Author Topic: voltage drop on 2KW Bitcoin mining power supply (Read 43625 times)

800908 · « **on:** March 26, 2021, 03:55:05 pm »

Hi
I am working on a 2kw power supply that used for mining.
the power supply brand is WhatsPower and the model is P5.
output should be 12.8v and 172a max.
problem is that voltage without load is not stable and rapidly change from 11.6v to 12.1v and when i connect three 100w car lights bulbs to it the voltage become 11.7v without changing.
i checked some components and voltage on the board, every thing was good.
where should i focus my attention to?
any help. thanks.

PKTKS · « **Reply #1 on:** March 26, 2021, 04:02:07 pm »

Sample network on the secondary side.

Transcoductance feedback elements and
primary side opto coupler polarization network.

All the components may be aged or drifted from specs..
or if GLUE is used on smd parts..

remove them CLEAN ALL GLUE and try again...

Paul

800908 · « **Reply #2 on:** March 26, 2021, 08:58:53 pm »

there is two smd 6 pin opto coupler on the back of the board.
the part number is not too obvious but seem to be 1801
how ever when i check pin number 6 to the primary side ground one of them show 250 ohm and the other one show very high value.
is that should be?
look at the picture below U7 & U27.

PKTKS · « **Reply #3 on:** March 27, 2021, 11:17:56 am »

2 or more opto couplers usually means that you have
(or you may have) independent sampling networks and
independent supervised power rails.

In this case you have doubled the check points.

First collect information about this supervisor (unreadable by
the photo) to check how protections are implemented...

Then check why the sampled feedback network is drifting..

IMHO components aged and usually in my case I replace
the whole network to check again... I have done this
before with *not very good* results because it is hard
to find EXACTLY REQUIRED components in some cases..

Although better.. some fluctuation still may persist

Paul

800908 · « **Reply #4 on:** March 27, 2021, 02:00:14 pm »

the PWM ic part number is ICE2HS01G
from my research on Internet this problem may cause by opto coupler or resistors connected to it.
i checked most of low value resistors and they fine.
i can't find opto coupler from number written on it.
first row is L1801 and second row is 314 and third row is Y
look at picture.

ace1903 · « **Reply #5 on:** March 27, 2021, 03:56:29 pm »

I would check primary DC voltage first. Sometimes stuff like faulty relay that bridges inrush NTC can cause unstable primary voltage.
If there is primary PFC circuit, check for capacitors with increased ESR. Think that in equipment that is used in though environment preventive replacement of all capacitors is half way job done.
Can you post pictures from component side also?

800908 · « **Reply #6 on:** March 27, 2021, 04:24:07 pm »

this is the front side picture.
voltage of big caps is 410v as it should be.
so i think PFC is fine.
and also i heat up all caps with heat gun while it was running to see if voltage get a little bit stable but no change.

PKTKS · « **Reply #7 on:** March 27, 2021, 04:36:02 pm »

i Have just overlooked the datasheet ....

This is one of these "modern" new resonant topologies...
I am not surprised at all one of these failing so shortly...

The basic bare bones are much the same ...
431 TYPE3 compensated feedback with current sensing resistor .

But.. too much "MODES" of operation..
the folks seems far more concerned with "efficiency" than load regulation..

The result is that it will be hard to spot the culprit of such poor
load regulation - or even NON REGULATION AT ALL..

** Did you performed a LOADED TEST REGULATION ?
trying to see the REQUIRED VOLTAGES on FB/CS pins?

From datasheet CS is independent and boasts a 4 stage comparator
While FB (aka LOAD) is fairly typical but it has a threshold of 1.8V to enter
decision algos and change operation modes...

I would check this threshold asap.. waveform comparison of
both FB/LOAD and CS too see why the CS is ramping wrong
and not triggering the proper duty-cycle..

Honestly these resonant designs seem too bloated of stuff.
hard to spot such simple drifts easily..

A half load is recommended as this IC changes states on light loads..
(perhaps to acquire these new BADGES.. )

Paul

800908 · « **Reply #8 on:** March 27, 2021, 09:32:40 pm »

unfortunately I don't have scope to see the wave form of IC.
there's digital pot on bottom right of board with part number of CAT5140ZI-50-GT3. looks like it's common failures in this model power supply.
i don't have the part or douner board to test so how can I be sure that this is the problem and then order the part.

PKTKS · « **Reply #9 on:** March 28, 2021, 07:04:28 pm »

Try
https://www.banggood.com/MUSTOOL-MDS120M-Professional-Digital-Oscilloscope-120MHz-Analog-Bandwidth-500MS-or-s-Sampling-Rate-320x240-LCD-Screen-Support-Waveform-Storage-with-Backlight-p-1759856.html

Cheap affordable - does the basic stuff wo hassle.
Portable, isolated from mains..

Without visual measure of the operating point..

Almost guessing and random replacement it will be.

Paul

800908 · « **Reply #10 on:** March 28, 2021, 09:41:55 pm »

base on datasheet of CAT5140ZI-50-GT3 pin 6 is potentiometer low terminal and here its connected to ground.
pin 7 is potentiometer high terminal. so its must be 50k ohm compare to the ground but if I put black probe to ground and red probe to pin 7 it show 4M ohm and when i swap the probes it show about 3k ohm.
also pin 5 show exactly like pin 7.

PKTKS · « **Reply #11 on:** March 29, 2021, 01:30:56 pm »

without the exact implementation of this "resistor"
this reading is dubious... it may be (perhaps it is)
a semiconductor reading.. and biased only hot.

A cold reading may be not trusty.. but this schema
is tricky anyway... with so many operating modes
some clever procedure to check the regulation
*should* be done.

Scope will help

Paul

800908 · « **Reply #12 on:** March 29, 2021, 01:39:29 pm »

OK thanks.
i ordered the potentiometer and I hope changing it fix the problem.
datasheet said it is programmable pot. so how to program it?

PKTKS · « **Reply #13 on:** March 29, 2021, 01:54:44 pm »

I2C EEPROM

like many modern (cheap) DMMs also use.

Paul

amyk · « **Reply #14 on:** March 30, 2021, 01:36:18 am »

"mining"? I was expecting something more industrial-looking than that...

dc101 · « **Reply #15 on:** March 30, 2021, 02:23:57 am »

Quote from: amyk on March 30, 2021, 01:36:18 am

"mining"? I was expecting something more industrial-looking than that...

I was thinking this is a power supply for some custom crypto-currency miner?

800908 · « **Reply #16 on:** March 30, 2021, 07:39:06 am »

Yes it is used for crypto-currency mining.

amyk · « **Reply #17 on:** March 30, 2021, 01:24:02 pm »

That makes far more sense...

Check the ESR of the secondary caps.

800908 · « **Reply #18 on:** April 24, 2021, 06:37:16 pm »

Hi
after a month today the POT arrived.
when I replace THE digital POT with new one the output changed but still act like before. so remove the POT and short the pin 5 and 6. turn it on and the voltage become stable at 13.0 volt. then I add 11k ohm resistor to pin 5 and ground. turn it on and voltage was like before changing from 10 volt to 11 volt rapidly. then I add 680 ohm resistor and turn it on surprisingly voltage is 12.8 volt as it should be and I tested it with 30a load and it is stable.
so what is the purpose of this digital POT?


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: voltage drop on 2KW Bitcoin mining power supply (Read 43625 times)

Share me