[R-390] 26Z5W replacement

[Charles Steinmetz]

John wrote:

>Rationale for replacing the 26Z5W:
>
>      *   *   *
>Rationale for getting rid of the 1N561 diodes:

Very nice analysis, John, although I think you brought nuclear 
weapons to a fistfight!  Showing your work was useful, and is 
something the rest of us should do more often.

As you concluded, there is nothing magical about 1N561s.  The 1N4007 
has 4x the current rating.  They are both just silicon rectifier diodes.

>2) The voltage drop across the diode is small, leading to greater 
>output voltage and potentially stressing the other components in the receiver.

Correct.  In 1966, the Navy ordered all shipboard 390As to be 
converted to 1N561 diodes per Field Change No. 6.  The higher B+ 
voltage was blamed (rightly or wrongly) for failures of 6AK6s at V603 
and V604 (I'm inclined to believe that the Navy bought a bad batch of 
6AK6s -- I have not observed 6AK6 failures in 390As with SS 
rectifiers in the 35+ years I've been working on them).  Accordingly, 
in Electronic Information Bulletin EIB-895, they published an 
*optional* procedure for adding a 220 or 200 ohm series resistor 
after the SS diodes to drop the B+ by 20 or 30 volts.

However, many thousands of 390As have been modified to use SS 
rectifiers by the Navy and by hobbyists, some with the dropping 
resistor and some without.  I've worked on several hundred 390As and 
of these, about 30% have been modified for SS diodes, and about 2/3 
of those do not have a dropping resistor (that is, about 20% of all 
the 390As I've seen have SS diodes with no dropping resistor, and 
about 10% of all the 390As I've seen have SS diodes with a dropping 
resistor).  I have not seen statistically significant evidence that 
even the worst case -- SS diodes with no dropping resistor -- reduces 
reliability compared to 26Z5s or to SS diodes with a dropping 
resistor.  This anecdotal evidence doesn't mean there is no decreased 
reliability with SS diodes -- just that if there is, it isn't a huge 
difference.

I once built a 390A for a friend with regulated B+.  The test mule 
had a pot that could dial the B+ from 180v to 300v.  I could not 
measure any difference in receiver performance regardless of where 
the B+ was set.  So, I concluded that the actual value of B+ in a 
390a is monumentally non-critical.  Accordingly, while curve-fitting 
the impedance of the 26Z5s was an interesting exercise, it was, IME, 
wholly unnecessary.  More than a few tens of volts of B+ variation 
make no difference in the operation of the receiver.  A few volts 
over temperature is way, way below the threshold of 
detection.  (Also, note that a 1v change in the line voltage will 
produce a B+ change greater than the variation with temperature that 
you found.)

>No attempt was made to do a gradual turn on to emulate the warming 
>of the heater.  The 82 ohm resistor will provide some current 
>limiting at turn on.

As we have discussed here before, the choke-input filter of the 390A 
is particularly subject to high B+ at turn-on (for 5-10 seconds 
before the receiver circuitry starts drawing significant B+ current, 
or during "standby" operation).  For many other reasons, choke-input 
filters are to be preferred over capacitor-input filters, so I do not 
mean that as a criticism.

All in all, this is an instructive look at the engineering that is 
often necessary to implement a "simple" modification to an existing 
design.  One can either: (i) leave the 26Z5s in, and let their warmup 
characteristic solve the "high B+ at turn-on"; (ii) change to SS 
diodes and don't worry about the high B+ at turn-on; or (iii) change 
to SS diodes and figure out how to prevent the high B+ at turn-on 
[note that adding a dropping resistor does NOT prevent high B+ at 
turn-on, because no current = no voltage drop -- E = I x R].

Frankly, I generally prefer to leave the 26Z5s in.  However, I've 
seen plenty of radios operating for decades with SS rectifiers with 
no problems, so option (ii) is fully viable, IMO.

If one opts for (iii), things get messy fast.  One could use a 
time-delay relay in any of several ways to delay the B+.  If I went 
that way, I think I'd be most inclined to have it switch a shunt load 
resistor across C606A, the first filter capacitor, that was switched 
out after warmup.  However, it would need to be switched back in 
during "standby" operation.  It would also need to draw well over 100 
mA, so it would need to dissipate some real power (>30 watts) for 10 
seconds during turn-on or during standby operation (which could be of 
indefinite duration).  Solid-state switching could do the same thing, 
with the same caveats.  Alternatively, one could try series 
switching, but series switching at high voltage is problematic -- 
particularly with an inductive load (the choke-input filter).  These 
are all really kludgey solutions.  A power zener may be the best of 
the "simple" choices, but that has its own problems.

One then graduates to less simple solutions.  Some of these are not 
kludgey, but they are complex (generally, far more complex than is 
warranted for retrofitting what is, after all, an old tube radio), 
would require VERY careful design, and would probably be more 
expensive than hams/SWLs would be willing to pay.

Best regards,

Charles