Capacitance in power supply filters

[Edward Cain]

I think I remember reading in one of the posts on this board that different values of capacitance should be used before and after the filter choke.
   Can anyone verify this and, if true, explain it to me?

Thanks,
Ed

[WD5JKO]

Ed,

   It is my feeling that these situations need to be looked upon on a case by case basis. With no modifications using vintage gear, the B+ with today's higher line voltage can be 10% or more higher than it was when originally designed. Then there is a push to solid state the rectifiers causing another boost in the B+. Now the DC voltage rating of the filter caps often gets exceeded, especially during the first 10-15 seconds after the gear is turned on.
   Back to your question, if the rectifier is tube, and the filter is a pi type, it is important to not allow the tube to see a 'hot switching' current surge higher than the tube manufacturer recommends. So yes keep the input cap stock, and if you want additional filtering, then beef up the downstream cap.
   You will find a wide variety of opinions on this topic. There is more than one way to tame a shrew..

Jim
WD5JKO
 

[k4kyv]

If you have two caps of different values, I would use the one with the most capacitance at the output of the filter. The output cap filters not only the ripple, by bypasses anything to ground that the load might try to introduce.  For example, in a plate modulated final the audio needs a path to ground if the secondary of the mod transformer carries the PA plate current.  This path is provided by the output cap in the power supply filter.  If the power supply is used for a class B modulator, the load on the supply will vary at twice the modulating frequency, and if an effective output cap does not exist at the output of the filter, 2nd harmonic distortion may be introduced into the modulated stage.

In pre-WW2 equipment, the filter caps were often only 1 or 2 mfd.  That was because capacitance was extremely expensive and took up loads of space.  To-day it is easy to obtain 25, 50 or more mfd.  I would use as much as I could for every capacitor but not enough to require step-start for the power supply, or as Jim mentioned, to exceed the peak current rating of tube type rectifiers.  Solid state rectifiers generally can withstand instantaneous current overloads better than tubes.

If you use too much capacitance, the charging current when the power supply is first turned on may be high enough to trip the breaker or blow a fuse, so a "step start" circuit is required to apply voltage to the filter gradually.

For my transmitters, running from 1250 to 2500 volts from the power supplies, with choke input, I have found around 25 mfd to be optimum the best compromise.

[WQ9E]

The input capacitor (the one between the rectifier and choke) will have a large effect on the output voltage and peak current draw (impacting the rectifier and power transformer).  Tube type rectifiers are generally not used with a large value input capacitor and MV rectifiers often feed a choke directly.  

I would suggest looking at one of the vintage ARRL handbooks (or even better the Orr handbook) which provide a good discussion and formulas/charts for choosing capacitor and choke configuration.  Terman is also an excellent reference.  Tube data sheets will also provide suggestions and limits on the size of the input capacitor, for example take a look at the N7JP site http://www.nj7p.org/Tube.php  for your chosen rectifier tube.  

The output capacitor is generally going to be larger than the input capacitor to take advantage of the greater ripple reduction without the problems of increased voltage and peak current draw.  Also keep in mind that for choke input filters, there is a minimum current draw that must be achieved (via the load and/or bleeder resistor) to prevent the filter from behaving as a capacitor input design with the resultant increase in voltage.  This is a key reason for the use of "swinging" chokes that provide a fairly high inductance at low current while still maintaining a reasonable size, price, and voltage drop (under normal load).

[Edward Cain]

   Thank you Jim, Don and Rodger.
You all seem to be in agreement about the downstream (output) cap being of larger value if needed and your explanations provide me with some insight as to why.
   While my question was a general one I have an immediate application in mind: my SX-62. In stock configuration, the input cap (cap between rectifier tube and choke) is larger, 30 uF, than the output cap, 20 uF. This would seem to violate the reasoning given above. Do I understand correctly?

Ed

[WQ9E]

Ed,

Hallicrafters takes the B+ for the output stage from before the filter choke so it needs decent filtering to avoid too much ripple thus the 30 uf input cap.  I would stay with this value for the input cap but for the output side you could either use a 22 uf or go up to a 47 uf value.

The power transformer in the SX-62 is running plenty warm as is so don't go any higher on the input capacitance side.

If you haven't done so already, be sure and replace C-109, C-111, and C-112.  C-111 is the cathode bypass and if it shorts the output stage will draw excessive current cooking the power and output transformer.  If C-109 and/or C-111 become leaky then the control grid of the respective output tube will become positive with the same ultimate result as a failed C-111.   C-109 also keeps B+ off your headphones so it is a definite safety consideration.

The SX-62 is a nice band cruiser and sounds great.  I used one for a week as a novice receiver and it wasn't a good choice for that but I have a couple of them around now for FM and shortwave listening.

[ke7trp]

The info is spot on here. Keep the input cap the same.  You can double or tripple the Output side cap.  IN the king. It had a 3UF cap. I use a 12 UF now that fit right in.

I used to double or tripple all of the caps. Until I learned the hard way.  I now study the schematic and only increase caps that are outboard of chokes.

Dont trust old oil cans in power supplys. Look at the ratings.  In the Champion, Mine had a 1000V sangamo oil can.  With solid state rectifiers and higher line volts, I was passing 1200+ through that old oil can. It did not mind.. But when I realized this, I put the 866s back in That Dropped it way down to 960. 

The result was a loss about 40 watts Am. It ran 56 years like this so I figure its ok. 

C

[Edward Cain]

   Thanks Rodger and TRP. I have already replaced all the paper and electrolytics except C111, C113 and C116. Oh, and the 3 paper with the braided leads are still in place. I have a 20 uF 50V for C111 but only (2) 33 uF 450V for C113 and C116; thus my question. I see now, from your explanation, that I needn't be concerned with replacing C116 with a 33 uF instead of the stock 20 uF.
   Now to check the resistors and micas in the discriminator and if cans and maybe I'll be listening to the radio instead of wrestling with it.

Ed

[KE6DF]

Here is a link to a freeware powersupply design computer program.

It lets you try different configurations of cap and choke input single and multi-stage filters.

You can select various tube and silicon rectifiers.

You can simulate the powersupply and determine the peak current to see if you are in spec.

http://www.duncanamps.com/psud2/index.html

It says it runs on XP, but in fact I run it on Vista and Win 7 with no problems.

[KM1H]

The real answer is easy, look up the tube specs on the web or an RCA manual. They list the maximum input C based upon voltage and the curves show the rest of the details.

Slightly exceeding the value to coincide with current values is no problem, so a 33uF is fine in place of 30 and is probably less C than the original which had a huge tolerance variation of -30 to +100% or more. Modern caps are 10-20%.

Id stay away from no name caps found at hamfests, some dealers, etc. Mouser has good prices on known brands and they ship the same day usually.

The 62 is a great radio, I have a pair of restored 62A's that get used daily, Im listening to one now.

Carl
KM1H

[WD5JKO]

One item not mentioned is something I have done quite a few times with low power tube rectifiers that came with a pi C-L-C filter. I switched the rectifier to solid state, and the filter to choke input while beefing up the capacitance at least 2x, and preferably 5-10X the stock value. A 200uf at 350v is pretty small these days. The result is a slightly lower than stock voltage. The boost from higher line voltage + lower drop with SS diodes is offset by the big drop when going from a pi C-L-C filter to a choke input filter L-C.

The problem I've seen with this method is that diode spikes from slow reverse recovery time (1N4007's do this) come right through at a 120 hz rate and might get into low level stages as a buzz. The cure here is to change the filter from a L-C to a L-C-R-C. Here the C values could each be cut in half and the R could be 50-100 ohms. The diode spikes are now gone as is the buzz.

So consider a stock radio that had 300v B+ back in 1955. Today the same radio might have 350v B+ or more. Convert to SS, and we have 400v or more. In both cases the radio runs HOT. My approach above will fix the high B+ problem, and the thermal HOT issue.

Another issue to mention is the power transformer. Some vintage iron goes into core saturation as the AC line voltage rises above 120V. This will increase the primary excitation current, and the transformer will get very HOT. In these situations we have to dial down the line voltage somehow. This could be with a variac, or other means such as using the unused 5v winding (SS diodes) to buck the primary voltage.

Nothing has been stated yet about filaments today often run at 7 volts.........

Jim
WD5JKO

[KM1H]

One step at a time Jim!

Hallicrafters are especially known to be light on the iron and Ive measured many in partial saturation at 122-123V and have mentioned this for years on many forums. National and others have their issues also with certain models. The solution is a 8-13V bucking transformer at the line input. The only criteria is that the secondary winding be able to handle the total current draw of the radio and throw in a 1.5 to 2X safety factor.

I have small transformers mounted in small boxes at every place I use a vintage radio, these plug into the wall and can handle at least 200W. In the shop and hamshack all benches are fed with their own big transformer in the 12-20A range. Surplus and hamfest iron of odd voltages as well as no CT are dirt cheap. At 108-113V everything runs cool and with filaments in spec. The secondary benefit is that going to SS rectifiers often doesnt require any additional fudging.

The bucking (or boost) idea is nothing new and there were commercial products available back into the 40's at least. Perma Power is one example that had a -10, 0, and +10 switch position and rated at 300W. Ive picked up several at hamfests for $1-2. One feeds both a SX-62A and a HQ-180 right next to me here at the computer desk upstairs and both are often turned on.

Ive used a single .001 to .005uF caps across diodes (only 1 cap across the whole leg if its a series string) and another at the output to ground to kill noise. Dont know if it was the buzz you encountered Jim as these were on modulator and amp PS. I use 1N5408's almost everywhere since I scored about 10K of them on reels about 20 years ago for $20 total 

About 2 years ago during a major utility upgrade my AC has dropped to 119V with the benefit of less lamp replacements. I still use the same bucking iron as when I was getting 125V.

Carl
KM1H