–{•Ά‚ΦƒXƒLƒbƒv

This site shows my DIY work mainly at imaginative level.

Making of small power electric welder


Making of small power electric welder

This is a description about a making of small power electric welder.
Electric power is stored to a capacitor over time, and is discharged momentarily.



Table of contents
1. Energy
2. Pulse width
3. Current
4. Capacitor
5. MOSFET
6. Power circuit
7. Specification
8. Circuit diagram and sketch
9. Making of power circuit
10. About signal circuit
11. Case


Make believe (Assumption) section

The major target is a stainless steel that have some electrical conduction but not have a good heat conductivity.
Furthermore, it is considered whether silver can also be welded.
The thickness for welding shall be 0.1 to 1 mm.

1. Energy
Since I feel that it is not tough enough if the diameter of a melting spot (diameter of a fusion face area)
is less than 1 mm, then the diameter of a target melting spot shall be 1 mm.
For example, the specific heat of stainless steel is 0.59 [J/(g-K)] and density is 7.93 [g/cm3],
the weight of 1mm diameter and 10 micrometer thickness is 62.2micro gram then it is 27213 [K/J],
if it thinks that it will be melted even for a short time
if it heats to about 1600 Ž since the melting point is 1400 to 1450 Ž,
then the required energy is estimated to 1600/27213=0.06J .
Since 1mm for diameter is too wide area for the case of say 0.2 mm thickness,
adjustment of the energy according to an object will be required.
If it is incidentally silver, for example,
since the specific-heat-capacity =0.232 [J/(g-K)], density =10.49 [g/cm3],
and the melting point = 1235 Ž,
therefore, if it says only about energy,
it can melt silver by the same energy to melt stainless steel.
It is same story for brass, which is 0.377[J/(g₯K)], 8.5[g/cm3] and 900Ž.
ReturnToTop

2. Pulse width
Thermal diffusion length is 4 mm/1s (12.6 micrometers/10microsecond) by stainless steel
with a comparatively small value.
And the silver is 26.4 mm/1s (83.5 micrometers/10microsecond) with a comparatively large value.
Since thermal diffusion length is a square root of time,
the time when thermal diffusion length is set to 20 micrometers
by stainless steel should be set to about 25 or less microseconds.
In the case of silver, it should be 0.6 or less microsecond.
ReturnToTop

3. Current
I do not want to care about insulation not much.
Even if it should touch directly, I would like to use voltage of the grade
which does not receive an electric shock severely (if it has not got wet).
Therefore, if it can do to 30V that it is high, I will make it at least 24V.
I think that 30V is maximum and about 24V will be better.
It will be 3.31milli-Ohms if it uses 1m of 5.5-mm cable (3.31ohms/(km)) for round circuit.
By estimating as roughly 10 milli-Ohms in total with some other resistance for internal circuit,
and if the 10V voltage drop is allowed for these resistance,
the current is limited to 1000A.
Therefore, it may be able to set 2000-3000A, though 1000A is likely.
When effective voltage to load (contact resistance) is temporarily set to 5V and it is 10 microseconds,
it is 1000A*5V*10microsecond=0.05J.
It is suit for stainless steel.
And no way for silver of 0.6 microseconds.
If the 10 microseconds pulse apply for silver,
the 1mm diameter and thermal diffusion length = 83.5 micrometers require
about 0.2J (Specific-heat-capacity =0.232 [J/(g-K)], density =10.49 [g/cm3],
and melting point =1235Ž and the melting point use as it is since silver will be easy to melt.
Since thermal diffusion length is a square root of time and the energy is proportional of time,
then it is 160 microseconds to melt in 0.8J.
It seems that silver welding requires cautions to a burn. Or, an electrode may be welded.
ReturnToTop
4. Capacitor
Because of Q=CV=it=10mC, at the 1000A and 10 microseconds,
it require 333 micro-F at 30V and 417 micro-F at 24V.
It require 5333 micro-F at 160 microseconds. Then silver have a difficulty in capacitor.
Considering about the kinds of capacitor,
a electric double layer capacitor is left out of selection at voltage or an ESR target,
usually, the electrolytic condenser which becomes a leading candidate is also slightly severe in ESR for this use.
Then I select a conductive polymer aluminum solid electrolytic capacitor.
It have rather lower ESR and larger capacitor with high voltage.
ReturnToTop
5. MOSFET
The lower voltage MOSFET have very low on state resistance,
though the terminal resistance make its limit in larger current.
Then I select a little higher voltage type and increase parallel number.
I select SUP85N15-21 (85A/150V, 17.5 milli-Ohms) in five parallel.
The on state voltage at 200A is estimated to 6V.
It is good balance value with 5 parallel and 2 serial of OS-capacitor (820 micro-F/16V, ESR=11 milli-Ohms).
And 150V of MOSFET will be able to stand the turn off surge voltage at very high current.
ReturnToTop
6. Power circuit
1 mm thickness is likely, though 0.5 mm thick is reasonable to make.
(Probably 0.2 mm thick copper sheet is enough thickness
to flow 1000A pulse with 1A average except contact point.)
ReturnToTop
7. Specification
<Specification>
Input voltage: 35Vmax (Limited by 7815)
Charge voltage: 30Vmax
Output current: 500A to 2000A (Effected by wiring)
Output energy: (0.37J*5)*0.5=0.9[J/pulse] (if the 50% efficiency)
Pulse width: 2 to 30 micro seconds
Frequency: Decided by charge current
Operation: Capacitor charge current limited to about 1A
Discharge capacitor while the set term
Wait re-charge and discharge again
Continue while SW on

<How to use>
Set pulse width by VR
SW on to welding
ReturnToTop
8. Circuit diagram and sketch




Although these circuit diagrams are visible like the buck chopper circuit of constant voltage, and a one shot circuit,
and of course, originally it is a circuit used for such a purpose,
substantially, that is not right in the purpose.
A protection circuit and normal operation are used conversely.
The 34063 operates as a constant-current-charge circuit by using its current limiter,
and original constant voltage operation functions as voltage protection.
( Since it was troublesome to have made sufficient heat dissipation structure,
not the resistance regulator but the switching regulator was used. )
Since charging current detection does not need accuracy,
it has been sensed as base current of a transistor.
10 ohms was put in in order not to operate by the charging current
(a part to discharge by partial resistance) under standby.
The diode linked to BE was put in in order not to flow a large current through BE.
Then, the diode for switching uses a shot key and the diode for protection between BE uses the PN junction one.
The current for balancing voltage shall be 0.5 mA,
then the partial voltage resistance of C is about 150k-Ohms x 5 parallels x 2 series.
The output of 555 begins to keep High, during the period whose Trig is Low.
Trig is temporarily set to High with a charging current detection signal,
and whenever re-charge is completed, a trigger starts, and a pulse comes out.
Therefore, this circuit is not a single shot but pulse comes out continuously during SW ON.
Although re-charge will sometimes start immediately and Trig will serve as High by charging current detection,
if a trigger starts and C begins electric discharge,
since timer operation will not stop once a trigger starts,
C is discharged during the pulse width set up by.
PNP and NPN of the buffer preceding paragraph were not changed under the upper,
because bias current is wasteful and VGS (th) of MOS is higher than 2*Vbe.
Although there is nothing in a circuit diagram,
7815 is required separately because of the voltage (15V) supplied to 555.
An electric power board (copper plate) is as it is shown in a figure.
Since a slightly larger insulated board than this enters in between, it is set to about 70*100 mm.



ReturnToTop

Actual making section

9. Making of power circuit
Usually, it is although it will just be going to carry out conversion treatment first
before touching and making it corrode when I buy a copper plate and it opens up the film,
it is the end material of 1-mm thickness which was used this time, since it was already very dirty,
first of all, it starts with acid washing.
Although it will probably be difficult to make if it is not 0.5-mm thickness or less, it will become somehow.
Oil-and-fats dirt is flushed with a part cleaner (the main ingredients are n-hexane),
and it washes with the liquid of weak acid. It is quicker to rub.
After washing away with a part cleaner again, it goes into conversion treatment.
When performing copper conversion treatment, it is the 1H-benzotriazole which I am always using.
It is the medicine generally mixed to a plastic etc. well because of protection from ultraviolet rays.
I purchased being sold as an ultraviolet absorber for the development of a photograph
(silver halide film instead of a digital camera).
A chelate action has this molecule to a copper ion
( = the settlement condition of molecules is good to the copper molecule
which makes the hand of combination remain and is unreserved on the surface of the copper plate),
and when this molecule is located in a line with the copper surface,
this molecule and copper have a near coefficient of linear expansion,
and then it will be in like a passivation film state.
Although placing close also has an evaporation antirust action,
if it thins with alcohol and applies to the surface, in general treatment, it will hardly corrode.
And it is the thickness for one molecule that adheres,
and since secession sublimation is carried out at a little more than 100 Ž,
there is almost no prevention to soldering.
(Since there is only thickness for one molecule, it is very weak physically,
and it does not become instead of a paint only now.
It is only for temporary use at the time of preservation and work.)
Since it will remain considerably rather than one-molecule thickness if it applies to the surface
even if it thins intensely,
a film is again formed in the portion of a crack and the grade
which is an evaporation antirust action has a thing like a self-restoration action.
For this reason, if a large area is not deleted or it does not wash away with a solvent,
even if it is cut, an antirust action continues while machining time.
After treatment, make a guide scribe line and cut by share and saw.



Although there was the rolling direction in a rolling board,
since I did not find well,
the mark was put on the portion to bend with the marker so that it might not be divided.
Although I thought that I would paste up a copper plate on an insulated board,
since heat-resistant adhesives did not sell at the neighboring home center,
I've bought heat-resistant double side stick tape instead.
Since this double side stick tape was thick,
it decided not to use an insulated board but to insulate it only with the tape.
Since this tape is thick, it does not care, even if there is a crevice.
Double side stick tape is arranged in on a large copper plate,
the copper plate is turned up and a small copper plates are arranged in the remaining tape sides.
Double side stick tape is turned up on the back side for a while at the end of a copper plate.



A polyimide tape is stuck on a portion not to attach solder to, and flux coating and solder are attached.
It would come so far and will not progress well.
Solder does not melt.
Is 1-mm thickness unreasonableness after all? .
Although it melted somehow by using the solder iron with a large tip area, it took time very much.@@
Temperature has been raised to the extent that a nasty smell comes out considerably.
It seems that however, the capacitor is valid
since a buzzer stops at an electrical connection checker by a definite period of time.



Oh, it is forgotten to attach the diode of the measure against output oscillating current.
It was the failure which was not drawn when a fair copy was made by CAD.
The Schottky diode (1N5822, 3A/40V) which it had was used.
Although power capacity of this diode is a little small,
inductance of an output line will not enlarged to burn this one when using as a pulse welding machine.
It is a useless spell although resistance (330 k Ohms) is attached to thing between GS(s) of MOS.
Although it seems that heat dissipation grease is used with the photograph of the making process,
it has changed to heat conduction double side stick tape in the end state.
ReturnToTop
10. About signal circuit
Since it was troublesome to have drawn a wiring diagram, it made suddenly on the universal board.
Therefore, there were many jumpers and the cable extraction position was also bad,
though it was settled somehow.
2SC3074, 2SA1244, 2SC2655, and 2SA1020 were used for the buffer transistor.
2A transistor was used to the preceding circuit in order to get enough hFE in large current.
Since there was no 0.22 Ohm stock, 1 Ohm was arranged in parallel four times (0.25 Ohm).
The diode used 1N5819 for Schottky, and used 1N4007 for PN junction.
The photograph while making does not remain.
The state where the lid was opened has become like a lower photograph.
It is in the state of insulating by a tape and fixing with double side stick tape mostly.




If a stainless wire is attached to a trial ... sparks do not shown -- ? ... but yes, it already welded.
It seems that welding intensity is slightly weak though acceptable.

It is welded like the following photographs.
The clip at the lower right of a photograph is a stand for photography, and it is unrelated to welding.@@



ReturnToTop
11. Case
I waslooking for some good and cheap case but I couldn't.
Then I made it by using aluminum plate as following photo.


Sorry, the scale is not inch but metric.

...It is although it is writing as if it was able to make completely and favorably,
but in fact, in a soldering failure and adjustment, a circuit is flown and part was exchanged several times.
Since parts break immediately in little unusual operation, a power circuit is troublesome.


ReturnToTop