MMC Capacitor: internal wiring and voltage distribution issues / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

MMC Capacitor: internal wiring and voltage distribution issues

Move Thread

LAN_403

gargamel

Tue Aug 11 2015, 09:09PM

Registered Member #56093 Joined: Wed Jul 29 2015, 04:57PM
Location:
Posts: 8

Hi there & thanks for enabling my account.

I'm building an MMC Cap for a little project.
My application demands the cap to be loaded to a voltage as high as the cap allows, kept there for a few seconds and then discharged.

To achieve the necessary electric strength and capacitance, I'll have to use several individual caps in series and in parallel.

Since cost is an issue, I'd like to utilise the full individual caps electric strength rating, as far as possible...

Now there are two things I'm not certain about.
First, there are two ways to wire them, see my attachments A and B.

Which one would you prefer?
My feeling tells me that variant B would be easier to service with respect to voltage distribution.
However, most MMC contructions I saw on the web are built like variant A.

And second, balancing resistors. Is there some rule of thumb for dimensioning them?
I'll need to rely on an even balance. Is that even practical?

My caps are of MKP type with a tolerance of 10%.

Dr. Slack

Tue Aug 11 2015, 10:20PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Configuration

The individual series strings is easier to make, but will need more resistors. If you try to do a 'how many am I going to fry in the event of this type of fault?' exercise, they behave differently to different faults (open, short, high leakage etc), I haven't yet come down strongly in favour of either.

Balancing

As ever, specifications. Do you want 10% balance, 1%, 50%? More power dissipated in your balancing resistors will control the intermediate voltages better. There is little point in aiming at a better long term, leakage controlled, spec, than the dynamic short term capacitance balance spec can deliver. If the caps have 10% tolerance, then the voltage distribution will be at worst 10% after the initial charge. Now, what's the worst case leakage current? It's not what you measure on the bench on a new cap at room temp. Is the cap well specified enough to give a worst case leakage current? If you assume the best case is zero current, then design for an imbalance current of worst case to zero, and balancing resistors to control the long term voltage to 10%. Is the calculated power OK? If so, job done, if not, revise expectations and safety margins.

gargamel

Wed Aug 12 2015, 11:11AM

Registered Member #56093 Joined: Wed Jul 29 2015, 04:57PM
Location:
Posts: 8

Thank you for your answer.

The caps are of some older type, NOS. I could not find any data sheet.

My idea is to use the second schematic and match the caps into equivalent parallel groups, using a capacitance meter.
If that works I might get away with higher resistance.
The balancing resistances should also function as a safety bleeder, giving a time constant of maybe 10s or so. I'd be happy if this is sufficient for balancing...

Since I do not have a capacitance tester, but want one for a long time, I wonder if these cheapo devices in the 15-25$ range are of any use?

DerAlbi

Thu Aug 13 2015, 08:47PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

imho you want the second version with the 3p4s configurations. you can measure the individual caps so that the praralleled groups match closely.

Discharging is no problem in such configurations. Only ballancing while charging and keeping the voltage can be hard.
If you use the proper techique for charging like a flyback charger with individual turns for each paralleled group, then you dont even need a ballancing network. A flyback automatically pumps the energy in the most discharged capacitor and the turn ratio (1:1:1:1) of the output windings will keep the ballance.

But that all depends on you application. Maybe your charge source is allready defined.

gargamel

Fri Aug 14 2015, 05:48PM

Registered Member #56093 Joined: Wed Jul 29 2015, 04:57PM
Location:
Posts: 8

Status quo is that the whole voltage is delivered by a single CCFL inverter + cascade.

I don't need very high effeciency for my project. In fact I don't mind to create useless heat in balancing. That is until my voltage source does not reach the right voltage anymore ;)

The MMC will be around 1ÂµF and should be charged to 4500V, ideally in less than 10 seconds.

Measuring the voltage is not easy, because the load of measuring will mess up the balance...

DerAlbi

Sat Aug 15 2015, 04:40PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

How much power does your charger have?
20J in 10sec are 2W. If you have 10W you can burn 80J inside the ballancing resistors........... just calculate how small they can become and try it

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob