Why would touching a DC/DC converter stabilize it?

Why would touching a DC/DC converter stabilize it

Hey all,

I'm seeing a weird phenomena with a DC/DC converter I'm using, specifically a quad TPS65400. It has four buck converters in one package, three of which are operating perfectly on multiple boards, the same three 1.2V 1.8V and 3.3V. The fourth, a 7.5V supply is acting strange. The switching node will look good with a waveform of the proper duty cycle but will then fall to a DC value about half of the input voltage for a few msec before starting up switching again. The result is a slightly unstable 7.5V supply making a nasty buzzing noise that is not there when the switching is working correctly. Now the strange part is when I touch the top of the IC with even just the softest contact from my finger the supply looks perfect. When I remove my finger the problem returns. I have two lines of thought. First I noticed when replacing one of these ICs that there is extremely little to no solder on the large ground pad beneath the IC and maybe my touch is making the connection. However, if this were the case I wouldn't expect the other supplies to work. Second, I wonder if I'm adding some needed capacitance to the circuit through my body but I can't imagine why touching the black part of the IC helps. Any ideas?

-Kevin

Re: Why would touching a DC/DC converter stabilize it

Make sure that you have a good bypass capacitor on the power pins. When you touch something like that, you are adding about 5pF of capacitance to the circuit. Either your bypass cap is bad (or non-existant), or putting pressure on the solder joints completes the circuit (bad solder joint).

Re: Why would touching a DC/DC converter stabilize it

KD494 said:

adding some needed capacitance to the circuit

Click to expand...

Capacitance is likely. Or else your hand's proximity might carry ambient mains hum near the IC. Who can know whether it stabilizes a borderline situation, or adds an instigator to that borderline situation.

Another possibility, your fingerpress restores a bad connection inside the IC.

Re: Why would touching a DC/DC converter stabilize it

KD494 said:

I'm seeing a weird phenomena with a DC/DC converter I'm using, specifically a quad TPS65400. It has four buck converters in one package, three of which are operating perfectly on multiple boards, the same three 1.2V 1.8V and 3.3V. The fourth, a 7.5V supply is acting strange. The switching node will look good with a waveform of the proper duty cycle but will then fall to a DC value about half of the input voltage for a few msec before starting up switching again.

Click to expand...

What is the input voltage? Is the 7.5V output lightly loaded? Could be that it's operating in a burst mode, in which case it may normally appear unstable, but that's just how burst mode works.

Re: Why would touching a DC/DC converter stabilize it

Whether 7.5V regulator is initially working and suddenly not working. Or It is not working at all in faulty board (without your touch)... For me it looks like a Dry solder or sometimes if regulator compensation is not designed or on edge we see this kind of behavior....

Re: Why would touching a DC/DC converter stabilize it

What is the input voltage? Is the 7.5V output lightly loaded? Could be that it's operating in a burst mode

Click to expand...

The input voltage is 15V and this problem occurs both with no load and a 1A load. Is burst mode associated with DCM operation? The chip defaults to CCM operation only and I've verified that by checking the registers.

Make sure that you have a good bypass capacitor on the power pins. When you touch something like that, you are adding about 5pF of capacitance to the circuit.

Click to expand...

Is this the input power pins you're talking about? I have bypass caps on input and output but I'm just curious. Whats confusing me about the capacitance is why touching the casing on the IC is what fixes it, wouldn't the fact that its an insulator negate any effects of me touching it?

Whether 7.5V regulator is initially working and suddenly not working. Or It is not working at all in faulty board (without your touch)...

Click to expand...

On all three boards I have the other three supplies function with or without touch but the 7.5V will only ever work correctly if I'm touching the IC. One thing I've noticed is that even though the datasheet seems to say I can make 7.5V with it all the demo circuits stay below 3.3V like my other supplies.

Or else your hand's proximity might carry ambient mains hum near the IC. Who can know whether it stabilizes a borderline situation, or adds an instigator to that borderline situation.

Click to expand...

That's an interesting thought, I wonder if there's an easy way to test this theory.

Re: Why would touching a DC/DC converter stabilize it

KD494 said:

Is this the input power pins you're talking about? I have bypass caps on input and output but I'm just curious. Whats confusing me about the capacitance is why touching the casing on the IC is what fixes it, wouldn't the fact that its an insulator negate any effects of me touching it?

Click to expand...

Possibly... It's possible that you are getting some RFI feedback that your finger is effectively bypassing. Another big possibility is a bad solder joint on the board.

Where are the photos?

Re: Why would touching a DC/DC converter stabilize it

Touching won't introduce bigger capacitance. If you have a circuit diagram like schematic or layout post it. If assembly issues are not there (pls confirm this from your side), then design is in edge can be Regulator is un-stable because of may be compensation or something else or bad layout....

Re: Why would touching a DC/DC converter stabilize it

Looking at it philosophically is fun but it will not make the circuit work ..... which is much of what we are paid for.
Either the IC is bad, one of the discretes is bad, the PCB is bad or the installation [soldering] is bad
I'd try replacing all the discretes first. Probabilities favor that.
Then replace the IC which will cover 2 more of 4 possibiities.
I would bet the touch sensitivity will be gone by then.

A bad PCB or loose solder sliver shorting something is the remaining possibility.
I assume your design calculations and component selection is solid.

Re: Why would touching a DC/DC converter stabilize it

KD494 said:

The input voltage is 15V

Click to expand...

So when you said the switching node goes to half the input voltage, you meant it goes to 7.5V, which is the output voltage. Sounds like the converter is just shutting down or going to sleep, like it would in a burst mode.

Is burst mode associated with DCM operation? The chip defaults to CCM operation only and I've verified that by checking the registers.

Click to expand...

Burst mode can be applied to basically any operating mode. It's just another layer of modulation/control at a lower frequency.

See page 28 of the controller datasheet:

At light load, the COMP terminal is driven by the error amplifier to the minimum clamp voltage. When the COMP voltage reaches below 0.6 V and the error amplifier is sinking more than5μA, both the high-side and low-side FET will be tri-stated to prevent the output voltage from rising above the set value. The FET function is re- enabled when the GM amplifier sinks less than 3uA. This results in a burst mode operation at light load. The low-side FET has a 200-ns one-shot ON-time to ensure that the bootstrap capacitor is charged before the normal function of the converter is resumed

Click to expand...

What is the voltage on you COMP terminal doing?