The AM Forum

Hello all,

What value of blocking capacitor do you like to use between your final amp and the pi network output tank? I have seen schematics with everything from 250pf to 4000pf. I have the option to use either a really large broadcast type 500pf@10kv or to use parallel type 850 doorknobs 500pf/5kv each (I have six of these I think).

Here are my operating parameters:
1. Bands - 160-80-40
2. Mode - high level plate modulation
3. Eb - 2000V
4. Ib - 250mA

Ron W8ACR

Hi Ron,

The coupling cap value requirement will be dependant on the plate impedance, of course. The lower the plate impedance, the bigger value the cap needs to be to keep RF voltage drop to a minimum at a particular frequency. Here's what I have used over the years:

For a standard 1500W+ linear amp for 160-10M, I have used .001 uf at 10KV.  Same for the plate bypass cap. Even .002 uf will work FB.  Whatever you have in high quality caps is usually OK.

However, for a plate modulated rig, we want to keep these two values to a minimum because the modulator audio highs can get shunted to ground thru either the plate bypass cap or the coupling cap. (thru the pi network)  These two undesirable audio paths are in parallel, shunting the audio, so be aware if your highs are down or you detect excessive audio phase shift on the modulated RF carrier..

So for plate modulating a 4-1000A rig, or any big AM rig, I use 500pf for both the plate bypass and plate coupling cap.  Offhand, this may not seem like enough for 160M, but it does the job FB and extreme audio highs see very little phase shift as a result.

In your case, with an 8K plate impedance,  500 pf at 1.885 mhz is about 168 ohms of (Xc) cap reactance. That is only about 2% of the plate impedance, so insignificant, which is FB.


BTW, what new creation are you building?

T
 

5000 volts a bit light. Heck a Viking 2 uses an 850.

Well Tom, as you know, I built this 254W rig, and have had it on the air enough times to be very happy with its performance. Since it was my first serious attempt at building something big, I purposefully kept it simple by keeping it a single band 160 meter rig. But to be able to talk to you guys on the east coast, I'm gonna need 75 and 40 meter capability. I've gone through about a hundred possibilities in my brain, but the only thing that really is practical is to modify the 254 rf deck to cover 160-75-40. So that is what I'm up to. I've got the grid circuit working on all three bands, but now I have to do the plate tank. I have this nice 500pf/10kv capacitor that I wanted to use, but wasn't sure it had enough capacitance, especially on 160. Hence my question. Once I'm done with this project, I'm gonna build a good old fashioned push pull transmitter using 35TG's. I've got B&W TVL coils for 160-10. It will probably run about 250watts. Too bad the 254W rig worked so well, now I've got the bug. ;)

On the 4-1000 it is 2500pF/30KV made of 5 doorknobs.

At 2KV plate voltage, 5KV is just making it, too light for me.  In my 813 rig at 1500V plate, I use two 510pfd 18KV doorknobs in parallel. I use two plate chokes in series, each bypassed with one 510pfd 18KV.  I don't see any problems at the higher audio freqs.

Even for the plate-choke bypass caps I think 5KV is not near enough.  Think about it, at 2KV Eb with plate modulation, at 100% modulation, the plate voltage hits 4KV.  Get some above-100% spikes and say bye bye to the 5KV caps.

In my 6146 rig at 650V Eb, I use a .001 6kv plate bypass cap and a 500pfd 10KV coupling doorknob.

Fred

If you want 160 and use electrolytics you may have to sacrifice a few highs if the plate choke doesnt have enough L. Another option is to add a big fat .0047 or .01 down at the PS to bypass any stray RF. Or use 2 chokes, each bypassed with 500pf.

One way around all of this is to buy an Ameritron 225uH choke o one of those old multi pi 1mH Nationals if just sticking to 160-40.

I wonder what happens to the audio waveform when there is RF on the mod iron or reactor which then travels back up thru the plate choke?

I use a pair of 1000pf @ 7.5KV in parallel in the linear for plate blocking (the amp is used a lot on 160) and roughly the same value in the Class C amps but using hockey pucks which seem to come in oddball values. Audio reports have all been good 160-10M and waveforms meet "spec".

Cool on the rig plans, Ron. Sounds like you're taking the right approach with the 10KV caps.

Yes, it pays to use caps that are several times the working voltage. I regularly use 10-15kv on most amps and would never go below 10kv for a 1 KW plate modulated rig or big linear amp. Again, it's a different world for linears where there is no X2/X3 modulating peak plate voltage to contend with.

Audio frequencies:   "Normal"  AM audio below 5kc is usuallly not affected by 500 -1000pf caps much. But lets take a few examples of extremes...

At 15khz, a 2000pf cap Xc is about 5.3K ohms.  In a normal plate circuit, this is significant, especially if a .002 is used for BOTH the plate bypass and plate coupling caps, since they are in parallel to the audio path. (The pi-net is in series with the plate coupling, so it will always be less that the coupling cap there.)
In this case at 15kc, the Xc could be as low as 3K.  This will produce some phase shift and highs attenuation depending upon plate impdeance.

At 5kc, this becomes about 9K of Xc shunting the modulator.

For example, even though we may use only 4-5kc of audio , the transmitter should sweep cleanly without phase distortion up into the ~8kc range  to keep the 4-5kc audio pristine.  Desired  audio bandwidth is a matter of opinion, but you get the idea...  Just scale up or down these ideas to apply it to your own situation.

Here is a good Xc calculator to use:
http://www.electronics2000.co.uk/calc/reactance-calculator.php

T

It doesnt quite work that way Tom, there is the matter of the plate choke between them. If its reactance is high enough the caps will not "see" each other. If it is way low as in many commercial amps the choke becomes part of the output tank on 160 and sometimes 80.

Even if the choke was electrically invisible the 2 caps are in parallel but the blocking cap is in series with the Tune cap so its total will be somewhat less than C Tune. On 80M this would be somewhere around 250pf in parallel with the choke bypass.

The Collins KW-1 uses 1000pf for both caps but I dont know its audio specs. The Hallicrafters HT-20 has a .0047 bypass and a .002 blocking yet responds well to opening up the audio. The choke is 4.3 MH!

Carl

Sure it works that way, Carl…  ;)  Let’s take your points one at a time…



A standard plate choke is almost a "short wire" at 5khz and close to that at 15khz audio. At 15khz, a 1Mh choke is 94 ohms.  At 5khz it’s 31 ohms. That is not enuff XL to isolate the two caps at audio frequencies in a plate modulated tube rig.  The choke HAS to be low XL at audio or the audio highs would have trouble getting to the tube plate for modulation without dropping across the choke.



Agreed. As I said in my previous post, the tank circuit is in series with the coupling cap so the shunt at that point will always be less than the coupling cap. Yes, these two circuits are in parallel as I mentioned.   I was talking about 160M as a worst case scenario.




Those rigs are probably designed for 3kc audio so can get away with those higher cap values. Even so, at .0047, I’ll bet there is some serious audio phase shift going on.

BTW, a 4.3 MH choke is very high for most big rigs, but even a 4.3mh choke is only  135 ohms at 5khz.   I still feel 500pf for both caps is a good ballpark if we are striving for the best hi-fi audio we can get using a plate modulated rig.

All of this does not apply to linears, screen modulated rigs or rigs in the “Apache” class using stock, restricted audio.


T

I lost a decimal place again Tom :o

How much Xc are you aiming for?

Yes the HT-20 is 3kc but the ones Ive heard with opened audio sound just fine. I dont care about hi-fi, 5kc is plenty.

A linear sure does offer some benefits, a modified LK-800 will run 800-1000W carrier and not break a sweat and is a small desktop package. The xfmr sits in its own small box on the floor. Most Class C AM rigs will need an attenuator to drive it. Even my lowly LK-500ZC will run the so called 375W limit with no sweat, it has the optional external xfmr.

I worked on the RF deck today, and got it working on 75 and 40 meters. 400 watts of carrier. Should be able to work you easterners now (if the world doesn't end tomorrow). The modification involved putting in a new plate tank coil - a really BIG one that is easier to tap. Problem is, it has a bit less inductance than the old one and it changed the tuning settings on 160, but I think I got it working OK on 160 after a bit of tweaking. Tunes real nice on 75 and 40, a bit touchy on 160 now. Was able to use that big 500pf/10KV coupling cap. It neutralized successfully on 40 mtrs, I assume that the neutralization should hold for 75 and 160 as well.   

Will be out of town for the next week. Going to Marion, Indiana to watch my daughter participate in the NAIA national championship track meet. Will put the transmitter through its paces when I get back.

Ron

Ron,

While I agree with Tom in principle, I doubt whether you have any highs in your voice that need a bandwidth of 20kHz so I’d go with Carl’s suggestion and use 1000pF coupling and decoupling capacitors, particularly as the rig is touchy on 160m. That would give you a -3dB bandwidth defined by the total capacitance that’s around 10kHz, so you could still add another 1000pF decoupling capacitor, if required for better overall stability on 160m, and still have in excess of 6kHz modulation bandwidth.

For those who have a need in the future, I’ve attached a table of minimum recommended values for blocking capacitors on 1.9MHz. You’ll notice they’re dependent on power output and the supply voltage used. I haven’t mentioned capacitor current ratings because that’s very dependent on the value of the plate choke used. If inadequate inductance is used for the plate choke and its reactance canceled by the tuning capacitor, then the circulating current between that and the plate choke goes through the blocking capacitor and increases the rating required for that. It’s possible to predict the increase in current rating required for the coupling capacitor with reference to a particular value of choke inductance and plate load on each band, but requires yet another table. I wanted to keep this attachment down to 1 page, so didn’t include any mention of current ratings. I can try and produce a short document on the effect of plate choke inductance on the coupling capacitor current rating if there’s enough interest.

Dave.

Thats a nice chart for driver stages Dave ;D.

Would you please expand that a bit for real tubes? There are several on here using tubes with handles ???

Thats the first time Ive ever seen that info short of having to calculate it, it should be a permanent part of the Tech Files.

Carl

Hi Carl,

No particular XC values - I think most of us use caps based on what is available to us since they are rather rare and can be expensive. But for a big, hi-fi plate modulated rig, it is a compromise to keeping the shunt XC value high enuff to pass clean audio highs without phase shift but still low enuff to give 160M RF a near-short path.

But as Bruce has said in the past, the XC/XL audio modulation path is very complex indeed and the final tube and associated circuitry add variables to all this too. The modulation xfmr, Heising reactor, Heising cap, etc, all have their effects. It's to the point where plate modulation is really an antique method and not all that good compared to the newer modern solid state PDM methods. But we  love it just the same... :-)

Dave, thanks for the chart and coupling cap info. I could be wrong, but I think those values are tailored more for linears and other non-plate-modulated rigs. (Or 3kc plate modulated rigs) This is good, though, it all depends on how far we want to push the RF side and how far we want to juggle the audio purity at the extreme highs. It is a compromise with plate modulation. I have personally seen where 1500pf had an effect on my extreme highs using a 4-1000A plate modulated KW. It can be seen when the rig is swept above 5kc or so.  It's just a matter of testing our rigs to see what they do for phase shift at the highest audio freq we choose to make a limit.

That said, yes, 1000 pf or even 1500pf would probably pass as "good opinion" audio in a KW plate modulated rig. My intention was just to point out that some guys may run 5kc audio, but they want their rigs to be clean all the way up to 10kc, etc just as a buffer so that 5kc is truly not affected by the Xc audio phase shift.  If we see some audio problems related to this, at least we have a trobleshooting path to try - being aware of this potential.

I know Tim/Wa1HLR, one of our resident AM gurus, always told me to use 500pf and even 250pf for both caps if I could get away with it on 75M. Overkill yes, as the XC becomes upwards of 60K++ at 5kc...  ;)

Good discussion.

T

Carl, we don’t know what real tubes are here in the UK. We think 813s are high power! If you want the chart to go higher in HV and power, I can do that. I had to guess how far to take it in the first place because I don’t have a very good idea of the range of tubes and HV you use over there. I thought I’d gone far enough, but obviously not. I’m not an amplifier expert, just an interested amateur who uses theory to make up for lack of test gear.

Tom, the chart is based on energy considerations and whether the plate current pulse is short and sweet (class C) or the full 180 degrees and lower in peak amplitude (single-ended linear) makes little difference to the change in voltage by the time you’re halfway through the plate current pulse. These charts will be pretty close for both linear and non-linear use because it’s the integrated energy to the halfway point that determines the drop-off in voltage swing and power. I agree that designing a power amplifier is a compromise and if you want an ultra-high audio response you’ve got to sacrifice some output power to do it. It’s a personal choice and I wouldn’t do it. I tend to go with nature and prefer to pre-emphasize my audio in the transmitter up to about 3.5 or 4kHz and then roll it off quickly. Then on receive the natural roll-off of old receivers makes the overall response sort of level. That way the HF noise to which the human ear is very sensitive doesn’t ruin the readability of the signal, even when it’s weak. I know you guys in New England prefer to widen the transmitter response at both ends and increase the response of your receivers as well. I disagree with that, but each to his own. You have the luxury of high power limits on AM and we don’t.

Dave.

OK Dave -

Good on the chart accomodating class C.

Thinking more about it, I'll bet an addition to the chart that includes the audio phase shift at chosen audio frequencies for each capacitor would be helpful.  For example, maybe there is a 30 degree phase shift using a 1000 pf capacitor with a 6K load impedance at 7kc. That wud add a nice feature for plate modulated rig designers. These numbers choices would be subjective -we would have to decide what is a meaningful phase shift is and what audio frequencies to use.


When I built my WA1GFZ-designed MOSFET audio driver, Frank spent time choosing coupling caps that had less than 20 degrees phase shift down to the single digits degrees for the lows. The highs also looked great. Considering the mod iron adds it's own phase shift, keeping the coupling and bypass caps low is a good idea.  It's also a good thing that the audio negative feedback is not greatly affected by these caps since the pick-off point is usually before this. ** But in my case, I pick the audio NFB at the OUTPUT of the mod xfmr secondary. I think those caps could affect the NFB phase adversely.

Yes, I understand your feelings about wider audio in the UK and I tend to agree with you. There is nothing worse than hearing a report that our transmitter is tearing up another QSO...  Though, I also like a deep low end response, since it costs nothing in bandwidth.    Personally, I have two AM transmitters (plate modulated 4-1000A and a 24 pill class E rig) that will easily sweep cleanly up to at least 8-10kc. However, I always run the audio tailored for between 3.5 -5kc, depending on how busy the band is. I use low level audio filtering and constanly keep an eye on the HPSDR spectrum analyzer to be sure I'm not beating up some AMer on a side channel. Running the cleanest audio and RF final (IMD) possible is my goal here.  Having a transmitter that will do X2 of the desired power will also help keep things cleaner.

Well, back outside for some more lawn mowing... ;)

T

Dave, Id think that expanding to 3000W and 5KV would do it for most and would cover the 4-1000A, 3CX3000A7, and YC-156 builders :o  Actually many hams I know are also MARS operators and they have no power limit.

Id also suspect many on here would be interested in hearing your views of the effect of the plate choke.

Carl

Carl,

I'll add 750W and 1kW to 3kW in 0.5kW steps. How low should I take the supply voltage at 3kW and what do you think the supply limits should be at the 750W level?

Predictions of what the coupling capacitor values should be are quite straightforward, but the plate choke inductance is a whole different story. It's hard to justify the way I treat it without a lengthy explanation, and that bores everyone to tears. The method works, though, and when my amplifiers are finished I know I couldn't have done any better if I'd spent ages cutting and trying different components. It saves a lot of time. The problem is how best to present it to other people. Obviously, a table would be more acceptable than a few equations, but the presentation needs to be well thought out to make it easy to use. I'm still thinking about the best way to do it, but I'm more than happy to discuss the role of the plate choke and the consequences of using too low a value while I'm deliberating how to present the information in a better form. You guys may be able to help me find a simpler way of presenting this stuff. 
   
Tom, the inclusion of phase shifts at different audio frequencies would add too much extra information to the table and might make it more difficult to follow. In addition, since the table is energy based, the impedance presented to the modulator at any power level and supply voltage could vary by over 20% just for class C. Most folks wouldn't be bothered about the phase shift anyway, unless it was high enough to change negative feedback into positive. My feelings are that the leakage inductance of the modulation transformer would change the phase shift significantly at the top end and make any predictions based on the load and capacitance alone pretty pointless.

Dave.

Thanks Dave. Id stick to the generally used voltages for common tubes which would be 1500-3500 for 750-1500W (4CX1000A/1500B, 572B, 813, 4-65/125/250/400, 3-500Z, 8877, 3CX800, and most vintage triodes and tetrodes/pentodes such as 100/250/304 TH/TL, 4E27A, 814, 4CX250 and family, HK 257 and load of others including the Ruskies.

The big bottles and tubes with handles in common use pretty much stop at 5-6KV and make the power with lots of current. There are always exceptions such as those playing with 10KV on a 4-1000A ???

As far as the plate choke discussion Id just let it rip in the format youre comfortable with and let us mere mortals have at it with questions, etc. It would be nice to have the end result be a paper to put in the Tech Notes on here that covers the math for those who care and down to graphs for the basement tin knocker. I leaned the subject on the job from those who learned the same way, it was nothing taught in a military tech school or later in university.

Carl

Dave,

You make a good point that the modulation iron would probably swamp out any choke bypass cap or coupling cap phase shift by a large degree.  Since it's cumulative, the best we can do is pick a minimum set of caps and hope our iron has minimum phase shift at the maximum audio freq we intend to use.

BTW, after installing the MOSFET audio driver, (transformerless) I was able to take my audio negative feedback from the mod xfmr SECONDARY back to the low level input without oscillation. This was about 8db of NFB. Evidently this is a tough task with more than one xfmr, but the SS driver makes it possible since the mod iron is the only xfmr in line.  Most schemes do it off one mod tube plate before the xfmr.

I've found audio NFB makes all the difference, thus my great interest in keeping phase shift to a minimum, in general.

Later -

T

It is fun to start a thread, and then just sit back and soak up all the info contained in the comments. Although I am a professional person, my profession has nothing to do with electronics, and all my electronic knowledge, such as it is, is self taught. I have read a lot of old manuals, but I think I have learned more on this forum than anywhere else, and for that I thank all you guys. The discussion has raised a few questions in my mind. 1. What is the minimal plate choke inductance recommended for 160 meters? My guess would be about 200uH. I know that a B&W 800 is not recommended for 160 and I think it is either 90 or 100uH. 2. For a medium powered RF deck running 300-500W with plate voltages of 1500-2000V, is a pie wound or single layer plate choke preferred, or is either OK? 3. I have some Sprague 500pf@20KV "doorknob" capacitors but they are not ceramic. They have what looks like a molded plastic shell, and do not look as high quality as true ceramic doorknobs. Are these caps ok to use as coupling/blocking caps? Furthermore, although these caps are all marked 500pf, they actually measure from about 350-450pf. Does this suggest a fatal problem, or are these most likely OK to use?
Thanks, Ron

I have (4) 650pfd doorknobs that all measured 510pfd.  All the ones you have measure low, probably just production tolerances.

All the doorknobs I've ever seen were all molded plastic shells.  These things were used in early TVs to filter the HV to the CRT.  Then some smart-ass figured out how to add conductive coatings to the shell of the CRT to act as the filter cap.  As a result, no more doorknobs in TVs.

Fred

Ron, are you measuring them with a LCR bridge (such as a General Radio, HP, etc) or one of those Chinese "capacitor meters"?

The latter will give you incorrect information regarding the capacity if there is sufficient leakage and/or ESR.  The accurate way to measure a cap is with a proper bridge.

If a capacitor deviates significantly from its marked value, rest assured there's something wrong with it and it needs to go over the shoulder, not in a circuit where it can cause damage or loss of life (as in the case of a B+ blocking cap).

Carl,

I’ve attached a one-page addendum to the original document rather than enlarging Table 1 to include all the higher power tubes. It allowed me to add some comments which wouldn’t fit on the first single-sheet version. I hope this goes far enough to cover all the cases apart from the odd one you mentioned at 10kV.

Ron, the fact that you’re a professional man must mean you’re reasonably intelligent and can learn from doing things yourself. I like to read up on the background information and then do experiments to fill in the detail and sort out any misunderstandings I have about what I’ve read. I’m not professionally involved in anything any more because I retired last year at 60. Before that I was a university academic specializing in materials science. I tend to question everything I’m told, regardless of who tells me and how famous they are. I guess that’s partly to do with my scientific training. I think the power amplifier experts find me a pain in the ….. Anyway, be very wary of doorknob capacitors that are not specifically designed for RF service. Some are OK and others are very poor. It all depends on what RF current they can take. In RF amplifiers the voltage across coupling and decoupling capacitors tends to be reasonably constant and the dielectric loss is not as important as the effective series resistance (ESR). TV-type doorknob capacitors normally have too high an ESR for RF use, but some are OK as long as the RF current through them is not too high. You can use an existing transmitter to pump some current through these capacitors at low impedance to try them out and see how hot they get at a given RF current. Alternatively, you can try one in a transmitter and see if it explodes!

Now, Carl has suggested I just dive in and start discussing the plate choke, so here goes. This could get contentious! The conventional explanation put forward by the amplifier experts is that the plate choke is merely a means of isolating the plate from the HV supply so it can swing freely at RF. A typical example of what they think is given on W8JI’s website. He’s the only brave soul who has dared to say anything on the subject, and most of them ignore it and don’t explain anything about how you can calculate suitable values of coupling capacitor and plate choke. They just suggest typical values with no regard to variations in the supply voltage and power output of specific designs.

The suggestion that the plate feed choke (PFC) is purely there in a passive role has several technical issues as far as I’m concerned. We know that getting the whole plate current pulse into the tank is important to delivering all the available power to the load, yet in the parallel-fed arrangement we use these days the plate current has to be shared between PFC and the coupling capacitor, CC, when the tube conducts. We also know that the portion of the plate current going through CC to the tank drains charge from CC. This charge has to be made up before the next cycle, otherwise the voltage across CC will continue to drop and the tube won’t be able to drive the tank circuit to its full negative peak. We know it does! So, the other part of the plate current has to be delivered to the tank while the tube is off, and in the absence of any other active device to do it the plate choke must play a role in driving the tank and re-charging CC. We know that inductors can generate very high voltages to keep the current going through them at the same level, but the experts steadfastly refuse to accept that the choke has any part in driving the tank circuit. Why? As far as I can see it’s because they don’t actually have a clue how parallel-fed RF amplifiers really work! If you put a current transformer in the lead to CC you’ll find that the waveform of the current going into the tank has an almost flat top like a square wave during the positive-going half of the voltage swing if the choke inductance is large enough. This flat-topped part of the current comes from the DC supply and re-charges CC at the same time as topping up the tank circuit. In the negative-going half you can see the share of the plate current pulse that goes to the tank when the tube’s conducting. Its peak level is pretty much what you’d expect if you subtract the DC choke current from the plate current pulse. The waveform is essentially the same whether the RF amplifier is working in a linear or non-linear mode. It’s what the coupling capacitor current waveform has to be to meet the laws of physics that govern capacitors and inductors, and electrical laws such as Kirchhoff’s current summation requirement. It can’t be a sine wave, though it begins to look more and more like one if you use very low values of inductance for the plate choke because the circulating current between PFC and the tuning capacitor goes through CC.

That’s probably enough for the first round. I’ll see what reaction I get to this before going on to discuss the increase in current through CC as a result of low inductance in PFC. The HV decoupling capacitor also suffers increased RF current if PFC is too low in inductance because that’s part of the return path for the circulating current to the tuning capacitor, so I must not forget to include that as well.

Dave.

Ameritron uses 225uH while most others are in the 100-150 range. Alpha copied the Ameritron in their current products.


 
The 800 is for 80-10. When the 160 rules for 1500W were allocated the 802 came along at 110uH and a higher current rating.


Both have series resonances that must be determined before usin. A piwound resonance is usually lower making them a poor choice above 40M. An exeption is the old National R-175 and 175A, the latter allowing 15M.


 
Those are the classic TV doorknobs used in all sorts of ham equipment, tolerance is all over the place They worked fine at power levels up to 400-500W or so and could be paralleled for higher power but they are far from RF rated. Try one and if it doesnt overheat you are OK.

There are other doorknob type caps used in ultrasonics and other industrial applications. Heath used a 1000pf @6KV in their SB series of amps that have held up well over the decades but there are signs they are slowly failing.

Carl

Carl,

It's interesting to see what the commercial linear amps use in the way of plate chokes on 160m, but surely they are working at a much lower plate impedance than Ron's class-C amplifier?

I make the optimum load for his PA something greater than 4.5kohm, whereas I'd have thought the Ameritron and Alpha linear amps would have been working around the 1.3kohm region. The inductance of his plate choke needs to be higher.

The RF component of the current through the plate choke is determined by the plate voltage swing and the reactance of the choke at the operating frequency. A useful reference is the inductance (Lc)that gives a peak RF current equal to the DC component. That's easy to work out and everything can be compared to that to judge performance fall-off.

For a class-C amplifier, Lc = 0.3 x Optimum Load/Fo, where Fo is the operating frequency. The Lc value for Ron's amp will be about 833uH. I try never to go below 0.33Lc, but 225uH would be 0.27Lc. It'll work, but the losses will be a bit higher than 0.33Lc. They may still be acceptable though.

By comparison, Lc for a linear like the Ameritron would be 185uH and so 225uH is 1.2Lc. Even 110uH would be about 0.6Lc.

In case you're interested in linear amplifiers, Lc = 0.25 x Optimum Load/Fo for Class AB or B.

I don't have any experience of American TV doorknob capacitors so my comments won't be very useful. I've had bad experiences with British ones, though, and stay away from them. Anyone wanting ceramic RF capacitors should look at qro-stuff on eBay. His prices are reasonable and although he's in the Ukraine, he has a contact in Texas for shipping within the United States. A friend of mine bought some 3300pF 10kV strontium titanate RF capacitors from him which had a current rating of 25A and they had an ESR of 0.094 ohms at 3.6MHz. Not bad for $10 each.
 
Dave.


   

The common TV door knob caps:

Many of us have had poor experiences with them. They were designed for high voltage and low current, so will certainly blow up in loading/50 ohms C2  padding service. They also "talk back" when used as plate coupling caps under heavy plate modulation. And, NEVER use them in high power tank circuits where the Q*I circulating currents can make them go off like cherry bombs.

TV door knobs and similar caps should be avoided for use in higher power service.

However, there are plenty of good quality caps from the Russians on the web and other sources. Be sure they are rated for high current. It's not easy to tell from the outside, but physical size is a good indication.

For plate chokes, the pi-wound chokes should be used only for lower power rigs, like <150 watts plate modulated, in my opinion. Thye will implode due to the close proximety of the turns to each other when high RF voltage/current is present. I like single layer wound plate chokes best.  We have had threads about winding your own using #24 or #26 insulated wire here.

T

Tom,

I'm amused by your idea of low power. We have a power limit of 100W of carrier on 80m through to 10m. 150W is high power for us. On 160m our limit is 8W of carrier above 1.85MHz where most AMers hang out. You guys just don't know how lucky you are.

In the west we've always had a poor opinion of Russian components and their reliability, but some of the RF stuff coming out of Russia and its former satellites these days is excellent. I've been quite impressed by it.

Yep, there's no reason why you shouldn't run with a single-layer plate choke down at 0.25Lc or so. It just means you need to use very low loss capacitors with a suitable current rating in both the coupling and HV decoupling positions to avoid increased RF losses compared with higher values of plate choke. It ain't impossible though, it just needs very good caps.

Dave. 

Dave,

Heheheh.... yep, it's all relative and we are fortunate to have reasonable AM rules here.  Though, listen to some of the UK stations down in the ssb 75M DX window and some of the contesters and draw your own conclusions...

Pictured below you will see a sample of the plate chokes I wind for my big rigs. This happenes to be an older 4-1000A modulated by a pair of 833A's. I build them with Plexiglass for viewing pleasure.  

I just wind a choke and try it out on the bands 160-20M. If it doesn't blow up, there are no resonances. If it does, I add or subtract some turns and try again. It's actually difficult to randomly wind a choke that falls into a small ham segment by luck.

Yes, the Russians have gotten their act together. They certainly are talented engrs when they want to be.

T

Hi Tom,

Well, there's ham radio and then there's DXing and contesting. If someone has to lie and cheat because he absolutely has to be the strongest signal on the band, then that's up to him. I'm saddened by the way "sport radio" has come to dominate amateur radio, but its certainly not progress in my book.

The amp looks impressive. I have some 4-400 tubes somewhere, but not being that interested in SSB any more I haven't done anything with them. One would loaf along at our SSB power limit.

I can appreciate that some people prefer the empirical approach. It's just that I like to get to the bottom of everything and understand it, so I develop little theories for this and that to help me find the optimum values for a given design. In the case of the plate choke, of course, there really is no single optimum value and a whole range of values can give quite good results. They all have different resonances, of course, and that tends to determine the "best" value for multi-band use in the end, so my little theories probably aren't much help to many folks apart from me.

I was writing to Ray, the editor of ER, when an email came in from the forum about your post, so I'd better get back to that.

73,

Dave.

Id make a 2000V@ .25A a 4K Rp using your countrymans spread sheet which has been my go-to source for a few decades Dave.
http://www.ifwtech.co.uk/g3sek/

Be it C or linear the typical 1200-1500W amp has a Rp of ~2K or lower.

Im busy taking a gas tank out of my sons car today so my comments here are hit and run until after dark.

Carl

OK Dave, Carl, Tom, Thanks for the comments. My current coupling/blocking cap is a large, black cylindrical Sangamo cap that came out of a 5K Gates broadcast transmitter. It must weigh 2lb or more. The end terminals are 3/8 inch machine screws. I'm not home right now or I'd put up a picture of it. It is rated 500pf@10KV, 10Amp RF current@1000kc, so I'm sure it's up to the modest task currently assigned to it. I was concerned that 500pf might be a bit meager for 1.9mHz, but from the comments, I think it should be usable.

Regarding the plate choke, I am currently using a high quality 5 pie National unit that was designed for amateur band use. I don't know the model number, it's not a 175 though. By it's physical appearance, I would guess it to be at least 1-2mH and 500mA or more. I do have a B&W 800, and another similar single layer plate choke. I suppose I could connect these in series - that works - right? If I do this, I should put a bypass cap between them, as well as one on the power supply side of the first choke in the supply line - right?

Ron

OK Carl, I wondered where you'd got to today. I'm not going to quibble about 500 ohms but the plate load depends greatly on conduction angle in class C and between 150 and 90 degrees there can be more variation than that. I don't know what G3SEK's spreadsheet is based on but I prefer to trust Terman's "Radio Engineering."

Ron, sounds like you've got the National R-154 or R-154U. Either way, that's a 1mH choke and should do fine on 160-40m. That's 1.2Lc!!!

Good luck with the project,

Dave.

Those caps are FB, Ron. I have used them (and still do) in many big rigs as both plate bypass and coupling caps. Never had a problem. They are overkill for your intended use.

The current rating goes up somewhat, to a point, as the frequency is increased above its 1 mhz rating.

BTW, a friend who knows RF said that those particular type caps start to look inductive above 40M or so. They were designed for use in BC rigs, so you can understand. I have used them on 20M, but have always used more RF-friendly ceramic caps above 20M. Just something to consider if you have problems later on the higher bands.

Yes, you can always use chokes in series, though remember that the more complex the choke, the more chances for possible combinations of resonances on the higher bands.  I would opt to use a smaller choke at the plate circuit  and then a larger one for the  lower bands behind it, if you need to series them. It's almost like the technique of winding smaller turns groups at the top of a choke, etc.

Adding a bypass cap between two chokes is OK, as  long as the first choke has enuff inductance for the lowest band. Worst case example wud be to have a 20uh choke first and a bypass cap to ground, then the bigger choke. It may be OK on 10M, but on 160M it wud be an RF short to ground. (Or, as Carl suggests, use a relay to switch the bigger choke in for the lower bands)
T

Dave, I agree about it not being worth a quibble, a pi-net set up either way will tune the other with no apparent difference.

Here is a nice little rig using a R-154.

Carl

Ron,

Looks like you're all sorted out and ready for 160m now. That Sangamo cap from the Gates rig sounds like a monster. The 500pF capacitance will only drop your efficiency on 160m by 2% compared with my recommended value of 735pF for the coupling cap, so your power output will probably only drop by 8W or so compared with 75m. With a 1mH plate choke, I make the effective rms current through the coupling cap less than 0.5A, so it is going to be overkill by some margin but if you haven't got a better use for it .........

I agree with Tom that you have to be careful about the inductance of internal capacitor connections and the external leads you use. You need to try and choose suitable values of capacitor that series resonate with the leads at the very top end of the frequency range you want to cover. That way they never become too ineffective as capacitors.

Tom's already commented on the use of an extra choke in the HV lead, so there's nothing more for me to add.

Dave.

I forgot the link yesterday!
http://n4trb.com/AmateurRadio/GE_HamNews/issues/GE%20Ham%20News%20Vol%2003%20No%205.pdf

If using 2 chokes run them at right angles, most use a vacuum relay to switch in the lower one as its nearly impossible to cover 1.8-30MHz without a nasty resonance at least for a QRO amp. Henry does this with their "big" amps.

Others have found with homebrew using tubes with handles that forgetting to switch in the lower choke for 160/80 the amp still worked OK as long as a big bypass was in the circuit to keep RF out of the electrolytics. The power loss wasnt enough to smoke the 40-10M choke.

Carl

Good idea about using the relay to switch in the lower band choke and possibly a bigger bypass cap, Carl.  That way we can optimize parts for the  6M-20M zone , if desired, and still have large values available for the 160-40M areas.

As you said, getting a fixed-value QRO amp to function from 160-6M is quite a task, though it can be done. 

Interesting on the 8 watt difference, Dave. Looks like 500pf is surely the lower limit in Ron's case.


Well, I've reached the limits of my knowledge on this subject, so we be black out... ;)

T