[R-390] R390-A Power On B+ Voltage and Choke Filter

[Charles Steinmetz]

Larry wrote:

>I certainly agree that the initial voltage will be higher with SS 
>rec[tifier]s, however, the purpose of the 'swing choke' input power 
>supply filter in the R390-A is to eliminate a high voltage surge at 
>'power on'.

A swinging choke is nothing magical -- it just acts like a larger 
choke at low current than at higher current (i.e., inductance 
decreases with increasing current).  In the case of the 390A, L601 
(the only choke that matters for the choke-input filter analysis, 
because it is the only choke before the first filter capacitor) is a 
2-12H swinging choke (12H at very low current, decreasing to 2H at 
rated current).  So, there still has to be enough load current from 
the moment the rectifiers are conducting to bring the filter 
capacitor voltage down to the choke-regulated value for the choke's 
maximum inductance.  And since the maximum inductance of L601 is 12H, 
6 times the inductance at rated current, that minimum required 
current is about 1/6 of the rated current of the choke.  That will 
NOT be just a few mA -- the rated current of L601 is at least 150mA, 
so the minimum load current for choke-input regulation to occur will 
be at least 25mA.  And, indeed, calculating the choke-regulation 
current for the 390A supply bears this out.

>I have done indepth testing on my 2 R390-A's with 26Z5's in this 
>area and have found that the B+ at C606A (1st filter cap) rises 
>quickly to a maximum of 255V in Standby and slowly goes down to 
>242V, where it holds.  *   *   *  Turning a R390-A on by switching 
>immediately to AGC yields a little lower initial maximum voltage 
>(250V), as you would expect and is an identical scope trace.

More or less exactly as I and others have said here many times, 
provided that the radio uses 26Z5s and that the 26Z5s warm up 
normally.  Not the case, however, if the radio has SS rectifiers 
installed or if the 26Z5s warm up faster than normal (as some 
do).  (But I think your measurement method missed a short excursion 
above 255v, for reasons explained below relating to the strike 
current of the 0A2 regulator tube.  If the voltage had only reached 
255v, the 0A2 would not have started.)

>Here's why:  The R390-A uses a 'swing choke' input power supply 
>filter, the immediate load on the supply in standby is 6.7 ma and in 
>AGC is 16.9 ma.  This load is from:  Standby: 4 ma - 150V reg, 2.7 
>ma - audio resistors to gnd;

Neither 6.7mA nor 16.9mA is enough current to hold the C606A voltage 
below 300v, even assuming that L601 still exhibits the full 12H 
maximum inductance at those currents (which it wouldn't).  However, 
the 150v regulator actually draws much more current than you have 
calculated (note that 5mA is the minimum operating current for a 26Z5 
-- see datasheet -- and that a shunt regulator cannot supply more 
than about half of its no-load current to the regulated circuits).

The 0A2 strikes at ~75mA [see datasheet], then holds at about 25mA 
for a C606A voltage of 240v [R617 + R618 + R619 = 3.6k; current = 
240-150v/3.6k = 90v/3.6k = 25mA].  The holding current will be higher 
if the C606A voltage is more than 240v (Ohm's Law).

To reach the 75mA strike current, there must be at least a short 
excursion of the C606A voltage to ~420v (75mA through 3.6k produces a 
drop of ~270v, plus the regulator voltage = 420).  In practice, most 
0A2s strike at less than 75mA, so the overhead is typically less than 
270v -- but the total voltage required with 3.6k of dropping 
resistance is still well above 300v.  Even a strike current of only 
50mA requires a C606A voltage excursion to ~330v.

25mA *is* enough (just barely) to hold the C606A voltage below 300v, 
once the 0A2 is conducting.

Note: the above analysis is somewhat simplified.  R619 is actually 
switched out of circuit (shorted) when the radio is in AGC mode, so 
the dropping resistor is R617 + R618 = 2.8k.  If you switch from OFF 
straight to AGC, after the 0A2 strikes it will be drawing 32mA (for 
240v on C606A; more if the C606A voltage is higher).  This explains 
why the C606A voltage peak is lower in AGC mode, even before the 
signal tubes are fully warm.

>The reason that the additional 10.2 ma load in AGC only reduces the 
>initial maximum voltage by 5 volts is the way a 'swing choke' input 
>power supply filter works.

Not so.  The reason is that the 25mA drawn by the 150v regulator, by 
itself, is enough to hold the C606A voltage down to its nominal 
choke-input value.  The extra 5v is due simply to the resistive drop 
in the power transformer secondary, the 26Z5s, and L601, as well as 
the increased 150v current due to shorting out R619.

>The following is an excerpt from Norman H. Crowhurst's book, 'Basic 
>Audio' from 1959.  I like this explanation of a swing choke as it is 
>short and clear:

It is a good explanation, but your interpretation makes it sound as 
if a swinging choke magically satisfies the current requirement to 
hold the voltage of a choke input filter down to its choke-regulated 
value with just a few mA of current.  It doesn't.  The best it can do 
is act like its maximum inductance -- in the case of L601, 12H.  Note 
that the diagram in Crowhurst's book shows no scale, so there is no 
basis to conclude that "it takes very little current to hold the 
output voltage down."

The amount of load current required to bring a choke-input filter 
down to its choke-regulated value can be calculated, which is what I 
have done wherever I have cited currents and C606A voltages.

>The way I verified I have no leaky caps is I removed all the tubes 
>except the recs, plugged it into my variac, while measuring the 
>voltage and current at F102, brought it up to 255V, in AGC I read 
>12.9 ma, as it should be (16.9 ma - 4 ma for the regulator).

As I noted above, all of your current estimates and measurements 
appear to be wildly off the mark.  Somewhere I have measurements of 
the B+ currents of quite a few 390As.  I would need to look in my 
notes to be sure, but ISTR that the total B+ current of an operating 
radio in good repair is around 125mA, perhaps a bit more.

>So, if your R390-A's are working correctly and using 26Z5 recs, it 
>will never see more than 255V DC on any capacitor

Except for the brief excursion to the high 300s or low 400s every 
time you turn it on, to fire the 0A2.  And assuming that your 26Z5s 
warm up slowly.

>This is why I believe the designers rated the 5 electrolytic filter 
>caps C603 and C606 in the power supply mounted in the 2 cans on the 
>audio deck at 300V and not higher.  They knew what they were doing.

As I and many others have pointed out many times on this list, the 
stock 300v electrolytics generally survive the even longer excursions 
to the low 400s at turn-on and to the low 300s (during operation on a 
125v AC line), when the 26Z5s are replaced with SS rectifiers.  That 
is true even after they have run for 60 years at the lower voltage 
developed by the 26Z5s.  So, yes, the designers knew what they were 
doing.  But what they were doing wasn't keeping the C606A voltage 
below 300v at every instant -- it was trusting the capacitor design 
margin to tolerate the excursions above 300v that they knew occurred.

Interestingly, that has implications for choosing replacement 
electrolytics.  Generally speaking, the technology of aluminum 
electrolytic capacitors has come a very long way since the 
1950s/60s.  However, note that today's electrolytics are MUCH 
smaller, value for value, than those of the '50s/'60s.  One 
consequence is that the dielectric layers are thinner these days, 
which means that today's capacitors are not as robust when it comes 
to dielectric failures from voltage surges.  So, I would NOT trust a 
300v capacitor you buy today to replace an original 300v cap in a 
390A.  I use 450v caps of the highest temperature rating I can find 
(125 degrees).

Best regards,

Charles