Output inductor design for 6 pulse scr rectifier

[falcosms]

Hi there.
Can anybody tell me what is the calculation of an output inductor in scr full wave 6 pulse rectfier?
I need to design a LC filter for a rectifer three phases with output of 135 Vdc @ 100 A, maximum ripple of 1 % at full load.
I now that when we have a full bridge three phase rectifier with diodes, the output ripple is 5%, but it is increased in scr bridge because of angle variations.
Sincerelly
Marcelo

 

[BradtheRad]

The filter is exposed to 6x nominal frequency. If 60 Hz is going to each phase, then your filter carries 360 Hz.
Your load is in the area of 1 ohm. Raw ripple amount is 5%.

Here's a crude estimate. Say you wish to reduce ripple to 1/10 th of the raw figure. Calculate an inductor value that offers 10 ohms inductive impedance at 360 Hz. The standard formula is:
2 π f L

--- Updated Oct 6, 2020 ---

As another approach, say you want 1% ripple at the load. This implies there's equal AC voltage drop at the inductor. That is, 1.35 VAC. Find the correct Henry value that does this carrying 100A @ 360 Hz.

--- Updated Oct 6, 2020 ---

My simulation indicates a mode is reached at heavy loads, where waveforms become a frequency of 720 Hz. You may find the same thing happening in your project.

 

[Easy peasy]

Step 1: make sure the LC res freq is well below 300 or 360 Hz ( else you will cook the L & C ).

Step 2: you generally want to minimise the L size and compensate with larger electrolytics as the caps are cheaper than a large L and you will need some pretty large electrolytics anyway to handle the ripple current.

Step 3: the exact calculation depends on the nominal firing angle for your 135VDC and 100A, which is based on the nominal transformer output, but let us suppose we allow for 10% rms volt ripple out of the SCR's and we want < 0.5% volt ripple ( of 135VDC ) on the output caps at full load - the calculations are pretty simple , we need to go from 13.5Vrms ripple to 675mV rms ripple with 100A in the L.

Step 4: Let us choose an Fo of 20Hz for the filter, i.e. 20.2.pi = 1/( sqrt (L.C) ), or 15,790 = 1/ (L.C)

Step 5: At 360Hz rectifier ripple we need Xc / ( Xc + X L ) = 0.675 / 13.5 = 0.05, i.e. a 20th, Xc = 1/2.pi.F.C X L = 2.pi.F.L

Step 5a: using simple algebra we can solve for L and C using the information in 4 & 5 above

Step 6: more empirically we choose a small-ish value for L, say 150uH ( 100ADC remember )now X L = 0.339 ohms ( 360 Hz )

we need Xc to be 1/20th of this or 16.9 milli ohm approx, this is 26,000uF for C, this will give the required attenuation.

Step 7: Fo for the LC pair is now 1/(2.pi.sqrt(LC)) = 80 Hz, which is a bit high, to get to 20Hz say, C needs to be 420,000uF

For the filter to work the ESR of the electro's needs to be well below the Xc ( 16.9 milli-ohm ) - so you may well need up to 9 x 47,000uF at 200V to get near this figure, else the ESR will be the main ripple voltage constituent.

Step 8: If we assume all the ripple volts are now dropped across the L ( as the C is large and low ESR ) the ripple current in the choke is Vac/X L = 13.5 / 0.339 = 40 amps, all of this ripple will be in the caps - so if we have 9 x 47,000uF they each need to be able to handle 40/9 = 4.5 amps @ 360Hz continuously - this is the major design criteria for the electro's.

looking at a typical cap from element 14, we see 200V 2200uF = 1.98amps and 200v 10,000uF screw terminal at 19.9A 18 milli-ohms

So 2 x Vishay 101/102 PHR-ST 10,000uF 200V gives 40 amp ripple current at 9 milli-ohms total

the combined volts from the Xc & ESR = SQRT (884mV^2 + 360mV^2 ) = 954mV - a bit higher than the 675mV we were hoping for - but never mind,

the overall attenuation is ~ 0.954 / 13.5 = 0.0707 ( -23 dB ) and 954mV / 135VDC = 0.7% ripple rms is inside the spec you wanted.

the 10000uF caps will dissipate 7.2 watts each , for 2 only - which is getting on the warm side ( so 4 might be better 1.8 watts each )

[ 4 x 10,000uF will give lower ripple, and a filter res freq of 65Hz, still a bit on the high side - but live-able ]

Happy designing...

 

[falcosms]

Hi there!

Easy peasy and BradtheRad, nice explanations! Thank you.

I will use this tips!

Sincerelly
Marcelo

 

[ardavan5]

dear easy peasy :
I am a little confused of your design , please explain more or introduce me any reference on scr rectifier output filter design .
thanks .