Mean Well RST-10000 10kW power supply with 24, 36 or 48 Volt output

[BUAROD]

Problem solved! See update Nov. 1st!

Hello!
I was able to bid for a 3 phase 10kW power supply for little money ... however it is a 24 Volt version.
As I need 50 ... 60 Volt instead of the 24V the first step will be some reverse engineering of the
secondary "low voltage" part. My idea is, only to buy some "spare parts" at MW to realize the
modification instead of spending ~2000 € for a new unit.
Does anybody have experience with these parts or an idea, where to get circuit diagrams and
spare part lists? I have attached a block diagran, which is common for all variants.

This is my first topic in this forum, and I hope I am at the righ place ...
Thank you for your support!

[SteveyG]

I would be surprised if you could get a replacement transformer as a spare part to be honest. It may be possible to modify the rest of the secondary side though. Without more details of the unit we probably can't advise on the complexity or potential success of such a modification.

[BUAROD]

Hello!
After receiving the power supply today, I took the attached picture:
there seem to be two identical 5kW blocks inside the unit, connected in parallel on the input (3x 400V AC) and output (24V DC).
My idea to double the output voltage: connecting the two 24V subassemblies in series!
Sounds easy, but I think the regulation circuitry will be the problem, especially if they use a commen ground for both parts ...
I will start reverse engineering the next days.
Any helpful ideas?     

[coppercone2]

meanwell sells special power supplies for series parallel operation AFAIK

[BUAROD]

Thank you for the information, but I do not want to buy new parts,
I want to modify the existing one to increase the output voltage!

[BrokenYugo]

Assuming the outputs of the two supplies have the negative connected to the case/earth, as they typically do, you'll have to alter one so the output is floating, in a safe manner (don't float the whole supply).

[NiHaoMike]

Start by trying to modify it into a pair of isolated 24V outputs, probably just a matter of removing the bus bars linking them and then disconnecting the load share circuit and getting each output to work independently. Then verify that the two outputs really are isolated from each other and connect them in series.

[mikeselectricstuff]

24 to 50-60v is going to be a stretch - probably not worth the hassle.
Look for ex -telecom 48V supplies - often available cheap.
 

[BUAROD]

Thank you for your comments!
Indeed I just try to separate the two subassemblies: attached you can see the "interface board", where all controll signals from both parts are switched together.
This is very similar to the method explained in the RST-5000 manual to connect up to four units in parallel. However, there is one big difference inside the RST-10000:
additionally the two controllers (PIC18F46K80) communicate via a 3-wire cable (GND, TX/RX, RX/TX)! My assumption: with this data transfere you get 100% output
power instead of only 90% when doing the wiring as shown in the second picture. 
In the next step I will replace this 3-wire link with TX/RX communication via opto-couplers. If this works, I will disconnect the ouputs, remove the interface board and
do the output voltage sensing and settings individually. Finally I will connect the two subassemblies in series ... and hope, that it works!
I will tell you in a few days ...     

[BUAROD]

The good news first:
replacing the 3-wire connection between the two PICs by the opto-coupler circuit worked fine to interrupt this GND-connection.
Then I separated the power outputs and replaced the interface PCB by individually wiring of the subassemblies.

Bad news:
the units did not switch on after these modifications, and even worse: after changing back to the original state the output remains dead!

Conclusion: (this assumption was wrong!)
I think it is rather unlikely that components were damaged during the procedure.
What I tend to believe: the "irregular operating condition" was stored as fatal error in the memory, forcing the system to a permanent shut down now!
The PIC18F46K80 has a 1 kBit EEPROM inside ...

Any helpful ideas how to procede without circuit diagrams and knowledge of the PIC`s software?
PS: the communication between the two PICs is still active, but it looks different on the oszilloscope now ...   

[NiHaoMike]

Figure out how the PIC is disabling the PWM and override it. Finding the datasheet for the main PWM controller would be useful there.

[BUAROD]

Hello NiHaoMike!
Thank you for your advice!

The PWM controller is a UCC2895DW with pin 19 (SS/DISB) allowing soft-start and disabling: below 0.5V the controller shuts down.
Also pulling pin 4 (REF) below 4V will cause a reset according to the TI data sheet (page 16).
With the conformal coating on the board it is not so easy to find the path to the PIC, but I will manage this ...

However, when overriding the controllers input, I am concerned what happens with the (safety) functions the PIC normally handles?
Are overload, overvoltage, short circuit detection etc. managed by the PWM controller and the PIC is just responsible for even current distribution? 

PS: a few days ago I asked Mean Well Europe for information about switching the RST-5000 in series ... so far no answer! 

 

[NiHaoMike]

However, when overriding the controllers input, I am concerned what happens with the (safety) functions the PIC normally handles?
Are overload, overvoltage, short circuit detection etc. managed by the PWM controller and the PIC is just responsible for even current distribution? 

On a power supply that well designed, it's unlikely they'll be relying on software for protection. You can always add your own protection circuit to it.

[BUAROD]

After checking several components I found two defective opto couplers. They enable remote switching of the output.
The input circuitry of the RST-10000 PCB seems to be different to the regular RST-5000 PCB: with identical external
wiring the LEDs inside the opto couplers were damaged. After changing the couplers the power supply works again!

I will now continue my attempt to separate the subassemlies to be able to switch them in series ...

Update Oct. 27th:
I took away the bus bars, connecting the outputs of the subassemblies. The two PICs communicate via the isolated Tx/Rx opto-coupler link (picture 1).
After removing the "interface board" (picture 6) each subassembly has its own wiring now with 2k Ohm in series for the ON/OFF opto-couplers (picture 2 +3).
When switching input power on, both subassemblies switch from red to green LED ... but only the right one (master?, picture 4 + 5) activates the output voltage.
Normally the ouptut voltage level can be set with one potentiometer on the "interface board" (picture 6). For testing I used a 1k Ohm resistor for each subassembly.
Measuring the impedance of the resistor inputs shows different values: so it seems that the left control board (slave) has a different population.
The next step will be to check the differences and change the slave PCB to be same as master PCB.

Update November 1st: 
Topic:  can you change a Mean Well RST-10000 series 10 kw power supply with little effort so that the output voltage is doubled?
The answer is: yes, this is very easy because the 10 kw series consists of two almost identical and interconnected 5 kW modules!
Of course the max. output current is halved: I have done the modification with a 24 V / 400 A version which has a 48 V / 200 A output now!
As tests relating to safety (OVP, excess temperature, current limitation, etc.) are still pending, accept the following listing as preliminary, please!
 
The attached pictures now correspond to the step numbers below:

1. swap the three-wire connection cable between the CN308 connectors on the control boards for an insulated version with optocouplers.
A 5 V supply voltage can be taken from the not used CN309 (picture 1_1 + 1_2).
2. remove the "capacitor - PCB" mounted between the bus bars: in the 24 V version there are 4x 2.200 µF / 35 V assembled, obviously to reduce the ripple voltage
(picture 2). Pay attention to capacitors voltage rating after doubling the output voltage!
3. remove the bus bars, the "interface board" and all its connections to the control boards (pictures 3_1 + 3_2).
4. switch the control board inputs individually, as indicated in the manual. Important note: an additional 2 kOhm resistor must be added in series to the supply line
of the ON / OFF optocoupler, because a 0-Ohm serial resistor is fitted on the control board. Without this external series resistor, the LED inside the optocoupler would
be destroyed when 12 V are connected. The left module is the slave (picture 4_1), the right module is the master (picture 4_2)! To get and ON/OFF indication on the
slave board, 2 LEDs can be soldered directly to the PCB (picture 4_1). On the master board connector pins can be used (picture 4_2).
5. the two control boards (master / slave) differ slightly: a 470 ohm pull-up resistor for setting the output voltage level by a resistor is not fitted on the slave control
board (picture 5_1). Add a 470 Ohm resistor to get same situation as on master PCB (picture 5_2).
To access this position on the slave control board, the complete right module including its EMI - filter PCB must be removed!
Remark: setting the output voltage with a control voltage has not been tested so far. Pay attention to keep the "grounds" separated!
6. connect the positive output of the left module with the negative output of the right module, ideally with a shunt resistor to allow easy current measurement (picture 6).
If the positive sense input of the left module is not connected to its positive output but to the negative output of the right module the voltage drop of the shunt
is automatically compensated!   
8. if ripple voltage is an item, connect additional capacitors to the output(s)!

Any questions?
 


 

[BUAROD]

... and finally a picture of max. load testing: 48 Volt with approx. 200 Amp!

Features added to make operation more comfortable and save:
- ON/OFF switch
- contact protection
- voltage adjustment (10 ... 55V)
- current limit adjustment
- voltage and current indicators