Power Supply 60v 250A Repair

[bdunham7]

I also see in the instruction manual for the ESS that inductor L1 limits the peak charging current.  Unfortunately, they don't specify the maximum surge. 

I was wondering about those chokes, which I would have assumed were to suppress HF noise from the inverters.  Do they give values for L1A/B and L4?

[CrazyTiger]

Maybe I missed it somewhere above, but have you confirmed the relay contacts across that resistor are clean and functioning?

The relay contacts are good was tested with 34401A to check for any significant resistance drop. The relay was also tested with a insulation tester to check for any contact leakage between NO and NC. Everything tested ok.

Did you check the coil to make sure it isn't open?  In a perfect world, put 24V on the coil and confirm that the relay operates and the contacts close.  There's a diode across the coil so mind the polarity.

I will be getting few 100W and 50W bulbs and sockets today. I will be also borrowing a 3 phase variac, at least probing can be done in a safe manner with a variac.

Careful!  A variac does **NOT** provide isolation from the line.  It only reduces the voltage.


For the bulbs, remember that you'll need two 220V bulbs in series for each of the three phases.  So a total of 6 bulbs & sockets.

By the way, you said that you were working in your other house that has 3-phase power.  Is it common for houses in Malaysia to have 400V 3-phase electrical service? 

1. The coil was engaged with external supply for checking.

2. Each phase will need two light bulb for sure.

3. A normal house in Malaysia usually comes with 230VAC Single phase, but when the house gets big, 3 Phase is provided with normally two Distribution boxes, with separate MCB&CB&RCCB provided for the ground floor and first floor, of course Electrical meter is one. Service provider assumes the entire house will likely be air conditioned, and water heaters are running, they calculate on worst case scenario, hence they started providing 3 phases. Its a overkill anyway electricians usually "try" to balance out utilization of each phase during wiring process itself.

I got the house so that I can shift all my instruments to this house and make a huge electronic evil lair

[edpalmer42]

For the bulbs, remember that you'll need two 220V bulbs in series for each of the three phases.  So a total of 6 bulbs & sockets.

Why can't he just put one bulb in place of the resistor?  It will be a bit over voltage, but unless there is a short it should just blink at the most.

When he flips the switch, a single 220V bulb will see the entire 400V line voltage.  It will be running over it's rated voltage until the capacitors charge to something like 250V.  I wouldn't expect a single bulb to survive that.  If the surge-limiting resistors were in place a single bulb might survive, but we removed them to troubleshoot why they died.  As I stated above, L1 limits the surge to some unknown value so that also might save a single bulb.  It just seems to me that two bulbs in series is cheap insurance. 

[CrazyTiger]

I also see in the instruction manual for the ESS that inductor L1 limits the peak charging current.  Unfortunately, they don't specify the maximum surge. 

I was wondering about those chokes, which I would have assumed were to suppress HF noise from the inverters.  Do they give values for L1A/B and L4?

I did measure the inductance at 1KHz for L1A/B it was about 1700uH each coil. But could not find any nominal value.

[edpalmer42]

I also see in the instruction manual for the ESS that inductor L1 limits the peak charging current.  Unfortunately, they don't specify the maximum surge. 

I was wondering about those chokes, which I would have assumed were to suppress HF noise from the inverters.  Do they give values for L1A/B and L4?

No, this is just a copy of the user manual.  It has a bit of circuit description, but no schematics or parts list.  I'm surprised L1 is even mentioned.  It says:

L1 controls the peak charging current which is inherent in capacitive input filters. Additionally, L1 also increases the conduction angle, thereby improving the power factor of the system.

The way they drew L1 with the line connecting the two cores, I thought it was like a common mode choke.

[bdunham7]

When he flips the switch, a single 220V bulb will see the entire 400V line voltage.  It just seems to me that two bulbs in series is cheap insurance. 

Two is fine with me, but the voltage from P-N is not 400V, that is P-P.  Phase to Phase, that is, not peak-to-peak.  It should be ~240VAC P-N.  But while rectified AC has the same RMS value as non-rectified, the sum of the three rectified half-waves will be higher by a bit.  20% or so if my math isn't flawed.

[vk6zgo]

I also see in the instruction manual for the ESS that inductor L1 limits the peak charging current.  Unfortunately, they don't specify the maximum surge. 

I was wondering about those chokes, which I would have assumed were to suppress HF noise from the inverters.  Do they give values for L1A/B and L4?

No, this is just a copy of the user manual.  It has a bit of circuit description, but no schematics or parts list.  I'm surprised L1 is even mentioned.  It says:

L1 controls the peak charging current which is inherent in capacitive input filters. Additionally, L1 also increases the conduction angle, thereby improving the power factor of the system.

The way they drew L1 with the line connecting the two cores, I thought it was like a common mode choke.

So it's an old school "choke input filter"?
They were common in the days of mercury vapour rectifiers to prevent damage to the tubes upon switch on.
They weren't usually the only inductors in the filter back then, though.

[bdunham7]

No, this is just a copy of the user manual.  It has a bit of circuit description, but no schematics or parts list.  I'm surprised L1 is even mentioned.  It says:

L1 controls the peak charging current which is inherent in capacitive input filters. Additionally, L1 also increases the conduction angle, thereby improving the power factor of the system.

The way they drew L1 with the line connecting the two cores, I thought it was like a common mode choke.

I did measure the inductance at 1KHz for L1A/B it was about 1700uH each coil. But could not find any nominal value.

At the value measured, again assuming no math errors on my end, the effect on the resistor would not be that great, although perhaps not totally negligible.  In the operating mode though, it would do what they say by spreading the intra-cycle conduction peaks and reducing harmonics.  It's actually set up as a differential-mode choke, which should also suppress switching noise from the inverters.  I'm not sure why L4 is separate and uncoupled with the other two, but there's probably a reason.

[edpalmer42]

When he flips the switch, a single 220V bulb will see the entire 400V line voltage.  It just seems to me that two bulbs in series is cheap insurance. 

Two is fine with me, but the voltage from P-N is not 400V, that is P-P.  Phase to Phase, that is, not peak-to-peak.  It should be ~240VAC P-N.  But while rectified AC has the same RMS value as non-rectified, the sum of the three rectified half-waves will be higher by a bit.  20% or so if my math isn't flawed.

Thanks for the education.  As I said, I'm not familiar with 3 phase.  I'm an electronics guy.  So Tiger only needs one 220V bulb per phase.  If he has two per phase he might be able to put them in parallel and leave them in place when he starts the entire unit.  Again, cheap insurance.  Of course, when he starts to load it up the bulbs will have to be removed completely.

[Jay_Diddy_B]

Hi,
Let me use an LTspice model to illustrate what should happen during start-up. The circuit is symmetrical I will only model half of the circuit. Under normal operation the currents in the Neutral conductor will cancel, except for the single-phase loads within the power supply. The single-phase loads are the fans and the auxiliary power supply.

Start-up



I have included the inductor, with a value of 1.7mH.

The waveforms look like this:





If I zoom in on the first 120ms:



If I replace the 1.7mH with a short, there is no significant change to the waveforms. The 35 resistor and the 4700uF are the dominant  components.


Full Power Operation


Again I am modelling half of the power supply. The other half is identical. I have applied a load of 8kW.

Model




With the 1.7mH Inductor



Without the 1.7mH Inductor



You can see that the voltage is higher, I adjusted the load resistor, to keep the load dissipation constant. The peak line current is higher.

Unless the inductor is faulty and has a short to ground it has very impact during start-up and is used to improve the line current waveforms at full power.

Jay_Diddy_B

[bdunham7]

Let me use an LTspice model to illustrate what should happen during start-up. The circuit is symmetrical I will only model half of the circuit. Under normal operation the currents in the Neutral conductor will cancel, except for the single-phase loads within the power supply.

Very nice!  On the currents in the neutral conductor, in full-power case with the inductor, they will partially cancel.  Without the inductor, it appears that they would not cancel and the neutral current would be 3X.  The conduction time appears to be <3.3ms and thus <60 degrees, less than 60 degrees => no overlap.  With the inductor, it looks like ~5ms and ~90 degrees, which gives you some cancellation, but not total.

I think L4 and the fact that L1A and L2A are coupled will also make a significant difference in this result.

[edpalmer42]

Jay_Diddy_B,

So what you're saying is that if it takes K1 more than about 0.5 sec. to operate it won't actually be limiting the energy dissipation in R1 because by then the capacitors will be almost fully charged and almost all the energy will already have been dumped into R1.

I'm surprised those resistors survived as long as they did.  It didn't look like they had any heatsinking at all.

This provides more evidence that both resistors should be replaced with higher wattage units.  But does it address the question of why the replacement resistor immediately failed?

[fzabkar]

Instead of a light bulb, why not use the heating element in an electric kettle? A 240VAC kettle rated at 2200W would have a resistance of 26 ohms.

[edpalmer42]

Instead of a light bulb, why not use the heating element in an electric kettle? A 240VAC kettle rated at 2200W would have a resistance of 26 ohms.

A heating element has a constant resistance.  A light bulb has a low resistance if the current is low (i.e. the power supply is operating normally).  But if the power supply tries to draw lots of current (i.e. it has a fault) the resistance of the light bulb will increase and limit the current.  It's a troubleshooting tool to help prevent the further release of magic smoke.

[Jay_Diddy_B]

Hi,

You might want the resistor to open like a fuse, especially if there is something wrong with the big capacitors.

Looking at the inrush current with a current probe would reveal a lot. It should look like the simulation that I showed earlier.

If there is no load on the output of the power supply, the capacitors should charge up and the line current should drop to a low value.

Jay_Diddy_B

[CrazyTiger]

Hi All,

More update, I am still waiting for parts, E14/Farnel been delivering things slow since covid19. Orders from UK and US takes almost 2 week to reach, it used to be 3 Days, pre-covid.

Anyway I have found and fixed more issues with control board. One of the UC3825 chip was no starting causing the main relay to never engage. With that fixed and probing with my Scope, the control board seems to be working now.

The relay works, It takes about 3 Second for the relay to engage. I have also replaced a defective FET(it was open) which is used to activate soft-start pins of UC38225 Chip (SS Pin).

I have also pulled all the Inverters out, tested the IGBT off circuit and all the components. Inverter 2 had one open IGBT, parts ordered. Seems like Murphy Law is very much applicable in the instrument.

Inverter 4 smells weird. Its not because of burned components but this is because of Rat Urine. I suppose Mickey Mouse managed to pee through the air inlet filter. I don't see any component damage on Inverter 4 and PCB seems ok, anyway I will be rinsing the board with IPA. Hopefully no further surprises.

[edpalmer42]

Sounds like you're making good progress.     Cascade failures are not surprising.  One component dies and takes out something else.  It's nice that you have multiple inverters because you can compare a working one to a dead one.

Keep us informed on your progress.

[CrazyTiger]

Update:
Finally I received all the components.

I have replaced all 4 UC3825 Chip on the controller board. On the inverter boards I replaced a pair of TC4429 with Max4429, Each inverter had a pair, I ended up replacing for all the inverters, so 8 in total. Both the 33ohm resistor was replaced for the soft start circuit.

After securing back everything, I turned on the power supply. For the first time ever, the relay clicked and 300V went to all the inverters. The Control Error Indicator turned off and the power supply came to life. Full voltage of 60V was measured and it was spot on. I was happy for a brief moment, till I started seeing the output voltage started dropping back to zero. 

Powering up again the PSU behavior is same, I tried isolating the issue by removing the control cable of each inverter. When only one inverter (any of the 4) is operated alone, the output voltage seems stable. But as I add more inverter the PSU output shuts down slowly.

Another observation I made is before the output dies, the output voltage fluctuates, example when PSU is set to 50V, the voltage fluctuates from 48-52V before shutting down.

Will keep looking tomorrow, my suspect now is at regulation and output rectification section, at least its good progress 

[edpalmer42]

Was 'a brief moment' seconds or minutes?
Could 'shuts down slowly' have been due to the output capacitors slowly discharging?

Some things to check:

- Remote sense jumpers
- Remote programming jumpers
- Overvoltage or Overcurrent protection settings
- Try with or without a load

[CrazyTiger]

Was 'a brief moment' seconds or minutes?
Could 'shuts down slowly' have been due to the output capacitors slowly discharging?

Some things to check:

- Remote sense jumpers
- Remote programming jumpers
- Overvoltage or Overcurrent protection settings
- Try with or without a load

With all 4 inverters, it takes about 2 minutes.
All sense jumpers and remote programming jumpers are made according to default setting stated in the user manual.
Overvoltage setting is set to maximum and this function works
Will put a load to test today.

[bdunham7]

With all 4 inverters, it takes about 2 minutes.

What are your 300V DC supplies doing at that time?

[CrazyTiger]

Final Update:

The power supply works,
The inverter works by two complementary PWM signal being fed into 2 mosfet driver chip and the output is fed into a transformer later switced with IGBTS and rectifiers. I found one mosfet driver dies causing the inverter working in half cycle, hence the reason for the power supply to turn on for brief moment and dies.

I have replaced all the original TC4429 to MAX4429 earlier after noticing one of chip had sign of being burned. Reason for placing max4429 was Farnel did not have stock for TC4429 and blind confidence that an equivalent would work.

I finally figured the inverters did not like MAX4429 as the transformer driver, when I placed back TC4429, the PSU was stable. I did not see any difference in datasheet between both the chips. But testing it on the breadboard revealed MAX4429 is very sensitive to stray capacitance causing the chips to overheat and die while TC4429 is very tolerant.  Another weird finding was the DCR of VCC and GND of a new MAX4429 is about 3MOhm while TC4429 is 12MOhm. Both the chips was purchased from Farnell and RS Component.

So after close to 2 months, with lots of waiting in between for parts. the PSU works. If anyone in the forum ever needs a 15KW power supply in Malaysia. I have one. Till then this is going to be a huge paper weight 

Thanks for the help

[bdunham7]

So what caused the burned resistor?

[CrazyTiger]

So what caused the burned resistor?

Burned resistor was caused by combination of resistor used (someone replaced the resistor before) and the death of pwm chip which caused relay not to engage. Now the resistor used is silicone coated type.