[Phil Allison[_3_]]

** Hi tubeheads,

these JPEGs were taken late in 2009 using my Rigol DSO.

The first is a 1kVA step down ( 240V to 120V) E -Core transformer with no
load.

http://sound.au.com/tmp/surge-002.jpg

The second is a commercial powered speaker with in built 200 watt SS
amplifier using a 300VA toroidal tranny.

http://sound.au.com/tmp/surge-001.jpg

The smaller current peaks are due to filter electros charging.

The moment of switch on was close to a zero crossing of the AC supply -
with the voltage going positive.

See the captions.

Note how the transformer current surges are all the same polarity and peak
as the AC voltage is crossing zero.



.... Phil

[Alex Pogossov]

"Phil Allison" wrote in message
...

** Hi tubeheads,

these JPEGs were taken late in 2009 using my Rigol DSO.

The first is a 1kVA step down ( 240V to 120V) E -Core transformer with no
load.

http://sound.au.com/tmp/surge-002.jpg

The second is a commercial powered speaker with in built 200 watt SS
amplifier using a 300VA toroidal tranny.

http://sound.au.com/tmp/surge-001.jpg

The smaller current peaks are due to filter electros charging.

The moment of switch on was close to a zero crossing of the AC supply -
with the voltage going positive.

See the captions.

Note how the transformer current surges are all the same polarity and peak
as the AC voltage is crossing zero.

Apparently, the huge curent surges are due to the core saturation. To avoid
them one must turn power switch on the peak(!) of the sinewave, not at the
zero-crossing. But then we will have possibly huge surges from charging of
the filter caps. However the later are easier to control by adding small R
or R+L in series with the diodes.

So which is the lesser evil?

[Phil Allison[_3_]]

"Alex Pogossov"
"Phil Allison"

** Hi tubeheads,

these JPEGs were taken late in 2009 using my Rigol DSO.

The first is a 1kVA step down ( 240V to 120V) E -Core transformer with no
load.

http://sound.au.com/tmp/surge-002.jpg

The second is a commercial powered speaker with in built 200 watt SS
amplifier using a 300VA toroidal tranny.

http://sound.au.com/tmp/surge-001.jpg

The smaller current peaks are due to filter electros charging.

The moment of switch on was close to a zero crossing of the AC supply -
with the voltage going positive.

See the captions.

Note how the transformer current surges are all the same polarity and
peak as the AC voltage is crossing zero.

Apparently, the huge curent surges are due to the core saturation.


** Just like pulling the iron core half out of the transformer for a few mS.


To avoid them one must turn power switch on the peak(!) of the sinewave,
not at the zero-crossing.

** Bit of a special skill, that one.

But then we will have possibly huge surges from charging of the filter
caps.

** Yep - but rather less peak current for a bit longer and on each half
cycle.

Cos the secondary copper losses are then involved, as well as the primary
ones.


However the later are easier to control by adding small R or R+L in series
with the diodes.


** There is no issue with silicon diodes or elector caps during start up -
both can easily cope with huge surges.

The problem is with the AC supply fuse !!!

Cos it forces you to use a much bigger rating fuse than would reliably
protect the tranny from burn out.



..... Phil

[Patrick Turner]

On Jun 4, 4:45*pm, "Phil Allison" wrote:
** Hi tubeheads,

these JPEGs were taken late in 2009 using my Rigol DSO.

The first is a 1kVA step down ( 240V to 120V) E -Core transformer with no
load.

http://sound.au.com/tmp/surge-002.jpg

The second is a commercial powered speaker with in built 200 watt SS
amplifier using a 300VA toroidal tranny.

http://sound.au.com/tmp/surge-001.jpg

The smaller current peaks are due to filter electros charging.

The moment of switch on was close to a zero crossing of the AC supply -
with the voltage going positive.

See the captions.

Note how the transformer current surges are all the same polarity and peak
as the AC voltage is crossing zero.

... *Phil

OK, if you are getting 115 peak amps at tranny turn on and mains
source resistance = 1.1 ohms, then at the max peak charge current
there is 126.5 Vpk across the source resistance. The winding input
*impedance* at the max peak current point would appear to be very low.
So what is the voltage being applied to the tranny, allowing for phase
shift? less than 340V pk probably. Your test tranny must have very low
winding resistance, and with the absence of any magnetic field the
voltage sees a short circuit at the instant of turn on until the
magnetic field establishes itself, no? The tranny winding resistance
in models appears in series with a primary winding.

If one had a 5 ohm resistance in series with the tranny, then would
not the peak current 340Vpk from the source be much limited? If
primary coil input Z = 2 ohms, then total Z + R might be 6 ohms with
added R and peak I max = 340/6 = 56A. ARC VT100 have 5 ohms
resistance with a relay to shunt the resistance after turn on. But
I've found sucha small R ineffective to limit the longer lasting
inrush current to rail capacitors. I have one amp with +/- 70V rails
and a total of 200,000uF so inrush to charge C up will blow mains
fuses unless the fuse value is high, maybe 10 A for 240V input.
Hence my prefered use of perhaps 100 ohms with relay shunting after 4
seconds when rails have come up to about 2/3 full value. I guess the
100 ohms cops about 3 amps at the first cycle of 240Vrms applied. The
instant power in such an R = 900 Watts, but only for less than 0.02
seconds. I've used 20W rated R with no troubles.
Even if one has not calculated or measured all the input currents one
soon finds the R value and type which won't fuse open and which will
last and which will allow
a smaller and more useful fuse value offering more sensitive
protection to excessive current draw from the mains in a fault
condition.

Patrick Turner.

[Phil Allison[_3_]]

"Patrick Turner"

Even if one has not calculated or measured all the input currents one
soon finds the R value and type which won't fuse open and which will
last and which will allow a smaller and more useful fuse value
offering more sensitive protection to excessive current draw from
the mains in a fault condition.


** That 115 amp surge from the step down tranny is not big enough to blow a
15 amp wire fuse or trip a 16 amp ( thermal magnetic) breaker as fitted to
domestic AC supply circuits here in Australia. But it IS big enough to
blow a 4 or 5 amp slo-blo glass fuse as would be needed to protect the
tranny.

Adding a permanent series resistor is not practical - it would dissipate far
to much heat.

The obvious and simple answer is to use a thermal breaker, rated at 4 or 5
amps - these have long opening times and will ignore most any half cycle
surge, eg:

http://au.element14.com/te-connectiv...-5a/dp/9659080

If it ever does trip, just push to re-set.

Avoids the possibility of users fitting an oversize fuse too.


...... Phil

[Alex Pogossov]

"Phil Allison" wrote in message
...

"Alex Pogossov"
"Phil Allison"

** Hi tubeheads,

these JPEGs were taken late in 2009 using my Rigol DSO.

The first is a 1kVA step down ( 240V to 120V) E -Core transformer with
no load.

http://sound.au.com/tmp/surge-002.jpg

The second is a commercial powered speaker with in built 200 watt SS
amplifier using a 300VA toroidal tranny.

http://sound.au.com/tmp/surge-001.jpg

The smaller current peaks are due to filter electros charging.

The moment of switch on was close to a zero crossing of the AC supply -
with the voltage going positive.

See the captions.

Note how the transformer current surges are all the same polarity and
peak as the AC voltage is crossing zero.

Apparently, the huge curent surges are due to the core saturation.


** Just like pulling the iron core half out of the transformer for a few
mS.


To avoid them one must turn power switch on the peak(!) of the sinewave,
not at the zero-crossing.

** Bit of a special skill, that one.

But then we will have possibly huge surges from charging of the filter
caps.

** Yep - but rather less peak current for a bit longer and on each half
cycle.

Cos the secondary copper losses are then involved, as well as the primary
ones.


However the later are easier to control by adding small R or R+L in
series with the diodes.


** There is no issue with silicon diodes or elector caps during start
p - both can easily cope with huge surges.

The problem is with the AC supply fuse !!!

Cos it forces you to use a much bigger rating fuse than would reliably
protect the tranny from burn out.

I can see two ways of killing of the inrush currents.
1. To avoid saturation on turn-on at zero-crossing a transformer shall be so
designed that the core would not reach saturation after one half-sine wave,
not after a quarter. In other words, the transformer primary shall have
double of turns per volt -- inrushless transformer shall be designed say for
240Vac to work in the US 120V environment or to 460Vac to work in Europe. Of
course this will make the beast inefficient, twice heavier. larger and
costlier for the same wattage. Bean counters would scream!

Since you are keen on experimenting, take two identical trannies, connect
theis primaries in series (to simulate one tranny with double turns/volt)
and do the in-rush test as before. You will see NO INRUSH at all! Not just
it will be halved. No. Not at all!

Hobbyists perhaps could afford designs where two identical trannies with
primaries and secondaries in series are used instead one tranny. But it will
be inrush free!

2. How to deal with capacitor charge in-rush. Well, Patric uses a delay
relay, others -- thermistors, SS current limiters, just series resistors,
etc...

A real answer is to reduce C in the filters. You guys seem to use hundreds
and even thousands of uFs in staggered RC or LC filters in the tube
amplifiers. It is crasy.

In fact, you need only one big cap of a value of about 1uF per 1mA of peak
load current. It will keep ripple at about 10V. You can feed output plates
from it. 90% of the home brewed amps can do away with 100uF or 220uF at
worst.

For all the other loads use active filters based on large R, relatively
small C and a Hi-V darlington of MOSFET. These can be taken from dumped TVs.
Typically they are in the nice insulated TO-263s and can be bolted to
chassis. (Note: transisors from the line deflection usually have 50R
built-in resistor between base and emitter. Use a lower power transistor in
front.)

This is another instance where transistors could be put to a good use in
valve designs, not just for making current sources (as Partick does).

Regards,
Alex

[Patrick Turner]

On Jun 5, 7:57*pm, "Alex Pogossov" wrote:
"Phil Allison" wrote in message

...







"Alex Pogossov"
"Phil Allison"

** Hi tubeheads,

these JPEGs were taken late in 2009 using my Rigol DSO.

The first is a 1kVA step down ( 240V to 120V) E -Core transformer with
no load.

http://sound.au.com/tmp/surge-002.jpg

The second is a commercial powered speaker with in built 200 watt SS
amplifier using a 300VA toroidal tranny.

http://sound.au.com/tmp/surge-001.jpg

The smaller current peaks are due to filter electros charging.

The moment of switch on was close to a zero crossing of the AC supply -
with the voltage going positive.

See the captions.

Note how the transformer current surges are all the same polarity and
peak as the AC voltage is crossing zero.

Apparently, the huge curent surges are due to the core saturation.

** Just like pulling the iron core half out of the transformer for a few
mS.

To avoid them one must turn power switch on the peak(!) of the sinewave,
not at the zero-crossing.

** Bit of a special skill, *that one.

But then we will have possibly huge surges from charging of the filter
caps.

** Yep *- *but rather less peak current for a bit longer and on each half
cycle.

Cos the secondary copper losses are then involved, as well as the primary
ones.

However the later are easier to control by adding small R or R+L in
series with the diodes.

** There is no issue with silicon diodes or elector caps during start
p *- both can easily cope with huge surges.

The problem is with the AC supply fuse *!!!

Cos it forces you to use a much bigger rating fuse than would reliably
protect the tranny from burn out.

I can see two ways of killing of the inrush currents.
1. To avoid saturation on turn-on at zero-crossing a transformer shall be so
designed that the core would not reach saturation after one half-sine wave,
not after a quarter. In other words, the transformer primary shall have
double of turns per volt -- inrushless transformer shall be designed say for
240Vac to work in the US 120V environment or to 460Vac to work in Europe. Of
course this will make the beast inefficient, twice heavier. larger and
costlier for the same wattage. Bean counters would scream!

Since you are keen on experimenting, take two identical trannies, connect
theis primaries in series (to simulate one tranny with double turns/volt)
and do the in-rush test as before. You will see NO INRUSH at all! Not just
it will be halved. No. Not at all!

Hobbyists perhaps could afford designs where two identical trannies with
primaries and secondaries in series are used instead one tranny. But it will
be inrush free!

2. How to deal with capacitor charge in-rush. Well, Patric uses a delay
relay, others -- thermistors, SS current limiters, just series resistors,
etc...

A real answer is to reduce C in the filters. You guys seem to use hundreds
and even thousands of uFs in staggered RC or LC filters in the tube
amplifiers. It is crasy.

In fact, you need only one big cap of a value of about 1uF per 1mA of peak
load current. It will keep ripple at about 10V. You can feed output plates
from it. 90% of the home brewed amps can do away with 100uF or 220uF at
worst.

I like to use 470uF caps in PSUs because they are small, and do a fine
job to reduce Vripple at a CT to negligible values. ARC uses far more
C than I do, but anyway, in class AB there is no CMRR in the OP stage
and in each class B wave cycle you have Vripple in series with OPT
load and anode signal so unless the amp works in pure class A you need
the CT voltage to be clean.

The CT must also be bypassed to 0V with a low impedance to allow good
AB operation, and what better than about 470uF which is 34 ohms at
10Hz. Its easy to allow for inrush currents.


For all the other loads use active filters based on large R, relatively
small C and a Hi-V darlington of MOSFET. These can be taken from dumped TVs.

I've found source follower used for B+ rails to be fragile. Too many
fail mysteriously after awhile. Better are more rugged BU208 or BU108
HV rated bjts, but they have very low hfe so you need to dalington
connect them and put all sorts of diodes and series resistances in to
prevent excessive currents into base and through collectors. I try not
to use any active regulation in B+ rails for input/driver stages and
just rely on good old RC filters. I don't have problems with any
motorboating, ie, LF oscillations.

Patrick Turner.


Typically they are in the nice insulated TO-263s and can be bolted to
chassis. (Note: transisors from the line deflection usually have 50R
built-in resistor between base and emitter. Use a lower power transistor in
front.)

This is another instance where transistors could be put to a good use in
valve designs, not just for making current sources (as Partick does).

Regards,
Alex- Hide quoted text -

- Show quoted text -

[Patrick Turner]

On Jun 5, 1:57*pm, "Phil Allison" wrote:
"Patrick Turner"

Even if one has not calculated or measured all the input currents one
soon finds the R value and type which won't fuse open and which will
last and which will allow a smaller and more useful fuse value
offering more sensitive protection to excessive current draw from
the mains in a fault condition.

*** That 115 amp surge from the step down tranny is not big enough to blow a
15 amp wire fuse or trip a 16 amp ( thermal magnetic) breaker as fitted to
domestic AC supply circuits here in Australia. *But it IS *big enough to
blow a 4 or 5 amp slo-blo glass fuse as would be needed to protect the
tranny.

Adding a permanent series resistor is not practical - it would dissipate far
to much heat.

The obvious and simple answer is to use a thermal breaker, rated at 4 or 5
amps *- *these have long opening times and will ignore most any half cycle
surge, eg:

http://au.element14.com/te-connectiv...d/w28-xq1a-5/c...

If it ever does trip, just push to re-set.

Avoids the possibility of users fitting an oversize fuse too.

..... *Phil

The full link is
http://au.element14.com/te-connectiv..._merc h=true&

Not a bad price, and not a bad device to be sure, which will fit
somewhere easily on any amp.

I didn't suggest a permanent series resistor ever be used to lower the
input current you mentioned. The consequent continuous current would
indeed fry such a resistance, which I like to shunt with a timed relay
after 4 seconds when the rails have come up to nearly the full B+/B-
value. The series R I've used seems to survive the high turn on surge
for the first few 50Hz cycles and the charge up currents needed for
massive capacitors. Once the caps are "full" the amp will rarely
produce huge mains current draw unless there is a serious fault. Where
multiple output tubes are used, complete bias failure in one output
tube may not increase mains draw enough to blow even a sensitive fuse
- especially if you have 12 x output tubes like in my 300W amps. Its
in cases like such amps where mains fuses have limited abilities and
responsibilities so one must use fuses in each cathode to 0V circuit,
or have active sensing so that if one or more tubes conducts more than
3 times the Idc for longer than 4 seconds the amp is turned off by a
relay interupting the mains supply to the primary. I fit such
protection even when only 2 output tubes are used. Such protection
saves OPTs and PTs from ever being over heated for some length of time
by a tube which has become saturated from bias failure.

A bright spark like yourself could give thought to designing a circuit
module which monitors amplifier output voltage and current and applies
the info to a chip based circuit which determines if the load
connected is below a setable value of say 2 ohms. If RL 2 ohms is
detected, even at low level, even for say 0.1 seconds, then the amp is
turned off by a relay and the red fault LED turned on. This would
protect all amps against shorted speaker cables, arcing ESLs and so
on, all of which cause many amp failures.
Such a protection measure is exactly what is missing from most
protection circuits which dismally fail to protect an amp against a
low RL at very low signal levels when excessive device currents are
not large enough to make any fuse blow but are large enough to
overheat and destroy the devices within seconds or minutes.

I would suggest it could be lucrative.

Patrick Turner.