Post by Phil Allison"Some Fool Wrote"
Post by Patrick TurnerIn cap filter input cases where silicon diodes and large cap values are
involved, its the
caps that cause the most inrush current and hence nuisance fuse blowings, not
the magnetic effects in the core.
** The *max* primary switch on surge current peak IS due to transformer
core saturation. Its value in amps closely approaches to the ratio of AC
voltage peak to primary ohms.
Summ-fule replies...
Well, gee, this could seem problematic.
Say I have a primary designed for 480VA input, 240V so I in working is 2A,
and load looking into the tranny is 120 ohms.
If I make the winding resistance so current density is 2A/sq.mm,
the Rw of the primary could be say 4% of the input load of 120 ohms,
so the winding has 4.8 ohms of resistance.
So with a peak voltage of
340V applied to a primary of say 4.8 ohms, there would be theoretical
peak current of about 70 amps.
if you are saying that the primary ohms is the 120 ohm load, then the
peak current would be 340/120 = nearly 3A.
Post by Phil AllisonEG : for a 500VA toroidal transformer operating from 240 volts AC with
2.8 ohms primary resistance plus 0.8 ohms from AC supply - the max peak
current value seen is +/- 90 amps.
Well now I see what you were saying.
But I have never seen such high input currents, even when I took measurements
with a peak and hold
volt meter across a 1 ohm series R to find out how high the peak current went,
and what sort of fuse i would have to select to be practical for an owner to
source,
and still be able to offer some protection against causing a house fire due to
flames from a transformer that wasn't fused properly.
Its serious stuff, this issue; I had to rewind a PT in an EAR509 amplifier which
I found had a 6.3A fuse where there should have been a 3 A fuse. The owner went
to get beer
at the shop and returned to a house full of smoke and a well fucked amp due to
bias failure
and imolation of all the internals.
Post by Phil AllisonWhen this *max* peak current value is flowing in the primary, the voltage
drop ( IR) loss is nearly 100% of the supply so there is almost no secondary
voltage developed. Filter caps on the secondary side simply do no come into
play at this time.
How come my 8A circuit breakers don't trip when i turn on large amps on my
bench?
I deliberately installed faily low amperage rated circuit breakers
to make sure they trip easily if a fuse in the gear isn't quite right.
If I have a 1kW heater going in winter, and then plug some other gear in then
maybe
the circuit breaker trips, but rarely.
Post by Phil AllisonAlso, if the AC supply is briefly interrupted during operation and filter
caps do not discharge, the same primary surge due to core saturation will
always happen.
The surge must be for a very short time....
Post by Phil AllisonWith SS diodes and large value un-charged electros, two kinds of surge go on
1. The core saturation surge, with current pulses of one polarity, half sine
in shape and decaying exponentially in 5 to 50 cycles. The magnitude and
duration depends on the actual point of the AC voltage wave at switch on.
I'l have to search with Google for a worst case wave form record of typical turn
on behaviour.
I ain't saying you are wrong, I but I ain't seen evidence of the huge turn
current.
Post by Phil Allison2. The electro charging surge, with current pulses of both polarities, half
sine in shape, that rises to a maximum value in a cycle or two and then
falls back to a steady value after about 5 to 10 cycles.
Maybe its more than 10 cycles, or 0.2 seconds.
Seems like the B+ rise takes longer than 0.2 seconds.
Post by Phil AllisonIn case 1, peak current limiting is provided by the primary resistance
**alone** !!
But what about the house wiring and lead to the amp, and perhaps
a series R or thermistor?
I used to like thermistors, but after blowing a few up I then switched my favour
to a 25 watt series R and a relay to shunt it which is operated by a 4 second
delay,
and if the power is turned off then on again within 1 second, the delay cycle of
4 seconds
start all over again.
I found that fuse values could be thus 1/2 what would otherwise have to be used.
Post by Phil AllisonIn case 2, the peak current is limited by primary and secondary
resistances, diode resistance and electro impedance all acting in series.
So, in case 2, the max peak current drawn from the AC supply is invariably
smaller - ie less than half.
Yes, the max charging current with say 1,000 uF and silicon diodes
would be less than 1/2 your values for max peak current estimates of magnetic
saturation
currents.
The max peak charge currents with silicon and large C and low winding R
is usually still quite high; maybe 6 times the average DC flow current.
The energy input must equal the energy output, so if the caps are being charged
during 60 degrees of the 360 degree cycle, then I must be high since the flow
time is
at perhaps 1/6 of dc flow time which is all the time.
To reduce this peak current and to increase the degrees of charge time and thus
reduce peak
charge currents I usually place a small resistance between the silicon diodes
and the large input cap
if I have one, about 33 ohms in series with 1,000 uF where Idc = 400mA seems
about right.
The ZC of the 1,000 uF cap at 100Hz is 1.6 ohms, but the 33 ohms limits this to
33 ohms.
The slight droop and poorer regulation of the B+ isn't a large issue in a tube
amp
which mainly works in class A, and ripple voltage shows that charge time is
increased,
and the CRO shows much reduced I peak and transformer runs slightly cooler
and the noise generated in the tranny is less due to the switching transients
from rectifiers.
But I digress from fuses....
Post by Phil AllisonHowever: The *max* time duration of the switch on surge event may very well
be increased by electros charging - this increases the total energy taken
from the AC supply by the surge event and all THAT energy has to pass via
the AC fuse.
This is when fuse sizing gets tricky and there needs to be some by trial and
error testing.
Indeed....
Post by Phil AllisonIf a slow blow or " T " fuse suitable for protecting the transformer from
overload proves to intermittently fail at switch on - then the PSU design
may need a NTC thermistor or a " soft start " system in the AC supply to
take the *sting" out of the peak surge value by adding series resistance
during that brief period.
Its routine to use a resistance + delay in my amps.
The combined cold and low resistance heaters and the rise in B+
with silicon diodes causes considerable primary load after the first
turn on surge from magnetic effects.
One benefit of the tube rectifier is the slow turn on for the B+.
Quad II have so much winding resistance that there is enough series R
to limit peak charge currents regardless how large the first C is.
I still use 47 ohms in series with silicon diodes.
See http://www.turneraudio.com.au/quad2powerampmods.html
About 1/3 the way down the page. R19 & R20 each = 47ohms
and in series with pairs of diodes in series to charge 100uF, then 100 ohms and
with 470uF at the CT of the OPT.
The ripple voltage at the CT is vastly reduced from the crummy original Quad II
PS.
The peak charge currents are also not excessive.
Notice that i fitted a 0.5A slow blo fuse in the amp schematic I have.
This pair of amps has been in daily use for about 12 mths now;
I think the owner blew a fuse maybe once or twice so i suggested he use
750mA or 1A slow blos.
The 0.5A slow worked fine during my tests during the mods done to the amp,
but fuses get tired; the wire fatigues, the weld or soldering at the wire ends
goes brittle;
nothing is forever, and they are made inTai-bloody-wan, so you can't expect
perfect
theoretical performance from such things forever.
Post by Phil AllisonAlso, if the max peak surge current exceeds about 60 amps, then it is
common practice to include such peak surge limiting - lest equipment AC
switches weld themselves permanently on or AC supply circuit breakers trip.
Well what is the time taken for a 60A surge to blow an 8 ohm circuit breaker
in a 50Hz 240V supply?
How come I have never tripped the breaker except perhaps during
while I was repairing a Phase Linear 700. This truly abysmally designed
amplifier
had the typcal huge magnetizing current and high hum noise while working;
the input current showed high saturation current peaks during normal op
and only stopped humming when the input voltage was reduced to 220V.
But we get 250V often here, so no wonder the PT was struggling.
Trannies should be able to take 280V input without saturation.
I use Bmax = 0.85T a lot less than the designs I see routinely with B = 1.25T.
I have not seen great evidence of huge magnetic peak input currents in amps I
have made with
E&I with say 125mm stack of 50 tongue material for a continuous 550VA draw.
Since the Rw of the primaries of such large trannies is so low, then peak IAC
should be quite high.
Patrick Turner.