[R-390] The Penultimate R-390* Ballast Replacement

[Drew Papanek]

On an external power supply to replace the BallasTube Bob Camp wrote:

<snipped>

>The good old style 78xxx regulators are pretty well behaved. Some of
>the newer parts are not so forgiving.

Of even greater benefit  is that most of us have those older parts already 
lying around in the junk box :o)

>If you do go with a solid state filament supply be sure to consider the
>inrush current. A quick check with an ohm meter on a cold tube should
>give you a pretty good idea what to expect from that particular tube.
>Common wisdom (often wrong ...) is to provide 4 to 5X the running
>current for inrush. Your 300 ma supply would have to source 1.5 amps
>while the tubes warm up.

>Current limit is one way to get around this. The two common options are
>constant current limiting and fold back limiting. A fold back limiter
>is not going to do any good in this situation. A constant current
>limiter actually increases the power dissipated in the regulator as it
>cuts back. Unless there is a big heat sink this generally either melts
>the device or puts it into thermal overload. If it goes into thermal
>overload you get the same problem as with the fold back limiter.

The 78Mxx series of three terminal regulators could be used in a constant 
voltage configuration.   those go into current limit at, IIRC, 500mA.  They 
are protected,as are virtually all 3-terminal regulators, from damage due to 
thermal overload.

Better still would be to not use voltage regulation at all but to use 
current regulation instead.  The LM317 or 7805 are well suited to the task.  
Dave Wise built such a constant current supply (before his sophisticated 
3DW7 analog and digital "tubesters").


[Dave Wise's text follows]

Here's IMO the simplest regulator that's also really good.

Parts list:
5ohm 10W resistor.
10ohm 10W resistor.
2.2K 1/4W resistor.
2.7K 1/4W resistor.
1K pot.
3000uF/50V cap.
Silicon rectifier.
LM317 on heat sink.

Vin goes to 5ohm resistor.
5ohm resistor goes to anode of rectifier.
Cathode of rectifier goes to cap and LM317 IN terminal.
Other end of cap goes to ground.
LM317 OUT terminal goes to 10ohm resistor and 2.2K resistor.
2.2K resistor goes to LM317 ADJ terminal and 2.7K resistor.
2.7K resistor goes to 1K variable resistor.
1K variable resistor and 10ohm resistor go to Vout.

This will adjust from 280mA to 335mA.  It has four big components, three of 
which are also hot, and it requires a ground. This was my first step on the 
road to the 3DW7.

How's it work?  The rectifier and cap give you DC. The 5ohm resistor softens 
the charging peak and takes on some of the heat load. The LM317 will do 
anything in its power to maintain 1.25V from OUT to ADJ.  This puts 1.25V 
across 2.2K
for 0.57mA, which also flows through the 2.7K resistor.  (The LM317's 
current out the ADJ pin is negligible.)  0.57mA * (2.2K + 2.7K) =3D is 2.78V 
.

The LM317 will do anything to make that 2.78V happen. In this case it 
punches 278mA through the 10ohm resistor. If you increase the 2.7K resistor 
to 3.7K, the
voltage is 3.35V instead of 2.78V for 335mA out. I can't remember what range 
of AC input voltage this will work over, but it's at least 25.2 +/- 5% .

[end Dave Wise's text]

Constant current regulation is advantageous because the PTO/BFO tube heaters 
never see more than their normal steady state (300 mA) current.  Possible 
shortening of tube life because of inrush transients becomes a non-issue.

Dr. Jerry's device is a (fairly) constant current regulator.

A  .1 uF disc ceramic across each rectifier diode will address any diode 
switching noise concerns.

>Twelve volt one or two amp supplies are not hard to find. They also
>won't set the bank roll back by much more than a nice dinner for the
>family.

David Wise's circuit described above was intended to use the 25.2 VAC 
available at the ballaSocket.  The diode, 5 ohm resistor, and electrolytic 
filter cap could be eliminated and the remainder of the circuit powered by a 
large wall wart.  Those commonly used to power cheap inkjet computer 
printers would be ideal, being rated at around 18 VDC at about an amp.   It 
would be well to connect a .1 uF disc cap across the regulators's "IN" and 
"ADJ" teminals to ensure stability.

>This would all be a bit easier to evaluate if we had some real data
>(1.2 Hz per 1% change) from several radios on the impact of heater
>voltage on the stability of the radio. The boys at Collins didn't take
>any data that they found convincing when they designed the radio ....

I believe some pertinent data appears in the "Pearls of Wisdom".  at 
r-390a.net

Drew