How much filtering?

[W4RFM]

In this modulator I am going to build, how much filtering would you folks suggest?  I plan on a 5 Hy choke, then 4x 80's @ 450 in a string with resistors, resulting in 20 mfd at 1800 volts peak for a 750 volt supply.  Whadda ya think?

[KB2WIG]

Check these  out.......  for the most part , the more C u got, the better it goes.... ..

http://www.amwindow.org/tech/pdf/geps1.pdf

http://www.amwindow.org/tech/pdf/rcaclassB.pdf

and also

http://www.amwindow.org/tech/htm/modheising.htm

http://www.amwindow.org/tech/pdf/rcamoddesign.pdf


klc

P.S. the search for oil is worth it.

[KA2DZT]

I think you could just use three caps in series for 750 volts DC.  That would give you about 25mfds which in enough filtering for a modulator.

[WBear2GCR]

A bit depends on if ur talking a choke input or cap input C-L-C...

My limited experience with choke input is that often you need to "tune" the filter with a parallel cap for lowest ripple... just adding C to the output end can result in more ripple under load rather than less.

The C-L-C is easier to get ripple down with somewhat random parts values and just slug C at the second position... of course that depends on the power iron.

               _-_-bear

[W4RFM]

Thanks to all, this is good info.  I am printing out the am window pages Kevin. Thats a lot of good info, shame of it all is, I was around during the tube era, but I have forgotten so much theory, having used these new fangled transistor things since the 70's.  When I was a bdcst engineer, I always told my younger co-workers, tubes are better, cause when they are unhappy with their conditions, they "show" it.  Unlike solid state devices that flake out and dont tell you. (Unless the paint burns off).

[flintstone mop]

In a BC TX I have seen as little as 10mf with choke input.
I googled a BC1H Gates TX 1KW out and 10MF was the highest value cap.
Maybe a linear amplifier around 25mf???

Fred

[ke7trp]

4mf will do it but 10 to 12 would good and have some overkill.

C

[W4RFM]

Great I could use some of these oil filled caps I have had around forever.  (I knew they would be needed some day). Thanks.

[ke7trp]

Agreed.  Find a big oil cap and use it.  There wont be any hum. Many of my transmitters all used 3 to 5 MF oil cans. More is better.  I would rather have an oil can then a string of electrolytic caps.  I have not had good luck using those.


C

[KA2DZT]

I only use oil caps in HV supplies.  With a 750VDC supply and using choke input filter (LC) you should use a 1000 - 1500 volt cap.

Add about 20ma bleeder resistor, about 50K, at least 50-100 watt.  Two 25K 50 watters in series is a good way to do it.

A 5hy choke may not be enough inductance to keep the unloaded supply voltage from jumping up to near the peak voltage.

Fred

[k4kyv]

I only use oil caps in HV supplies.  With a 750VDC supply and using choke input filter (LC) you should use a 1000 - 1500 volt cap.

Add about 20ma bleeder resistor, about 50K, at least 50-100 watt.  Two 25K 50 watters in series is a good way to do it.

A 5hy choke may not be enough inductance to keep the unloaded supply voltage from jumping up to near the peak voltage.

Fred

A 5 hy choke and 10 mfd might be enough for an AM rig that uses a common power supply for modulator and final.  But the modulation reactor also acts like an additional filter choke to reduce ripple still further from what comes out of the power supply. Most BC rigs and many ham rigs use a  reactor, but if you run the PA plate current through the transformer without a reactor, you may lose some of the advantage of this filtering, although the mod transformer secondary should add some filtering.

If you use separate power supplies for modulator and final, much more filtering is needed. The same applies to a class-C CW rig and a SSB linear.  You must maintain critical inductance in the choke at all times to prevent the DC output voltage from soaring towards peak voltage at minimum  load.  Critical inductance is calculated as a minimum of 1 henry for every 1000 ohms of load resistance. For example, take a CW rig with the final running class C biased to cut-off. Let's say it runs at 2500v/400 ma. With no additional bleeder resistor, the load is 6250 ohms key-down.  The minimum inductance to maintain critical would be 6.25 hy.  But under key-up condx, the voltage would soar to about 4 kv.  Add a 50k bleeder resistor, and you still need 50 hy of inductance. A better solution would be to use a 8-40 hy swinging choke, and reduce the bleeder resistance to 40k.  But a 8-40 hy swinging choke may be expensive and hard to find.  A more common value would be 5-25 hy @ 500 ma. Reduce the bleeder to 25k.  The voltage regulation should stay within 10% of the nominal output voltage.  But that is the static regulation.  Say you have 8 mfd of filtering. Monitor the output voltage with a scope or led voltmeter as you key the transmitter, and you will find that the voltage actually bounces up to almost 4 kv and down towards 1500v as you send code.  Looking at the envelope of the signal on as you key the transmitter will show a horrendously ugly keying waveform.  The easiest way to correct the voltage bounce is to use as much filtering in the power supply as you can, without having to resort to soft-start.  From my experience, the optimum value is about 25 mfd.  Even then, you will have some voltage bounce, but it will be tolerable. To completely eliminate it, you need about 100 mfd, which would definitely call for soft-start.

If the transmitter is to be used only for AM with a common power supply for modulator and final, the power supply requirements are less severe, since the RF final acts  as a heavy bleeder, and the bleeder resistor is used only for that - bleeding off the charge on the filter caps when the supply is turned off.  If that cw final as described above is plate modulated with the modulator connected to the same power supply, the constant load will keep the DC voltage at nominal, and the regulation will be excellent with modulation.  Of course, the power supply  components have to be capable of supplying the load drawn by the final, plus the modulator, plus the bleeder resistor.

One disadvantage to using a common power supply is that all tubes have to run at the same plate voltage, and this makes the mod transformer turns ratio critical to allow 100% modulation with some head-room.  The turns ratio should be somewhere between 1.2:1 and 1.5:1 to avoid too much step-down that  would limit the modulation  capability of the transmitter.  This ratio is regardless of the nominal plate-to-plate load impedance given in the tube charts for the modulator tubes.  If you cannot achieve an acceptable p-p load impedance within that ratio, the only satisfactory way to adjust it is by increasing or decreasing the plate voltage and adjusting the final plate current accordingly.  This may alter the power level you had expected to run with the transmitter.  P-p load is not magical; most modulator tubes will tolerate a substantial departure from the nominal load per the tube charts, but there are limits to how much variation you  can use without exceeding either the plate voltage or plate current  ratings of the tubes to maintain the  desired power level.

A SSB linear or separately powered class-B modulator acts more like the CW rig, except that the static plate current acts as additional bleeder current. In any case, it would be a good idea to use filter caps with a voltage rating at least a little above the peak voltage from the transformer secondary, just in case the final tubes fail, you lose filament voltage or the bleeder opens up, and voltage inadvertently spikes up to peak value.  Caps rated just a little above working voltage could fail due to overvoltage in the event the load fails.

[KE6DF]

 In any case, it would be a good idea to use filter caps with a minimum voltage  rating somewhat above the peak voltage value, just in case the final tubes crap out or  the bleeder opens up, and voltage inadvertently spikes up to peak value.  Caps rated just a little above working voltage could fail due to overvoltage in the event of a failure.

Another point is that with a choke input filter, without a soft-start, the transient voltage on startup on the output capacitor is nearly 1.4 times the RMS transformer output. Simulations show this occurs even with a bleeder of critical resistance.

The transient lasts only for a couple cycles, but still the output cap -- and all the circuits powered by the supply -- have to survive it.

[KA2DZT]

In this modulator I am going to build, how much filtering would you folks suggest?  I plan on a 5 Hy choke, then 4x 80's @ 450 in a string with resistors, resulting in 20 mfd at 1800 volts peak for a 750 volt supply.  Whadda ya think?

Don,

The OP indicates that the PS he is building is for a modulator.

No mention of a Class C stage being run from the same supply.  A class C stage adds plenty of steady state load.

Of course, for a modulator-only PS, a swinging choke is far better (5/25hy) than a 5hy choke.  I'm assuming the OP does not have a swinging choke.

So, he should try to use a larger choke or put two chokes in series to increase the input inductance.

We still don't know if the PS is going to be left on during stand-by or the PS will be keyed on and off.

But, for anyone building HV supplies, it's always a good idea to build the supply to withstand unloaded voltages.  Some bleeder current will help to keep the unloaded voltage from reaching the peak voltage.

Of course, if the bleeders fail, then you have no load (not good).  So, I always use bleeder resistors that have a wattage rating 4 or 5 times the dissipated power.

Fred

[k4kyv]

Another option is to resonate the 5 hy choke.  That makes it function like a brute force filter choke of much higher inductance.  Collins uses the concept in the KW-1.

The problem is finding the exact amount of capacitance needed and then acquiring a capacitor the proper value with enough breakdown voltage rating. This is the reason so few hams use it in homebrew supplies.  The a.c. voltage rating should exceed the peak ripple voltage by a substantial safety margin. For safety's sake I would use a voltage rating in excess of the peak a.c. voltage from the transformer. If one has a large collection of small value oil caps, one might be series-ed up to the appropriate capacitance, found by experimentation, and then begin the quest for a single cap of the proper value.  They required capacitance is usually less than 0.5 mfd.

[KA2DZT]

Caps from microwave ovens are a good start.  They are usually slightly under 1 mfd and have a few KW ac voltage rating.

I've found that you really can't rely on the rated inductance of most commonly used chokes.  I've done a lot of testing on these chokes and found that some are poor.  The worst ones are the UTC S series, in contrast, UTC PA series are some of the better chokes.  The older Thordarsons are very good but the last series they made are fair at best.  Stancor 300ma and 500ma chokes do a good job of filtering.  UTC's CG series are also good.  I've tested many others.

Fred