Capacitor and Resistor to Suppress Voltage Spikes in Welder

If Dodie was really toasted then I don't think it was by spikes, Normally it will go without any toasting effects if its due to high voltage spikes, Reason for that is instant permanent breakdown of junction, it may happen even if there was no heat generation,It looks that you were drawing over current from welder,

About diode on output, I would say if you are using a bridge rectifier then its not going to work as two Dodie making negative terminal will be forward biased with the any spikes generated on output,

The better solution is to not buy low end junk welders to start with. I worked as a welder and plasma cutter service tech for the Praxair corp and I am now self employed and also do welder and plasma cutter repair work. Unfortunately far to often in my repair experience I have seen welders just like yours with diodes rated for about half the peak amperage and with only around 100 volt ratings.

The easiest way to determine what happened is to look at how your diodes went bad. A high voltage spike will most often cause a dead short across the diode and an over current or overheat will cause an open circuit condition in your diode.

The best recommendation I can give you is to use diodes with at least a 10X maximum working voltage rating and at least a 1.5X maximum working current rating.

Thank you for your replies. I had a Lincoln model that would surge and stop and start, ruining welds. Funny thing about that, when I used my friends Lincoln it does the same thing, when I rented one it did the same thing, and when I told another friend about it he complained his did the same thing. My comparably priced Italian made Westard never did it and produced superior welds every time. The diodes were rated for the out put and the new ones I have are rated at 50% more. The diodes fried because they over heated due to a short circuit. I simply needed to know how to protect my new set of diodes from voltage spikes. Yes, it is a bridge rectifier configuration... Please help.

Reverse diode with out resistance on output may sink all the reverse current generated due to spikes, but again diode itself will go in seconds if not rated for spike voltage or reverse current may burn it, Putting a series resistance may save it but in that case some current will pass through bridge diodes,Probably you can find out some optimum value of Resistance with diode But all is about Hit and trial, starting point may be using the same diode that is used in bridge with high current rheostat to adjust resistance,One more method that may work is LC network with a diode on output, and tuned near to frequency of spikes,

The reason for surging and stopping was that Lincoln used the welding power to drive the wire feed motor rather than a separate power supply. In theory and in practice this can work but in the majority of cases but the settings are fiddly re wire size, wire speed,power setting,tip condition,local mains voltage drop etc it can lead to poor control. The diode ratings in your bridge are not equal to your 75Amp output [due to cost saving] as the the diodes are operating in a 50% duty cycle hence they can get away with a 50Amp rating. In the case of small cheap welders with a small heat sink and poor cooling sometimes no fan this can eventualy give thermal break down as the rating is much diminished at a higher temperature , it is hardly ever a voltage spike break down.Maximum DC voltage out put would not be more than 32 Volt probably 28 Volt in a small welder. Suppression is usually by strapping a transorb across the diode. If your diodes are rated at 50% more than the original so long as your heat sink arrangement is good enough you should be ok.

Diode can heat up and essential thing to do is:

Select diode from standard manufacturer and rarely pick from market by numbers. They may not be good one.

Mount the diodes properly and couple to the heat sink with good heat coupler such that transfer of heat is proper. Even terminal connectors can be cleaned and low resistivity coupled properly. These are the points of heat source.

Look at the low resistance copper wires connected to the diodes and their current rating and heat generated in the wires. Are these wires hot? or junction point hot. If poorly coupled then all heat will be generated in joints and then this will partially heat up and will also not deliver current into load.

You can add current limiting resistor and charge storage capacitor to deliver uniform current. Even 10 Milli ohms with 100A will drop voltage by 1V. So you really don't need any great resistance and only charge storage capacitors of 50V 10000uF are to be added in parallel and terminated at one point. Poor electrical connections are often the cause of problem.

There can be inductive high voltage in your secondary coil exceeding the limit of the diodes. Current spike in load can do that. Hence, put a voltage control resistor across secondary of the transformer to kill excessive voltage spikes there. RC snubber can't kill it but it can reduce the effect to some extent. Use floating oscilloscope probe to see the waveform at the transformer output while you do welding.

Hi PIV rated diodes are the cheapest and easiest way to protect the system.

You can also use MOV's to absorb transient overvoltages and keep it below the breakover voltage of the diodes.

30 years ago one paid a large premium for higher PIV devices. The cost difference now is not much of a consideration. I agree with the other poster that suggested using devices with a 10x margin (300 volts in your case), but you could readily go to 1000 PIV.

Hi GW,

I have tested diodes from market rated for 1kV and 50% failed to reach even 100V. Some of them had high reverse leakage for under 30V. Only few were good enough and worked up to 1000V, 1200V and 1500V.

Most of the people do not test the device and they just put something they get from market. This is bound to be of problem. Manufacturers do not test all parts they produce and only do batch testing and also let few or many bad part go into sale basket.

I have specially built this high voltage test setup for reverse voltage testing of diodes up to 15kV and it trips on reverse current reaching threshold current of 50uA to 500uA preset level for tripping the HV supply. This set up works fine in weeding off defective rectifier diodes and Zener diodes.

Same problem is with MOV and I have tested 1kV little fuse MOV and they conduct at voltage 1500V-1600V and above. Be sure you select the right voltage ratings for these parts also as they must operate when you want these to operate.

Shyam:

Do you have a schematic you could post for the test jig? I suspect many of us could and would use such a jig, if we had one we could trust.

I have always done PIV testing using my megger being it has 250, 500, and 1000 volt test ranges.

If I have a 1000 volt rated diode and the megger cant give me a resistance reading while set on the 1000 volt test range its good enough for me!

Practical electronics!

I stick with my semiconductor curve tracer!

Allowabbe leakage for device and application can have wide margins.

The thing our poster doesn't want is a reverse breakdown or overheating of the diode.

He can tolerate considerable leakage (mA). I suspect thermal rating is the bigger problem. A bigger device with heat sink will give him more head room.

Our high voltage power supplies are having current and voltage monitoring facility and over current trip mechanism. We increase high voltage output slowly not more than 10V/s for 1kV device testing. Leakage current trip is set to desired level before raising voltage.

Our 15kV test set up is very useful for testing many devices and also high voltage cable. Perhaps for testing high current diodes reverse trip current between 100uA to 10mA will be ideal. Forward current of 100A device may have this level of reverse leakage current.

I think 6kV test system with 3mA maximum current capacity will be good enough for testing most of the devices commonly used.

I too have a high voltage tester. We 100% test our MOV's, capacitors, snubber resistors, and SCR's in power circuits. Any rebuilt device is tested to manufactures limits before we return it to service. I too have seen devices that were supposed to be 1200V breakdown prematurely. In the power semiconductor field ( I am talking hundreds to thousands of amps) every device is typically tested. Leakage on these large pucks are often up to 10mA at 1200V, and that is within the manufacturing specs. On signal level devices this would be considered a dead short!

The new power devices we use for 600VAC are tested to 2400V. OEM from 25 years ago were only 1600V. I also remember on incoming inspection 25 years ago we had a problem with batches coming in and being only good for 1200V.

The voltage level required by the poster is so low that he can easily purchase a very high PIV device and even if it does come in under spec he will have so much head room it will be academic.

IGW,

I think you are a professional engineer and you are doing things properly.

f MOSFET switches are used along with inductor / transformer then surge voltage from 24V supply by switched inductance can go as high as few hundred voltages. Modern electronics require lots of protection mechanism by looking at the waveforms.

I think you guys way over analyzed a common diode burn out problem that happens to all most every small welder that gets over used.

Now it we could just get the engineers who design those welders to put half the thought and efforts into this as was done here they would never need guys like me to fix them!

Very well said. This is a fantastic forum and an invaluable resource. What caused the diode to short was an accidental torch to work strike that resulted in a half second bridge that caused an almost instantaneous failure. I just want my diodes to last as long as possible, given the amount of work I am putting into fabbing the heat sink and re-arranging the transformers to allow a fan. I appreciate all of the well thought out replies. Thank you all.