Re: Phase Control of SCR's; Hi Current

--- In CAD_CAM_EDM_DRO@y..., jmw@c... wrote:

> I know this is a little off-topic, but the talent level of the group
> is so tempting...
>
> Can someone steer me to a circuit for triggering SCR's at a user-
> selectable point in the AC cycle? I've seen circuit snippets in asst
> cookbooks, but not being an ee I'd like to find something a little
> more complete.

Greetings list!

This might be off-topic (I'm de-lurking to reply) but here goes
nothing...

I've done it using a zero-crossing detection scheme to reset an R-C
element. The RC voltage will start to rise once the zero crossing has
passed. If you feed this RC voltage into a comparator which is
comparing it to some other input 'contrl' voltage - you will find that
the time at which the comparator's output 'flips' will be delayed past
the zero crossing in (rough) proportion to the controlling voltage.

In this way the comparator's output can be used to trigger an SCR at a
known selectable phase.

The RC reset is as simple as full wave rectifying the incoming power,
resistor-zenering the voltage to limit it, and using it to control
some small transistor (2N2222 etc) which shorts out the cap in the
main RC timing element. If you are paranoid some opto-coupling is
never a bad idea too!

I think Motorola makes a nice opto-coupled zero crossing detector used
for firing scrs etc. In reality it is a non-zero inhibitor -
precluding the firing of the scr once the mains voltage is 1-2 volts
either side of zero...

>.The rectifiers will be working at 50v 250amps max in

a

> mig welding application; duty cyle at max output would be in the

nbhd

> of 20-30%. Ideally, I'd like to be able to chop out the middle of

the

> AC cycle rather than the initial segment--but I guess SCR's don't
> play that way. Triacs do, right, but not at 250A?

Both triacs and SCRs will stay ON until the current flowing through
the device drops below a certain threshhold value.

A triac acts as two back-to-back scrs. Generally using triacs in
inductive applications (transformer) is a bad idea since the
switch-off dv/dt tends to turn on the other scr (they share the same
silicon die) and power control is lost - followed by device failure.
:(

If you want to extract the center chunk out of the AC waveform - I'm
told that IGBT devices could do that - since they act more like a
conventional transitor - ie - can be turned off before the end of the
cycle... But I haven't tried them yet.

> The block diagrams I've seen for commercial migs using transformers
> with electronic control (instead of multiple taps) use SCR's as
> described above.

Yea I'm pretty sure most of them are operating in 'lamp-dimmer' mode.
That is - they use a trigger circuit which closely resembles that
found in a residential lamp dimmer (open loop phase control).

It is possible to use a residential lamp-dimmer as your phase
adjusting component - but the dimmer will have to in turn fire another
pair of SCRs which pass the main current. Residential dimmers can't
handle inductive loads for more than 16 mS. :)

I'm running a large pair of SCRs in 'lamp-dimmer' phase control mode -
powering a large stick welder. A foot pedal controls the phase and
thus the welding current. A large diode bridge, large filtering
inductor and a $100 TIG torch complete the job! DIY TIG on the cheap!
:)

> But wouldn't it also be possible just to use
> ordinary diodes on the secondary and control the primary side? It
> sounds easier to switch 240VAC at 40A rather than 250A.

Agreed. I found that it is easier to switch the higher voltage side
than the lower. Be sure to properly snub your devices to keep the
turn-off inductive dv/dt from destroying the switching component!

>The
> transformer doesn't mind getting an AC waveform with a bite taken
> out, does it?

Not as long as the transformer doesn't see any long term DC bias. For
long term - I'd suggest using one full period of the incoming power.
If you feed the transformer any DC - it'll saturate the core. This
will cause a large current spike into the transformer which will
likely kill your SCR etc. Ask me how I know this! :)

> One final question. MIGS generally run "constant voltage" volt-amp
> profiles, meaning that the output voltage tends to change relatively
> little even though amperage may change significantly.

Confirmed! Mig power supplys are this way. Stick and TIG are
constant current.

> So far as I

can

> tell by looking at several units, this is accomplised with a big
> honkin' iron core inductor in series with the output. The inductor
> tends to be about 2/3 the size of the power transformer with roughly
> the same number of turns as the transformer secondary. A healthy

bank

> of caps on the order of 100k uf, is also part of the equation.

You may note that within a TIG (which is more of a constant current
power suppply) you won't find the huge bank of capacitors. From my
dealings it is apparent that it is this large cap bank which
contributes to the MIG's constant voltage characteristic. The
inductor tends to smooth out the cap voltage as well as average out
the current swings within the rectifier bank (but not the welding arc)
to some extent.

>(The
> inductor air gap, if any, I don't know about.)

The inductor is certainly special as far as the fact that it carries a
large DC component w/o saturating etc. The gap is important!

> My question is this:
> Can the volt-amp curve be tuned--its slope changed--by phase

control,

> possibly active phase control during operation?

I'm not 100% sure but my Guess would be NO.

The commercial units with programmable slope etc are all high
frequency switchers - a different design than the old style
scr/transformer units.

>This would make the
> design of inductor less critical. My intuition is that there're too
> many df in the problem to conrol with single parameter (trigger
> point.) But maybe not ... I guess a scheme like this would require
> microprocessor control and amps / volts feedback. There are some
> interesting Nuts and Volts articles back in Dec / Jan about
> microcontrollers and SCRs.
>
> Thanks in advance.
>
> --Jack

-Bill
DIY EDM