180 capacitor swapout

[Rig Pig]

just replaced the capacitor bank in my old Automig, the cheapest option was good old eBay - unfortunately the closest quality ones I could find were half the value- so I bought 10! Good job the case has plenty of room! It's made a massive difference to the way the machine runs!

 

[normspanners]

you are actually better having more caps of a smaller size to make up the large value that you want, because they each charge and discharge quicker. That therefore makes the setup more efficient.

 

[rtbcomp]

you are actually better having more caps of a smaller size to make up the large value that you want, because they each charge and discharge quicker. That therefore makes the setup more efficient.

How does that work? A smaller cap will charge/discharge more quickly than a bigger one, but you're using more smaller ones so the resultant charge/discharge time will be the same. Capacitance is a measure of the time a cap takes to charge, so the charge/discharge time should be independent of the number of actual capacitors, I would have thought.

 

[normspanners]

I was always told it was like filling a bucket from a tank if you have a 1" hose to fill a 10 gal bucket it takes it takes say 10 sec to fill, but if you had 10 x 1gal buckets and each bucket has a 1" hose to it from the tank it takes 1 sec to get the same 10gal.

 

[rtbcomp]

I was thinking along the lines that if you have 1 hose delivering 1 gallon a minute it takes 10 minutes to fill 10 1 gall jars and 10 mins to fill a 10 gall bucket. Which is nearer what we have in a welder, the ripple current available to be absorbed by the capacitors is the same no matter how many capacitors you have. It's the total capacitance that matters.

Electrolytic capacitors do have internal inductance, but I don't know if that depends on the capacitance but I reckon it wouldn't be a problem at 50/100 Hz ripple. Sometimes you see a paper capacitor connected across an electrolytic to filter out high frequencies & spikes the former can't pass.

 

[MattH]

The charging time constant for an RC network is a result of the resistance and capacitance in the circuit. Paralleling smaller capacitors together allows them to act as one cap of the combined value. The independent size of each cap makes no difference to the overall charging time of the circuit.

 

[GW8IZR]

Depending on the quality of manufacture and design the ESR of an electrolytic capacitor can be very different for large and small values, this can affect circuit performance beyond simple RC time constants.

I dont believe you can make' statements such as 'lots of small are better' or 'same capacitance value will be the same' - you need to look at the specific components used.

But for this application if you have used good quality components the C and ESR should be OK and importantly remain OK for a long time.

If you had used cheap Chinese unspecified components you might be ok but they may change significantly over a short time, particularly in a hot environment like the inside of a welder.

HTH

 

[Rig Pig]

It took a bit of hunting to find British made caps I didn't like the idea of far eastern ones, the originals are probably 20+ years old, so if these last half as long as that then I think the welder will have done well!

 

[rtbcomp]

What made you decide the caps needed replacing?

 

[normspanners]

I was thinking along the lines that if you have 1 hose delivering 1 gallon a minute it takes 10 minutes to fill 10 1 gall jars and 10 mins to fill a 10 gall bucket. Which is nearer what we have in a welder, the ripple current available to be absorbed by the capacitors is the same no matter how many capacitors you have. It's the total capacitance that matters.

Electrolytic capacitors do have internal inductance, but I don't know if that depends on the capacitance but I reckon it wouldn't be a problem at 50/100 Hz ripple. Sometimes you see a paper capacitor connected across an electrolytic to filter out high frequencies & spikes the former can't pass.

But you havn't got one 'hose' each cap is connected separately, therefore lessening the resistance to and from and the inherent internal resistance in each cap ?? and heat build up in the capacitors system overall, a little like having ten battery chargers each charging a small battery, if the ten batteries where connected parralel to one charger it would take longer to charge them.
Also the heat build up would increase the resistance, etc etc

from http://www.google.com/patents/US8824175

In higher power welding systems, a single capacitor may be insufficient to smooth the output voltage because the capacitor does not have any limitation on current other than its own equivalent series resistance (ESR). As a result, the capacitor may generate a substantial amount of heat as it charges and discharges during the welding process. To dissipate the heat generated, a bank of multiple capacitors is typically used to share the load. Unfortunately, the bank of capacitors adds to the cost, size, and complexity of the welding system. Since the output capacitor is coupled directly across the rectifier, the charging current flowing into the capacitor from the transformer via the rectifier is only limited by the impedance of the AC source providing power to the transformer along with the impedance of the transformer, the internal resistance of the capacitor, the impedance of the rectifier and other components inherent in the circuit. In general, it is desirable to minimize these inherent impedances and resistances in the circuit to minimize losses that cause heating and reduce overall efficiency and capacity of the welding system. Thus, the peak charging current may be of substantial magnitude and may result in a substantial RMS ripple current flowing in the capacitor. The internal losses within the capacitor and thus the internal heating may be directly related to the RMS current flowing in the capacitor. As the current increases, the heating may become substantial.

This was the way it was explained to me when I inquired as to the best way to improve my single phase machine and it did appear to work VERY well.

 

[normspanners]

Scott Henion's Welding pages.

Adding a Capacitor Bank
When running on single phase, the output power is a rectified AC with a lot of ripple. This will decrease the arc stability a bit. Adding a bank of capacitors will smooth out the power and provide a smooth arc. Capacitance from 20,000 to 150,000 micro farads will work. More is better for a smooth arc at higher currents, but will increase the surge when the contactor turns on. I used 60,000 micro farads (uF).

A small motor run capacitor can be used (100 to 200uf.) That will increase the stability a bit but there will still be a "buzz" of the rectified AC.

One big capacitor will likely just burn up as we will be pulling huge currents through it. So, a bank of smaller capacitors is better. These need to be rated at 50 volts or higher. I used two, 30,000uF capacitors. On my DIY-welder, I used three, 23,000uf capacitors. I would have used 5, but I could not fit them. These should be "Computer Grade" or "Low ESR" capacitors. Each capacitor should be 20,000 to 30,000uf. Larger may not handle the current well.

The capacitors need to be connected together with a heavy copper or aluminum bar. I have used flattened 1/4" ID copper tubing or 1/2" x 1/8" aluminum stock.

 

[rtbcomp]

I take Norm's point, especially about distributing the heat between multiple caps. I suppose it depends how far back in the circuit you define your 'tank', if the mains is your tank then the 'hose' would consist of the transformer and the wiring from it, the rectifier &c, ie one 'hose', but then it splits into smaller hoses where the caps are connected.

Clearly the considerations go beyond the actual capacitance of the arrangement. I don't really believe you achieve a faster charge/discharge time with several caps as most of the circuit is common to them all, certainly not for practical purposes, also bear in mind the quoted value of an electrolytic can be +/- 50% of its actual value.

So in conclusion m'lud, I'd go along with multiple caps. This sort of application where you bang a low impedance power source straight across a large capacitor is probably the harshest you can get.

 

[normspanners]

By no means am I claiming to be an expert on this, after all I am a mechanic, but I try to learn from various people and have some friends that are electrical engineers in their own right and one that works on big sound rigs etc is well respected in his field, and that is how it was explained to me, about the set up of multiple caps.

BTW if you look in many older welders they use multiply caps often shaping and sizing the cabinet to accommodate them, take butters for instance I was working on a old single phase butters and the thing was full of caps and that was a delight to use. sounded more like spray transfer when welding normally.
They would not do it for no good reason, expense didnt come into it when that old girl was made, by the look of it.

but then it splits into smaller hoses where the caps are connected.

Yes but those 'hoses' arn't 'small' in their own right they are each the same size as the one big cap, would be.

The capacitors aren't connected separately, they are interconnected, so in effect you do have one 'hose'.

But they are by the mere fact EACH is connected and that is the 'multiple hose'
we are talking virtual instantaneous charging and discharging and anything that helps this occur, is surely a good thing, so lots of smaller caps, EACH connected separately gives an easier route from the source and then to the load.
After all any voltage traveling in or out will create heat in some fashion, that heat is the cause of problems in many ways, the main one being increasing resistance and in this situation that is the last thing you want.

 

[normspanners]

The battery and charger analogy I believe to be false.

OK maybe that was not the best analogy, but it is a fact smaller batteries charge faster and if you have lots of them the max capacity is gained quicker and better for the batteries by having multiple chargers than one big battery and charger hence, why you run a slow equalizer charge every now and again.

 

[normspanners]

They also don't have individual connections,

It is the fact of the actual internal resistance of each that is the limiting factor,
if you read the evidence that is abundant on the net most plum for more smaller rather than one large unit for the reasons that I have given and the butters machine I believe was built for performance not monitory limitations.

given the same available charging current an array of small batteries will take 'about' the same time to charge as one big one of equal capacity.

I did admit that the batteries are not the best analogy, If you took the time to read what I have written that is what I have said.

Also I think Scott Henion would not have done it that way if it was not the best way forward.
That my view of it anyway

 

[Rig Pig]

What made you decide the caps needed replacing?

The welder power seemed to have dropped, when I opened it up I spotted this. I found a nasty looking corroded joint in the output as well.

 

[rtbcomp]

A bit of giveaway!

 

[arther dailey]

Really interesting , so it needs someone who is a bona fida top dog welder to take a machine and rig it with various caps of differant sizes and ratings to see which is best ? my geuss would be you would find differant machines(plant) would be better with differant set ups,so surely if you replace with multiple and it works better,or indeed single and it works better, thats the point isnt it ,end result is all that matters .
Will bear it in mind if I change any in future thats for sure, my monies would be on multiples though!

 

[normspanners]

The welder power seemed to have dropped, when I opened it up I spotted this. I found a nasty looking corroded joint in the output as well.

Look at the size of those terminals they are huge, are they double ended ie a terminal at either end ?, that would be so that the plates act as heat sinks to rid the system of unwanted and restrictive heat build up.
Are your leads just push on lucar type terminals, if so you might be better soldering the leads direct to the terminals, because the blue are only good for 15/20 A, you may find that they become hot and burn at that point.
Again only my view, on mine I used yellow crimps but mine where screw terminal caps to test the principal and they got warm so the blues might be over their limit.

As per @shenion. welding pages that is why I think Scott advises using flattened copper pipe or ally buzz bar arrangement, I think.

 

[Rig Pig]

Yes the connections are push fit, I will keep an eye on them and the banks temperature when I give the Mig some serious use to make certain I don't get any nasty surprises! If the connections are rated 15-20A each I have 150-200A rating, the Mig is 180A max so I hope I won't have a problem.