[Collins] {Collins} Question on 516-F2 PS

[Dr. Gerald N. Johnson]

On 9/14/2015 8:01 PM, Garey Barrell wrote:
> Jerry -
>
> According to the manual I have, the original looks to be a 5% part. so
> they 'really' wanted 24k! :-) Especially since R4 and R5 were 25k and
> from the part number look to be simple standard part numbers. It takes
> some 'juice' to get the bean counters to add a 'new' part. "Why cantcha'
> just use the same part as the other two". Heard it before... :-)

EVERY part used at Collins has a part number and a specification or a 
drawing. Even parts bought from the Allied or Newark catalog (we could 
get overnight delivery to Cedar Rapids from them before they 
computerized) for experiments or prototypes came in with a Collins 044 
family part number. The 500 to 599 prefixes were reserved for Collins 
drawings, bills of material, and manuals. Bean counters didn't specify 
designs, engineers did but Art had to approve radical design concepts, 
though he didn't have any degrees. If he didn't like it you didn't build 
it. If he wanted it built he would spend no end of money despite always 
being deep in debt. That's how Rockwell bought the company in the 70s 
for $232 million about 1 year's gross sales. Really cheap for a high 
technology company. In the years before that, he would go to the banks 
for a $100 Million or $200 million loan and they would require him to 
reduce the white collar work force by 10%. Clearly the banks didn't 
realize new high tech requires engineers, not production line workers.

There was a design standards manual that often suggested what parts to 
use and there was a book of standard specifications for parts used most 
often. When I was designing metering for the 821A-1, I found a need for 
1% resistors but I found the 1% resistor specification had only every 
other value and working out my circuit I found I needed the missing 
values. So the project paid for a new 1% resistor specification covering 
all commercially available values. The component specification 
department had to create the specification and qualify manufacturers. 
Incoming parts were then tested to meet the specification before going 
to the warehouse or production line. The circuit I was developing was 
for a panel meter with 100 microamps full scale and 1000 ohms meter 
resistance. It needed two ranges selected by which input was used that 
gave it full scale readings of 3.0 volts and 10.0 volts and each had to 
provide a 10K load to the meter point. I didn't want to switch scales, I 
wanted it to be automatic according to which input wire I used. So I had 
to include two meter multiplier resistors and two input shunt resistors 
that were always in the circuit, so the multiplier and input shunt not 
in use were shunting the meter changing its sensitivity. I spent a day 
or two with a Friden calculator going around in circles as the second 
input always changed the first input load requiring changing every 
resistor as I iterated the final solution. Today I'd write a spread 
sheet set of equations or a C program and let the software do the 
iterations until the results were fixed. And I'd program in the C 
version a list of 1% values for resistors.

As for carbon composition resistors the specification was fairly simple 
and the vendors list was "QPL except for IRC." IRC resistors were 
actually film on a glass tube encased in a molded case where evey one 
else made true carbon composition resistors. The IRC had different RF 
characteristics and when overheated failed by going down in resistance 
as the molded part charred. True carbon composition resistors failed by 
going up in resistance and with extreme overload burned open. So they 
were sometimes used as fuses and IRC didn't burn open so weren't allowed.

Wire wound resistors have tended not to fit in the RETMA standard 
values. I just checked a 1956 "Reference Data for Radio Engineers" by 
ITT. It shows two standards in effect. ASA and RETMA. ASA standard 
sequence was based on a multiplier between values of the 5th root of 10 
for the "5" series and the 10th root of 10 for the "10" series. So the 
number sequence for 5 series ASA was 10, 16, 25, 40, 63, 100. And for 
the 10 series was 10, 12.5 or 12, 16, 20, 25, 31.5 or 32, 40, 50, 63, 
80, 100. The RETMA 20% sequence is 10, 15, 22, 33, 47, 68, 100. The 
RETMA 10% sequence is 10, 12, 15, 18, 22, 27, 33, 39, 47, 56, 68, 82, 
100. And the RETMA 5% sequence is 10, 11, 12, 13, 15, 16, 18, 20, 22, 
24, 27, 30, 33, 36, 39, 43, 47, 51, 56, 62, 68, 75, 82, 91, 100. In 
RETMA the multipliers are based on the 6th root of 10, the 12th root of 
ten and the 24th root of 10. When made carbon composition resistors had 
a wide distribution of values and were selected by automated machines. 
The tolerances and the target values mean that EVERY resistor made fits 
some standard value whether being selected for 20%, 10%, or 5% values. I 
have detected that sometimes 20% resistors tended to not be within 10% 
of the marked value, but always more than 10% away hinting that the sort 
was biased by demand and most demand for 10% resistors was still for the 
values in the 20% list. Same thing for sometimes a middle hole in the 
10% values shipped missing the 5% resistors. There are few carbon 
composition resistors made today and most metal film are laser trimmed 
to the desired value by varying the turns of the helical cut.

This book says the ASA standard is widely used for wire wound power type 
resistors and the RETMA standard is used for small electronic 
components, as fixed composition resistors and fixed ceramic, mica, and 
molded papers capacitors. I believe these standards go back to the 30s 
when many a resistor color code was body, end, dot.

Searching Mouser.com last night I found only 25K 25 watt wire wound 
resistors in 5%. No 24K which would have been RETMA values.

R4 and R5 are listed as 25K but use different mounting studs. Probably 
because one end of R4 is at 800 volts above ground it used a Collins 
fabricated stud. But R5 and R6 use the same hardware and the manual 
picture doesn't show them close enough to see whats different. The 
difference might be because of the high voltage to ground of R4 hot end 
and different manufacturers had different standards and different 
voltage ratings. Likely the low voltage bleeder had a lower price with 
lower voltage insulation though in general price was a minor 
consideration while designing at Collins. The price criteria was make it 
work better than anybody else's product.


>
> I'm sure that would have been a custom part at that time. 24k would be
> more likely to be a standard now, but when this design was done 25k
> would have been more likely....

They were BOTH standard values.
>
> 73, Garey - K4OAH
> Rome, GA
>
> Drake 2-B, 2-C/2-NT, 4-A, 4-B, C-Line
> and TR-4/C Service Supplement CDs
> <www.k4oah.com>
>
73, Jerry, K0CQ, Technical Adviser to the Collins Radio Association.
> Dr. Gerald N. Johnson wrote:
>> 24K is a standard 10% value. Wire wound resistor makers often stuck
>> with their pre standard values, like 25K.
>>
>> The 25K has a tolerance too and could be high or low. 5% would be +/-
>> 1.25K. 10% would be +/- 2.5K. The true value for best voltage
>> regulation (not jumping too high at no load) depends on the filter
>> capacitor value and it has a much wider tolerance, sometimes +40 -10
>> or other wide variations.
>>
>> There is a graph in vintage ARRL handbooks that they may have used. I
>> know I showed it to my boss at Collins who had been in charge of the
>> group designing the S-line and power supplies.
>> http://www.geraldj.networkiowa.com/papers/Capfilterps.jpg
>> It clearly shows the effects of transformer, rectifier and filter
>> capacitor on capacitor input filters. The filter choke complicates
>> things but eases the every half cycle peak current.
>>
>> 73, Jerry, K0CQ, Technical Adviser to the Collins Radio Association.
>>
>>
>> On 9/14/2015 3:59 PM, Garey Barrell wrote:
>>> I don't know that much about the 516F-2, but choke input supplies
>>> require a minimum load for best regulation.
>>>
>>> I suspect that Collins engineers slipped their slide rules and came up
>>> with 24k as that minimum load.
>>>
>>> That said, the original part was +/-5% so 25k is JUST within
>>> tolerance! :-)
>>>
>>> 73, Garey - K4OAH
>>> Rome, GA
>>>
>>> Drake 2-B, 2-C/2-NT, 4-A, 4-B, C-Line
>>> and TR-4/C Service Supplement CDs
>>> <www.k4oah.com>
>>>
>>> Stanley M Miln wrote:
>>>> Hi Guys,
>>>> Typically with Tube Equiment 20% tolerance is
>>>> acceptable. Probably with Frequency Components
>>>> that may not be so, like with the PTO.
>>>> All the Bleeder Resister does is discharge the Caps
>>>> after the Supply is shut off. It also may help
>>>> with Regulation. I can not see the Value being that
>>>> Critical.
>>>> Stan K6RMR
>>>>
>>>>>
>>>>> How important is R6 (24K) replaced with a 25K. A 25K resistor with 5%
>>>>> tolerance is 23.750K so would that be close enough to 24,000 to work?
>>>>> Comments please. Thanks.
>>>>>
>>>>> JAC
>
>
>