certain IC inputs.
https://en.wikipedia.org/wiki/Latch-up

Also, never apply more than the positive rail voltage to any IC inputs/outputs, while not so troublesome, especially if the current is limited by a resistor, for example, it's still best avoided. Use of a diode ( preferably a Schottky type for low forward voltage drop ) can protect IC pins from these mishaps.

There is a bizarre phenomena where an IC can be functional, with no positive power supply, and I have actually witnessed it myself. It was a 4000-series CMOS gate package. When one of the inputs was logic high, an internal parasitic diode conducted to the positive power supply, hence powering the chip. It was effectively running at 11V instead of 12V, but was fully functional. Only if there were no inputs High would the power disappear. The cause of the missing power? The leaded IC was inserted in a socket, and the power pin had 'dog-legged' during insertion, and had not made contact with the socket.

I haven't looked at the datasheet for this driver chip, but I assume it monitors the current using a series resistor in the Source of the FET, and uses this to set the serially-programmed current control. This obviously is incapable of working the wrong way round.

MOSFETs can be used the 'wrong way round', but only in certain situations, that the designer is fully aware of. Two common cases:
* switching low level voltages, generally below 200mV. Such as small AC signals that may go +100mV to -100mV , in audio equipment.
* high-power rectification, sometimes called synchronous rectification. The FET is used as a diode, with both the body-diode and the D-S channel conducting in one direction, but the D-S turned off and the diode reverse-biased and non-conducting in the reverse direction.

[waltr]

I looked at your schematic and the data sheet.
Skippy got what happened, there are always intrinsic diodes in MOSFETs.

The data sheet is a bit skimpy but the clue is in the MAX ratings.

Vo Output voltage -0.5 to 20 V

this implys that the Voltage on an output pin must never go below ground.

Can you flip the LEDs around and then power them from a positive 5 to 12V?

[dbanner]

The application notes I've seen have the LEDs connected the other way around, with the anodes connected to VCC 1, which is a separate supply provided by L5970d.

[SaltyDog]

The application notes I've seen have the LEDs connected the other way around, with the anodes connected to VCC 1, which is a separate supply provided by L5970d.

Guys, I am sorry I brought it up.

I know the correct way to use this driver, I just could not do that with a ground based on +12v. I need to use the 12v to drive the led's to get the brightness I need, I can't use the +5v rail, as it is derived from a capacitive switcher (LT1054) that is already maxed out supplying the rest of the circuit.

I could use another LT1054 to give me another +12v that is above GND (Battery +12v) and driver the LED's of that, or I could put an isolation chip on the SPI lines (Adum160), that way I can run the LED's and driver straight off the battery and isolate drive them ...

@Skippy
You are mostly correct, except for the fact that you can run a MOSFET at full rated current in reverse (as long as the mosfet does not have a body diode) I have done that many times (I have been an Electronics Engineer of some 50 years ... so know a couple of tricks ..) I just wasn't expecting the driver to have body diodes on the Fet's, my mistake, but you are correct I had forgotten about the current limiting circuit ... the datasheet does not have any detail on the method of current limiting, and the schematic of the chip shows no body diode on the fets ... very bare info ..

[waltr]

Is there any way you could reference the LED driver to the -12V rail?

[waltr]

Is there a way you could reference the LED driver to the -12 Rail?

Looks like it could be done if you find an easy way to level shift the SPI lines.
That could be done with transistors without too much difficulty.

I agree the LED driver data sheet lacks important details.
Also work as an EE and have also gotten caught with mistakes like that.

[Skippy]

Quote:"@Skippy, you are mostly correct"
What's with the 'mostly' ? Did you read my note about synchronous rectifiers, using power FET's the 'wrong way' round ? And if you're using it the 'wrong way', it doesn't matter if there's a diode internal to the FET or not.

And yes, the IC datasheet will skimp on such details like body diodes, as it's assumed there will be such diodes, and they will never conduct, as the output pin will never go negative by more than 0.5V, as that is stated in the design absolute max figures.

I'm curious as to what power FET's you've come across that don't have body diodes. I don't ever recall seeing one. I've used small-signal MOSFET's, the type with seperate gate, source, drain and substrate, and there's a 4000-series CMOS part ( the CD4007 ) that can be used that way, they are diode-free, and dual-gate RF FET's like 3N201's also may be diode-free. But all power devices seem inherently lumped with a diode. I've seen ones with additional diodes, either fast p-n junction or schottky types, as the intrinsic body diode is slow.

35 years of professional electronics design experience speaking here.

Edit: I had to Google the subject of body diodes ... looks like transistors made with the Silicon-on-Insulator process can get round having body diodes. SOI is reserved for rad-hard parts, eg for spacecraft and EMP-resistant military hardware, so not really applicable to our little toys.

[Skippy]

And regarding a fix for your design problem:

Make your own higher-power +ve rail generator. One that works well is using a 555 timer IC as a free-running oscillator, and then use a diode/capacitor charge-pump circuit to create a voltage of about 10V above the 0V rail. There's plenty of examples of this on the web. The bipolar NE555 is the most capable, though the LinCMOS TL555 will be adequate for a few LED's. Don't use the feeble 7555 CMOS version, it can't drive anything. A simplified oscillator with 50/50 duty cycle can be arranged by connecting the output ( pin 3) to the RC charging circuit, so it runs like a high-power inverter. It's not a particularly efficient solution ,as the LED's only need 2V to drive them, so a 10V supply means dropping 8V across a 'resistor'. But it will give a powerful +ve rail.