DC to DC converter harmonic ringing

[NDani]

Dear Forum Members,

I hope you can give me some hints regarding my question about DC to DC converter noise. In order to power an op-amp, I have built up a switching boost converter, which converts the voltage of a 9V battery to 12 volts using an LT1615 controller chip. (Then I split this voltage to +/- 6 V using a TLE2426 virtual ground chip.) Although the desired voltage is generated, there seems to be a lot of harmonic ringing on the power rail at the beginning of each ramp-up of the ripple voltage. I have attached 2 oscilloscope screenshots of the 12 V power rail, both using AC coupling, but different bandwidths:

• A.png with 20 MHz bandwidth limit on
• B.png using the full 1 GHz bandwidth of the scope.

The measurements were taken directly across the output filter capacitor using the aligator clip on the scope probe for ground.

The main problem is that the harmonic ringing also gets to the output of my op-amp, where I would like a clean voltage. I have attached 2 screenshots of the AC-coupled output voltage of the op-amp:

• C.png with 20 MHz bandwidth limit on
• D.png using the full 1 GHz bandwidth of the scope.

This time, the ground point where the probe aligator clip was attached, was a few cm's away from the op-amp output.

The DC to DC converter is built up on a small veroboard and the op-amp is connected to it by means of a solderless breadboard, into which the veroboard is plugged into. The opamp has 100 nF decuopling capactors on both of its power pins to (virtual) ground.

I would expect that this amount of harmonic ringing is not normal and is not acceptable in most applications, although I'm not entirely sure, since this is the first switching converter I have ever built up. What am I doing wrong? Can you help me figure out how I could get rid of the harmonic ringing seen on the screenshots and get a relatively clean output? I am aware that DC to DC converters always have ripple and noise but this seems excessive. (If I substitute the DC to DC converter with a lab power supply to output the 12 V, then everything is OK.)

Thank you! I am also attaching the schematics of the power circuit.

[Audioguru]

If the 10uF output capacitor is an electrolytic type then it works poorly at high frequencies. A tantalum capacitor is better or add a 0.1uF ceramic capacitor with very short leads across the 12V.
What circuit do you have that needs a virtual ground providing a high ground current?

[NDani]

Thanks for your reply. The 10 uF capacitor is a multi-layer ceramic one. The circuit I am powering is a non-inverting amplifier with an LT6015 op-amp amplifying the voltage across a 0.1 ohm current sense resistor. I don't think the circuit generates high ground currents.

[PChi]

The peaks don't look too bad. Some of the noise can be due to the Oscilloscope probe ground lead and input capacitance ringing. I think that there is an old EEVBlog video showing this. If you haven't done it already probe with a really short ground lead,  piece of copper wire wrapped around the probe ground (probe tip (witches hat) removed). There was an old Linear Applications Note that showed how to filter most of the noise.

[NDani]

Hi,

indeed, if I use a short ground lead and turn on the 20 MHz bandwidth limit on my scope, as explained in the EEVBlog video you referred to, then the peaks on the power rail don't look that bad. But the ringing on the op-amp output is still very high (in the order of 0.1 V) even with the scope BW limit on. I haven't tried using an as-short-as-possible ground connection when probing the op-amp output yet, but that's not really possible anyway. The purpose of the circuit is to measure the op-amp output voltage (to get the current through the sense resistor), and I can't really ask the "user" (even if that's just me) to use a short scope ground connection every time when doing that.

So my feeling is still that something is wrong. Or should I just forget about clean analog singals entirely when using a DC to DC converter? I would like to point out that the ringings have a very high peak-to-peak value (to my understanding), ~0.1 V.

[Benta]

You certainly need a low-value cap across the output 10...100 nF. You also need 10 nF decoupling directly at the opamp.

[NDani]

Thanks, I'll try the low-value cap in parallel with the 10 uF. The op-amp laready has 100 nF decoupling capacitors on its power pins to virtual ground.

[PChi]

An old Linear Technology Application Note Minimizing Switching Regulator Residue in Linear Regulator Outputshttp://www.analog.com/media/en/technical-documentation/application-notes/an101f.pdf may be of interest.

[ogden]

In order to power an op-amp, I have built up a switching boost converter, which converts the voltage of a 9V battery to 12 volts using an LT1615 controller chip. (Then I split this voltage to +/- 6 V using a TLE2426 virtual ground chip.)

If you can live with +/-5V, then use LDO for +5V and inductorless cmos voltage converter like LM2664 for -5V Much lower cost solution than yours.

[NDani]

An old Linear Technology Application Note Minimizing Switching Regulator Residue in Linear Regulator Outputshttp://www.analog.com/media/en/technical-documentation/application-notes/an101f.pdf may be of interest.

Thanks! This application note suggests using a ferrite bead and demonstrates its effect. Is that a common method with small DC to DC converters on a PCB however?

In order to power an op-amp, I have built up a switching boost converter, which converts the voltage of a 9V battery to 12 volts using an LT1615 controller chip. (Then I split this voltage to +/- 6 V using a TLE2426 virtual ground chip.)

If you can live with +/-5V, then use LDO for +5V and inductorless cmos voltage converter like LM2664 for -5V Much lower cost solution than yours.

I need +/- 6 V. An LDO and a voltage inverter would be a less noisy solution, that's true. I'm still wondering though if the amount of ringing I'm measuring on the power line and, particularly on the op-amp output is normal.

[ogden]

Thanks! This application note suggests using a ferrite bead and demonstrates its effect. Is that a common method with small DC to DC converters on a PCB however?

I would like to rephrase. When reduced noise out of switching supply is needed, it is common to use additional LC filters on the output. Ferrite bead is just another type of inductor, commonly used in supply rail filtering.

[David Hess]

The measurements were taken directly across the output filter capacitor using the aligator clip on the scope probe for ground.

Now make the same measurement with the probe tip connected where the ground clip is connected.

[ogden]

The measurements were taken directly across the output filter capacitor using the aligator clip on the scope probe for ground.

Now make the same measurement with the probe tip connected where the ground clip is connected.

For ground don' t use alligator, but spring:

[T3sl4co1l]

Useless to speculate without layout.

Tim

[ogden]

Useless to speculate without layout.

Before we speculate or dig into layout, we shall ensure that ringing problem is actually there. Probe ground lead is known source of "ghost ringing" (best seen in C.PNG). As OP says that output capacitor is MLC, thus small - it does not cost anything to do measurements again, correctly.

[T3sl4co1l]

It's there, regardless.  I'm betting it's common mode, caused by poor layout, and the pulse is visible almost anywhere in the circuit even with a grounded probe tip.

Tim

[NDani]

Thank you for your valuable responses. I have made some new measurements as per your suggestions, please see the screenshots. All were done with AC coupling and 1 GHz oscilloscope bandwidth.

Now make the same measurement with the probe tip connected where the ground clip is connected.

• E.png: Probe tip and alligator ground connection both on the negative side of the output filter capacitor.
• F.png: Same as E.png, but with short (spring) ground connector.

For ground don' t use alligator, but spring

• G.png: Measurement across the output filter cap with the spring ground connector.

So the ringing can be seen even if the probe tip is connected to the same point as the ground connector. And it gets smaller if a spring ground connector is used.

It's there, regardless.  I'm betting it's common mode, caused by poor layout, and the pulse is visible almost anywhere in the circuit even with a grounded probe tip.

Tim

The layout is indeed poor, since the circuit is built up on a veroboard, as a prototype. I'm attaching a photo of it, as well as a screenshot of the planned PCB layout (H.jpg, I.png). But would proper layout on a PCB really significantly reduce ringing?

And my main concern is the ringing on the op-amp output. I can't use the spring ground connection there. (Or, to put it in another way, I wouldn't expect that just because I use a DC to DC converter to power my circuit, I should use an alternative grounding technique on every singe point of the circuit. If the op-amp produced a clean DC voltage, as I'd like, then no special grounding techniques would be necessary.)