T
The Phantom
Guest
On 27 Aug 2005 17:12:14 -0700, "abekas67" <stuart@usaor.net> wrote:
would be to come up with a well-regulated 4.0 vac for the filament.
This could be done in several ways, but an easy and effective method
might be to get a ferroresonant regulator with 120 volts output,
followed by a filament transformer of somewhat more than 4 vac output.
You're asking for just a little more than 1% regulation. A typical
ferro usually isn't spec'd for 1% regulation, but their spec is for no
load to full load. I suspect that with a fixed load, you might be
able to get 1% regulation. It's worth a try. And the fact that the
output is distorted (somewhat like a square wave on the uncompensated
ferros) won't be a problem for you since you're just heating a
filament. Just be sure you measure the voltage at the filament with a
true RMS meter. Let's say you get a 120 to 6.3 volt filament
transformer for the output of the ferro; you say you have a pot for
adjustment in there now. You could do that with the ferro setup. But
it might be better to have a small variac before the filament
transformer rather than a pot (rheostat?) after it. Then you can
adjust the voltage applied to the filament to exactly 4.0 vac, and it
won't vary as much if the 1 amp current changes a little with ageing
of the filament for an applied 4.0 volts as it would if you use a
variable resistance for the final trim.
To verify the performance, you would want to set all this up, trim
for 4.0 vac at the filament, and have another variac in front of the
ferro and vary the ferro input voltage over the expected line voltage
range and see if you get 1% regulation at the filament. (There may be
some ferros that have tighter regulation. I wonder what would happen
if you cascade two ferros?! You only need 4 watts of power; perhaps
two small cheap ferros in cascade would give the regulation you need,
if a single one isn't good enough.)
And, if ferros aren't good enough, you'll have to solve this part of
the problem some other way, Whatever works.
The second step in this problem is to get the slow turn-on. I would
get a couple of low Rds power FETs. Low voltage N-channel FETs with
Rds of 10 milliohm are common these days. Connect the sources and
gates together. The two drains now form a two-terminal resistance
which can control AC; put that arrangement in series with the output
of the filament transformer that is feeding your filament. It might
be good to match the threshold voltage of the two FETs to avoid DC in
the filament transformer. Now all you have to do is turn the FETs on
slowly. Put a low leakage film capacitor of perhaps 10 uF between the
gates and sources (which are already connected together pairwise from
the two FETs). Charge the capacitor through several megohms and you
will get the slow turn-on you want. With a high enough resistance
here, you should be able to get as slow a turn-on as you could want.
The 20 milliohm resistance of the fully turned on FETs won't affect
your regulation much, and anyway you can do your final trim of the
voltage at the filament with the FETs in circuit, fully turned on.
You will need some mechanism to turn the FETs off when you power
down so they're ready for a slow turn-on at the next power-up. I
think I would put a reed relay contact in series with a low value
resistor (1000 ohms) across the capacitor, arranged so that the
contacts are open when power is applied to the circuitry, and closed
when the power is off, thereby discharging the cap and turning the
FETs off.
How to charge the capacitor, since the FET gate-source circuit is
floating? I can think of several ways, but for easy simplicity, how
about this. Since you will be charging the capacitor through several
megohms, it won't take much current to do the job. Put two 9 volt
transistor radio batteries in series, put the several megohms in
series with them, and apply to the capacitor which is across the
gate-sources (if you use logic level FETs, one battery may suffice).
When the power is on, as soon as the capacitor is charged up, the
current from the batteries drops to zero (except for leakages, of
course). It's only while the reed relay shorts out the capacitor
(through 1000 ohms) that the batteries have to supply a constant
current of several microamperes. Replace the batteries every few
months. And if you want to reduce the battery drain to zero while the
power is off, arrange another relay contact in series with the
batteries to open when power is off (or maybe use a FET).
I think I would approach this problem in two steps. The first
would be to come up with a well-regulated 4.0 vac for the filament.
This could be done in several ways, but an easy and effective method
might be to get a ferroresonant regulator with 120 volts output,
followed by a filament transformer of somewhat more than 4 vac output.
You're asking for just a little more than 1% regulation. A typical
ferro usually isn't spec'd for 1% regulation, but their spec is for no
load to full load. I suspect that with a fixed load, you might be
able to get 1% regulation. It's worth a try. And the fact that the
output is distorted (somewhat like a square wave on the uncompensated
ferros) won't be a problem for you since you're just heating a
filament. Just be sure you measure the voltage at the filament with a
true RMS meter. Let's say you get a 120 to 6.3 volt filament
transformer for the output of the ferro; you say you have a pot for
adjustment in there now. You could do that with the ferro setup. But
it might be better to have a small variac before the filament
transformer rather than a pot (rheostat?) after it. Then you can
adjust the voltage applied to the filament to exactly 4.0 vac, and it
won't vary as much if the 1 amp current changes a little with ageing
of the filament for an applied 4.0 volts as it would if you use a
variable resistance for the final trim.
To verify the performance, you would want to set all this up, trim
for 4.0 vac at the filament, and have another variac in front of the
ferro and vary the ferro input voltage over the expected line voltage
range and see if you get 1% regulation at the filament. (There may be
some ferros that have tighter regulation. I wonder what would happen
if you cascade two ferros?! You only need 4 watts of power; perhaps
two small cheap ferros in cascade would give the regulation you need,
if a single one isn't good enough.)
And, if ferros aren't good enough, you'll have to solve this part of
the problem some other way, Whatever works.
The second step in this problem is to get the slow turn-on. I would
get a couple of low Rds power FETs. Low voltage N-channel FETs with
Rds of 10 milliohm are common these days. Connect the sources and
gates together. The two drains now form a two-terminal resistance
which can control AC; put that arrangement in series with the output
of the filament transformer that is feeding your filament. It might
be good to match the threshold voltage of the two FETs to avoid DC in
the filament transformer. Now all you have to do is turn the FETs on
slowly. Put a low leakage film capacitor of perhaps 10 uF between the
gates and sources (which are already connected together pairwise from
the two FETs). Charge the capacitor through several megohms and you
will get the slow turn-on you want. With a high enough resistance
here, you should be able to get as slow a turn-on as you could want.
The 20 milliohm resistance of the fully turned on FETs won't affect
your regulation much, and anyway you can do your final trim of the
voltage at the filament with the FETs in circuit, fully turned on.
You will need some mechanism to turn the FETs off when you power
down so they're ready for a slow turn-on at the next power-up. I
think I would put a reed relay contact in series with a low value
resistor (1000 ohms) across the capacitor, arranged so that the
contacts are open when power is applied to the circuitry, and closed
when the power is off, thereby discharging the cap and turning the
FETs off.
How to charge the capacitor, since the FET gate-source circuit is
floating? I can think of several ways, but for easy simplicity, how
about this. Since you will be charging the capacitor through several
megohms, it won't take much current to do the job. Put two 9 volt
transistor radio batteries in series, put the several megohms in
series with them, and apply to the capacitor which is across the
gate-sources (if you use logic level FETs, one battery may suffice).
When the power is on, as soon as the capacitor is charged up, the
current from the batteries drops to zero (except for leakages, of
course). It's only while the reed relay shorts out the capacitor
(through 1000 ohms) that the batteries have to supply a constant
current of several microamperes. Replace the batteries every few
months. And if you want to reduce the battery drain to zero while the
power is off, arrange another relay contact in series with the
batteries to open when power is off (or maybe use a FET).
about 'How