Maker Pro
Maker Pro

Back again with some questions...(long - yep, you know it)

T

Tim Williams

Jan 1, 1970
0
Induction heater project. It's looking like I need maybe 4 x 50A 600V
highspeed IGBTs, or a 100A single phase module. If someone has some
low-cost suggestions, I'd be glad to hear them.
I still haven't produced the formulas for peak current in terms of supply
voltage, matching inductance and output power. It doesn't help that the
last time I tried reading current waveforms, I got some nasty 27MHz
parasitic shit that will absolutely not go away, despite a ceramic disc at
the probe!!

I also need some chunky generic bits and pieces, like a switch and
contactor. I have on hand a three-pole 30A switch, which is probably
insufficient (I want to make full use of the 240V 50A circuit for my welder,
if possible), and a contactor which is confusingly rated as:
-=-=-
B25 (type I guess; BBC Petercem manufacturer)
I(sub)th = I(sub)e AC1=45A, J(sub)i = 660xx (something scratched out)
IEC 158-1 VDE 0660
....(other ratings)
U(sub)e | 220 | 380 | 500 | 660V~
--------+-----+-----+-----+-------
AC3 | 6,5 | 11 | 11 | 11kW
-- -- -- -- -- -- -- -- -- -- -- --
(symbols, possibly international AC1:
UL/CSA orginizations) 25A
--------------------------------------
(backwards-"R"U) 25A, 600V.a.c.
( symbol ) conductors: 60/75°C
U | 120 | 240 | 480 | 600 | Va.c.
--+-----+-----+-----+-----+---------------
P | 3 | 7,5 | 10 | 10 |hp/3ph/3 poles

aux. cont. (if provided)
10A, 600 Va.c. A600
-=-=-
(Hm, that was longer than I wanted to type..) So it's 45A, but 25A, *but*
10A? WTF?

If 25A in reality, can I effectively parallel two of the four poles for 50A
switching capacity?

-- I'll add a timer circuit to sequence main power and control circuit power
so the full 50A load (plus power supply capacitors!) is never switched on by
any contacts (though it can be turned off at the main power switch).

Speaking of power supply, to get anywhere near 50A RMS, I need a good power
factor. Instead of the several kilojoules of capacitance required for a
smooth DC supply, I was thinking of using just 1000uF or so, which should
give around 50% current duty cycle (for a power factor of say 0.75 or
0.707?). I could also use a PFC input, but that would be a whole other 10kW
inverter in the system and would pretty much double my cost.

Back to the AC line, any thoughts on a breaker? I would like to have a 50A
resettable double pole breaker, fast blow so it goes instead of the breaker
back at the panel. I don't want to hassle with one-shot fuses, since with
my odds they'll be going too often for that. ;-)

What else... about the circuit, I've settled on the same SG3524 PWM circuit,
I'll use seperate drive transformers for each IGBT [pair] so I can generate
a gate waveform of positive and negative voltage (unlike the current
up-zero-down-zero cycle that works well with MOS), adequate to sink the
current to get the devices turning on and off promptly. Since duty cycle
will be near 50%, I can go with a straight square wave and get the requisite
+/-15V drive waveform.

About IGBTs and speed: devices with under 0.5uS total turn-off time are
often spec'd for use well under 20kHz!? My MOSFETs commutate in 1-2uS in
the circuit as-is, plenty fast for me, and they go all the way up to 100kHz!
What am I missing here? Not to mention the only difference between related
classes of e.g. IRG4PC5 - it's like, they're all almost the same speed, yet
some are labeled "WARP", "High speed" and the rest regular. The data sheet
shows different Ic(max) vs. F curves, but the parameters in the data tables
are essentially identical!

About the circuit: for control, I'll drum up frequency, phase, voltage and
current limiting feedback, so that if any of phase, voltage or current rise
above a set, variable value, frequency is increased [further] above
resonance so that all parameters decrease. Of course for frequency, that's
just an open loop control and would set minimum frequency. Reason being,
fixed frequency control is essentially useless, especially as resonance
changes with loading and material. Constant tank voltage is useful for
effectively constant temperature, and constant phase for constant power
output (maximum power output being limited by supply voltage and matching
inductor). I'm not sure what use limiting DC current will have (besides
protecting the transistors), we shall see.

Tim
 
T

Tim Williams

Jan 1, 1970
0
*Stands, tapping foot*

Too long again, or what?

Tim

--
Deep Fryer: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

Tim Williams said:
Induction heater project. It's looking like I need maybe 4 x 50A 600V
highspeed IGBTs, or a 100A single phase module. If someone has some
low-cost suggestions, I'd be glad to hear them.
I still haven't produced the formulas for peak current in terms of supply
voltage, matching inductance and output power. It doesn't help that the
last time I tried reading current waveforms, I got some nasty 27MHz
parasitic shit that will absolutely not go away, despite a ceramic disc at
the probe!!

I also need some chunky generic bits and pieces, like a switch and
contactor. I have on hand a three-pole 30A switch, which is probably
insufficient (I want to make full use of the 240V 50A circuit for my welder,
if possible), and a contactor which is confusingly rated as:
-=-=-
B25 (type I guess; BBC Petercem manufacturer)
I(sub)th = I(sub)e AC1=45A, J(sub)i = 660xx (something scratched out)
IEC 158-1 VDE 0660
...(other ratings)
U(sub)e | 220 | 380 | 500 | 660V~
--------+-----+-----+-----+-------
AC3 | 6,5 | 11 | 11 | 11kW
-- -- -- -- -- -- -- -- -- -- -- --
(symbols, possibly international AC1:
UL/CSA orginizations) 25A
--------------------------------------
(backwards-"R"U) 25A, 600V.a.c.
( symbol ) conductors: 60/75°C
U | 120 | 240 | 480 | 600 | Va.c.
--+-----+-----+-----+-----+---------------
P | 3 | 7,5 | 10 | 10 |hp/3ph/3 poles

aux. cont. (if provided)
10A, 600 Va.c. A600
-=-=-
(Hm, that was longer than I wanted to type..) So it's 45A, but 25A, *but*
10A? WTF?

If 25A in reality, can I effectively parallel two of the four poles for 50A
switching capacity?

-- I'll add a timer circuit to sequence main power and control circuit power
so the full 50A load (plus power supply capacitors!) is never switched on by
any contacts (though it can be turned off at the main power switch).

Speaking of power supply, to get anywhere near 50A RMS, I need a good power
factor. Instead of the several kilojoules of capacitance required for a
smooth DC supply, I was thinking of using just 1000uF or so, which should
give around 50% current duty cycle (for a power factor of say 0.75 or
0.707?). I could also use a PFC input, but that would be a whole other 10kW
inverter in the system and would pretty much double my cost.

Back to the AC line, any thoughts on a breaker? I would like to have a 50A
resettable double pole breaker, fast blow so it goes instead of the breaker
back at the panel. I don't want to hassle with one-shot fuses, since with
my odds they'll be going too often for that. ;-)

What else... about the circuit, I've settled on the same SG3524 PWM circuit,
I'll use seperate drive transformers for each IGBT [pair] so I can generate
a gate waveform of positive and negative voltage (unlike the current
up-zero-down-zero cycle that works well with MOS), adequate to sink the
current to get the devices turning on and off promptly. Since duty cycle
will be near 50%, I can go with a straight square wave and get the requisite
+/-15V drive waveform.

About IGBTs and speed: devices with under 0.5uS total turn-off time are
often spec'd for use well under 20kHz!? My MOSFETs commutate in 1-2uS in
the circuit as-is, plenty fast for me, and they go all the way up to 100kHz!
What am I missing here? Not to mention the only difference between related
classes of e.g. IRG4PC5 - it's like, they're all almost the same speed, yet
some are labeled "WARP", "High speed" and the rest regular. The data sheet
shows different Ic(max) vs. F curves, but the parameters in the data tables
are essentially identical!

About the circuit: for control, I'll drum up frequency, phase, voltage and
current limiting feedback, so that if any of phase, voltage or current rise
above a set, variable value, frequency is increased [further] above
resonance so that all parameters decrease. Of course for frequency, that's
just an open loop control and would set minimum frequency. Reason being,
fixed frequency control is essentially useless, especially as resonance
changes with loading and material. Constant tank voltage is useful for
effectively constant temperature, and constant phase for constant power
output (maximum power output being limited by supply voltage and matching
inductor). I'm not sure what use limiting DC current will have (besides
protecting the transistors), we shall see.

Tim
 
T

Terry Given

Jan 1, 1970
0
Tim said:
Induction heater project. It's looking like I need maybe 4 x 50A 600V
highspeed IGBTs, or a 100A single phase module. If someone has some
low-cost suggestions, I'd be glad to hear them.

have you looked at any of the Semikron stuff?
I still haven't produced the formulas for peak current in terms of supply
voltage, matching inductance and output power. It doesn't help that the
last time I tried reading current waveforms, I got some nasty 27MHz
parasitic shit that will absolutely not go away, despite a ceramic disc at
the probe!!

I also need some chunky generic bits and pieces, like a switch and
contactor. I have on hand a three-pole 30A switch, which is probably
insufficient (I want to make full use of the 240V 50A circuit for my welder,
if possible), and a contactor which is confusingly rated as:
-=-=-
B25 (type I guess; BBC Petercem manufacturer)
I(sub)th = I(sub)e AC1=45A, J(sub)i = 660xx (something scratched out)
IEC 158-1 VDE 0660
...(other ratings)
U(sub)e | 220 | 380 | 500 | 660V~
--------+-----+-----+-----+-------
AC3 | 6,5 | 11 | 11 | 11kW
-- -- -- -- -- -- -- -- -- -- -- --
(symbols, possibly international AC1:
UL/CSA orginizations) 25A
--------------------------------------
(backwards-"R"U) 25A, 600V.a.c.
( symbol ) conductors: 60/75°C
U | 120 | 240 | 480 | 600 | Va.c.
--+-----+-----+-----+-----+---------------
P | 3 | 7,5 | 10 | 10 |hp/3ph/3 poles

aux. cont. (if provided)
10A, 600 Va.c. A600
-=-=-
(Hm, that was longer than I wanted to type..) So it's 45A, but 25A, *but*
10A? WTF?

If 25A in reality, can I effectively parallel two of the four poles for 50A
switching capacity?
yep


-- I'll add a timer circuit to sequence main power and control circuit power
so the full 50A load (plus power supply capacitors!) is never switched on by
any contacts (though it can be turned off at the main power switch).

contactors are usually rated for opening inductive loads at rated
current, umpteen times - its the splat that eventually snots them. If
you sequence it & open/close at no current, they'll pretty much last
forever.
Speaking of power supply, to get anywhere near 50A RMS, I need a good power
factor. Instead of the several kilojoules of capacitance required for a
smooth DC supply, I was thinking of using just 1000uF or so, which should
give around 50% current duty cycle (for a power factor of say 0.75 or
0.707?). I could also use a PFC input, but that would be a whole other 10kW
inverter in the system and would pretty much double my cost.

<smartass>
its easier to get 50Arms with a ratshit power factor and humongous crest
factor
Back to the AC line, any thoughts on a breaker? I would like to have a 50A
resettable double pole breaker, fast blow so it goes instead of the breaker
back at the panel. I don't want to hassle with one-shot fuses, since with
my odds they'll be going too often for that. ;-)

try this:

5 litre plastic bucket. 2 ceramic standoffs, about 5" long protruding
down into bucket, filled with de-ionised water. Use 5A fuse wire. The
water keeps it cool, but at a critical current the water vapourises, the
thermal impedance skyrockets and the fuse blows. IIRC 5A wire blows at
about 300A or so - a buddy of mide used this trick for his PhD, as he
had no budget for semiconductor fuses.
What else... about the circuit, I've settled on the same SG3524 PWM circuit,
I'll use seperate drive transformers for each IGBT [pair] so I can generate
a gate waveform of positive and negative voltage (unlike the current
up-zero-down-zero cycle that works well with MOS), adequate to sink the
current to get the devices turning on and off promptly. Since duty cycle
will be near 50%, I can go with a straight square wave and get the requisite
+/-15V drive waveform.

good call. -ve gate bias helps deal with Cmiller
About IGBTs and speed: devices with under 0.5uS total turn-off time are
often spec'd for use well under 20kHz!? My MOSFETs commutate in 1-2uS in
the circuit as-is, plenty fast for me, and they go all the way up to 100kHz!
What am I missing here? Not to mention the only difference between related
classes of e.g. IRG4PC5 - it's like, they're all almost the same speed, yet
some are labeled "WARP", "High speed" and the rest regular. The data sheet
shows different Ic(max) vs. F curves, but the parameters in the data tables
are essentially identical!

AIUI you have either ZVS or ZCS (cant recall which, ZVS works best for
FETs, ZCS for IGBTs) in which case the datasheet is less than helpful.
suck it and see.
About the circuit: for control, I'll drum up frequency, phase, voltage and
current limiting feedback, so that if any of phase, voltage or current rise
above a set, variable value, frequency is increased [further] above
resonance so that all parameters decrease. Of course for frequency, that's
just an open loop control and would set minimum frequency. Reason being,
fixed frequency control is essentially useless, especially as resonance
changes with loading and material. Constant tank voltage is useful for
effectively constant temperature, and constant phase for constant power
output (maximum power output being limited by supply voltage and matching
inductor). I'm not sure what use limiting DC current will have (besides
protecting the transistors), we shall see.

peak current limit & trip is always a good idea.

Cheers
Terry
 
T

Tim Williams

Jan 1, 1970
0
Terry Given said:
have you looked at any of the Semikron stuff?

Not yet, just been browsing the IR and whatnot from the usuals (Digikey,
Mouser).

And breaking at the same exact time isn't a problem or anything?
contactors are usually rated for opening inductive loads at rated
current, umpteen times - its the splat that eventually snots them. If
you sequence it & open/close at no current, they'll pretty much last
forever.

Okay, so a snubber across it (if I were switching inductive) would help a
lot, besides the turn-on spike, no?

But since I'm not, then I can switch the full 45A, which is close enough to
my 50A requirement to be okay?
(Come to think of it, an electric current limit to 40-45A would be a good
idea to keep the breaker from tripping. Heck, I could put an anti-SCR
(..since I don't have a GTO on hand..) on the contactor so it latches when
current exceeds 40 or 50A for a few cycles. That would be nice, though
perhaps less fool-proof than a mechanical device.)
<smartass>
its easier to get 50Arms with a ratshit power factor and humongous
crest factor
</smartass>

Which is why I'd rather not go with smoothed DC, cap input filter...
5 litre plastic bucket. 2 ceramic standoffs, about 5" long protruding
down into bucket, filled with de-ionised water. Use 5A fuse wire. The
water keeps it cool, but at a critical current the water vapourises,
the thermal impedance skyrockets and the fuse blows. IIRC 5A wire blows
at about 300A or so - a buddy of mide used this trick for his PhD,
as he had no budget for semiconductor fuses.

Interesting, seems rather cheezy but if it's good enough for a doctorate I
suppose that'd work. Not replacable though, has to be opened.

By the way, how fast are fuses of the outdated "fuse panel" type?
Cylinder and plug types are still quite available.

Hmm. I just realized that a). I have absolutely no idea where I would find
such information, and b). I can't find anything with a few keywords online.
AIUI you have either ZVS or ZCS (cant recall which, ZVS works best for
FETs, ZCS for IGBTs) in which case the datasheet is less than helpful.
suck it and see.

ZCS, ZVS at turn-on. Voltage and current go wild on switching (flyback
towards the opposing diode). You know, half bridge switching mostly
inductance.

Tim
 
T

Terry Given

Jan 1, 1970
0
Tim said:
Not yet, just been browsing the IR and whatnot from the usuals (Digikey,
Mouser).




And breaking at the same exact time isn't a problem or anything?

I've seen plenty of contactors wired up this way.
Okay, so a snubber across it (if I were switching inductive) would help a
lot, besides the turn-on spike, no?

in theory yes. In practice, not really. think current density, rough
surface.
But since I'm not, then I can switch the full 45A, which is close enough to
my 50A requirement to be okay?
(Come to think of it, an electric current limit to 40-45A would be a good
idea to keep the breaker from tripping. Heck, I could put an anti-SCR
(..since I don't have a GTO on hand..) on the contactor so it latches when
current exceeds 40 or 50A for a few cycles. That would be nice, though
perhaps less fool-proof than a mechanical device.)




Which is why I'd rather not go with smoothed DC, cap input filter...

that measn of course you get 100Hz (120Hz) modulation on your load, but
your load is an integrator so who cares.
Interesting, seems rather cheezy but if it's good enough for a doctorate I
suppose that'd work. Not replacable though, has to be opened.

but just the bees knees to get the system up and running, during which
time you will blow a *lot* of fuses :) When its done, use HRC fuses.
By the way, how fast are fuses of the outdated "fuse panel" type?
Cylinder and plug types are still quite available.

slow, very slow or astonishingly slow.

even semiconductor fuses are slow. They are really there to prevent
catastrophic damage, never to protect semiconductors (I^2t invariably
ensures they cannot).
Hmm. I just realized that a). I have absolutely no idea where I would find
such information, and b). I can't find anything with a few keywords online.

Ferraz-Shawmut (or whomever they morphed into) would be a good start.
ZCS, ZVS at turn-on. Voltage and current go wild on switching (flyback
towards the opposing diode). You know, half bridge switching mostly
inductance.

Tim

so give it a whiz, and see which IGBTs run cooler. 2 basic types - PT
and NPT. One has low Vcesat, high switching losses, the other
vice-versa. Fred C. Lee et al at VPEC have written a whole bunch of
papers on soft-switching IGBTs, they are on-line if you look hard
enough. Sorry, I'm too lazy oopps I mean busy to dig up a link for you :)

Cheers
Terry
 
T

Tim Williams

Jan 1, 1970
0
Terry Given said:
I've seen plenty of contactors wired up this way.

Fair enough...
in theory yes. In practice, not really. think current density, rough
surface.

Oh, so it's the amps (which would need a series inductor instead of a
parallel cap) that kills it? Gotcha.
that measn of course you get 100Hz (120Hz) modulation on your load, but
your load is an integrator so who cares.

Yeah, plus the growl will let you know it's in progress. <g>
Likewise for the control loop I'll give that a 100ms or so time constant,
except current protection, which is going to want peak protection in the 1ms
area I'm guessing? (Probably rhetorical Q, I can plug a datasheet for
that...)
but just the bees knees to get the system up and running, during which
time you will blow a *lot* of fuses :) When its done, use HRC fuses.

True... ;)
slow, very slow or astonishingly slow.

even semiconductor fuses are slow. They are really there to prevent
catastrophic damage, never to protect semiconductors (I^2t invariably
ensures they cannot).

Good point.
so give it a whiz,

I am open to any $50 donations you may have on hand. ;-)
and see which IGBTs run cooler. 2 basic types - PT
and NPT. One has low Vcesat, high switching losses, the other
vice-versa. Fred C. Lee et al at VPEC have written a whole bunch of
papers on soft-switching IGBTs, they are on-line if you look hard
enough.

Alrighty...

Tim
 
T

Terry Given

Jan 1, 1970
0
Tim said:
Fair enough...




Oh, so it's the amps (which would need a series inductor instead of a
parallel cap) that kills it? Gotcha.




Yeah, plus the growl will let you know it's in progress. <g>
Likewise for the control loop I'll give that a 100ms or so time constant,
except current protection, which is going to want peak protection in the 1ms
area I'm guessing? (Probably rhetorical Q, I can plug a datasheet for
that...)

10us for a desat. Do you have desat detection? if not, stick it in. A
good gatedrive will allow you to bolt a flat copper strip directly
across the C-E terminals, and *not* shit itself.

I've always used 3 protection circuits:

current limit - turn IGBT off for some minimum time: 50-100us or so,
usually done digitally to ensure a minimum off time. that way, in hard
current limit the switching frequency doesnt get out of control.

current trip - shut down, user-restart required

desat trip - ditto.
True... ;)




Good point.

actually, whopping great rectifiers often have enough IIt that you can
protect them if you are *very* careful
I am open to any $50 donations you may have on hand. ;-)

My fiance beat you to it :)
Alrighty...

Tim

Cheers
Terry
 
T

Tim Williams

Jan 1, 1970
0
Terry Given said:
10us for a desat. Do you have desat detection?

Oh yeah, forgot that! Lemme go toss in another quarter LM339 on the
drawing...
if not, stick it in. A
good gatedrive will allow you to bolt a flat copper strip directly
across the C-E terminals, and *not* shit itself.

A copper strip bolted to the C-E terminals is fine with me... as long as
it's applied only when that transistor is "ON" and the bolts are removed
within the ten microseconds before the other transistor turns on. said:
I've always used 3 protection circuits:

current limit - turn IGBT off for some minimum time: 50-100us or so,
usually done digitally to ensure a minimum off time. that way, in hard
current limit the switching frequency doesnt get out of control.

Since I'm going to use this variable up to a maximum (of dubious use, we'll
see), I can just wire this to increase frequency - assuming something isn't
wildly wrong with the output section (like I said, it's seperated from the
outside world with a honkin' choke).
current trip - shut down, user-restart required

desat trip - ditto.

Yeah, like I said - SCR on the shutdown pin on the IC, or if not that then
shut down the power control circuit (which controls the relays - a bit slow
so a chipside electronic solution will still be required).

Tim
 
T

Terry Given

Jan 1, 1970
0
Tim said:
Oh yeah, forgot that! Lemme go toss in another quarter LM339 on the
drawing...




A copper strip bolted to the C-E terminals is fine with me... as long as
it's applied only when that transistor is "ON" and the bolts are removed



Since I'm going to use this variable up to a maximum (of dubious use, we'll
see), I can just wire this to increase frequency - assuming something isn't
wildly wrong with the output section (like I said, it's seperated from the
outside world with a honkin' choke).

sounds good. The desat trip will catch any serious faults - shorted
choke, shorted IGBT
 
Top