Why Sendust torroids used in output of 2.4kW SMPS?

[cupoftea]

Hi,
Supposing you had a 24V, 100A isolated PSU. 3-phase input via 3 phase rectifier. (No neutral to rectifier). Post rectifier capacitance of 24uF (MAB MKP). Being switched by a fairly simple looking high frequency converter, with 4 primary side switching (1200V) IGBT’s (IXER35N120D1). Some kind of transformer of size approx. 12cmx7cmx6cm (can’t see it, its in plastic enclosure). The transformer is off the IGBT PCB, and is wired to it. The transformer’s isolated output goes a to bus bar pair on its way to the output capacitor bank (approx 20 x 35V 1200uF El caps) . As the bus bar pair makes it way to the output capacitor bank from the transformer, (quite a way, about 20cm) one of the bus bars has seven CS467090 sendust rings round it.

The other bus bar goes through a LEM LA306-S Current monitor.

CS467090 sendust rings
https://www.alibaba.com/product-det...-Transformers-CS467090-77442_62351494118.html

CS467090 sendust rings are Fe-Si-Al material. This surely does not have the high frequency properties necessary for use in this place? Wouldn’t you want NiZn rings? Why are the Sendust ones used?

I am guessing the switching frequency is low. -We think this because the primary side wiring off to the transformer from the IGBT PCB is just two “loose” wires, which form a much wider loop than they would if they were twisted pair. Also, the primary side IGBT and control PCB is all Thru hole. The PWM controller is a UC2845B. (I am guessing the output diodes of the transformer are inside the plastic enclosure which contains the transformer, i cant see them anywhere else)

The only feedback seen from isolated secondary to the primary is from a 7 way ribbon cable which runs about 40cm back to the primary side from the secondary...presumably the isolation of it is by two V0943B opto's which i can see on the primary side IGBT board.

 

[cupoftea]

One of the busbars on the secondary side, coming out of the transformer, has a break in it...and it goes through a PCB for a bit.....this PCB comprises what looks like 70 or so 1206 "resistors" with gold terminals. I presume these are paralleled PTC's, (or resettable fuses?) and offer a good way of fusing the output? The "resistors" have some number on them, something like 0965. There are also some 0805 MLCC caps on this PCB. There is no visiable output inductor anywhere, and i suspect that this "function" is provdided by leakage inductance in the transformer. ...either that, or its inside the plastic enclosure that houses the transformer, along with the output diodes. I wouldnt be surprised if it was a current doubler rectifier output.

 

[Easy peasy]

You need to sharpen your eyes - the resistors are snubber R's for the diodes you also failed to see/mention

Having a large output choke = very low ripple current - hence low Rac losses in the copper and low delta-B losses in the material - also sendust is cheap and moderately effective - a plus is that the L goes down with current on a slope - unlike gapped ferrite which has a much sharper drop off ...

--- Updated Dec 4, 2021 ---

Sendust and Kool-mu are very similar materials .... the 7 toroids whould make a reasonable filter inductance - given that vout ac = 2 x the sw freq ...

 

[cupoftea]

Thanks, -am now trying to find the inductance of 7 stacked CS467090 sendust rings with a wire going through it carrying 10A/50A/100A.

Ill use this core as the "equivalent" for now, as its got a proper datasheet

https://www.mag-inc.com/Media/Magnetics/Datasheets/0076617A7.pdf

At 10A:
N = 1
Inductance = N^2/Reluctance

Reluc = length/(u0.ur.A) = 88.4E6 ________{l=144e-3, A(total) = 0.00216, ur=60}

Therefore, inductance = 1.13uH. (not greatly accurate but this is definetely a max figure, ie, it couldnt be any more than this)....(maybe the CS467090 has much higher ur than the 60 of the MagInc 0076617A7 torroid that i depicted though?)
.....This is very low inductance, and is telling us that these 7 sendust rings are not the output inductor of the Full Bridge (or whatever it is) SMPS stage here. As discussed, it must be very low fsw because of IGBTs and very "cost-effective" wiring to the transformer.

 

[FvM]

According to datasheet, Al is 202 nH. Expect less than 10% inductance drop at 100 A.

 

[cupoftea]

Thanks FvM for the torroid datasheet. I recalculated the inductance of the seven torroids on the bus bar...it now comes to 1.47uH maximum. This isnt enough for an output inductor for this converter. Also, as we discussed, sendust isnt much good with 100's of MHz, so it can't be to reduce radiated emissions either. So i wonder what these torroids are for?

 

[FvM]

Output filter cutoff frequency below 1 kHz, doesn't sound completely wrong.

 

[cupoftea]

Output filter cutoff frequency below 1 kHz, doesn't sound completely wrong.

Thanks, but this "7 torroid inductor" (7 adjacent torroids skewered on the bus bar, rather like a "shish kebab" type of thing) is upstream of the 35V output capacitor bank, its not part of an output filter.

 

[FvM]

Output filter = filters rectified AC to clean DC

 

[cupoftea]

Output filter = filters rectified AC to clean DC

Thanks. This converter is going to be very low frequency fsw....so 1.47uH sounds a little low for the output filter

 

[Easy peasy]

very hard to make any guesses when one can't see the power ckt ...

 

[FvM]

Also need to know the operation frequency.

 

[cupoftea]

Hi,
Tried to get some more time on this...
I removed the transformer from this PSU today. It was using 4 E core halves. (so two closed cores, side by side) Each E core half was of dims 55mm length, 25mm thickness, and 28mm "height". The centre leg was "cuboid" (not rounded), and of width 17mm. The "side legs" are 9mm thick. The bobbin was not one of the standard ETD, or PQ things. The Ferrites of the transformer are very closely surrounded by an aluminium covering. This is about 0.7mm thick. The one side of the ferrite is very close to the heatsink, which again is right next to that ferrite "face".
BTW, the primary capacitance is made up of two 12uF 400VAC caps (MAB MKP) in series.

There are two paralleled IGBTs switching from the upper cap, and two paralleled IGBTs switching from the lower cap. So i guess this is a Half Bridge converter. (There are no resonant capacitors to be seen, (but see later) and the driver IC is a simple UC2845). Not had chance yet to measure the leakage inductance.

The Primary side driver board has more than 2 layers. I am not sure how the UC2845B is managing to drive both lower and upper FETs....i cant see any circuitry for providing the alternate Hi_side/Lo_side gate drive signals.

As far as IGBT gate drive isolation goes, i can see two orange modules about the size of an ice cube. (~2.5cm by 2cm 2cm). One is slightly bigger than the other though, which is odd, as you'd think they'd use the same part twice) Each has 4 pins. I assume these are gate drive transformers. (but see later)
There are 2 DIP8 chips with marking..
"3120
V0943B"
...and also a character which looks kind of like an "F". Guess to be Fairchildsemi.
..I assume these are gate drivers for high and low side.

I cannot see any primary side current limiting at all....but since the two orange modules described above are slightly different size, i suspect one of these is a current sense transformer.

The UC2845B timing components appear to be 8k2 and 1nF. (This would give 100khz=fsw and Dmax = 0.477). The only other capacitor near to pins 1-4 of UC2845B looks like a mini film capacitor, and so i doubt its a timing component. This mini film cap has "u1k63" written on its top bit. I think that means 63V and 1uF(?). This film cap is 6mm x 4mm x 1.5mm. I can't yet identify an RC filter for the current sense pin of UC2845B.

The lack of an appreciable output inductor though, makes me think this may somehow be an asymetric half bridge. However, i cannot identify a resonant capcaitor yet.....unless its that second orange module roughly ice-cube size, but then if its that, then where is the primary current sensing.
The output diodes are 3 ISOTOP packages by st.com. So thats six 80A diodes all told (3 for each of the split coils). The primary is a single coil. Secondary is split.

 

[Easy peasy]

All sounds pretty standard - what id the o/p capacitance ? p.s. u1k63 = 100nF 63V

 

[cupoftea]

Thanks, C(out) is approx 40 pieces of 1mF, 35V El caps.
It does seem like a standard half bridge, but hard switched at 2.4kW, and from a HV 3 phase rectified input (no neutral). And 100khz. And the unusually low inductance looking output inductor....raises more questions.

 

[Easy peasy]

100kHz on the osc = 50kHz actual sw freq ..... 100kHz on the rectified o/p.

that many caps can absorb a fair bit of ripple current ...

 

[cupoftea]

Thanks, but Sorry, there's two supplys in one case...i meant to say 20 pieces of 1mF 35V el caps, not 40.
Must admit i havent got into this one properly yet....i must be wrong about the half bridge....theres nothing to drive the upper and lower fets with.....it must be a two transistor forward.....and relying on the leakage inductor, because theres no visible output inductor.....but i havent yet found the reset diodes that a 2TF would have.

 

[cupoftea]

Tried to get some more info on this "mystery converter" today.

I now think that the attached (2 tran forward) may be what it is?....have you seen anything like this before?.....its either the attached, or a half bridge....but as discussed, in the minmal time i had i wasnt able to find the flipflop which the half bridge would have needed to convert the UC2845 gate drive to half bridge style.

So anyway...this is it....or possibly.............the output 1uH is by way of 7 or 8 torroids round the busbar.

BTW i now know the proper spec.....its 12V, 300A....there are two of these in a unit and they can be paralleled. But they are variable voltage and can be any vout from something liek 5V to 24v, as long as pout is no more than 7200W.

 

[FvM]

Nothing mysterious, two-switch forward is pretty common for higher power converters, e.g. used by most simple arc welders.

 

[cupoftea]

Thanks, but what is shown in post #18 is a "precision" 2TF...change the leakage from k = 0.997 to k = 0.996 and it drops out......change the output inductor to much below 1uH and again it drops out....thats a bit of a nuisance when that output inductor is seven torroids round the busbar....a bit "how ya doin'" by anyone's measure....

I wonder how many winders can do a transformer where all parts always have k = 0.997 rather than k = 0.996?

Also Hard switched at 3.6kW..........and it works...even with the HV input voltage. Driver board is all PTH.....no nice tight SMD layout here. Transformer is wired to the driver board by a pair of "any old how" wires....not even twisted. But still works fine.

You need to sharpen your eyes - the resistors are snubber R's for the diodes you also failed to see/mention

Thanks, you were bang on with this, are such PCB snubbers common at high power?......with what seems like the world's allocation of chip resistors and capacitors.