Switching regulator output fluctuates with a touch by bare hands

Thread Starter

p699

Joined Oct 8, 2021
14
After many silly mistakes over a period of two days I got the following circuit to work. This is the second time I worked with switching regulators. I am happy that I succeeded, well at least partially. :)
Screenshot_2021-10-15_18-58-42_edited.jpg
Circuit is from from the data sheet of XLSEMI, XL6009 ic.

I still havent connected any load.

Now I realize it behaves wierd when I touch it with my bare hands or even with the help of any metallic conductor held in my palm. I have marked the red circles where I touched and the output swings wildly between 2V to 90V+, no electrical shock though, the input voltage was just 13Vdc but I find that even this ip voltage fluctuates. I can see this because I have connected 2 multimeters at input and output points. When I remove the contact the voltages come back to the original values. The points marked blue are areas where I touched but produces no such change. The voltages are steady in this case. I also see no heating issues anywhere.

My layout and soldering is very amateurish:( , I have basically reproduced the circuit diagram on a perforated veroboard but tried to keep components as close as possible.

Do you think this is a known problem with such regulator ICs ? My LM317 implementations never behaved like this !
Do you think excessive heat while desoldering this ic from another PCB might have damaged it ?
Do you think I can ignore this problem because this circuit is going to be working in an isolated box and wont be touched ? The voltages are steady when touched any where else apart from the red areas marked on the circuit diagram.
If this is a serious issue and needs to be corrected before I connect it to any precious load, please give a clue as to how to rectify this. It will be helpful of you if you post any links to reading material to upgrade my knowledge regarding this issue.

Thanks for your time :)
EDIT : I am sorry, I forgot to add an important point, The recommended inductance is 47uH. I wound till 90uH, was too lazy to unwind, so its running with higher inductance. Is this the cause of instability ?
 
Last edited:

LowQCab

Joined Nov 6, 2012
1,340
This is largely due to using such a High-Frequency and trying to reduce
the footprint down to "Postage-Stamp-size".

Manufacturers spend ridiculous amounts of time in tweaking to achieve
stable operation for products similar to this.
The fact that You got it to work the first time out is pretty amazing.

A big Capacitor and a LDO-Linear-Regulator is much more reliable
for a DIY Project such as this,
or just drop the Frequency down to ~25-kHz and start over.
.
.
.
 

Thread Starter

p699

Joined Oct 8, 2021
14
I think I am facing problems a bit similar to this, excessive spiking at input and output :
https://electronics.stackexchange.c...ut-under-load-and-extreme-input-distorti?rq=1

At no load the output is stable. At higher boost voltages it starts oscillating.
When a 200mA load is connected both ip and op multimeters give crazy oscillating readings, but my TRMS meter shows stable numbers. I guess there are excessive spikes being induced into both ip and op. At 700mA load there is output voltage sag.

I decided to redo the circuit with different components and better soldering. I have paralleled capacitors, may be the very low ESR is causing problems. I will tell later how this one worked out.


This is largely due to using such a High-Frequency and trying to reduce
the footprint down to "Postage-Stamp-size".

Manufacturers spend ridiculous amounts of time in tweaking to achieve
stable operation for products similar to this.
The fact that You got it to work the first time out is pretty amazing.

A big Capacitor and a LDO-Linear-Regulator is much more reliable
for a DIY Project such as this,
or just drop the Frequency down to ~25-kHz and start over.
.
.
.
I consider it barely working sir. Until my cheap 2$ multimeters read a constant voltage I wont consider it a success. I will try for some more time before I give up. This is probably beyond my capabilities.
 

crutschow

Joined Mar 14, 2008
28,201
Your problem may be the layout, since high frequency switching regulators are very sensitive to that.
Just connecting the parts "as close as possible" is not necessarily sufficient.
The high current paths must be separated from the voltage sensitive points.

The ground should be a copper plane (use a copper clad perfboard), not just a wire.
The high current paths (input line, output line to and from the transformer and to the output capacitor and output should be larger gauge wire as short as feasible.
R1 and R2 should go directly to the output pins.
 

Thread Starter

p699

Joined Oct 8, 2021
14
Tl;dr : I gave up !
It was not a total failure, I succeeded partially :)

I built a second version following @crutschow advice. I kept a straight input and output path, kept the IC further away connecting to this high current path with shielded wires. I even build a shield for the inductor. The performance actually improved a bit. Till 200mA both input and output were stable. At 800mA the same old behaviour started. It sometimes reached 1000V ip during fluctuations :) I think the main point which helped this improvement was placing the feedback resistors R1 & R2 just next to output capacitors as crutschow told.

During all this I observed my DT830D gave these fluctuating readings while my M830BZ (slightly better multimeter) gave stabler readings. The other TRMS multimeter (similar to Aneng 8009) also gave stable readings. However these mask the spikey-ness of the power being measured. Only the dumb DT830D is my yardstick to see if the output was clean or not. I dont have an oscilloscope:( Getting an oscilloscope and tuning the circuit would be a nice project for another day, not enough time for that now. I had hoped to power a couple of hardisks in my other server project. Now I changed the motherboard and went with a professional server power supply, consumes ~70W where ~30W would have accomplished the job :(

The actual dealbreaker for me was the immense effort I would have to put in to clean not just the output, but even the input. The spikes would have propagated thru the transformer and onto the mains and onto other computers.

I modified the SEPIC circuit into the boost only circuit given in the datasheet. The performance was even better. I got upto 34V from an input of 13V at 200mA current. After this I packed up this project as time was up. Some day when I get an oscilloscope I will powerup these circuits again. I havent dismantled them. I now have developed an interest in these switching regulators and power filters, till now I was a LM317 guy.

EDIT : For the benefit of others regarding unstable outputs this note from LM2596 might help:

LM2596.png

Thanks to those who gave their inputs. Bye.
 
Last edited:

Juhahoo

Joined Jun 3, 2019
250
Post your actual schematics, layout and component values on it. You might have just horrible PCB design or totally absurd component values.
 
Top