Any known issues with TPS71xxx regulators?

I was using the 71518 a while back and I had one occurence of fizzle and die. I think you are thermally killing the device with output surge - that's what I concluded with my problem. With about 17V across it, you'll reach the nominal power rating with about 15mA, and the surge current to charge the decoupling caps (which ultimately comes out of the beastie) may be too high for _just_ too long for it to handle.

Certainly that was what apparently blew up the one I lost - and I was operating it at about 20V in when it happened.

Cheers

PeteS

Hello Pete,

Hmm, interesting. Could a 0.1uF cap really kill it? I mean, it would get most of its initial charge out of the 1uF that's already at the output. Also, the data sheet says it's supposed to have at least 0.47uF ceramic on the output pin and that one would get charged instantly when you connect a battery or the power supply.

It starts up fine with that 1uF on the output. Only when the 0.1uF plus a wee load of a few mA is connected via a FET does it die.

I deliberately kept the bypasses to 0.1uF total and also provided a

Joerg wrote: The output is

Can you add a capacitor on the FET gate so it turns on sloooowly? A bit of a kludge but might help.

That is already there. 10k into a 0.1uF at the gate to provide a ramp of around a millisecond. Should keep the inrush under a milliamp.

I don't buy that the surge current into the 0.1uf total decoupling caps is blowing the LDO. Even so, you could temporarily remove these caps to make sure.

Did you double-check your MOSFET wiring? If this is wrong then you could be shunting the regulator output to your control signal via the drain-source diode or the gate zener. If this is the case, and your control device has a low impedance output, then this might sink enough to toast the LDO. Have you tried connecting the regulator directly to the load, instead of through the MOSFET?

================================

Greg Neff VP Engineering

Hello Greg,

Takes about a millisecond to come on full.It's unlikely in this case because I did slow the turn-on via a gate RC. I've seen people get burned by LDOs so I was a bit careful during this design.

Yes, went over the layout several times. It's correct and other than the occasional regulator blow-out it does its job nicely. They don't always blow, it's hit or miss.

I've tried it here on eight prototypes, banged them around pretty good but I never had one blow in my lab. It happened at a client. They might have to send some over so I can do all those kinds of tests but I was wondering whether there is some suspected pathology with these devices. As usual, the timeline is tight and there aren't any 2nd sources that I know of for regulators like this one.

How does the battery get connected? Maybe there is a surge voltage or ESD at the battery terminals?

We are using the TPS71550 to power an LTC1694 SMBus accelerator. The LDO can be fed from either a 19.6V power supply or a 14.4V Lithium Ion battery pack. No problems so far...

================================

Greg Neff VP Engineering

Hello Greg,

It's basically just being plopped in. That part works fine, the regulators provide a nice 5V afterwards. Some of the (very low current) sense circuitry is running at that time. About 7uA in addition to the

That is good to hear, especially since you guys do industrial and mil grade stuff. At least there is hope that we just caught a bad batch or so. Order for another pre-series round is placed but unfortunately there were no lot numbers on the last Digikey shipment so we won't know for sure if they were from the same batch.

The load transient response is specified with a 49.9mA load step. The load transient response graph appears to be incorrectly labeled, but I wouldn't expect smoke from your load step. There has to be something else going on here.

By the way, the data sheet was updated in September, so make sure you have the latest rev.

Well, let us know how this turns out. I am thinking about using the

================================

Greg Neff VP Engineering

Didn't you post about this about 3 months ago?

Maybe TI are being a bit optimistic with the "any output capacitor >= .47uF" gives stability bit.

When you turn on your MOSFET slowly you are sweeping an R-R//C network across the output, maybe stability is marginal somewhere in that sweep and sometimes some devices will honk and kill themselves.

Did you look carefully at the regulator output and maybe the input current during the ramp?

Hello Greg,

The load transient graph shows around 350mV excursion on a full swing. Ouch. And they conveniently did that with a 10uF cap, not the recommended 0.47uF low limit. Also only 1V Vin-Vout. It probably wouldn't have looked good enough otherwise ...

On light loads they do fine though. I just wouldn't use them anywhere near full current.

Thanks for the pointer. I have the latest but when clicking the data sheet on Digikey you still get the Jan-2004 version. Wish they had some kind of automatic update mechanism in place with key manufacturers.

So do I, in an MSP430 application. But first I'd like to see these things behave consistently, else I'll roll my own again. I just don't like this whole LDO concept. Seen too much grief there. And I wish they had put a thermal limiter into the devices, can't be rocket science.

I'll post when I know more.

That was about a different problem. The devices blew instantly when input voltage (within range) was applied. Turned out to be a faulty series of bench supplies, not the part. Unbelievable, a $1000+ supply generated load change spikes while my old Maybe TI are being a bit optimistic with the "any output capacitor >=

Hopefully they weren't ... but I'd like to know.

Well, since there is a 1uF on the output it would only sweep from 1uF to

Unfortunately I can't since it's far away at a client. But I may have to.

(snip) I recommend you add a 10 or 15 ohm resistor in series with the 1 uF output capacitor. This will make the output load change from capacitive to resistive just about at the frequency where the output impedance peaks. I would also raise the input capacitor to be a couple times larger than the total capacitance that is switched at the output. This will increase the current spike, but lessen the input voltage collapse.

Hi Joerg,

Joerg a écrit :

Did you check this? (the few mA) I mean with a real wire loop and current probe. It might happen that your hand_crafted_analog_circuit_costing_no_more_than_half_a_coffee_:-) draw some highish transient current during power ramp up.

Hello Fred,

Yes.

With a current sense resistor and diff probe.

It doesn't. Actually, it wouldn't be able to :-)

Hello John,

I am sure TI wouldn't endorse this. The ESR needs be low and placing a

The input capacitor could be increased a bit but it's on the HV side so it would cost some space. And right now it is all pretty much what the data sheet recommend. Let's see what TI comes back with tomorrow.

Whatever the cause, the source of current that does the damage is either the battery or even that 0.1uF. It it practical to put in a higher voltage zener, or even a low Idss jfet (as a diagnostic current limiter).

BTW: You had a similar thread some months ago. Any chance it is the same company, maybe even the same technician?

How about you use a four-trace digital scope watching the regulator input and output voltages and currents? With that setup you could potentially figure out if it's excess current in the input or output loop, and at what step in the voltage ramp.

Oh, here's another possibility-- with the FET off, the input capacitor could charge up to 24V thru the zerner's leakage current. Then when you turn on the FET, kablooie.