Making square wave AC from DC? (inverting)

Let's say that I have a source of high current 300A 40V DC. Let's say that I want to make square wave AC from it. I am thinking about
building a device on the cheap that could do it.
I could buy 40 power transistors like these:
http://cgi.ebay.com/2N5302-Power-Transistor-30A-60V-NPN-Qty-10_W0QQitemZ7543195186QQcategoryZ4666QQrdZ1QQcmdZViewItem
Item 7543195186
Say I have DC inputs A and B. Say I want square wave AC on outputs C and D. All I need is connect A to C through 10 transistors, A to D through 10 transistors, B to C through 10 transistors, and B to D through 10 transistors.
I would then make appropriate AC input to signal input of transistors so that A to C and B to D are turned on when AC signal input is positive, and A to D and B to C are turned on when AC signal input is the opposite.
That would generate a square wave AC between C and D.
The cost of this project, not counting time expense, would be around $60 or so.
Are there any fatal flaws with it?
Would this be sufficient to use it in welding applications?
i
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

http://cgi.ebay.com/2N5302-Power-Transistor-30A-60V-NPN-Qty-10_W0QQitemZ7543195186QQcategoryZ4666QQrdZ1QQcmdZViewItem
How many MORE transistors do you need for the TIME MACHINE? (I'll go with you... we could go back and slap Cliff's mother and tell Gunner to quit smoking.)
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

We could just strip Cliff of electrons and watch him explode from positively charged particles pushing away from one another...
i
--


Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Ignoramus11916 wrote:

Can you make that Time Machine a three-seater? This is one ride I'd really like to go on..
John
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
make that Time Machine a three-seater? This is one ride I'd

Just to slap Cliff's mother!
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Tom Gardner wrote:

I'll slap her good enough that her kids'll be simple.. no, wait, that's bad.. Aw, dang, this time travel stuff is tough.
John
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Great!!! Now you've done it...or will do it...or something like that. All I know is that you are to blame!
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Tom Gardner wrote:

Sorry.. I'm so sorry about that..
No, wait- new information just in- read the post I brought in from misc.survivalism; "OT Neocons in China"- cliffy's a *clown*!
WooHoo! It wasn't me, he's just a clown! No, no.. he's a *she* clown!
Heh-heh-heh-heh. Heh. Hah.. hahahahahaha..
Whew.. I really feel better now.
Johh
She's a clown!! Hahahaha!
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Not that this really matters, but that would be enough explosive power to destroy the earth.
Interesting trivia: I vaguely remember reading in the Feynman physics lectures that if you took the electrons away from 10% of the atoms in a human body, it would be strong enough to move the earth a few miles...
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
So you plan to connect 10 transistors in parallel to drive one leg of your output waveform? There may well be a problem with thermal runaway. No two physical devices are actually 100% identical. If you have two transistors fully "on" then one's conductance from collector to emitter will be slightly different from the other. In other words, one transistor will conduct slightly more current than the other. All solid state devices will change conductivity as they heat up -- some will become more conductive, some less. If your devices become more conductive as they heat up, then the one that starts out carrying an infinitesimal amount more current can wind up trying to carry it all, and blowing, and then they can all fail zipper-fashion. Google on "thermal runaway".
That's the only thing I can think of. You will have to be a little careful of your triggering circuit.
GWE (once an electronics engineer)
Ignoramus11916 wrote:

http://cgi.ebay.com/2N5302-Power-Transistor-30A-60V-NPN-Qty-10_W0QQitemZ7543195186QQcategoryZ4666QQrdZ1QQcmdZViewItem
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Wed, 07 Sep 2005 20:35:02 -0700, Grant Erwin

A bit of resistance in series with each emitter will make them share nicely. MOSFET's don't need that; they're routinely used in parallel.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Thanks Grant, some deep stuff for me to digest...
i

--


Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Thu, 08 Sep 2005 02:37:38 GMT, Ignoramus11916

Well, maybe a couple of little problems.....
What you are describing is an H-bridge. Bipolar power xstrs like 2N5302 are kinda slow. You could get "shoot thru" if transistors say, from A to C don't turn off before transistors from B to C turn on, etc. POW. (They really do go "POW", sound like a pistol shot)
You're gonna need about 60 amps of base drive (3 amps per transistor) to saturate your transistors.
Vce(sat) on 2N5302 at 30 amps is 3 volts -- which will dissipate 90 watts per transistor, or 1800 watts with 20 transistors conducting. That's just on-state dissipation, doesn't count switching losses. You are gonna need either water cooling or a BIG heatsink and fan.
You can buy IRFZ48V 60-volt 72-amp MOSFET's for $0.84 each (10 or more) from Digi-Key. If you ran these at 30 amps they'd only dissipate about 11 watts per device -- and they are a LOT faster than bipolars so switching losses are considerably lower. They also need much less drive power. You would still need to think your drive scheme thru to be sure you avoid shootthru. There are H-bridge drive chips available that handle hi-side drive and contain anti-shootthru cctry. You could probably run them at 50 amps each, needing only 6 per leg. Dissipation would then be 30 watts per part or 360 watts total.

Not unless your power supply is capable of constant current operation. Even then, 40 volts open circuit would be marginal for stick (SMAW) or TIG (GTAW), which is where you might want squarewave.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
wrote:

OK.
okay, so they are wrong for my application.

That's nice. I am thinking, regular sine wave AC should, according to my uneducated logic, be able to control that nicely. The times when the voltage is below actuating voltage in absolute value, should hopefully allow both sides to become turned off without "shoootthrough".

I am thinking about buying one that should be capable of CC.

40 volts was an example... I am not sure of the capabilities of that welder...
i
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
(I have read 3 replies to your inquiry)
The transistor, 2N5302, is only good for 60 volts. No more. Your DC levels plus your signal level x 2 will far exceed that. So unless you are going all the way with circuit protection your circuit will go up in smoke.
I am not sure where you are getting your transistors. Unless they are a documented source they will be a large part of your problem simply because the chance they are rejects from a quality source.
There are better designs for what you want to do. Use some 600V Triacs or the like. And there are switchers that do this also.
Bob AZ
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Bob, can you give me some detail on that?
i
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Triacs and SCRs don't shut off on a DC source. They need to see 0V across them in order to shut off. There is a process that can shut off a SCR that is designed for that purpose but it ssn't an easy process. Better to use MOSFETs for this job. Bipolar transistors (npm or pnp) transistors have a large (relatively) on voltage across the ends (E to C) and they need to have a fairly large current into them (B terminal) in order to work and the larger devices also tend to be slwo relative to what is wanted. In addition, bipolar transistors have a positive temp coeeficent (the current goes up with temp) so you have to do emmiter resistors Iincreasing the power losses) in order to keep them from being unruly. MOSFETs have a negative temp coeeficent so they will decrease the current in the device when the temp goes up, making paralleling devices easy. The only downside to using MOSFETs is that they require 10-12V to turn them on fully but that requirement isn't difficult to achieve and the current requirements are minimal as the control terminal is basically a cap that needs to be driven to a particular voltage. The basic H bridge is what you are discussing and there are controller chips for doing that, sorry but I don't have any chip numbers for this, which can make your design easy to implement.
-- Why do penguins walk so far to get to their nesting grounds?
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

exactly.
I have some big ass SCRs, but, unfortunately, cannot see how I would use them.

If you can help me find such a controller chip, that would be greatly appreciated. I just do not know what to search for.
I realized that I want to make this system variable frequency, not 60 Hz. I have an HP 204C oscillator, but need to amplify it to drive the mosfets.
i
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Thu, 08 Sep 2005 19:19:34 GMT, Ignoramus965

Most offerings are in surfacemount packages, as http://pdfserv.maxim-ic.com/en/ds/MAX5062-MAX5064.pdf It has dead time for shoot-thru prevention.
http://www.irf.com/product-info/datasheets/data/ir2112.pdf is available in 14DIP, no shoot-thru dead time incorporated.
Other suppliers include Allegro and STM. Look for half-bridge controllers.
These all work well with resistive or inductive loads and voltage source for power, but dealing with an inductive or constant-current power source (as a welder) will introduce some interesting issues. Where does the current go when all switches are off, and how high does the voltage get while it's trying to do it?
This is not as simple as it might look.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
wrote:

Thanks, that's very useful. I saved your post.
i
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.