pulse transformer

[dr pepper]

Mots are deliberately designed to have high leakage, thats why all the e's and the i's are aligned and have magnetic shunts.
Might still work though, it would be interesting to see if you can get enough energy through it to get a weld, at a low welding voltage you will get lot higher current, maybe 1v @ 1000a, but whether you can keep that up long enough to make a weld is going to be the tricky bit, volt seconds will be against you here.
Take care with transformer saturation, once the core saturates the primary will turn into a short circuit, so you'll get lots of amps in the primary at that point, and not a lot in the secondary.

 

[throbscottle]

Chris - yes, that's the one I found, pretty sure it is John's document.
DP - we will see. I took the shunts out at the same time as the secondary. Might have time to play in the next few days. I think primary resistance might actually be my biggest enemy.

 

[throbscottle]

Right, I have the week off so can have a go at building.
I redrew the schematic with some timers in it to fire the SCRs and set the interval between preheat and weld pulses. I'm showing the preheat and weld windings going opposite ways. Hoping this dispenses with the need to to de-energise the core after the weld, especially considering it is a matter of seconds between welds.
I'll try with just one cap for experimenting and use a spark gap instead of SCR - I have an old flash from a disposable camera I can use to trigger it. If the SG works well I might just use those and not bother with SCRs at all since it's massively cheaper.
The PMOS is to isolate the caps from the charging supply during the weld so the SCRs can turn off. Maybe won't need it with SG's.
The oscillator attached to the output is supposed to detect if there is a piece connected to the welding tips or not. I'm starting to wonder if it wouldn't just be easier to have a pressure switch.

 

[dr pepper]

Your schematic looks a little like a forward converter, probably acts a lot like one.
If the forward & reverse times and voltages are equal then the core should reset Ok, still it might be a good idea to put some Mov's or voltage protection around the circuit if you use Scr's, as if the timings are not quite right or the windings are not perfect the energy left in the core might produce some high voltage spikes.
Spark gaps are not a thousand miles away from ignatrons I mentioned earlier, ignatrons work at lower voltages than a spark gap as they use a mercury atmosphere, the merc arc rectifier in my avatar isnt a millino miles away from an ignatron either.

 

[throbscottle]

I tried with the MOT yesterday. I roughly counted about 260 turns on the primary, about 2.5 ohms. High but I tried it anyway since getting it off would be a major undertaking. I tried secondaries with 3 turns, 7 turns and 20 turns. I tried the cap (4700uF) charged to 100v, 200v and 300v but only at 300v did the transformer give a nice satisfying "thump" but still no mark on the test-piece.
I also got the "inverter" microwave transformer apart with the ferrite core. Quite a big air gap on it, over a mm I think. Much less mass in the core than I thought. So that won't be any good.
So I'll see if I can squeeze in a smaller primary on the MOT, maybe 50 to 100 turns, which is tricky because I have no magnet wire...

I'm understanding why the commercial machines use such a high voltage, now

 

[dr pepper]

The lump of energy transferred each cycle of the trans is less on a higher freq transformer such as an inverter microwave trans because it runs at much higher freq, because there are more specks of energy per second they are a lot smaller.
50hz mot's tend to run close to saturation, if you reduce the number of turns but still drive it with 300v then you are going to saturate it, so some capacitor energy will just heat the core up.
Getting enough energy from the core in one cycle looks like the issue.
If you can get 2 mot's the same you could stack them side by side and put the windings around both, then you'd double the core area.
Another enemy you'll get is inductance, the current in the primary is going to build up slowly, using up energy in the cap reducing the 'thump', increasing the cap voltage will combat this to an extent, at the expense of wasting energy saturating the core.
Some fiddle factor involved.

 

[throbscottle]

The lump of energy transferred each cycle of the trans is less on a higher freq transformer such as an inverter microwave trans because it runs at much higher freq, because there are more specks of energy per second they are a lot smaller.

Ahh, what a perfect explanation! That makes so much more sense now!
Fiddle factor indeed.
I found some fairly thick 16 way ribbon cable (old IBM printer cable) so put 5 turns of that on, connected the ends together offset by 1 so giving 80 turns. Put on 8 turns of stranded wire as the secondary. Still not the slightest mark on the test piece. I thought I saw a bit of a spark but it must have been my imagination.
I guess I'd better not try any less turns than that on the primary, going by what you say.
Maybe I should just look for a better core. Those "site transformers" discussed in another thread suddenly become attractive...
It also become apparent that I need a much bigger capacitor

 

[dr pepper]

Site transformers are usually potted so you cant do much with them.
Maybe if your going to change the caps, get lower voltage higher capacitance ones and do without the transformer altogether and dump the caps into the weld, just change charge voltage for weld current.

 

[throbscottle]

Low V high C is the usual way hobbyists do it, and there are some really nice projects out there - I just wanted to try this way, especially when you look at the prices of very large capacitors or making a large bank of smaller ones. You get a greater amount of energy storage per £ spent by going up in voltage, at least up to 400V caps anyway. A 1F 12v capacitor commonly sold for car audio (for the people who like to make your house shake when they go past) commonly used for this costs at least £30. The energy stored is 72J which I'm using for my baseline, commonly used to weld battery tabs. A series chain of higher value lower voltage caps is cheaper but the ESR becomes too high. Compare my 4700uF 400V cap, cost me £10 (seller says normal price is £55) energy stored at 300v is 211J. Getting over 300v gets to be very slow at the moment because I'm only using 1/2 wave rect + 390R in series.
So most of that energy is going somewhere itt's not wanted. Either my wire's too thin or my core's too crap. Probably both.
Think I'll just get back to sorting the shed out (it was going great, I had room to swing an actual cat at one point! (I didn't try it, I hasten to add, cat ran away...)) and do some more on other projects I still haven't finished, then get back to this one. Maybe I'll have a brainwave...

 

[throbscottle]

If the core saturates, does that mean I will get nothing out of it at all?

 

[dr pepper]

450v caps will be cheaper as they are in most household appliances nowadays, thankfully there are not as many loud boy racers with big caps in their boot.

If the core sats you'll loose some of the charge in the cap not all, if the cap holds 100j and the core saturates after 70j has gone through it then the remaining 30j will pretty much be wasted as heat, some of the 30j will be 'transmformed', but most wasted.

Swingin a cat refers to a cat-0-9 tails which is a whip, dont think you'd be doing that in the shed, or would you.

 

[throbscottle]

Ok, I don't mind wasting some energy whilst I'm experimenting. I think you may be right - 450v caps being cheaper than 400v ones. Perhaps I should look out for one of those "discharge capacitors" big ugly things.
I thought as you do, but according to QI, it does refer to an actual cat. And who am I to argue with Mr Fry?

 

[dr pepper]

Well then keep experimenting.
There must be 2 meanings cat swinging then, to me its having enough room to whip someone with the cat-o-9 tails.

 

[throbscottle]

I just found some specs of a commercial unit on fleabay **broken link removed**
Welding power source Aluminum Electrolytic Capacitors
Capacity 66.000 uF
CHARGING VOLTAGE 50 to 200 volts, infinitely adjustable, accuracy +/-1 volt
Max. Charging power 1320 WS at 200 V
Loading times ø3 - ø4: 0,8 s / ø4 - ø5: 1,8 s / ø6: 2,0 s / ø8: 3,5 s
Welding sequence ø3 mm - 40 studs / min. ø8 mm - 10 studs / min.
Charging circuit Electronically: Primary switched, short-circuit protection, temperature controlled
...
Weight 6,5 kg
Dimensions 155 x 185 x 285 mm (WxHxD)
So it charges between 82.5J and 1320J. I suppose from the weight it might be possible to guesstimate the size of the transformer. Maybe 2 - 3 kilos out of that 6.5?

 

[dr pepper]

They cost a few bob, I wonder if they mean 66,000uf instead of 66uf.
I also wonder if the weld is all in one thump, or it goes through an smps to get to the stud.
Hmm 6.5kg, stainless steel box, and comes with leads, I reckon theres just a big cap and a little board in there.

 

[throbscottle]

Yeah if you work it out it's 66,000. I hadn't even noticed it was a dp not comma
Dunno, they output many 100's of amps - can weld M8 stud after all. Can't imagine they'd let there be 200V on the output
I guess It's likely pulsed output since that looks to be pretty standard. Could there be a buck converter (or something like) in there?

 

[throbscottle]

I think it's because it's German - they use a comma instead of a dp. Very confusing when you get invoices from there.

 

[dr pepper]

It would take a few amps to weld an M8 stud, probably around a couple of hundred.
If I was to build something like that I'd use a buck reg, but then I'm not a pro in such things.
Your right I wouldnt think they'd stuff 300v on the output, some youtube numpty no doubt will try shoving it in their gob.