1
\$\begingroup\$

I built a boost/inverting converter using TPS65130. The output voltage is +15V and - 15V. Below is a picture of the schematics: enter image description here

I know the schematics shows a zener diode symbol, but that's just a mistake. I'm using a schottky diode. Also, the base of U37 is not connected because I have a zero ohm resistor across R50. U41 is an LT3094 and U42 is an LT3045, they are both linear regulators used to filter noise. What I'm trying to do is to measure the noise of the power supply. Here is the procedure that I followed to measure the noise of this power supply:

  1. used a 1:1 probe that has a max bandwidth of 35Mhz.
  2. used a bandwidth limit of 20Mhz on the oscilloscope.
  3. used a ground spring instead of the long ground lead. I also made sure there is nothing else connected to the oscilloscope picking up common mode noise.
  4. I measured directly at the output capacitor C152 for the boost and c179 for the inverting converter.
  5. the load current is 80mA.
  6. AC coupling.

The problem is I noticed some weird overshoot and undershoot. The first picture shows what I see when I measure directly across C152 which is the positive supply: enter image description here

The next picture is what I see when I measure across C150 which is after the linear regulator: enter image description here The Next picture is what I see when I measure across C179 which is the negative supply: enter image description here

The final one is if I measure across C1 Which is after the linear regulator: enter image description here

I have checked everything including the inductor and output capacitors. Checked things like saturation current, rating current, etc. These waveforms repeat and when I measure the frequency between two repeating waveforms, I get 6Hz. This is the same for both the positive and negative supplies.
Any Idea on what this could be?

Edit: I noticed that the problem disappear if I lower the input voltage to 4V. Not sure why does it work with 4v better than 5V. Edit2: zoomed picture of the ripple on the positive power supply. enter image description here

\$\endgroup\$
13
  • 1
    \$\begingroup\$ Is this a problem? It looks to be only a few millivolts, which shouldn't be a problem for most things. What's your load? \$\endgroup\$
    – Hearth
    Commented May 29, 2021 at 18:53
  • \$\begingroup\$ To be honest, I'm not sure how much of a problem it is. I'm using it to power an op-amp circuit. I just think usually with sensitive analog circuits, you want the lowest amount of noise possible. Also, I just honestly found it to be weird. \$\endgroup\$
    – Analog
    Commented May 29, 2021 at 20:45
  • \$\begingroup\$ What is the output voltage from the boost converter? Maybe I'm misreading the values in the diagram, but I get 1.213*(1+976k/110k) = 12v, which can't be right if you're generating 15v. Also, 121k (I think, hard to read) on the LT3045 is also 12v, so I think you aren't regulating? \$\endgroup\$ Commented May 30, 2021 at 3:18
  • \$\begingroup\$ Instead of powering an op amp circuit, try connecting a resistive load instead and make the noise measurements. Maybe your op amp circuit is drawing bursts of current at 6kHz. Also, check if your DC-DC converter and linear regulator have been compensated properly and have an adequate amount of output capacitance. \$\endgroup\$ Commented May 30, 2021 at 10:20
  • \$\begingroup\$ As far as the output voltage, you are absolutely right. The output voltage is 12V not 15V. That was a mistake in the schematics. That shouldn't cause any problem for my circuit though. \$\endgroup\$
    – Analog
    Commented May 30, 2021 at 19:41

1 Answer 1

1
\$\begingroup\$

Starting with your positive rail and reading off the resistor values, I get that your output voltage is 1.213*(1+976k/110k) = 12v (rather than 15v).

Proceeding onto the linear regulator, you put the set pin to 121 kohm, which is 12v output. Hence, you are feeding an LDO 12v and telling it to generate 12v. That isn't possible, so the LDO is in dropout and so you see lots of noise on the output. You need to fix that.

Before you do, take a look at the PSRR as a function of voltage for your linear regulator:

PSRR vs voltage

The regulator turns on at about 300 mV and starts regulating, but its performance is extremely degraded. Increasing that to 1V (so 13v in) gives you 20-30dB better PSRR. Increasing it to 2V (so 14V) gets you another 10dB rejection at your boost converter frequency. If you're buying an expensive high PSRR regulator, you almost certainly want that extra 30-40dB! Set the output voltage to +/-15v like you said originally and (assuming good layout) you'll reduce that noise voltage by a factor of 10,000, which will put your noise into the low microvolt amplitude.

\$\endgroup\$
2
  • \$\begingroup\$ I finally understand what you were saying. I even forgot that I put a 121k ohm on the set pin, that's why I was confused. I was fixated on knowing the reason behind the odd waveforms on the boost converter that I didn't notice the mistake. Hopefully, this will help filter all that noise. Thank you very much for your help. \$\endgroup\$
    – Analog
    Commented May 31, 2021 at 4:29
  • \$\begingroup\$ @Analog No problem, it actually jumped out at me since I was laying out something similar with the TPS65130 recently (but using cheaper regulators). \$\endgroup\$ Commented May 31, 2021 at 4:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.