Llc hb smps 2000w

MyFirstSMPS

New member
I only ask and I see that this interest :)

This is example Epcos medium power application DC-link cap data:
https://www.tdk-electronics.tdk.com/inf/20/20/db/fc_2009/MKP_B32774_778.pdf

They are not always cylindrical shape, rectangle shaped is better to solder right for pcb. There are some much better types than this Epcos. I have even one commercial circuit diagram where input filter uses only this type filtering capacitor and if output power is mentioned 8kW then this filter capacitor must be only 320uF because (important part) they are much faster, high dV/dt, permissible currents are also higher, very low ESR (some milli-ohm) compared to elcos, long life compared to elcos.
I'm not a salesman though it may seems but I thought maybe someone had tried it and OK, I have halfway one experiment mockup at this moment and if this work I maybe speak more of it.
-)
 

res_smps

Member
This DC link capacitor works at a frequency of 100/120Hz (2x line frequency for bridge rectifier).
in my understanding what is needed is the amount of capacity for the ripple voltage that is allowed when the load is maximum.
I use this formula (attached)View attachment capacitance.zip or by rule of thumb 1uf per watt

that "faster" epcos capacitor more suitable for secondary capacitor
 

MyFirstSMPS

New member
This DC link capacitor works at a frequency of 100/120Hz (2x line frequency for bridge rectifier).
in my understanding what is needed is the amount of capacity for the ripple voltage that is allowed when the load is maximum.
I use this formula (attached)View attachment 6979 or by rule of thumb 1uf per watt

that "faster" epcos capacitor more suitable for secondary capacitor


Yes, this input filter part works just this way but why you miss important differences between electrolytic and film type capacitors?

Why you need this DC filter at all? To ensure this DC supply as smooth as possible, right? But which prevents it? If you need more power to convert and if you filter is not adequate then this DC begins to fluctuate and this may be end a catastrophic failure for converter...
Look closely electrolytic capacitors and this DC-link, film type capacitors datasheets, you see that their charging and recharging cycle speeds is very diferent - this input filtered DC fluctuation depends how fast you recharge your system...

And no, this is not Lab Gruppen, this is more different thing.
I have this circuit on a paper form.
Also if I can find it now then I can share some documenys which describe this DC-link film technology more closer.

https://www.kalbeck.com/asset/630/Whitepaper SalconeBond.pdf
- is also one technical paper, read page 4-5, I must say that this selection must be done according to circuit design and many other things. And this DC-link caps is not very good for output filters.
 
Last edited:

Silvio

Well-known member
If I understood correctly is that film capacitors (MKP) are good at high frequencies. In my case of the smps input voltage frequency is only 100Hz which is double the line frequency. From the examples I have seen the film capacitors are used where the frequency is much higher than 100Hz. I guess it would be a good choice here to use them.

In an input filter for a normal audio HB smps the frequency is no more than 120 Hz for a line frequency of 60Hz. So why is need to use them here. In my case the input current will be no more than 12-13A considering a PF of say 0.8 at maximum load of 2000w at the output.
Practically the current is lower than that stated as this only shows due to the voltage and the current not being in phase. In reality the power absorbed is less than what shows when calculating the voltage and current at the input.

When it comes to dividing the 320v for the half bridge then here is where to use them this time these will be working at the operating frequency. Sustaining the filtered input voltage with capacitance is where it all comes into play. In my case I have 1.35uF per watt (2700uF) (4 X 680uf X 450v) I will post some screen shots of the ripple current at various loads in a couple of days. Today I tested for the off load noise which I think it was very good having an RMS value of 12mV.

noise floor 2K.JPG
 

MyFirstSMPS

New member
No, they are good in both cases. Because it´s not my nature language (english) then it is hard to explain probably :(

OK, I´ll try once more:
If you rectified incoming grid voltage then you get direct pulsing current and you need to smooting which you say, used bulky electrolytic capacitors because if you converter work under heavy load you need more stable incoming direct power to its converter circuitry and during this voltage and current begins to fall on the filter capacitors, they will of course, recharge during this and so on and so on, but this of coare take some time (although, it is short time) and if you compared to that time with film capacitors then you see that they can react much faster than electrolytics. OK, I have nothing against elcos and I also use them but just tought that maybe someone had used film capacitors in filter stage. I worked couple of years ago electronic industry and this corporation produces high power converters, they used very much this dc-link film capacitors, some converters worked 500kHz and put out tens of kW power. This time DC-link capacitors were quite expensive and half a year ago I watched and they are now much cheaper and I thought that I must try them :)
 

Silvio

Well-known member
No, they are good in both cases. Because it´s not my nature language (english) then it is hard to explain probably :(

OK, I´ll try once more:
If you rectified incoming grid voltage then you get direct pulsing current and you need to smooting which you say, used bulky electrolytic capacitors because if you converter work under heavy load you need more stable incoming direct power to its converter circuitry and during this voltage and current begins to fall on the filter capacitors, they will of course, recharge during this and so on and so on, but this of coare take some time (although, it is short time) and if you compared to that time with film capacitors then you see that they can react much faster than electrolytics. OK, I have nothing against elcos and I also use them but just tought that maybe someone had used film capacitors in filter stage. I worked couple of years ago electronic industry and this corporation produces high power converters, they used very much this dc-link film capacitors, some converters worked 500kHz and put out tens of kW power. This time DC-link capacitors were quite expensive and half a year ago I watched and they are now much cheaper and I thought that I must try them :)

Oh ok I understood you now. Just one more question so for the same power in the case of an smps working at the grid voltage and the frequency is 100hz in the input. Do I need less capacitance using DC-Link film capacitors then using normal electrolytic capacitors?
 

res_smps

Member
apologize if I still have different opinions, I think if the industrial kilowatt power supply must have a Pfc boost converter frontend that works at high frequencies so it requires smaller capacitors
 

Silvio

Well-known member
apologize if I still have different opinions, I think if the industrial kilowatt power supply must have a Pfc boost converter frontend that works at high frequencies so it requires smaller capacitors

For your last comment I agree with you Res. Power factor is also an issue with conventional smps I guess a PFC will correct this as long as it is purely PFC and not a boost converter. Catering for two things at the same time is quite a task to achieve having minimal EMI and good ripple rejection on the output. As you may well know an smps has a high capacitve reactance and the PF is not so nice in the input especially on light load.
 

MyFirstSMPS

New member
Oh ok I understood you now. Just one more question so for the same power in the case of an smps working at the grid voltage and the frequency is 100hz in the input. Do I need less capacitance using DC-Link film capacitors then using normal electrolytic capacitors?


If I remember correctly then in one cases where half bridge converter frequency is 200kHz they used multiple parallel connected 40uF film capacitors, and one book explain that it is possible 20 times smaller capacitance than using instead electrolytic. I haven´t tried this technology at home but soon I will try this.
I have also some 40uF 1100V DC caps but I found that one Italian surplus store sell 200uF 1000V used but good capacitors ~15€ 4pcs (they are soldered in board which is taken probably from a wind farm inverter) but shipping to Estonia is expensive - 30eur :/
If you have interest to try then look here:
https://www.electronicsurplus.it/en...KNG1914-MKP-,-metallized-polypropylene-n-4pcs
 

Silvio

Well-known member
If I remember correctly then in one cases where half bridge converter frequency is 200kHz they used multiple parallel connected 40uF film capacitors, and one book explain that it is possible 20 times smaller capacitance than using instead electrolytic. I haven´t tried this technology at home but soon I will try this.
I have also some 40uF 1100V DC caps but I found that one Italian surplus store sell 200uF 1000V used but good capacitors ~15€ 4pcs (they are soldered in board which is taken probably from a wind farm inverter) but shipping to Estonia is expensive - 30eur :/
If you have interest to try then look here:
https://www.electronicsurplus.it/en...KNG1914-MKP-,-metallized-polypropylene-n-4pcs

Thank you very much for the information. I think I have something similar but mine are cylindrical shape. They where also from a switching power supply. I need to check what exactly they are. I have to look in my junk boxes and try to find them. I will take a photo of them and show them to you.
 

res_smps

Member
I tried to do a simulation
1. sinusoidal input 50Hz 300Vpeak, 300uf capacitor, 1A load current
Vpp ripple = 26.5V --> I think this proves the rule of thumb is 1uf per watt with 10% ripple voltage
2. sinusoidal input 50kHz 300Vpeak, 150uf capacitor, 1A load current
Vpp ripple = 50.8V

input with high frequency like pfc frontend
3. pulse/rectangular input 50kHz 300Vpeak, 300uf capacitor, 1A load current
Vpp ripple = 33.2mV
4. pulse/rectangular input 50kHz 300Vpeak, 50uf capacitor, 1A load current
Vpp ripple = 196mV

maybe this is the ideal condition, in fact there is series resistance from the capacitor and other parameters that might affect the result

sim1-2.jpgsim3.jpgsim4.jpg
 
Last edited:

Silvio

Well-known member
I guess using DC link caps will be suitable where high frequency is present. They seem more reliable than electrolytic but as res proved the amount of capacitance is always needed. There size is also a disadvantage. They are large compared to electrolytic capacitors for a given capacity. We will wait till our friend tries them out and see the outcome.
 

MyFirstSMPS

New member
Hm, is a simulation and is a "real world".
Res, probably you get this DC-link film capacitors in Indonesia much easier and try them in real application also?

Yea, if you compare 20uF DC link film capacitor and example 180uF 400V electrolytic - this film cap is larger, also, 10 times more expensive but how long they lasts compared to elcos?
20uF/900VDC Kemet C4AE (which I found the cheapest ~5€/pcs), ESR is down to 4 milli-ohm, Ripple current is 15A/10kHz.

and 381LQ181M400H032 Cornell Dubilier 180uF/400V electrolytic which may compare ESR is 0.458 ohm/20kHz, Ripple Current is 1.33A/20kHz, lot of parameters for film capacitors is relative flat to several hundred kHz.
I dont say like electrolytics is bad (I use them myself) but I was just wondering if somebody suddenly used a newer solution in amateur construction, yes there maybe is some thing to overcome but - several heads are still several heads :eek:

And even because if their using gives saving in the long term then they probably won't be seen in consumer electronics for a long time because their price.

OK - why do I hope them? Silvio, you see that 2kW SMPS filter capacitors is pretty large, yes, if you want to try to make it thinner then you can use lot of smaller filter capacitors (elcos) but they also need more space and pcb development is also somewhat more complicated to ensure their traces and placement is good. If you want to place this 2kW smps example 1U thick case, then you will see some problems. OK, there are some low profile elcos which I have 560uF/400V 35mm high but they are in the same price range than 40uF-50uF film caps and they dont seems last very long.
OK, if I get all components then I`ll try them.
 

res_smps

Member
Hm, is a simulation and is a "real world".
Res, probably you get this DC-link film capacitors in Indonesia much easier and try them in real application also?

Yea, if you compare 20uF DC link film capacitor and example 180uF 400V electrolytic - this film cap is larger, also, 10 times more expensive but how long they lasts compared to elcos?
20uF/900VDC Kemet C4AE (which I found the cheapest ~5€/pcs), ESR is down to 4 milli-ohm, Ripple current is 15A/10kHz.

and 381LQ181M400H032 Cornell Dubilier 180uF/400V electrolytic which may compare ESR is 0.458 ohm/20kHz, Ripple Current is 1.33A/20kHz, lot of parameters for film capacitors is relative flat to several hundred kHz.
I dont say like electrolytics is bad (I use them myself) but I was just wondering if somebody suddenly used a newer solution in amateur construction, yes there maybe is some thing to overcome but - several heads are still several heads :eek:

And even because if their using gives saving in the long term then they probably won't be seen in consumer electronics for a long time because their price.

OK - why do I hope them? Silvio, you see that 2kW SMPS filter capacitors is pretty large, yes, if you want to try to make it thinner then you can use lot of smaller filter capacitors (elcos) but they also need more space and pcb development is also somewhat more complicated to ensure their traces and placement is good. If you want to place this 2kW smps example 1U thick case, then you will see some problems. OK, there are some low profile elcos which I have 560uF/400V 35mm high but they are in the same price range than 40uF-50uF film caps and they dont seems last very long.
OK, if I get all components then I`ll try them.

Dear MyFirstSMPS,

sorry I didn't find the MKP capacitors with a large enough value with a large volt, there are only audio boutique capacitors sold here

in my opinion the capacitor in the simulation is an ideal capacitor that can work at all frequencies

my analysis in the formula below
vripple = I / (2 * f * C)
C = I / (vripple * 2 * f)
I see the magnitude of C is only determined by I (load discharging current) and frequency, while charging frequency f is only around 50 / 60Hz
with higher frequency we can reduce required capacitance

I appreciate if you will try it, it will add to our knowledge in the DIY world
 

MyFirstSMPS

New member
OK, I believed there were quite a lot manufacturing companies in Indonesia, so maybe it's easier to get them there but seems like I am wrong. They are quite expensive when you buy them from some electronics store - Mouser, Digikey and so on, Farnell maybe but only this Kemet which I mention previously is cheaper one, but if you want buy, say 50-75uF and so on, they easily are expensive, like 20 and over EUR/USD.
One Panasonic example, this is 50uF https://eu.mouser.com/ProductDetail...GAEpiMZZMv1cc3ydrPrF7yRxWMKYHonRMG0LdR%2BEXg=
and 75uF/650VDC Kemet https://eu.mouser.com/ProductDetail...c3ydrPrF95ZAHiPXMKgkNKWrF%2BHuecRZE9uqEyFQg== is also cheaper but they seems older models.


Right, that 100% of course obviously it is not possible to dispense with the use of an electrolytic capacitors because this Film type due to this relatively low capacitance value and because high reactance/resistance factor this foil capacitors may easily resonant, this is what I fear can be an disadvantage but I dont know this right now.
If I think then all this converters what I've seen and what we made is 3 phase - this is also advantage which needs smaller filter capacitance and sure, they are complex PFC system - I do not want to imagine if multi-kW converter switched to grid without a PFC :D

So far, I have not used PFC in my experiments which have been simler design - TL494, IR2153/IR2156 etc. One IR2153 design is over 1kW and without a PFC I think is not a good solution but there is one bad thing because in my region there is no low volume PFC magnetics available with hobbyists :(
I have high power devices (FETs, IGBTs) and power magnetics (ETD, EE, EI, PQ, even planar) components but no PFC comparable material..
 

res_smps

Member
here we can get smps part cheaply.
we have online shop website like www.tokopedia.com and www.bukalapak.com
I can get ferrite core at a very cheap price, for example eer4220 is only 1 usd, 1000u/200V elcos just 0.5usd, recycled pfc core just 1 usd
but i didn't find this dc link capacitor. I only found capacitors like this
https://www.tokopedia.com/naokielec...=1_ob=4_q=kapasitor+panasonic_po=23_catid=577
https://www.tokopedia.com/dutairama...=1_ob=4_q=kapasitor+panasonic_po=32_catid=577
which are commonly used for air conditioners
 

res_smps

Member

the problem is shipping costs to Indonesia are too expensive.
if not mistaken a few years ago I had been shopping at a mouser with a shipping cost of 50USD, from digikey 75USD, not including tax.

I also wonder why the components are also taxed, making (home) industries difficult to develop, on the other hand, Chinese goods flood the market at very cheap prices. finally, a lot of DIYers uses original recycled spareparts
 
Last edited:

MyFirstSMPS

New member
This two seems like Power Factor Correction capacitors but

one is here: https://www.tokopedia.com/marconiel..._page=1_ob=23_q=capasitor+MKP_po=42_catid=577

and datasheet you get here:
https://eu.mouser.com/ds/2/212/F3303_C4A-540888.pdf


And yes, you are completely right this shipping prices :(
This is expensive even in Europe like I mention one previos post - If I want to buy components from Italy then this component maybe costs 10 euros and shipping is 30 euros and if you add taxes which is here 20% then all this goes very easily over the head.
 
Last edited:
Top