Whats a 12V, 100A buck converter look like?

[acannell]

I'd like to make a 100A buck converter that operates in constant current mode, and takes a nominal 12V +/- 2V as input.

This is for a welder, powered by lead acid batteries. So heavy input ripple and big EMI are okay. Even extreme heat dissipation is okay because this device overall only needs a duty cycle of about 20%, say 30s on, and 120s off. Its an emergency-use tool and doesn't need to be friendly, but it does need to be reliable.

I am looking for some help to ballpark what this sort of device would realistically look like, if its realistic at all.

Voltage does not need to be regulated. This device is really there only to keep the current at safe levels, otherwise 500+amps would probably be drawn during a true short like electrode stuck to metal with no arc. There may be times when even directly connecting the batteries to the electrode would not exceed 100A, and in that case, I will try and make it so there is a bypass around this SMPS to increase efficiency and overall duty cycle.

Does this sound realistic for a single SMPS, or would multiple ones be required? What do you think is the sticking point in designing this?

Thanks for any help!!

- - - Updated - - -

At first glance the LT1339 might fit the bill..at least up to a few dozen amps, but that might be close enough to get an idea what 100A would take.

http://cds.linear.com/docs/en/datasheet/1339fas.pdf

And maybe its in LT power designer too..

 

[BradtheRad]

You'll probably have an easier time if you interleave two or more buck converters.

To illustrate, here is a simple simulation. (I do this for relaxation.)



As you can see, each coil carries tens of Amperes. Although it is conceivable to do the job with a single coil, it will need to carry current peaks between 120 and 200 Amps. So it is easier to split the load among multiple converters, and multiple switching devices.

Interleaving gives you an advantage of drawing a smoother current waveform from the power supply, if you stagger the clock signals. The circuitry for doing that is absent in my screenshot.

 

[acannell]

You'll probably have an easier time if you interleave two or more buck converters.

To illustrate, here is a simple simulation. (I do this for relaxation.)



As you can see, each coil carries tens of Amperes. Although it is conceivable to do the job with a single coil, it will need to carry current peaks between 120 and 200 Amps. So it is easier to split the load among multiple converters, and multiple switching devices.

Interleaving gives you an advantage of drawing a smoother current waveform from the power supply, if you stagger the clock signals. The circuitry for doing that is absent in my screenshot.

Thank you very much!!

What would a single coil system look like? I am trying to keep things as simple as possible. Its probably okay to draw large current pulses because the source is a lead-acid battery, and the intended duty cycle for this supply is something like 10% to 20%.

The supply is really only acting as a current-reducer for safety and convenience limits. Directly connecting car batteries as a welder can result in unpredictable and very high currents, maybe 1000+ amps, especially if a short occurs like a welding rod sticking to a plate. So this supply will limit the current to 100A or similar.

There are also times where due to welding conditions and/or battery charge state, 100A may not be drawn by the welding load, and in that case, this supply would ideally go to 100% duty cycle and basically connect the batteries straight to the electrode.

 

[Dan Mills]

I would still go polyphase, MUCH easier to handle a few tens of amps five times then it is to handle 100A or so, especially when you consider the voltage drop limits implied by only having 12V or so to start with.

Current mode control is the way to go with a part that will happily let you program the current easily, and I would actually be thinking buck/boost here (a rather higher open circuit voltage will make striking an arc very much easier), Linear tech LT3790 looks like a sane option to me, dead easy to parallel and you should be able to use off the shelf magnetics if you use enough phases.

The simple minded alternative is a paint tin full of transformer oil housing a couple of hundred watts of low value resistor.... Kind of crude, but clearly going to be at least as reliable as anything active.

Regards, Dan.

 

[asdf44]

Except this is a pretty crude application, so I could understand a reluctance to go polyphase. It sounds like more of a current limiter than a regulator.

I could see a design with a very large inductor and very large heatsinked mosfet or IGBT with a fast but potentially crude current limiting control scheme. Perhaps shut off at 120A and turn on X time later.

Though I'd search pretty hard for what other people have done for this particular application as welding is pretty common and undoubtedly someone has approached this same problem. Bear in mind that high currents combined with large inductors can be quite dangerous. You'd need a very robust voltage protection scheme.

 

[BradtheRad]

To see how the theory works, here is a simplest possible buck converter.



A 555 timer IC can provide the switching pulses. Duty cycle can be altered by raising and lowering voltage at the 'control' pin.

It's almost certain that you would have to parallel two or more mosfets.
Two or more diodes.
Etc.
Once you start doing that, you might as well adopt the interleaved design.

I inserted a resistor in the supply wire, to show that the ohm value needs to be minimal, or else it can reduce Ampere flow. I made the load the same value for no particular reason.

It may take some searching to find a power inductor rated 200A.

As asdf44 states (post #5), large current flow through a coil can generate a lethal voltage spike if it is cut off suddenly.

 

[acannell]

Dan Mills, asdf44, and BradtheRad, thank you very much for your thoughts!!

I will look into the LT3790. I am still getting my head around the fundamentals of buck and boost in circuitmaker, so it will be awhile before I can evaluate SMPS IC's.

I have also considered just using a big resistor. The problem is, it needs to vary in resistance I think, depending on the welding activity, and would require another battery to be used, to get to 24V. 12V welding is not really possible with SMAW by hand, apparently.

asdf44: yes, this is definitely a crude application. And what I have noticed is that almost universally, SMPS app notes and reference designs are all geared towards low emi, very high efficiency, 100% duty cycle (overall), and very small size, none of which are concerns in this application, which is what makes me think there is room here for it to work out.

In other words, it can be big, get very hot, only work for 20% overall duty cycle, waste 1000W (plenty of electrical energy to go around from lead acid + vehicle charging system, the waste heat is another issue of course)..it can have all those faults and be a success.

There are several vehicle powered welding systems, check out this link for a somewhat disorganized but thorough write up comparing a bunch:

https://www.parksoffroad.com/miscinstalls/welderreview/weldermain.htm

These designs require either lugging around either one (or TWO!!) additional lead acid batteries, besides the one in the vehicle, or adding an additional alternator to the vehicle.

They do not have any real current or voltage control. You even have to manually control the engine speed on the alternator version to get to the right voltage. As battery voltage drops, welding performance drops.

I think these designs are clumsy and very overpriced, and do not take advantage of SMPS technology.

In my opinion, a vehicle powered welder should connect to the existing battery in the vehicle, with no changes to the vehicle. The welder should draw whatever current it needs to output roughly 24V at around 100A. This should allow stick welding (SMAW) of 1/4" steel plate using a simple rod holder, with the chassis acting as return. Electrode positive is actually ideal for this particular welding type.

1/4" steel is about the biggest you would find in an emergency situation and would allow pretty much anything that needed a weld to be repaired.

Winches routinely draw 3kw+ at similar duty cycles, directly from the battery/charging system, at higher duty cycles, so I dont think its a question of the electrical power from the vehicle. If you look at the electrical source you basically have an unlimited amount of current available at 12V, backed up by a charging system that probably puts out another 100A. At 20% duty cycle there should be no issue drawing 200A for 30 to 60s.

Its just a matter of getting that power conditioned to the right voltage and amperage limit.

I am still not sure if buck or boost or buck/boost is more appropriate.

In order to use a buck-only, you will need an additional lead acid battery to get things up around 24V. This is possible. A smaller lead acid battery could provide the current at this duty cycle and not be too cumbersome to carry around as part of the welder. Circuitry could be added to charge it when welding is not taking place. The buck would act on either the second battery voltage of 12V or the entire series voltage of 24V, whatever is easiest.

A boost-only system is the nicest because it requires no additional battery which can lose charge or become damaged from storage. It would likely have the greatest currents though, to boost 12V to 24V.

Buck boost may also be the most appropriate, but Im not sure. It may be the worst option if it requires the most complex SMPS design as well as an additional battery.

I did some stick welding yesterday to see what it was like. I do lots of TIG welding but have never done stick. I set the welder to 100A. I measured the open circuit voltage and it was 65V. The welding went VERY easy. However its extremely common for the electrode to stick to the plate, and then you essentially have a short circuit composed of the cables, ground clamp, rod holder, and rod. This is where you would want welding current limited, otherwise who knows how much current would be drawn from a lead acid battery.

You also need welding current reduced during welding. If its too high you can blow through thinner metals. You can also damage the batteries by drawing too much current, not to mention increasing the design requirement for the SMPS to pass all that extra current that isnt even needed. So I say minimum current needed is the goal here.

Something I have not done yet is weld with 24V being the maximum voltage. I am not sure how easy or hard that is. I cant sent my welder to a fixed output voltage.

BradtheRad, thank you for the example.

Regarding finding components rated at such high currents, arent most of those component ratings assuming air-cooling and 100% duty cycle? If I have some kind of liquid/oil bath possible with a pump, operate at 20% duty cycle, and allow for things to get very hot, would that open up some more doors as far as component availability?


Here are the welds I did, the joint between the two pieces. All the other stuff wasnt part of it. The question here is what voltage the welder was maintaining. Ill have to do it again and get my meter connected to the welder output terminals.

**broken link removed**

**broken link removed**

- - - Updated - - -

What I think the right direction here is a real dirty, crude design, but careful design, that just slams electrons to the right voltage and is probably immersed in coolant of some kind. Not pretty, not EMI friendly, gets hot, can only run for a short time, etc...but reliable and wont fail. Pretty much a complete deviation from almost every typical SMPS design.

One more option I didnt mention is possible using a 2kw+ transformer and push-pull the vehicle 12V through it to get 24VAC out. That could then be rectified to 24VDC, and the current limit would come from transformer saturation, just like every buzz-box AC welder. Im not sure if 24VAC could be used directly, eliminating the rectification.

This may be an attractive option because you really could submerge the transformer in oil for cooling. And, I think transformers of that power are realistic for the application. My harbor freight MIG welder weighs about 25 lbs I think and consumes about 2.4kw. Thats about right for this. So instead of a SMPS design it would be a very simple push-pull bridge and a transformer design.

 

[Dan Mills]

How about a hysteresis converter, place a few milliohms of current sense in series with the inductor, then you need a dual comparator, a set/reset latch and the usual mosfet driver and butch catch diode shenanigans.

When the inductor current falls below say 90% of the setpoint one comparator crosses threshold and sets the S/R latch turning the mosfet on, when the current exceeds 110% of setpoint the other comparator crosses threshold and resets the latch turning the mosfet off....

It will need a butch catch diode, and probably a fairly serious snubber, but it is simple, I would suggest operation at somewhere around 10KHz or so as a reasonable trade off between weight and switching losses.

There is no voltage regulation, only current is regulated, and the frequency is set by the voltage drop and inductor value.

I suspect that MMA welding at 12V is going to be a pain in the arse, there is a reason most sets have an open circuit voltage somewhere in the 40 - 60V region.

Regards, Dan.

 

[acannell]

How about a hysteresis converter, place a few milliohms of current sense in series with the inductor, then you need a dual comparator, a set/reset latch and the usual mosfet driver and butch catch diode shenanigans.

When the inductor current falls below say 90% of the setpoint one comparator crosses threshold and sets the S/R latch turning the mosfet on, when the current exceeds 110% of setpoint the other comparator crosses threshold and resets the latch turning the mosfet off....

It will need a butch catch diode, and probably a fairly serious snubber, but it is simple, I would suggest operation at somewhere around 10KHz or so as a reasonable trade off between weight and switching losses.

There is no voltage regulation, only current is regulated, and the frequency is set by the voltage drop and inductor value.

I suspect that MMA welding at 12V is going to be a pain in the arse, there is a reason most sets have an open circuit voltage somewhere in the 40 - 60V region.

Regards, Dan.

Yep 12V is definitely out of the question for the output voltage..way too low. Is what you are suggesting a crude buck converter? Because that might work out if I add a second small lead acid battery to the system to get to 24V, and then use your idea to limit current. I'll try to whip that up in circuitmaker and see what it looks like.

 

[mtwieg]

For welding you generally need a voltage between 20-40V on the output to maintain a decent arc, even higher to start the arc.

 

[acannell]

For welding you generally need a voltage between 20-40V on the output to maintain a decent arc, even higher to start the arc.

You can strike an arc and stick weld successfully at 24V, i.e. two lead acid batteries in series:

https://www.youtube.com/watch?v=PV5oLPLUzrM

And actually more like 10V is definitely enough voltage to maintain an acceptable arc at 100A for stick welding.

From Millers website:

**broken link removed**

**broken link removed**

Not having the higher open circuit voltage available will make things a little different, but for the purposes of an emergency welder I think it will work fine.

Just curious, where are you getting those numbers from?

 

[FlapJack]

home made 12v welder.

https://www.motherearthnews.com/diy/portable-dc-arc-welder-zmaz80ndzraw.aspx?PageId=1

 

[acannell]

home made 12v welder.

https://www.motherearthnews.com/diy/portable-dc-arc-welder-zmaz80ndzraw.aspx?PageId=1

I dont see any mention of modifying the alternator voltage regulator to output higher than 12V, so I guess that really is a 12 or maybe 14V welder. I wonder how much current flows when the electrode sticks.