Inrush into 4.3mF capacitor bank

[treezleef]

Hi,
We have a 4kW PFC we are doing. (100-265VAC, +/-180V)
The input is a Boost PFC...it has 4.3mF of output capacitance.
We want to inrush into it through a resistor, then short out the resistor with a relay.
We want the resistor to be 20 Ohms. Whats the smallest 20r resistor we can use for this.?
We dont want an NTC as they dont work when hot.

We are finding that resistor datasheets give Peak repetitive power pulse ratings, and continuous power ratings, but absolutely nothing which pertains to one-off inrush events.

 

[pc1966]

We dont want an NTC as they dont work when hot.

You put a relay (SSR or mechanical) to short the NTC out once supply charge is established, then it cools and is ready for immediate re-use.

 

[treezleef]

Thanks, but the NTC will be getting hot during the inrush event , and wont present the full 20 Ohms throughout the inrushing time.

 

[pc1966]

Thanks, but the NTC will be getting hot during the inrush event , and wont present the full 20 Ohms throughout the inrushing time.

That is kind of their whole point! As the bank charges you need less R to limit the current.

Any limiter is going to get hot, at least NTC thermistors are rated for it and to a large degree self-protecting. You can put two lower R ones in series for a bit more energy handling as well.

A zero-crossing SSR is a good choice for shorting out, less dV/dt when it kicks in, and ideally switched from the AC side of the rectifier, maybe with short delay, so a charged bank is not bypassing the NTC for medium term interruptions.

 

[treezleef]

OK thanks, thats v good info...the thing is, we want a solid 20R throughout the inrush.....NTC's very quickly loose R and inrush rises more than we want......its impossible to calculate exactly what R an NTC will be giving at what stage of the inrush.....so we just want to back out of NTCs

 

[marconi]

This may help you:

https://www.vishay.com/docs/48516/_ms9702509-2003-vishaychecklistpulseload.pdf

As well as Vishay you could contact this UK supplier too - Arcol:

https://static.rapidonline.com/pdf/62-8192.pdf

For your application you would require Arcol‘s NHS resistors which are non-inductive.

Note Arcol resistors may need a heat sink.

 

[pc1966]

A search for using the PFC inductance, etc, for inrush current limiting comes up with various options. This is an interesting example where the input bridge is dual-use and they suggest using thyristors to progressive charge the reservoir capacitors:

https://www.st.com/content/dam/is20/document/PE4-2_Erin_Wang_3-6kW_Totem-Pole_PFC_with_ICL_Solution_V02_EN.pdf

However, complex. Unless already a design example it might be a bit much to just poke around to get working. Also you need to take more care on input EMC filtering here, and probably also bleed resistors for input X capacitor as it might be significant size.

 

[pc1966]

Seems you can buy a demo board for $618 + tax/postage/etc

STEVAL-DPSTPFC1 - STMicroelectronics

STEVAL-DPSTPFC1 - 3.6 kW Totem Pole PFC with inrush current limiter reference design using TN3050H-12WY and SCTW35N65G2V, STEVAL-DPSTPFC1, STMicroelectronics

www.st.com

 

[Lucien Nunes]

I don't think there can be a hard-and-fast general answer. Thermal impulse characteristics vary widely according to the thermal mass, thermal resistance, maximum temperature and maximum temperature gradient of the components of the resistor. This will require detailed information from the manufacturer.

Don't forget to allow for what happens if the relay fails to close! In a xenon lamp rectifier (3kW SMPSU) made in Japan in about 1980, that I repaired last year, there was a dry joint on the PCB in the trigger circuit for the bypass triac - the inrush limiting resistor had gone quietly open-circuit. OTOH on a particular crane power supply if the bypass relay failed the resistor used to explode or catch fire.

Side note; I can never get on with the notation mF. I think the reason is that historically, MFD was used for μF on components and the lazy would sometimes use MF. Illogical though it might be in theory, I would always write 1,000μF for 1mF.

 

[marconi]

I state first that audio amplifiers and their design is not something I am that familiar with. And I stopped playing the piano as a teenager so music is only something I listen to and generally classical music not thumping bass playing or drumming.

The switch on surge current problem is because of the requirement to have capacitors for storage and smoothing. Have you thought about using a battery in their stead? The mains supply is used in this idea only to charge the battery and the battery provides the high power capability in terms of peak power and average power - it deals with the crest factor of the bass playing. The charger provides a current to maintain the battery in a sufficiently charged state for the duration of the performance and is a nice steady waveform which can be slowly ramped up from zero at switch on - thus no difficult switch on surge current. One can also continue to play during a power cut!

Just a thought.

 

[marconi]

I knew the academic definition of crest factor but wanted something which applied it to music and provided some numbers which might aid the design team. Here is a piece I researched today:

https://www.mc2-audio.co.uk/wp-content/uploads/mc2_crest_factor_and_amp_power.pdf

In it it says:

When specifying a system -(DPG's earlier point) , it is important to not only know the dynamic performance and limits of the loudspeakers and the amplifiers, but also the expected signal type to be used. Please bear in mind that any signal source that has a lot of constant tones (CF between 3.0 and 1.5) will require twice as much available power than a typically ‘heavy’ dynamic music signal (CF between 3.0 and 5.0), and about ten times the power of a system mainly used for speech and ‘light’ music.

Thus it seems to me that one would need to determine the range of crest factor of the bass music to be amplified before designing the amplifier. This may require different selectable modes by the power supply for classical, pop, heavy metal, jazz.....

 

[Lucien Nunes]

And therein lies the difference between a professional-grade amplifier where you don't have to think about these parameters during use - you just use it - and a 'DJ-grade' amplifier that can survive and deliver power within a specified envelope but lets out the magic smoke when you make the slightest move outside of that envelope. With bipolar output transistors there was typically a headroom limitation imposed by the SOA of the output devices. With MOSFET it's more a PSU and overall thermal limit. Having become accustomed to MOSFET in the late 1980s, I recall our attempts to destroy Amcron Macro-Tech 2401s and 5000s (bipolar BTL output with flying rails) at university, with some very loug gigs. We failed, because the Amcron was a well-engineered, heavy duty amp.

 

[marconi]

So that Lucien and I educate - SOA - is safe operating area. Here is a graph for a transistor. So when you read a 'rating' for component there is more to say to fully describe it and even this diagram does not completely describe its rating. When selecting components one wants to be some margin away from the boundaries so that the device is in its comfort zone. I am old school that does not believe in operating things at their limits.

It is not wrong to minimise the margin but it has consequences to bear in mind eg reduced life but it produces a more economical product - perhaps.