The AM Forum
October 09, 2024, 09:00:23 AM *
Welcome, Guest. Please login or register.

Login with username, password and session length
 
   Home   Help Calendar Links Staff List Gallery Login Register  
Pages: [1]   Go Down
  Print  
Author Topic: Swinging Chokes  (Read 19867 times)
0 Members and 1 Guest are viewing this topic.
KB3DKS
Member

Offline Offline

Posts: 176



WWW
« on: March 02, 2009, 03:26:45 AM »

 Is there any real advantage to using Swinging Chokes in a homebrew rig these days? I rarely see them in more modern designs.
  I find that I have several good size ones, 4/20 mh 225, 400, 550, ma. Mostly UTC "S" series. Others as well.
  Would they be good for choke input L-C-L type supplies as they used to be used or are they really just a waste of space when L with a bigger C will do these days?

  I know that they can't be used for heising, obviously, but does the size of a regular supply choke influence the size of the heising choke at all?
  Say the supply is a L(swinging), a C, 20-40 ufd. then a 10 mh L.
Unless there is another C would not the output L add to a heising somewhat if that type of modulation was used?

 Excuse me, I'm all choked up right now  Wink

Bill, KB3DKS in 1 Land
Logged
N3DRB The Derb
Guest
« Reply #1 on: March 02, 2009, 08:14:49 AM »

I think a PS with a swinging choke is needed in one circumstance: when you need a modulator PS. You need the added regulation and scrotefulness when dealing with large current excursions that typically happen in a plate modulator : going from 20~30ma zero drive to swingin the monky to 400~500 ma along with yer voice peaks. with a RF deck its not an issue because you have no fast big current changes like you do in a modulator supply. I sez use yer swingin chokes in yer mod supplies and just use smoothies + caps in yer rf deck supplies.

summary: any rapid zero or near zero to full scrote current swings, USE THAT SWINGER!

http://www.resourcerags.com/astro/sst_swingmonkey.shtml <--- gotta get me one of these
Logged
flintstone mop
Contributing
Member
*
Offline Offline

Posts: 5048


« Reply #2 on: March 02, 2009, 08:53:50 AM »

Hows about a schwangin choke in a leanyear?? That swings the munkey too from the voice peaks.
Phred
Logged

Fred KC4MOP
N3DRB The Derb
Guest
« Reply #3 on: March 02, 2009, 09:08:38 AM »

on slopbucket, but on am there's always some current draw due to the carrier. not as critical an issue.

I'd also say use the choke if you use one supply for both RF and modulator service.
Logged
WBear2GCR
Member

Offline Offline

Posts: 4142


Brrrr- it's cold in the shack! Fire up the BIG RIG


WWW
« Reply #4 on: March 02, 2009, 09:25:50 AM »



Not so sure about the swingin' chokes, but them groovy chokes be happenin' baby!!

                          _-_-bear
Logged

_-_- bear WB2GCR                   http://www.bearlabs.com
Tom WA3KLR
Contributing
Member
*
Offline Offline

Posts: 2127



« Reply #5 on: March 02, 2009, 12:03:32 PM »

Bill,

I consider the swinging choke thing to be mostly folklore, the culture carrying over from the old days in ham radio.  I think that the intent of the swinging choke design was to make a more economical choke for a given design; a choke that had higher inductance and realizing that you can get away with lower inductance at higher currents.  However, the old catalogs that I have looked in and picked out a swinging choke and a corresponding smoothing choke that would be chosen instead to do the job, the price was not significantly different!

Yes the part exists but they are rare today.  Most chokes made are not swinging chokes, they are called smoothing chokes.  There is NO ELECTRONIC DESIGN REASON that actually REQUIRES a SWINGING choke.  There is the requirement for the choke to be above critical inductance at all conditions in a choke-input filter.  (And the more inductance you have, the more ripple reduction you get.) Since the smoothing choke does not drop in inductance as much as the swinging choke, there is more ripple filtering at the higher currents, you do get what you pay for.

The thing you have to remember about the B+ L-C filters is that they are an L-C circuit; have a low frequency resonance and the output voltage will ring.  As Derb says, applications where the load has extreme changes, the B+ voltage swings will never settle down.  For receivers and steady current stages of a transmitter, the L-C filters are fine.

In the old days, the filter capacitance available was low - paper caps were big and the aluminum electrolytics were new and big for the value.  Today we can get very high capacitance electrolytic filter caps, something that was not available in the old days.  A capacitance-only filter is the way to go for high current swinging loads.  This is what the linear amplifier B+ supplies are.  This gives the stiffest voltage regulation for a passive supply.  With this capacitance-only filter, the peak repetitive currents the rectifiers see may become high, you can add a small inductance value of say ¼ to ½ the critical inductance value ahead of the capacitor bank.  This is rarely done, but the filter will still act as a capacitor input filter and is a way to get the peak repetitive currents within spec. 

The circuit simulator programs are great for examining the B+ filter situation – the voltage swings at key-up and key-down and the peak currents of the rectifiers.  To accurately model the situation, the d.c. resistances of the transformer secondary and the filter chokes must be included in the circuit.

…My take on the use of swinging chokes today.
Logged

73 de Tom WA3KLR  AMI # 77   Amplitude Modulation - a force Now and for the Future!
W4EWH
Member

Offline Offline

Posts: 825



« Reply #6 on: March 02, 2009, 12:32:34 PM »

IIRC, swinging chokes were traditionally used with MV rectifiers, to limit inrush current before the tubes fired.

YMMV.

Bill W1AC
Logged

Life's too short for plastic radios.  Wallow in the hollow! - KD1SH
N3DRB The Derb
Guest
« Reply #7 on: March 02, 2009, 03:54:47 PM »

Quote
the culture carrying over from the old days in ham radio.

well, yeah. he's got a bunch of parts from the old days in ham radio. He should use them since he already has them.

I submit that todays amp makers use cap input supplies because caps are cheaper than good iron, not because cap supplies are inherently better. I agree with 98% of what ya say otherwise. With today's caps, you dont need swinging chokes - but if ya got the real deal and you're buildin a HB rig, why not?
Logged
W2XR
Contributing
Member
*
Offline Offline

Posts: 859



« Reply #8 on: March 02, 2009, 05:47:04 PM »

Bill,

I consider the swinging choke thing to be mostly folklore, the culture carrying over from the old days in ham radio.  I think that the intent of the swinging choke design was to make a more economical choke for a given design; a choke that had higher inductance and realizing that you can get away with lower inductance at higher currents.  However, the old catalogs that I have looked in and picked out a swinging choke and a corresponding smoothing choke that would be chosen instead to do the job, the price was not significantly different!

Yes the part exists but they are rare today.  Most chokes made are not swinging chokes, they are called smoothing chokes.  There is NO ELECTRONIC DESIGN REASON that actually REQUIRES a SWINGING choke.  There is the requirement for the choke to be above critical inductance at all conditions in a choke-input filter.  (And the more inductance you have, the more ripple reduction you get.) Since the smoothing choke does not drop in inductance as much as the swinging choke, there is more ripple filtering at the higher currents, you do get what you pay for.

The thing you have to remember about the B+ L-C filters is that they are an L-C circuit; have a low frequency resonance and the output voltage will ring.  As Derb says, applications where the load has extreme changes, the B+ voltage swings will never settle down.  For receivers and steady current stages of a transmitter, the L-C filters are fine.

In the old days, the filter capacitance available was low - paper caps were big and the aluminum electrolytics were new and big for the value.  Today we can get very high capacitance electrolytic filter caps, something that was not available in the old days.  A capacitance-only filter is the way to go for high current swinging loads.  This is what the linear amplifier B+ supplies are.  This gives the stiffest voltage regulation for a passive supply.  With this capacitance-only filter, the peak repetitive currents the rectifiers see may become high, you can add a small inductance value of say ¼ to ½ the critical inductance value ahead of the capacitor bank.  This is rarely done, but the filter will still act as a capacitor input filter and is a way to get the peak repetitive currents within spec. 

The circuit simulator programs are great for examining the B+ filter situation – the voltage swings at key-up and key-down and the peak currents of the rectifiers.  To accurately model the situation, the d.c. resistances of the transformer secondary and the filter chokes must be included in the circuit.

…My take on the use of swinging chokes today.

Tom makes a very good point with regard to the power supply ringing issue associated with LC filtering.

This characteristic is also referred to as power supply bounce. Cost issues aside, I think it is also a major reason as to why high voltage power supplies of modern design do not use LC filtering for applications involving highly dynamic loads.

Correlated to this, it is worth noting that many plate modulated rigs cannot easily pass the clipped output waveform of some of the more sophisticated state-of-the-art audio processing schemes utilized by the likes of Orban, Omnia, et al, due to the ringing and overshoot that is created by the inherent complex non-linear impedances within the transmitter input and driver transformers, modulation transformer, modulation reactor and DC blocking capacitor, power supply LC filter network, and attendent power supply bounce issues, etc. For example, the Orban AM processors employ a soft clipping scheme that adds significant density to the modulation, but transmitters with the complex L or LC networks described above may have difficulty achieving a consistently high percentage of positive-going modulation peaks without hitting the baseline. This is due in large part to power supply bounce.

There was an excellent technical white paper written in the mid-60s on this very subject by an engineer who was employed by Electro Engineering Works. Electro wound much of the modulation and power supply magnetics used in the plate modulated broadcast transmitters of that era. This paper essentially described the advantages of eliminating power supply chokes from this type of transmitter design and going with a capacitor-only filter network, so as to significantly reduce the power supply bounce that held down the positive peak capability in these rigs. This would only work if the power supply ripple frequency was high enough to provide adequate ripple reduction on the order of -60 dBc or so.; as such, 12-phase rectification was required to get the ripple frequency to 720 Hz as I recall.

Such a power supply change would also permit the successful use with plate modulated transmitters of the more aggressive and effective broadcast audio processors that were just begining to come to market at that time.

Obviously, 12-phase rectification is not practical in an amateur radio transmitter, where we are limited to single-phase residential prime AC power. But with inexpensive banks of electrolytic capacitors configured to provide adequate voltage rating consistent with sufficient capacitance to provide the necessary amount of ripple reduction in a full-wave rectification circuit, it is indeed worth considering such a design to eliminate the power supply choke in a homebrew plate modulated transmitter.

73,

Bruce
Logged

Real transmitters are homebrewed with a ratchet wrench, and you have to stand up to tune them!

Arthur C. Clarke's Third Law: "Any sufficiently advanced technology is indistinguishable from magic".
Steve - WB3HUZ
Guest
« Reply #9 on: March 02, 2009, 07:10:58 PM »

Dynamic regulation.

http://www.amwindow.org/tech/pdf/geps1.pdf
http://www.amwindow.org/tech/pdf/geps2.pdf


No chokes.

http://www.amwindow.org/tech/pdf/electroimpulse.pdf
Logged
k4kyv
Contributing Member
Don
Member

Offline Offline

Posts: 10037



« Reply #10 on: March 02, 2009, 08:44:54 PM »

The swinging choke is useful when the current load on the power supply varies from a very low value to the full rated load.  Examples would be a CW transmitter where the plate current of the final is at or near cut off under key-up conditions, a class-B modulator in a transmitter with separate rf final and modulator power supplies, and a slopbucket leen-yar.

I found this out the hard way, by trial, error, observation with the scope and experimentation, before reading the G-E Ham News article, which would have saved me some long hours of grief.

I used the swinging choke with a CW transmitter, and no matter what, the keyed waveform at the transmitter output sucked to say the least.  The plate voltage meter showed about a 10% variation in plate voltage from key-up to key-down.  The problem is that the mechanical meter movement cannot follow the actual instantaneous variations in the plate voltage, but the actual plate voltage bounces around over a much wider range and the meter movement averages it out.  I found that the best way to achieve a half way decent keyed waveform was to use a very large filter capacitor in the output, and to use as low a value of bleeder resistance as possible.

I ran into the same situation with my Gates BC1-T when I modified it for CW capability.  With plenty of inductance in the filter choke (65 henries or so), the meter showed the voltage dropping from 2600 volts down to 2500 from zero (key up) to 600 mills (key down).  But the waveform was horrible looking, so I rigged up a scope to indicate plate voltage, and I discovered that it was momentarily dropping down to about 1500 volts when the key was first pressed down, and kicking up as high as 4000 volts when the key was released.  I changed out the stock 8 mfd PS filter cap to 25 mfd, and replaced the 100k bleeder resistor with a 50k one, and that put the dynamic regulation under control.  In the HF-300 rig I use separate power supplies, and each one uses a swinging choke that runs about 8-35 henries over a range of 500 ma to 50 ma load.  Each choke is nominally rated at 5-25 hy at 100/1000 ma.  The modulator supply, which runs at higher voltage, use a 30k bleeder resistor, while the PA supply uses a 25k bleeder.  Each supply has 28 mfd of power supply filter capacitance - as much as I could get away with without having to resort to step-start.

If the same power supply is used to power the modulator and rf final, there is no need for a swinging choke, because the RF final bleeds down the supply heavily enough that the load variations induced by the modulator are of negligible consequence.  My 8005 rig shares a common power supply with the 805 modulators.  I use 50 mfd of filter capacitance, a 10 hy 1-amp choke (forget exactly how much bleeder resistance - about 25K I would guess), and there is very little voltage variation resulting from the class-B modulator load.  In cw mode, the class-B driver (a pair of 845's) maintains about 150 mills of minimum load, plus the load of the bleeder, and the CW waveform is better than that of either of my two other rigs, with less than 100 volts drop key down according to the meter.

Some of the pre-WW2 transmitter circuits I have seen published used no more than 2 mfd of power supply filter capacitance, along with a swinging choke.  In those days an oscilloscope was rarely available to the average amateur, so ignorance was bliss.  The cw waveform and modulation peaks would have sucked big time if the operator could have only seen them.  That explains one reason why so many of those old rigs sounded like crap even though high quality components might have been used, and why in those days, a common power supply for modulator and final should have always been used.

But with a choke-less power supply, voltage regulation may be an issue, particularly with an older plate transformer designed for use with a choke input supply.  Unless the leakage reactance of the plate transformer is extremely low, the capacitor input supply tends to have very poor regulation, with the DC output varying wildly between the peak (at minimum load) and r.m.s. value (at maximum load) of the a.c. output from the transformer.

With older iron, better to use choke input, swinging if necessary, with plenty of filter capacitance while maintaining as much load on the supply at all times as possible.  This type of supply tends to crap out solid state diode rectifiers, so I would recommend tube type rectifiers or a set of commercially made modules designed to directly replace M.V. rectifier tubes.  I never had much luck with strings of solid state diodes as rectifiers in HV power supplies.

If the transformer is  good enough to maintain good regulation without a filter choke, solid state rectifiers or high-vacuum tube type rectifiers should be used, because the extremely high peak current pulled through the rectifier by the filter capacitor will destroy gaseous rectifier tubes in no time flat.
Logged

Don, K4KYV                                       AMI#5
Licensed since 1959 and not happy to be back on AM...    Never got off AM in the first place.

- - -
This message was typed using the DVORAK keyboard layout.
http://www.mwbrooks.com/dvorak
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8290



WWW
« Reply #11 on: March 02, 2009, 10:15:34 PM »

I got to side with Don's findings on general regulation, although I did no pulse load testing (CW test). Having a chance to totally experiment with an adequate power supply and a modulator capable of eating 3000V/500mA. I found:

C filter - 40uF supply had poorer regulation when modulator was driven even when bled with 88mA (34KOhms at 3KV).

LC filter - 4H-40uF supply had better regulation with same bleed, as long as I kept the chokes above the current for the Lcrit, which is about 60mA.

I'm running CLC with 8uF-4H-32uF and the 88mA bleed. The regulation is OK and I get a few more volts.

None are swinging iron. The 4H is made of two 8H 400mA chokes in parallel. I do not know what would happen if I put them in series, and hit the 600mA peaks. Probably nothing bad. If I had swingers I would put them in the modulator.

Here is a chart "pevsrl.gif" showing actual measurements done scientifically on a choke input and capacitor input power supply. The same hardware was used, and the primary voltage of the HV transformer was adjusted to give the same starting points for HV. To be sure there was no error from other impedances and the like, the efficiency of each component was measured (not shown here) and the input voltage of the HV transformer was kept constant, so this test is as scientific as I know how to do. For the load, a 4-1000A and a pair of 3-500Z's all in parallel were employed as a variable resistor. It is easy to see that the C-input filter does not regulate nearly as well as the L-input filter.

The chart "pevsrl.gif" shows clearly that other things being the same, the LC filter delivered more voltage (and wattage) to the load. The load was varied from 8000 Ohms down to 3000 Ohms.
L-input filter = -5.4% regulation
C-input filter= -11.5% regulation

The chart "volts-vs-res.gif" shows how the output voltage varies (steady 120VAC to the HV transformer in each case and same variable load) as the C and LC filters are loaded. Considering 110mA to be the resting load, the C filter has a far more pronounced voltage drop with increasing load.
The regulation from about 112mA to 682mA for both filter configurations in this test:
L-input filter = -6.25% (2250 to 2047V)
C-input filter= -13.43% (3350 to 2900V)

This is why I will never be a fan of C-only filters where the load varies. In a slopbucket leenyar, whith no voice processing, it's not too much of a problem. But when processing is used and the voice starts looking like single tone signals with little mumps on them and dips between them, thats the end of it for the iron-less filter.

The variation in power supply voltage over the audio cycle has a definite impact on the overall linearity of the transmitter and /or amplifier.


I have no charts for swinging chokes. Because I do not have any swinging chokes with the gusto to be placed into the "tester".


* pevsrl.gif (42.22 KB, 1240x841 - viewed 540 times.)

* volts-vs-res.gif (61.92 KB, 1240x841 - viewed 576 times.)
Logged

Radio Candelstein
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8290



WWW
« Reply #12 on: March 02, 2009, 11:11:26 PM »

back to the question - Terman and Everitt have said nothing about swinging chokes, but in "The Radio Manual (5th ed. 1950)", Sterling and Monroe say that the "...high cost of a choke with such high inductance at light load, but large enough to carrythe full-load current, has led to the 'swinging choke' popularized by F.S. Dellenbaugh" (Jr.).

So we owe this quandry to Mr. Dellenbaugh Jr.. According to "Modern Radio Reception, CHARLES R. LEUTZ, (1928)" He was head of the Research Division at MIT. Among other things he investigated ways to measure the distributed capacitance of Vitrohm resistors. The capacitance was found to be less than the parasitic or error of the finest instruments of the day. If that ain't slick I don't know what is. I suppose we can take the gentlemens' words written in the ancient and dusty tomes at face value on the swinging chokes.
Logged

Radio Candelstein
W2XR
Contributing
Member
*
Offline Offline

Posts: 859



« Reply #13 on: March 02, 2009, 11:31:54 PM »

In a slopbucket leenyar, whith no voice processing, it's not too much of a problem. But when processing is used and the voice starts looking like single tone signals with little mumps on them and dips between them, thats the end of it for the iron-less filter.


Hi Patrick,

Could you clarify this statement for me? I do not understand your observation as to how the human voice with audio processing can look like a single tone signal. I'm not trying to be argumentative here; I just need some elaboration!

Please note that I am not claiming to be any kind of technical expert on the subject of high voltage power supply design and the subtleties associated with it. I do in fact use LCL filtering in my homebrew 2x 4-400As modulated by a pair of class B 833As. I'm using a pair of Wilkinson plug-in solid-state rectifier stacks pulled from a Gates BC-1J transmitter, the input swinging choke is a 5-16 Hy at 1.2A choke out of a Gates BC-1G, and the smoothing choke is an 8 Hy at 1A choke from some unknown  GE broadcast rig. The swinging choke supports the 833A modulator, and the combined swinging and smoothing chokes support the the pair of 4-400As HPA stage. The capacitance behind the swinging choke is 38 uf, and I have 8 uf after the smoothing choke. The bleeder resistor value is 100K ohms. The plate transformer is from a Gates BC-1G; it is rated at 6200 VCT at 1.2 amps.

I am not in any way stating that the LCL filter arrangement I am using is the optimal design for this particular application. In fact, I do have known issues with power supply bounce in my rig (not to be confused with carrier shift, which is very small in my transmitter); as such, and unfortunately, this does not allow effective use of my Orban Optimod AM processor. For me to eliminate the existing LCL filter and add the very large amount of filter capacitance required to provide adequate ripple suppression and dynamic regulation would entail replacing the plate transformer with one that would provide about 2700 VDC into a capacitor input filter. I don't have one presently. If I did, I'd certainly give it a try. I have not calculated the minimum required value of C to support these requirements, but you can safely assume it will be very large, probably on the order of at least 250 uf for a ripple frequency of 120 hz. That's a lot of energy storage at 2700 VDC.

However, I still believe that the technical white paper I described in my earlier post (and which Steve/HUZ graciously provided the link to) does make an extremely compelling argument based upon the sound engineering logic defined by the author for the elimination of filter chokes within a high-power plate modulated transmitter.

I do agree with Don/K4KYV's statement that LC filtering is essential in a radiotelegraph transmitter. The dynamic load variation is much more severe in this case, compared to a situation where a common power supply is supporting both the class B modulator load, and the high/constant current demand class C HPA load which obviously improves the regulation.

Just my 2 cents.

73,

Bruce
Logged

Real transmitters are homebrewed with a ratchet wrench, and you have to stand up to tune them!

Arthur C. Clarke's Third Law: "Any sufficiently advanced technology is indistinguishable from magic".
Tom WA3KLR
Contributing
Member
*
Offline Offline

Posts: 2127



« Reply #14 on: March 03, 2009, 08:07:42 AM »

There are 2 "arguments" going on here.  One is choke versus no choke.  The other is the topic of this thread - smoothing choke versus swinging choke.  I haven't seen any objective A/B comparison and results.
Logged

73 de Tom WA3KLR  AMI # 77   Amplitude Modulation - a force Now and for the Future!
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8290



WWW
« Reply #15 on: March 05, 2009, 01:27:36 AM »

Bruce,

I'm no power supply expert either. I merely experiment.

On the slpbucket comment what I meant is that the average power consumed by the slopbucket linear goes up alot when the voice is loud and its driven hard. In extreme cases, the output looks almost like full CW output but with little peaks going up and steep troughs going down - or worse when some yahoo whistles into the mike. I do not have a picture to describe what I am trying to express. Anyway, just speaking of the load increase from hard use and its effect on regulation.

My comparisons above for the power supply, are very specific to DC regulation differences between LC and CLC filters and not to other problems like bounce. I can't argue the point of bounce but I am reading this with much interest. It would be interesting to check that characteristic in my own rig.

There is also the transformer angle. An old transformer may have higher resistance windings than desired for the best regulation with a C input. None of it was intended to disparage against the article in question. If I could find a pair of HV transformers with 5000VCT and a DCR of no more than 50 ohms or so per side, I could consider swtching to C filtering. I got to run what I brung..
Logged

Radio Candelstein
k4kyv
Contributing Member
Don
Member

Offline Offline

Posts: 10037



« Reply #16 on: March 05, 2009, 03:31:52 AM »


There is also the transformer angle. An old transformer may have higher resistance windings than desired for the best regulation with a C input. None of it was intended to disparage against the article in question. If I could find a pair of HV transformers with 5000VCT and a DCR of no more than 50 ohms or so per side, I could consider swtching to C filtering. I got to run what I brung..

Leakage reactance can degrade the voltage regulation of C input just as much as DC resistance.  A xfmr with high leakage  reactance will work better into a choke input filter.  High leakage reactance means loose coupling between windings, and conversely, low leakage reactance means the transformer has very tight coupling between primary and secondary.
Logged

Don, K4KYV                                       AMI#5
Licensed since 1959 and not happy to be back on AM...    Never got off AM in the first place.

- - -
This message was typed using the DVORAK keyboard layout.
http://www.mwbrooks.com/dvorak
KB3DKS
Member

Offline Offline

Posts: 176



WWW
« Reply #17 on: March 05, 2009, 04:09:58 AM »

 It has been a good thread ! and lots learned.
So swinging chokes are obsolete and really only of use for period gear restoration and building. Large capacitors are plentiful and Iron is expensive.
  A smoothing choke is still useful especially to take up some of charging current of the caps, to a point. For Amateur use an LC filter of proper design still has applications for most good classic design transmitters. Broadcast and other types of always on amplifiers/finals using step start can get away with large amounts of capacity with or without the L.
 I'm simplifying greatly here but my initial question has been answered.
Now, as I have more of a focus on what is needed it is probably best to start a new topic regarding methods of T/R switching since large capacity filtered supplies cannot be easily just PTT due to the high charging current and somewhat long voltage decay upon switching to receive. Here is where the leakage inductance and actual supply impedance comes into play. I can't see just slamming on a plate transformer without having to deal with quite a current spike.
  This can be really an issue of correct sequencing and makes break in difficult. So will start another topic soon about this issue rather than diverge from the initial subject.

I am only able to use the stuff that I have accumulated and that is it. Just figuring out what pieces will work best together to assemble a good working 250 watt class or so rig.

Thanks all !
Bill, KB3DKS in 1 Land
Logged
k4kyv
Contributing Member
Don
Member

Offline Offline

Posts: 10037



« Reply #18 on: March 05, 2009, 03:50:42 PM »

There are 2 "arguments" going on here.  One is choke versus no choke.  The other is the topic of this thread - smoothing choke versus swinging choke.  I haven't seen any objective A/B comparison and results.

It's a matter of economy.  There would be no reason to use a swinging choke if you had a smoothing choke with enough inductance to maintain critical at minimum load.  But such a choke would likely be much larger, heavier and expensive than an appropriate swinging choke.

At full load, you want more than critical inductance.  I'd have to re-look up the formulae to calculate it, but I recall that there is an optimum inductance for a power supply under load.  Using optimum inductance, which is substantially higher than critical inductance, reduces the peak current to the minimum possible value at maximum current draw, which is important if you are using gaseous rectifier tubes.

The classic circuit used a LCLC filter network, in which the first L was a swinging choke and the 2nd was a smoothing choke.  But as pointed out in the G-E Ham News article, you will probably get better dynamic regulation by seriesing the two chokes and paralleling the two capacitors to make up a single section filter.

However, for filtering out ripple voltage at the output of the PA power supply, the two section filter is much more effective.  This is somewhat of a moot point for an AM transmitter that uses a modulation reactor, with the modulation transformer secondary returned directly to ground through the coupling capacitor.  In this case, the mod reactor provides additional filtering of the ripple voltage.  But if the final is used for CW, RTTY or slopbucket leen-yar service, the hum problem may be more apparent. If the mod transformer secondary is returned to the B+ instead of directly to ground, the power supply ripple voltage will will appear at the PA via the modulation transformer secondary.

Ripple hum is less a problem for a push-pull modulator, because a small percentage of ripple in the DC to the modulator is inherently cancelled out by the balanced push-pull circuit; the voltage is simultaneously raised and lowered at the plates of both tubes simultaneously, so there is no differential in plate voltage on the two tubes to show up on the other side of the modulation transformer.  But a large amount of ripple in a class-B modulator will show up as 120~ intermodulation mixed in with the audio, because in class-B, under high amplitude modulating conditions, only one tube is conducting at a time and for that half-cycle the amplifier is acting like it is single ended, and a poorly filtered supply will cause intermodulation of the audio, even though the ripple is balanced out and no hum is audible in the absence of a modulating signal.

All chokes are swinging chokes to a  certain extent, because even a smoothing choke is subject to some core saturation under DC load, which reduces the inductance.  The difference between a swinging and smoothing choke is the size of the gap in the core.  The smoothing choke uses a larger gap, which reduces the inductance, but that inductance holds up well at maximum current load.  A swinging choke uses a much smaller gap, which allows the inductance to rise to a higher value at little or no load, but with the minimal gap, the core has a greater tendency to saturate even under moderate load, so at full load the inductance may be a small fraction of what it is at minimum load.

Many of the popular xfmr manufacturers like UTC made power supply chokes in pairs - swinging and smoothing.  The coils and cores were identical, and the only difference was the spacing of the gap.  For example, the PA or CG 1s and 1c is a typical example of this.  The smoothing choke is rated at 10 Hy @ 1 amp, while the swinging choke is 5/25 Hy @ 1000/100 milliamps.

I use a couple of those swinging chokes in one of my homebrew rigs, but I  run it at reduced ratings, more like 500/50 milliamps, and the best I have been able to measure gives me a figure of 35/8 Henries over that range, which is much better than the typical 5/25 Hy 500 ma choke.

Logged

Don, K4KYV                                       AMI#5
Licensed since 1959 and not happy to be back on AM...    Never got off AM in the first place.

- - -
This message was typed using the DVORAK keyboard layout.
http://www.mwbrooks.com/dvorak
Tom WA3KLR
Contributing
Member
*
Offline Offline

Posts: 2127



« Reply #19 on: March 05, 2009, 06:56:38 PM »

The 1954 GE power supply article part 2 made a “deluxe” 1500 Volt  250 ma. dc power supply by putting 3 chokes in series.  Only one was a swinging choke, thereby diminishing the total inductance swing ratio.  The one swinging choke is a Stancor C-1720 4 - 20 Henry and the other two are UTC S-31 smoothing chokes which are rated at 6 Henrys in the 1960 UTC catalog, incorrectly stated in the article as “20H”. Error, or a spec change by 1960?  (I don’t see any 20 H. choke in their list above 100 ma.)  They were tested for temperature rise and pushed to higher current than rated so the drop in inductance would be greater and something reasonably lower than 6 Henrys, perhaps a 2:1 change over the current range ( ~ 10 – 5 Henries?) for a total of ~ 40 to 14 Henrys.  So they were expecting more than 60 Henrys at no load, bleeder current only, but only about 40 Henrys.  QTF?  The bleeder current should have been about 20 ma. no load.  At 1550 V. this is 77.5 Henrys for the critical inductance.  At full load, 250 ma + 19 ma bleeder is a critical inductance of 5.6 Henrys.  This they exceed by a factor of  2.5:1.

In the 1961 Stancor catalog the C-1720 300 ma.is $18.35.
In the 1960 UTC catalog the S-31 is $6.90.
They could have picked the swinging choke from the UTC catalog also:
S-32 20/4 H @ 225 ma. $6.90 also, (same price as the smoothing choke) or,
S-34 20/4 H @ 300 ma. $9.60.  UTC much better value than Stancor apparently!  (Was GE “spreading out” the business in that article?)  So here it looks like the swinging choke is a little more expensive, but not much more if you stuck with UTC instead of Stancor. 

Below is a pdf of a choke study I did a couple years ago.  All iron chokes appear to follow the same generic line shape; one generic line can be scaled to any choke.  The smoothing chokes reach their rated current at a point just at the knee of the curve.  Swinging chokes operate to much farther down past the knee.  I have seen swinging chokes with rating of up to a 6:1 change in inductance at the rated current test points.  Smoothing chokes generally have about a 1.5:1 change in inductance over the rated current range.

The red line for choke 8 is the only known bonifide swinging choke in my study; a Stancor C-1402, $11.62 in 1961.  It is rated at 12 Henrys @ 25 ma. (I measured 13.5H) and 2 H. @ 250 ma. (I measured 3 H.).  You can see that the red line and magenta line chokes are headed to an asymptotic inductance value at the highest currents measured..  I pushed a number of the chokes past their rated maximum currents for the measurements.

Graph information:

Choke 4  Thordarson T20C54 rated 6H. @ 150 ma. R meas. 148 Ohms.
Choke 5  SNC 2P144 rated 10.6 H. @ 113 ma. R meas. 200 Ohms.
Choke 6  “C1355” R meas. 313 Ohms.
Choke 7  20C89? Rated 12 H @ 110 ma. R meas, 237 Ohms.
Choke 8  Stancor C-1402 swing choke 2H @ 250 ma., 12H at 25 ma. R meas. 59 Ohms.

Most chokes were receiver types.

* choke_study1.pdf (7.99 KB - downloaded 268 times.)
Logged

73 de Tom WA3KLR  AMI # 77   Amplitude Modulation - a force Now and for the Future!
Tom WA3KLR
Contributing
Member
*
Offline Offline

Posts: 2127



« Reply #20 on: March 05, 2009, 07:13:31 PM »

Graph information:

Choke 4  Thordarson T20C54 rated 6H. @ 150 ma. R meas. 148 Ohms.
Choke 5  SNC 2P144 rated 10.6 H. @ 113 ma. R meas. 200 Ohms.
Choke 6  “C1355” R meas. 313 Ohms.
Choke 7  20C89? Rated 12 H @ 110 ma. R meas, 237 Ohms.
Choke 8  Stancor C-1402 swing choke 2H @ 250 ma., 12H at 25 ma. R meas. 59 Ohms.

Most chokes were receiver types.
Logged

73 de Tom WA3KLR  AMI # 77   Amplitude Modulation - a force Now and for the Future!
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8290



WWW
« Reply #21 on: March 06, 2009, 02:37:08 AM »

Tom, I find your choke study very enlightening. I see no reason it could not apply to larger chokes like the UTC S-35, an 8H 450mA 60 Ohm choke rated 5KV.

As the load increases and the voltage from the choke input filter drops due to the transformer etc, then even the non-swinging choke could help if overloaded slightly. That's one thing I take from your study.

I also went back and simuated the power supply as a LC filter, and there was an unaceptable bounce when the L was high and the current was stepped.
Logged

Radio Candelstein
Tom WA3KLR
Contributing
Member
*
Offline Offline

Posts: 2127



« Reply #22 on: March 06, 2009, 07:47:01 AM »

Yes Patrick, thanks, I thought people would find it interesting.

Unfortunately don’t think I can get my simulator to do a swinging choke.  Maybe it can be done, but I haven’t put thought into it so far.

I would like to see the bounce difference between L-C filters where it is say, the 5 - 20 H. choke versus the 20 H smoothing choke.

The reason I went back to the study is to see if I could match up a pair of smoothing and swinging chokes on hand for a comparison.  I don't have a storage scope or digital scope here so would have to use my digital camera on long exposure like done for the 1954 GE article; perhaps possible but inconvenient.
Logged

73 de Tom WA3KLR  AMI # 77   Amplitude Modulation - a force Now and for the Future!
KB3DKS
Member

Offline Offline

Posts: 176



WWW
« Reply #23 on: March 06, 2009, 01:10:25 PM »

Tom, I find your choke study very enlightening. I see no reason it could not apply to larger chokes like the UTC S-35, an 8H 450mA 60 Ohm choke rated 5KV.

As the load increases and the voltage from the choke input filter drops due to the transformer etc, then even the non-swinging choke could help if overloaded slightly. That's one thing I take from your study.

I also went back and simuated the power supply as a LC filter, and there was an unaceptable bounce when the L was high and the current was stepped.

 Pat,
If you increase the C is there a point where the transient sag levels out?
  The point of diminishing returns.... for Amateur use.

Bill
Logged
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8290



WWW
« Reply #24 on: March 08, 2009, 09:09:41 PM »

I used Duncan Amps PSUD II. It's free and lets you put a current source as the load and step it.

These combinations were tried in the simulator, stepping from 100mA to 800mA:

16H 40uF
16H 400uF
4H 40uF (actually 32uF)
4H 400uF

I did not run a series to discover at what point the capacitance adequately fixes the bouce for a given inductance. It does not permit swinging chokes.
If you go to www.duncanamps.com, you can DL it free and play all day.
Be sure to also DL the updated rectifiers.txt file because it contains more rectifiers than the default.
The BBS would not permit the upload of the simulation file due to an extension prejudice, but the program is so easy to use that you can make your own just be specifying the components.


* 620ma startup LC 4H 32uF.GIF (401.04 KB, 998x772 - viewed 583 times.)
Logged

Radio Candelstein
Pages: [1]   Go Up
  Print  
 
Jump to:  

AMfone - Dedicated to Amplitude Modulation on the Amateur Radio Bands
 AMfone © 2001-2015
Powered by SMF 1.1.21 | SMF © 2015, Simple Machines
Page created in 0.077 seconds with 19 queries.