Snubber for capacitive loads

[ricko_uk]

Hi,
I need to charge a large capacitive-only load of 3,000uF at 250V from the mains using a triac (to switch at zero crossing). Do I need to use a snubber across the triac or are snubbers used only for inductive loads?

Thank you

[Circlotron]

If your locality has 240v mains then your 250V capacitor is likely going to charge to 339VDC. Let's see a schematic of what you are proposing to do.

[ricko_uk]

Thank you Circlotron

Please find schematic attached. The steps/sequences are as follows:

1) At zero crossing I enable the triac for a full cycle (or maybe even 2 full cycles if required) and charge all 4 caps so that across MAINS_GND and UV_SIDE_A there are almost 680V and the same between MAINS_GND and UV_SIDE_A.

2) Disable the triac to isolate the system from mains

3) Discharge the top pair of caps (i.e. across MAINS_GND and UV_SIDE_A) into a coil and wait for full discharge of those 2 caps (still working on the schematic of the discharge part - I am thinking of using a half bridge configuration)

4) Discharge the bottom pair of caps (i.e. across MAINS_GND and UV_SIDE_B) into a coil and wait for full discharge of those 2 caps

5) start the cycle again

Any suggestions or improvements?

Thank you for your feedback

[Circlotron]

First off, I would put the live side to the junction of the two diodes and the neutral to the junction of the two capacitors. The neutral that would now be at the capacitor junction is at nominally the same potential as mains ground, so grounding the lower capacitor as you have done is going to cause a bit of unwanted excitement.  Looking forward to your discharge cct. 

[TimNJ]

This is surely what was intended..just a schematic typo!

Voltage doubler:

[Circlotron]

SiC schottky is totally unnecessarily. Conventional diode with sufficient surge current rating would be fine.

[TimNJ]

SiC schottky is totally unnecessarily. Conventional diode with sufficient surge current rating would be fine.

SiC schottky is totally unnecessarily. Conventional diode with sufficient surge current rating would be fine.

Indeed. And, on that note, now's a good time to point out that SiC schottky generally are far less robust compared to traditional Si when it comes to forward surge current.

Here's a model of your circuit:



Notice over 130A peak with Rsource = 0.4ohm, and ESR of the cap = 75mOhm. IDWD10G120C5 has a repetitive surge rating of 40A, and a non-repetitive rating of 140A. In either case, very close to the limit. 

I don't know your average current requirement, but how about something like Vishay S8CM or Vishay VS-20ETS12-M3 if you need higher current?

Also, if you're going for some sort of smart control, maybe consider switching in some inrush current limiting as to not continually be asking for 130A peaks (per cap) from the AC line? I'm not sure what your charging time requirements are but perhaps there's a compromise. Pulse rated wirewound would be a good option. Maybe around 5 ohm per cap? You can bypass with a relay or semiconductor device after 1 or 2 cycles to speed up the charging rate, while keeping the initial peak down.

Just a thought.

[TimNJ]

Could be way off base on the inrush limiter idea. (How do these kinds of devices usually handle this?)

[ricko_uk]

Thank you Circlotron and Tim, much appreciated help, suggestions and simulation!

I updated schematic (see attached) with all the suggested mods: diode changed to a rectifying one, added current limiting resistor, swapped neutral and live. And added the discharge stage.

For the discharge stage I decided to keep very simple with just:
- 2 P-mosfets on top top side (as they can be switched on without any major timing requirements - they just connect the caps to the coil/igbt for subsequent discharge). Maybe I should use IGBTs for the top side too?
- and 2 IGBT for fast switching when discharging the caps into the coil

Ref the current limiting, my limit is the 13A plug fuse we have in the UK.

Few questions:
1) with ref. to the inrush current limiting. My limit is the 13A plug fuse we have in the UK. Ideally the system has to fire up to 20Hz so can only charge the caps with a maximum of 2 mains cycles. (SIDE NOTE: the resistor shown there is clearly not 1W - I just noticed that)

2) is the discharge stage a good way to implement it?

3) if instead of a resistor I use this inrush limiter like this one, would it be ok and would it accelerate the charging during after the initial inrush is limited? https://www.ametherm.com/datasheetspdf/SL220R712.pdf

4) is the electrolytic type and the one I chose a good choice for continuous full charge and full pulsed discharge (https://www.digikey.co.uk/product-detail/en/kemet/ALS80A162KF500/ALS80A162KF500-ND/6872250)?

5) With reference to your picture below, yes that's what is intended but cannot see any difference between that and my schematic? Could you please point it out?

6) P-Mosfets seem to be more expensive and much harder to find with those high voltages high current rating. is it a better idea to ise IGBTs instead of of the P-mosfets?

This is surely what was intended..just a schematic typo!

Voltage doubler:

Many thanks again for all help and suggestions!

[ricko_uk]

Hi,
could anybody provide feedback/help about the above questions?

Many thanks

[Circlotron]

The triac would be better to be in the live line rather than the neutral line.

The MAINS_GND net should all be connected together of course, but definitely not connected to the actual mains ground. Mains ground is at neutral potential so you would effectively have a short across C15 and C28.

What is the inductance of L1? Depending on it's value, you may have crazy high currents.

Where do you expect the energy from the discharged capacitors to go? Least worst thing that can happen is when the caps are at zero volts and the inductor is at maximum current, when you turn off the IGBT it will recharge the caps via the upper mosfet body diode to near the original voltage, minus losses.

What is the purpose of this circuit? Practical application or just an experiment to see what happens with stuff?

You are about to discover that power electronics is rarely easy or safe, hence the old adage  - "with power electronics you are always a microsecond away from misery."

[TimNJ]

If you need to fire at up to 20Hz, you’re really going to be slamming the mains very frequently trying to recharge the caps. You’ll need to check the I2T and repetitive surge ratings of applicable components.

Have you considered a PFC boost stage instead of just “passively” charging by switching on mains? This should keep the peak current down, inrush only when you turn it on. This would, in general, reduce stress on the

But even with that...

To me it seems like your power requirement is too high for normal mains. I’m not sure exactly the timing of your system, but let’s say the cycle is to charge and discharge 2*1500uF caps @ 350V each every 50ms (20Hz). Each cap stores 90J @ 350V. So if you fully deplete both, you need to supply 180J every 50ms. That’s 3600W! With a perfectly efficient converter, the average current is 15A. With losses, 17-19A average. And that’s just for one pair of caps.

You’d need to slow down your fire rate to keep the RMS current low enough to not blow a fuse or breaker.

But before getting to that...

This is a pretty dangerous sounding project with lottttsss of stored energy and technically quite difficult due to very high currents in pulse application. A few people have tried to DIY a welder on this forum, and although usually not insanely complicated circuits, very hard to do right.

——

To answer your question about the inrush limiter, those are nice but only provide any limiting when they are cold. If you’re repetitively drawing high current through the NTC, it will get hot and stay hot, and the resistance will practically go to 0.

Regarding capacitor type, if you need very high capacitance, you practically do not have another choice but to use electrolytic. You could check the temperature rise of the cap to see if it’s being ovetstressed due to self heating. Low ESR is best.

[OM222O]

Like others said, this seems like an odd project. Why would you need that much energy discharged into the coil so quickly? Depending on the inductwnce and parasitic resitances, the coil might take longer than 50ms to discharge. Also what does the coil interact with? Is it a motor winding near magnets? Is it hanging freely in the air? Part of an actuator of some sort?

Overall this seems like a high voltage H bridge across the coil, is that the goal? Because that means you can simplify things A LOT.

As it stands you need insane anount of cooling and power requirement. I suggest if this is the actual circuit you need, start with smaller caps for testing and then increase the values.

P.S: like others said switching the live with triac is the safer configuration.

[ricko_uk]

Thank you Tim, Circlotron, and Om2220 for all the feedback!:)

Answers to all your questions...:

Tim,
yes, I did think about PFC but only briefly because don't have much experience in power electronics and thought the solution in my schematic was the easiest implementation. But after all your feedbacks perhaps that's not the case.

After your reply I looked at ready made PFC modules but they are 300 USD to do what could be done (if I had the knowledge) for maybe 50 USD or less of components (I am referring just to the capacitor charging stage). This is the one that might be suitable while allowing me enough room to try experiments requiring more power (just in case): https://www.artesyn.com/assets/aif_-_pfc_series_11feb2015_a8e76167e7.pdf   
Do you think this is a better way to go about it?
Do you know any other parts/modules perhaps from other suppliers?

A (minor) issue there is that from the datasheet it is not clear what other electronics is needed because there is a large number of pins I have no idea what function they perform and the datasheet does not say - I have emailed them asking for more infos (pins like CMon, CShare, PVAux, PFWadj, LDenable, PFenable).

OM2220,
I looked at full bridge solution in the various suppliers (digikey etc) and this seems the one offering the highest peak current: https://www.st.com/resource/en/datasheet/pwd13f60.pdf

Do you know where/how I could find a higher current rating one? Are there perhaps solutions/components called something other than full bridge?

OM2220 and Circlotron
it just for general experimentation (various ideas like CDIs, discharge coils, spark gaps, engine ignition, etc). Just a bunch of things and ideas I had through the years and never got to try them.

Thank you all again!

[duak]

R22 is going to dissipate a fair bunch of power depending on how often the charge/discharge cycles occur.  Your caps will also get fairly warm.  Are they rated for this type of service?  Ie., pulse discharge or photoflash?

Crydom made a nice soft start relay setup using two modules years ago that I used in an off line power supply.  I see they have added other features and functions to their line: http://www.crydom.com/en/products/control-relays/mcs-series/  If you don't want to spring for one of these, you should be able to work up a similar function with your existing triac.

Will you be isolating the mains from this circuit with a transformer of some sort?  If not, you may end up with some surprising voltages here and there.

Bleeder resistors and hazardous voltage indicators on the big caps are a Darn Good idea.

[OM222O]

If it's for experimentation then where did the 20Hz switching frequency come from? Either relax that to lower cycles per second or use smaller caps (both of which reduce the power needed). There are half bridge FET drivers, I'm not sure how well they work for IGBTs though, I never used them like that, but you can use two half bridges to make an H bridge. Also adding 1 Meg resistors across the caps seems like a good idea. They won't interfere much with normal operation but discharge the caps when not in use.

Now to the simplification: why not use a bridge rectifier to get a high voltage DC input from mains, which is then smoothed by your high value capacitors, connected directly to an H bridge that then carries out the pulsing? This way a lot of the current comes directly from the mains, and the caps only compensate for when voltage is below peak. This eliminates a lot of the complexity (zero crossing detection, opto triacs, normal triacs, high inrush currents, pulse grade caps, etc etc). Is there specific reansons for which you need to isolate the caps before discharging them? They'll be at mains potential with a nasty shock regardless so isolation doesn't really add much safety.

[Circlotron]

OM2220 and Circlotron
it just for general experimentation (various ideas like CDIs, discharge coils, spark gaps, engine ignition, etc). Just a bunch of things and ideas I had through the years and never got to try them.

Thank you all again!

Ignitions??? Oh my goodness! Why didn’t you say so??? My favourite subject! I have made literally thousands. Let’s talk ignitions.   

[ricko_uk]

Thank you Circlotron, Duak and OM222O

Sorry for replying only now but took some time off.

Replies below:

Circlotron:
LOL, I started my electronics journey making ignitions, but that by now I need quite a bit of refreshing. Good that you are passionate about them, maybe we can have separate conversations about those when I start playing with them!!

OM222O:
Yes, it looks like I will have to do either of those, smaller caps and/or lower frequencies. I am putting together a spreadsheet to try to balance all the variables. I will share it here when done.

Now to the simplification: why not use a bridge rectifier to get a high voltage DC input from mains, which is then smoothed by your high value capacitors, connected directly to an H bridge that then carries out the pulsing? This way a lot of the current comes directly from the mains, and the caps only compensate for when voltage is below peak. This eliminates a lot of the complexity (zero crossing detection, opto triacs, normal triacs, high inrush currents, pulse grade caps, etc etc).

That's a good idea but:
1) it would half the voltage available...?
2) how can I make sure that when discharging I always get the same voltage level fed into the bridge?

Is there specific reansons for which you need to isolate the caps before discharging them? They'll be at mains potential with a nasty shock regardless so isolation doesn't really add much safety.

3) what do you mean by "isolate the caps before discharging them"? Do you mean the 2 diodes or the "charging" mosfet?
4) If you are referring to the "charging" mosfet than to avoid shorting the mains. Is that wrong in some way?


DUAK:

Are the caps rated for this type of service?  Ie., pulse discharge or photoflash?

5) I could not find any pulse-rated caps that size. Only ones I could find are electrolytic and those I found do not say anything about pulse rating. Are there any electrolytic rated for pulsing applications?

Will you be isolating the mains from this circuit with a transformer of some sort?  If not, you may end up with some surprising voltages here and there.

6) I assume you mean a 1:1 isolation transformer?
7) I wasn't planning to. Would Why would that cause "surprising voltages here and there" and how would the transformer help?

Many thanks again to you all!

[OM222O]

Why would it half the voltage? In the other case you have 1 cap charged to mains peak and one connected basically to GND, it'll be the same with just using an H bridge. If you want double the voltage, then you should use two bridge rectifiers and charge a set of caps to + mains peak and another set to - mains peak. If using 240vRMS, that's about ~700v. What the actual fuck would require that amount of power? You'd blow the coil up / melt it.
To ensure you always get the same voltage, you can use a peak detect circuit and switch on the H bridge at peak voltage.

I was refering to the triac and disconnecting the caps from mains. Seems pointless since there will be high voltages involved anyways. It is a good idea to have a switch for fault conditions, like short circuit, but those switches are called circuit breakers (ideally followed by GFID) not triacs.