OnSemi 30160 chip review. cut short.

[HackedFridgeMagnet]

Well I didn't actually want to do a review but I seem to have a problem with this chip.

http://www.onsemi.com/pub/Collateral/NCL30160-D.PDF

I wanted to use this chip to drive a string of 10 cree leds  from 36 volts dc for domestic lighting.
I drew up a board and sent it to olimex for fabrication, it hasn't arrived yet, 3 weeks since it was sent, I am losing hope.
I bought five of the chips, a couple extra, in case I burnt one.

Anyway when I got the chips I went into prototype phase, and one by one said bye bye to my chips.
Chip 1. Sloppy handling, no esd protection, never worked, my fault.
Chip 2. I had no esd mat but I observed as well as I otherwise could ESD safety. Chip worked, went into hysteris mode but doesn't dim. my fault.
Chip 3 soldered onto dip converter board, was trying to hurriedly connect wires and wife called me to get into car, just as I was connecting 28 volts to power pin, but I got the wrong pin. my fault.
Chip 4 soldered into dip board, placed into breadboard everything good, we have light. we have dimming, I left it on for a few hours, once I cranked it up to 800mA, I checked the chips temperature, it did seem surprisingly cool. I put it back down to 300mA. All good until I disconnected one banana plug from the psu. Simulating a switch turning on and off. When I reconnected after say 1 second, the chip went short circuit. I assumed it was the Fet but it was the internal voltage regulator that is short.
Chip 5 was going fine for one hour, I left it running at 130mA for another hour, when I came back into the workshop, the workshop was quite smoky, the fire alarm hadn't gone off and the power supply was showing 450mA and was being current limited.

I contacted OnSemi just to see what they would say, they said it probably overheated.

I'm not worried, I might go back to a Amtel solution with discrete power components, but I am disappointed as I had high hopes for this chip.
Also I realise breadboarding fast power circuitry is not good but I don't think that is the main problem.
My circuit was very close to the reference circuit.
I generated the square wave with a 555 at about 3 volts.
the 555 was regulated by a 7805 rated at 35 volts or so, which had no problems at the same vin as the NCL30160.
Lastly I have never before wasted more than 1 of an individual type of chip on one project in my life. Now I have done 5.

In short either me or the NCL30160 rate a Fail. Probably both.

[Jeff1946]

I built a led lamp to replace the two automobile tail light bulbs in a light for my front yard.  The light was supplied by a 24 VAC transformer.  I used 3 strings of six warm white leds.  Each set has its own resistor and is driven with DC from a full wave bridge.  Obviously you can find the correct resistor size by trial and error, but I decided to play around with a spreadsheet to see what current the leds were experiencing.  Attached is the spreadsheet I developed.  NOTE:  It is an XLS excel file but I changed the extension so that it will be accepted by this forum.  You need to change it back to XLS to use it.  The voltage across the resistor matches what I calculated with the spreadsheet.  The leds use about 2.5 watts vs 36 watts for the two bulbs and should last about 50,000 hrs according to Cree.   The bulbs needed to be replaced about twice a year.  And cost about as much as the leds.  Definitely the way to go.

[T4P]

Even the resistor charred ... what happened there ? Looks like the aggressive short/thermal shutdown protection wasn't helping at all and instead it caught fire and the SOIC>DIP board leaked FR-4 epoxy

[muvideo]

It's difficult to make working switchers on breadboard, I suppose, usually I don't
use breadboard.
You should draw the schematic and annotate the value and partnumber of components,
for example what did you use as recirculation diode?

As for this part:

All good until I disconnected one banana plug from the psu. Simulating a switch turning on and off. When I reconnected after say 1 second, the chip went short circuit. I assumed it was the Fet but it was the internal voltage regulator that is short.

the inductance of the input leads can generate enough voltage to kill the ic when disconnected, it happened to me before
with a led driver, I wanted a low input capacitance and this makes the problem worse, I put a TVS to input and
solved the problem.

Also for me 36V out of a buck that accept 40V input max is too little headroom, but you said you tested at 28V.

Fabio

[HackedFridgeMagnet]

I want to do the lights for my whole house in probably 36 volts with current controlled leds. I want a dimmable solution so the series resistors are not an option for me.

I think the problem is the internal voltage regulator, and when that goes short the rest of the circuit doesn't matter, so no thermal shutdown.

It's difficult to make working switchers on breadboard, I suppose, usually I don't
use breadboard.

I used vero board, for chips one and two, but I thought I would try a breadboard and see what happened.
I should've waited for my real boards I guess.

You should draw the schematic and annotate the value and partnumber of components,
for example what did you use as recirculation diode?

Sorry here are the components.
I used a 5819 for the diode, 4.7uF ceramic for the input cap.
.22uF for the regulators output cap.
200k for the ROT resistor, 33uH wurtz ferrite inductor (Freq > 8Mhz)
Rsense was 1ohm for the the chip in the photo,chip 5.
The leds were 10 x Cree xpg warmish white I think.

The bottom left pin in the photo is pin 1.
the 200k resistor is not actually charred. It's the camera lying. I took the photo through my magnifier as I don't have a good macro camera.

the inductance of the input leads can generate enough voltage to kill the ic when disconnected, it happened to me before
with a led driver, I wanted a low input capacitance and this makes the problem worse, I put a TVS to input and
solved the problem.

for chip 4 I think you are right here, though it did have 4.7 uF on the input. Still the voltage was nominally 29 volts,( my bench power supply only goes up to 30.5 volts). That is 25% headroom.
I will definitely use a TVS next time.

Also for me 36V out of a buck that accept 40V input max is too little headroom, but you said you tested at 28V.

Yes you must be right, what sort of headroom do you suggest for Fets, is that the same for Voltage regulators, and other types of components?
I was planning to use a good quality switched mode to provide the 36 volts. This is why I thought I could get away with being so close to 40V.

[alexkiritz]

Did you eventually get this working?

[HackedFridgeMagnet]

I just reread the thread, I didn't progress using this chip but I had decided it was too fragile and I didn't want a hysteretic converter after all.
In retropect the last two chips must have had too much inductance in the main current path and caused overheating.
Anyway I moved on to an Attiny micro solution with transistors as gate drivers, the control loop in the micro and it works fine. Never missed a beat.

[tszaboo]

Oh no. Never Breadboard a switcher working more than 100KHz. Never breadboard power electronics. Never run power electronics without supervising the first time. Never  run switchers without proper bypassing. D'oh!