Thread Rating:
  • 2 Vote(s) - 5 Average
  • 1
  • 2
  • 3
  • 4
  • 5
48V 5000W Oil cooled inverter build
(02-25-2019, 09:54 AM)completelycharged Wrote: Found out the transformers in the Victron are in parallel but one is apparently switched in as the load increases, also part of the toroid is left exposed to increase leakage to provide some additional inductance that then eliminates the need for an second in line inductor. learnt something new on that one...

I'd be interested in links to this info being a Victron owner! :-)
Running off solar, DIY & electronics fan :-)
This is the post,
The victron starts about 1/3 the way down with a lot of photo's of the unit as well.. the unit was a teardown of a failed unit, interesting read. There are several other threads that are also quite interesting...

The two toroidal transformers in the case.... the explanation of the switching arrangement I have no reason not to believe as this is a common approach for inverters with multiple transformers (switching extra transformers in as needed with increasing load). The aim is to save on no-load losses while also making construction a little simpler.

I like the transistor stuck onto the PCB as a hot fix with the resistors for the gate voltage, neatly lined up.... not quite sure this is what I'd put in a top tier product quality list. Everybody rants about Victron and that's great, however they are just like any other manufacturer in the end... where reputation can be damaged by thier own products quality or service feedback (or lack of) quite easily.

The interesting lossy (leakage inductance) transformer design to eliminate the need for an external inductor - like the E type transformers, but without the energy efficiency hit.

Forgot to add, with inverters that have switched in transformers they may have an overall maximum capacity of X kVA but if they have N trnasformers the surge power handling capability is reduced to

(Rated kVA / No Transformers) X Short overload rating Factor

i.e. (3000VA / 2 Transformers) x 3 = 4500VA

If the unit only had a single transformet then the surge handling capability is then 9000VA - but the larger transformer has more no-load losses as it is magnetised all of the time.

Some other inverters have 3 or more transformers, nice and efficient for random load handling, but not great for starting larger loads or running for long periods of time at high load on a lot of occasions.

i.e. A Single 3000VA transformer would be more efficient handling 3000VA than two 1500VA units..

Depends on what the useage of the inverter is intended to be, infrequent random loads like a washing machine or EV charging....
Korishan and Redpacket like this post
Wow, nice. Thanks for the lesson Smile

Could you have several transformers of varying sizes to handle ever increasing loads? Or do they need to be identical with each other? I suppose it depends on the circuitry in the end. I was thinking if the main transformer is designed for 1000W, then the second unit can hand 2000W, so together they could do theoretically 3000W. Or does it not work like that?
completelycharged likes this post
Proceed with caution. Knowledge is Power! Literally! Cool 
Knowledge is Power; Absolute Knowledge is Absolutely Shocking!
Certified 18650 Cell Reclamation Technician

Please come join in general chit-chat and randomness at Discord Chat (channels: general, 3d-printing, linux&coding, 18650, humor, ...)
(this chat is not directly affiliated with SecondLifeStorage; VALID email req'd)
You can have several transformers but I have not seen any with more than 3 switched stages as the economics and no-load efficency no longer works - unless they are high frequency and then you dont have the large transformers issues to deal with (energising surge)...
Design is a compromise of economics of costs, energy through efficiency and static losses.

Could be a case for a 3000VA + 1000VA + 500VA design BUT, which one do you switch on first and then the surge is limited to 3x this rating if you want to avoid a brownout. The second transformer to switch in also creates an energising surge that has to be dealt with on top of your load... 1,000VA unit active, 2000W kettle switches on... switch in 3000VA unit, switch out the 1000VA unit. Savings would probably add up to "two fifths of nothing" as the saying goes...

Now this is a "monster" of a DIY toroidal based inverter.... but only 2,000W (using 400VA toroidal transofrmers) and all the transformers are always energised so no-load losses will be interesting. Works...
I like the copper busbars...

Inside a 2000W GTIL unit... I was planning to supplement the output of the large 5kVA unit with (an / multiple) additional 2kVA grid tie inverters like this one and still need to confirm how well load handling of surges occurs. If I switch on a 1800W grid tie output when the load is at say 2000W then the transformer would back off to only 200W load level (leaving 4800VA available)... and then I switch on a 6000W EV charger for a "short" 4 hour charge..
Korishan likes this post
That thread made a very interesting read.
One thought: old school parts = can be repaired.
Turns out that the transformer design isn't "leaky" but uses a combination of careful filter design & good quality silicone steel laminations to get the low power draw.
They would have to have something serious wrong with the drive circuits to blow the FETs like some of the pictures.
I was also curious about how they may have been switching on the inverter & the ATS connecting it to grid mains while still running (loud hum comments) not going to be a good outcome if not a grid-tie design....
completelycharged likes this post
Running off solar, DIY & electronics fan :-)
The board they are using in the inverter is an off-grid only board, so if you have a feed-in tariff where your getting paid for any power generation then I would agree these are totally not right for the job. Apparently there is a grid-tied equivalent board (one to research).

This is a detailed paper explaining the lossy toroidal winding format used in those victrons
Figure 4 and Figure 5 are the good visual picture summary of the paper... saves reading it !
The axis on Figure 5 should be uH (as per the numbers in Table 1 on the same page) also confirmed inthe text..
Redpacket likes this post
Another great link. Learn something every day!
completelycharged likes this post
Running off solar, DIY & electronics fan :-)
After a bit of a pause for thought to re-check everything, finally ordered the transformer(s), two actually. Hopefully collect them next week, all 35kg worth...

Ordered a single 1kVA unit to initially use as a test transformer while finishing and checking the capacitors and inductors have the right ratings for the unit to avoid any resonance and harmonics appearing in the output. This smaller 1kVA unit will end up in a second inverter to power some other equipment separate to the house and should be good for surges upto around 2.5kW.

The larger unit that was ordered is now 6kVA rated (relatively small price increment and increase in core losses) so the oil cooling may be put on the shelf for a while.... this will have different capacitors to the 1kVA unit and the input chokes will also need to be slightly different. So, should hopefully be good to handle surges upto around 15kW and will have a prime test starting the 3hp compressor before daring to try the 280A MIG welder. The 12kW plasma cutter has a high crest factor load pattern so would blow the FET's so that is one load staying grid connected for now..... until I have a bag of spare FET's and control board parts.

They are both rated 240V / 30V and have the cores left unpotted (open center) with a few reason why.

Firstly any oil cooling used would benefit from the increased cooling surface area while air cooling also has more surface area

Secondly I wanted to have the option to add a winding or two (to the primary) if I needed to drop the 30V secondary if clipping of the output shows up under load when the battery is closer to a 46V charge state. Adding a winding or two to the primary (240V) would alter the turns ratio (increasing it) and therefore reduce the secondary (30V) level. Adding a winding (or more) to the primary also reduces the magnetising flux and core losses (very slightly).

The last option is to add a winding to the secondary (30V) to raise this level because if the battery pack is at a higher voltage level the PWM duty cycle is reduced because the current pulses are at a higher amp level with the larger voltage difference between the transformer and battery. The net energy is the same (baring I2R, etc.) but the full load impact on the FET's is higher stresses when the battery is fully charged. How high the pack voltage can go and still deliver a 15kW surge could become a critical consideration and a pre-emptive order for a bag of replacement FET's... lol.

Hopefully at the end of this will be a consistent list of "standard" off the shelf parts that can then build an inverter without the need for a scope, inductance meter, learning chinese, signal generator, variac, etc. Which will deliver 6kW all day and provide surge rating in the 15kW range (seconds, not the milli second FET vapourisation timing of over priced HF inverters).
Korishan likes this post
I am very interested in this project.
It would be fantastic to have a affordable (kit) inverter to build.
Keep up the good work!
Next up in the research list was the control boards as these are the real engine... or not..

Several boards appear to be in production, however not all may be what they seem,

Taking a look at the International Rectifier specification sheet (page 18) detailing the chip markings for example.

Some boards seem to have a different, or unique marking style.... could be a change in marking styles as the original Infineon spec is from 2005  but unchanged.

While others do not appear to have the logo, a little more suspect ?

And some appear to be the real deal.... we have a winner ?

While all of the boards appear to be the same basic layout the components are not, with the capacitors C4 and C12 taking on many forms for example (at least 3 variations).

The last issue with any of the EGS002 boards appears to be when your using a large toroidal transformers because  the shutdown process does not quite go according to plan and tends to result in magic smoke from all your FET's.

By removing the LM393 and shorting 7 and 8 it apparently solves this issue but it is not necessarily the complete solution and only occurs with large toroidal transformers.

The board I have ordered has an additional wire.... or randomly placed piece of heat shrink just for fun.

The function of this may be to allow for the correct shutdown with a toroidal unit without the removal of the LM393.

This is also not the only board to include the additional (unplanned) wire jumper as other inverter boards have this wire as well.

The smoothing capacitors used in the board I have ordered, may need an upgrade through as tests by someone else with a different 4FET board with 4 x 5600uF 80V capacitors can see them running at 40-50C with a 3kW load. This may in part be due to the battery type though and higher voltage sag under load.

The board has 7 x 1500uF 100V capacitors (10500uF) and replacing them with 5600uF 80V caps may help if your using a power source with a higher inline resistance (more voltage sag under load). If your battery supply has a very low resistance the size of the capacitors might be ok. One test I will be carrying out with the scope and a few kW load attached to the inverter.
Redpacket and Korishan like this post

Forum Jump:

Users browsing this thread: 1 Guest(s)