EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: Tom2 on May 01, 2023, 08:57:14 pm
-
First hello,
I am a long time forum reader (as a guest) and stuck now on a problem. So I created an account and this is my first post.
I ordered the split powersupply designed by "Phil's Lab" on JLPCB. For the order I used the pre-packed gerber files and the assembly files from the github repository.
Video Link: https://www.youtube.com/watch?v=qXWYXxDokv4 (https://www.youtube.com/watch?v=qXWYXxDokv4)
Git Link: https://github.com/pms67/BananaSchplit (https://github.com/pms67/BananaSchplit)
Some parts were not available anymore and I tried to select matching components. (I am not an electronic engineer, btw)
L1 original C169398 replaced with C718283.
L2,L3 original C139506 replaced with C408412.
After the PCB arrived I hooked the the board to my lab power supply (10V, 100mA Current Limit) and checked the output with a multimeter.
The +5V and +12V measured fine, but the -12V was measuring as -15V. So I attached my oscilloscope to the -12V rail and saw a sawtooth waveform from -12V to -19V. (Images SDS00001.png, SDS00015.png)
Under light load (100mA) the -12V rail jumps to -21V, but the sawtooth waveform is gone and the RMS noise is 41mV. (Image SDS00016.png)
I also removed the R11 resistor, but this didn't make a difference.
The +5V and +12V looks fine on the oscilloscope with and without a load.
All of the boards behave the same and the soldering looks fine to me.
So I was wondering if one of the replacement components are the issue or I made an error in the order process?
Thanks in advance and best regards,
Tom2
-
The part numbers for L1 , L2, an L3 are meaningless. What are the values? Are those Values the same as the originals?
Did any other values change from original design? Like resistors and capacitors.
-
First, welcome to the forum proper! I was a long time lurker, but it's much more fun if you get involved.
When asking for help, it would be good to provide some more background info directly in the post - easier to get "buy in" from the forum experts here. A basic requirement for this kind of question would be a schematic. Similarly, it's more useful to provide part descriptions/characteristics, rather than catalog numbers - some of the most senior (i.e. most experienced) forum members may not recognise LCSC part numbers, and even if they did, its hard to expect them to go searching the LCSC website for your query - not to say they won't, but I'd say it would only be polite to provide the data for them. As this is a board that you had assembled, the full, as built, BoM (i.e. your specific parts list) should be easy to provide, and finally, a photo of the actual board under investigation would be nice.
OK, we've all been newbies so:
1) see below for the schematic (which I could only find in the youtube video - would be nice if there was a pdf version on the associated github...)
2) L1 original: Sunlord SWRB1204S-470MT, 1.9A 47uH ±20% SMD Power Inductor 160mΩ
replacement: YJYCOIN YPRH1204-470M, 1.9A 47uH ±20% SMD,12x12mm Power Inductor 150mΩ
3) L2, L3 original: Sunlord MWSA0503-100MT, 3.2A 10uH ±20% SMD Power Inductor 125mΩ
replacement: Sunlord MWSA0503S-100MT, 3.2A (rated) 10uH ±20% 3.5A (sat) SMD Power Inductor 125mΩ
To be honest, in my personal opinion, those substitutions seem fine to me - I doubt that they'd be the cause of your problems. If it was a self built board I'd be more worried about whether the thermal pad of TPS65131 is properly soldered, however, under light load, you might get away with it being floating, as the datasheet seems to suggest that it is "just" for mechanical and thermal puposes. As it sounds like you had it professionally assembled, one would normally expect this pad to be fine.
Given that the negative output is oscillating, you could start by investigating the negative feedback newtork - R8, R9, C11. C11 is supposed to be very small, it's worth temporarily removing it for troubleshooting purposes.
-
The part numbers for L1 , L2, an L3 are meaningless. What are the values?
Could you please refrain from giving such absolute bullshit advice to beginners? It's exactly the other way around, especially for switch mode power supply inductors: only part numbers are meaningful; values like inductance are meaningless alone.
-
When asking for help, it would be good to provide some more background info directly in the post - easier to get "buy in" from the forum experts here. A basic requirement for this kind of question would be a schematic. Similarly, it's more useful to provide part descriptions/characteristics, rather than catalog numbers - some of the most senior (i.e. most experienced) forum members may not recognise LCSC part numbers, and even if they did, its hard to expect them to go searching the LCSC website for your query - not to say they won't, but I'd say it would only be polite to provide the data for them. As this is a board that you had assembled, the full, as built, BoM (i.e. your specific parts list) should be easy to provide, and finally, a photo of the actual board under investigation would be nice.
Yeah sorry, there were to many implicit information. When ordering an assembled PCB from JLPCB the parts will be stocked from LCSC, therefore I posted the LCSC part number instead of the part name/description. The github link shows a KiCAD project, but I understand that downloading/cloning the repository to view the schematic is a hassle. The next time I will provide more information, so that a reader don't need to "hunt" around for the information. I attached the PDF version of the schematic.
Given that the negative output is oscillating, you could start by investigating the negative feedback newtork - R8, R9, C11. C11 is supposed to be very small, it's worth temporarily removing it for troubleshooting purposes.
I had my eye on the same area, but before removing components I was looking for a mistake on my side. I assumed that the BOM and part selection was tested.
So I re-attached R11 and removed C11 ... aaand the negative 12V rail works like a charm. The -12V is stable loaded (up to 420mA) and unloaded. (Images SDS0000xx)
I measured C11 as 8.9pF with my LCR meter, so it is close to the specified 8.2pF.
The TPS65131 datasheet has an equation to calculate the "Feedforward Capacitor". (Image DatasheetNotes)
Maybe the value is a little to high, if I am not mistaken the recommended value would be 7.5 pF (7.5 us/1 MOhm).
The layout of the board follows the application note. (Image Layout)
Should I leave the C11 unpopulated or try a smaller capacitor?
EDIT: Uploaded missing images.
-
Honestly, I was expecting that the feed forward compensation capacitor, C11, was the incorrect value, resulting in the instability. Sometimes the parts get mixed up on the production line. Given that you have confirmed the value (well done, by the way - for these small values it can be tricky to measure. I generally just substitute a known good part), its a bit of a mystery.
At the end of the day though, you are only interested in the performance of the circuit. As it sounds like you are quite well equipped, you might be better off doing some load testing to assess if it matches your particular requirements without the capacitor, or to determine the best cap size.
-
On my next order I will add some SMD capacitors with various values for testing, but for the time being I will leave the capacitor off.
Thank you Swainster.