Author Topic: Lipo to 5V boost converter ~1A, overshooting a lot  (Read 688 times)

0 Members and 1 Guest are viewing this topic.

Offline hsquaredTopic starter

  • Contributor
  • Posts: 16
  • Country: us
Lipo to 5V boost converter ~1A, overshooting a lot
« on: June 06, 2023, 09:48:26 pm »
I am trying to run a small servo, MG90S, using an ESP32 development board.

I read the stall current is 700mA, so I bought this small boost converter on Amazon, rated to 5V and 1A.
https://www.amazon.com/dp/B091CRCZ65?

With no load, I noticed the servo was occasionally glitching.


I hooked up my scope, and measured 800mA peak.
However, 3ms after the load ramps down, the 5V output shoots up to 7V for 1 ms, before coming back down.

I tried adding a 10uF capacitor on both 3.7V, as well as 5V. But the problem is still there.

Is there something else I can do, or a suggestion of a small simple boost converter for my application I can buy?

Edit:
Also, I see "4R7" on the inductor on the board. Given this is 4.7uH, then the energy should just be 1/2*4.7uH*(400mA)^2 = 4.7uJ
If I wanted to make sure the voltage does not overshoot above 6V, if the load is suddenly disconnected, then 1/2*C*[(6V)^2 - (5V)^2] = 4.7uJ will calculate the minimum capacitance? I calculate C = 6nF.

So it does not seem like there is insufficient capacitance to soak up the energy. The converter itself isn't responding for some reason and continues switching..?
« Last Edit: June 06, 2023, 10:16:47 pm by hsquared »
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 21611
  • Country: us
Re: Lipo to 5V boost converter ~1A, overshooting a lot
« Reply #1 on: June 06, 2023, 10:41:24 pm »
First thing I'd do is look at the datasheet for the IC the converter is based on and then compare the application notes to the circuit used on that board. In my experience with inexpensive Chinese modules, they have usually skimped on component values, so the inductor is probably marginal and so are the capacitors, the layout may be less than optimal. You could try something large on the output, like 100uF or even 330uF and see if that helps.
 

Offline hsquaredTopic starter

  • Contributor
  • Posts: 16
  • Country: us
Re: Lipo to 5V boost converter ~1A, overshooting a lot
« Reply #2 on: June 06, 2023, 10:57:46 pm »
I think it might be this, based on the listing photos:
https://datasheetspdf.com/pdf/1088422/SHOUDING/SDB628/1

I actually found out two separate issues.

1) If I tap a 56Ohm resistor to ground, the ~100mA causes overshoot where about 50 cycles (40us), the converter continues switching. the ~100mA causes an overshoot. Then converter stops switching, and voltage settles back.

2) With the actual servo, the converter stops switchin when the load diminishes. However, even with zero switching, the 5V rail rises. If there was unsuppressed back-emf, would it be a dip in the 5V rail..? thus this is something else? Edit: If place my finger on the servo as it moves to position to provide some resistance, it obviously is slower, I can make the overshoot decrease. Picture of the overshoot below.

Edit:
3) If powered off of USB 5V on the ESP32, there is no overshoot on that 5V rail. Maybe the rest of the ESP32 is using the 5V power, so there is somewhere for the backemf energy to go..?

Pictures for 2) and 3) added below.

Adding approximately ~77uF, the overshoot drops from 8V down to 6.5V.
I just placed an order for 100uF 1206 capacitors that I can solder right on top of the existing ones.. Two capacitors cost as much as the entire board   :D
« Last Edit: June 07, 2023, 03:56:05 am by hsquared »
 

Offline hsquaredTopic starter

  • Contributor
  • Posts: 16
  • Country: us
Re: Lipo to 5V boost converter ~1A, overshooting a lot
« Reply #3 on: June 14, 2023, 02:10:30 am »
I added 2 100uF 10V X5R 1206 capacitors on the output.

I still get the overshoot. It is sometimes generally lower, but also sometimes just as high.

I put a 0.1Ohm shunt to measure the servo's 5V current, and it goes negative exactly when the 5V overshoots..

« Last Edit: June 14, 2023, 03:13:32 am by hsquared »
 

Offline MathWizard

  • Super Contributor
  • ***
  • Posts: 1333
  • Country: ca
Re: Lipo to 5V boost converter ~1A, overshooting a lot
« Reply #4 on: June 14, 2023, 05:48:12 am »
What does it do if you just have a resistor as the load ? IDK near enough equations for this stuff, but if it was made to work across a big wide range of conditions, maybe some other values could be changed to work better for this specific use. But yeah it's not like there's many external parts.
https://datasheetspdf.com/pdf-file/1088422/SHOUDING/SDB628/1

What is the full circuit of the converter ?

Is there an idle resistor on the output besides the feedback divider? Maybe a snubber would help more too, just a cap and small resistor in series, from Vo to GND. You don;t want to make an oscillator tho.
« Last Edit: June 14, 2023, 06:17:46 am by MathWizard »
 

Online sleemanj

  • Super Contributor
  • ***
  • Posts: 3018
  • Country: nz
  • Professional tightwad.
    • The electronics hobby components I sell.
Re: Lipo to 5V boost converter ~1A, overshooting a lot
« Reply #5 on: June 14, 2023, 07:47:46 am »
Try large-ish electrolytic capacitor on the output, 100 to 470uF.
~~~
EEVBlog Members - get yourself 10% discount off all my electronic components for sale just use the Buy Direct links and use Coupon Code "eevblog" during checkout.  Shipping from New Zealand, international orders welcome :-)
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 21611
  • Country: us
Re: Lipo to 5V boost converter ~1A, overshooting a lot
« Reply #6 on: June 16, 2023, 10:59:42 pm »
I added 2 100uF 10V X5R 1206 capacitors on the output.

You should try tantalum or aluminum electrolytic.

Multilayer ceramic capacitors such as those you used have a property that is not nearly as widely known as it ought to be and that is the capacitance drops, often substantially when there is a DC voltage present across the capacitor. Those 100uF capacitors may only be around 20uF in the application you are using them, this has bit me a few times in the past.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf