The AM Forum

Does anyone know of a simple online  Pi-L  network calculator that is not Excel based?

I don't see anything on the web or in the AMfone archives.


TNX.

T

Look up WA2WHV site.  I think he has an application there if the site is still there.

Thanks, Jim, but that site has just a pi network calculator, not a Pi-L.


In the meantime, my question is,  does a pi-L help reduce the C1 capacitance requirement vs: a standard pi network?  I have a tube for 10M that has a high plate shunt-C.  This internal shunt value is about what is needed for C1 already on 28.7 mhz.  If I add a C1 for tuning, this total capacitance will be too high.  I was hoping a Pi-L wud help reduce this problem...

T



  

T if you have an Orr book, say the 21st edition, there is a chart that will take care of you.  Usual design impedance for the Pi portion is 300 ohms then the last inductor is enough to get to desired impedance, 50-75 ohms.  If you don't have the book I can scan the page for you and email.

What is the load impedance?  C1 values range from 56 pf @ 1 K down to 13 PF @ 4.5 Kohms for a Q of 10.  

I think there is a calculator in this sw package.

http://www.cq-amateur-radio.com/cq_ham_calc/cq_ham_calc.html

why not bypass C1 on 10?   I think on 10 there is high circ. current through the plate blocking caps.   Avoid anything dinky.

Thanks  for the suggestions.

Well, after reading what Bill Orr says, the C1 cap will still remain the same value in either a pi or Pi-L. All other components change value, however.   I don't know how I got the impression C1 would be less. I've seen Pi-L's used by the good buddies on 11M so figgered this was a solution for the shunt C problem.  **  Update ** (However, my later post using a web calculator shows C1 requirement to be much lower with a Pi-L, thus a standard pi was used in the end)

Anyway, for a pi-L, L1 is usually 1.25 times it's pi-network counterpart.     C2 becomes 1/2 to 2/3 of the normal C2.  Also C2 must have its voltage increased by X3.

The well known Pi-L advantage is the extra 15db of second harmonic attenution.

Since I already wired it up as a pi-L, we'll see if I can get away with it as-is.  John, the reason for C1 is to have something to tune. If I eliminate it I wud have to tune L1, which may be the next alternative if the total C1 is too high and I need to disconnect it. However, the minimum value of my mini C1, a vac variable, is only about 5pf, so not so bad.


T

ah yes of course--the thing might be to parallel some cap with C1 that will bring its range way down to some low value so you can still vary it.

Sounds like you will be ok.

If you can stomach DOS programs, this might be useful or maybe entertaining. They are freeware and come with a money back guarantee.

http://www.montagar.com/~patj/rf_read.htm

Back in the day I wrote these programs in Turbo Pascal to calculate power supplies and Pi and Pi-L circuits. The UI is simple text entry, one line after another, and it prints a simple tabulation of answers on the DOS screen.

The PIout.exe program allows one to insert the conduction angle in degrees.

The source code might still exist if there is any interest (the horror of that!).

The calculations and conventions are derived from the ancient and weighty tomes of the Old Masters.

Pascal is responsible for my current dislike of programming.

 ;D

It's also one reason (of many, I might add) I abandoned all thought of pursuing a computer science degree and went with electronics engineering instead.   With the strict syntax rules and structure you'd think the Germans invented Pascal.

Hi K1:

Could you try this method:

http://amfone.net/Amforum/index.php?topic=29105.0  ?

If you match to Rmin it produces an L network.
Please let me know if it works for you.

TNX

Giorgio

Tnx for the info, Giorgio.   I have everything figured out at this point so can start testing and optimizing this amplifier now.


Funny comments about Pascal...  I first learned Basic in 1980 and then went on to Turbo Pascal.  I didn't know any better back in 1988 and wrote many programs using it, mainly for trading markets.  Later, "Easy Language" came out by Tradestation, which is also Pascal. That's all I've ever used and program using it every day..  I realize C and other languages have since dominated, but I speak Pascal here and that's all..  ;D    Thank God I never had to earn a living programming exclusively.

T

Try K8KK...he wrote a nice article in ER while back and has a webpage/on line app--can google it up.  I used his program do design the Pi in my HB 813 rig--worked great.  He also has Pi-L program.

73's
Geo W8VG/6
Rancho Palos Verdes CA 

if you think pascal is rigid and inflexible try Fortran.  i have not used either since college. everything is Perl, Python, database and shell scripts now.

a web based one, that might be useful:
http://home.sandiego.edu/~ekim/e194rfs01/jwmatcher/matcher2.html

Yes!   Just what I was looking for.  That link has calculations for both the Pi-L and the Pi, in addition to 14 more networks.  They are all figured out at the same time... cool.

This should be added to the AMfone website archives. Best I've seen.

It shows that the standard pi-network is best for a tube with big plate shunt C.   For similar conditions, it showed C1 needing 56pf for the Pi and only 9pf for the Pi-L.   In testing, I found this to be true and the Pi was superior to matching this tube with big shunt C.   Total C1, (tube plate shunt + variable capacitor) network requirement needs to be as large as possible to give adjustment range.

I'm all set now.

Thanks for posting that, OM.

T

...and low impedance, yes ?

Peter

Note that the source and load required inputs are all series form. If your tube design parameters are in parallel form (typical for semiconductors), you will need to convert those impedance sets from parallel to series before using the applet.

Alan

???   I don't understand, Alan.  Please explain.

In the link, let's take a look at the third picture down on the right. This shows a standard pi-network.  For a standard linear amplifier tube stage at 29 mhz, (29e6) using 1200 ohms source+ J0,  50 ohms output+ J0, Q=12, it gives the  following approx values:

C1= 55pf
L1= .64uh
C2= 246 pf

How would this differ according to your thoughts?

Tnx.

T

You are correct T. However, you chose the case where the load and source are REAL. Not complex. If their is reactance present or you wish to handle any stray C, or in fact the device has an associated C or L, then the load is COMPLEX. Now, the applet wants the entry to be in SERIES form.  Usually for power devices ahe load line (device output) is expressed as some RL//CP... and if that is the case, just need to convert to its series equivalent. I wrote an applet like this some time ago. The first entry point was....

IS YOUR LOAD COMPLEX?
IS YOUR LOAD SERIES FORM or PARALLEL FORM?

And then I would make the appropriate conversion dependent on the topology of the matching network.

Hope that is clear. Again what you did is on the money, it only gets sticky if the termination is complex.

Tnx for the explanation Alan.  I guess it's close enuff for now... ;)

At this point the amplifier appears to tune with the network parameters given, at least according to the MFJ-259B going into the 50 ohm pi-network backwards. The real fire test may be tmw.  I may post some pics and results.  Might even get it on 10M by the weekend with some luck.

Right now got the tube's fil on for 24 hours to burn out the gas.

T

Tom,
Let me know if you want to do any on air tests. I'll be around all weekend.

Sounds good, Frank.

Then I'll make it a goal to have it running on 10M this weekend.  I'd like to have you look at it on your HPSDR and check the bandwidth and cleanliness, etc.   If it works out and the band is good, maybe we can hold some court SW or NE.  With the 10M corntest this weekend, shud be some big activity on ssb too.

I'll let ya know the results.


T

Hi T

A pi net work consist of C1 for resonance an L and  a C2 loading coil.

So we always thought that the resonance must be made by C1,actually if We have a fix C1 (not variable,in this case the tube internal C) but close to the C needed for resonance,with the right value of L ,we can dip it by C2 ( the loading  Capacitor),if the current of the tube is to high after we find a dip,
change the value of the L make it longer(more turns} dip it again with C2 ,than the current will be smaller than before,by changing the value of  L and C2 setting we can have the current expected by the Tube.of course the Q factor will also change ,but not to great.

But this condition ,the C1 must have the value close to the Tube capacitance.needed for the Pi network(after calculating it).


Gito.N

10M monoband amp update:

Amp is running.

At first I thought it had VHF parasitics.  The Bird wattmeter read full scale in both for/ref into a dummy load.  But Dave/VW suggested that a dummyload will absorb power in VHF so it didn't make sense.   This should only happen with a monband antenna.

Anyway, turned out my Bird peak reading wattmeter board had RF getting into the 12V leads.  I tried an old Midland CB swr meter and the swr was flat... :-)

Frank/GFZ gave me some good ideas to prevent parasitics before I knew the wattmeter was the problem. I'm adding all these mods today and hope to test further.

Dave, you were right as usual  - the C1 variable was too much C and I am modifying to eliminate it and go with a tuned L instead.

Might have it running on the air early next week.  I don't hear much AM activity today so far.

T

You're welcome!

What dont you like about Excel Tom?  There are also free clones.

Carl

The link to the online calculator under discussion has been added to the Transmitters section of the Handbook forum here.

Hi

Following what I wrote before,I attached a picture to explain what I wrote earlier

The C1 in the picture is the plate tube capacitance,As T wrote before the C of the plate value is what the Pi Section needed( C resonance} .omiting the reactance of the coupling capacitor since it  have a low reactance,than the equivalent circuit is C1=56 pf .L =.64uH and C2 =246pf.

Say that C1 and L has the right value,so C2 must have a 246 pf value

If C2 has a value higher  than 246 pf like 300 pf ,the Amp is out of resonance
also if the value is lower than 246 pf the Amp is also out of resonance.
Only if the C is 246 pf is the Amp resonant ,and there's a dip in the plate current.

The point is when we know the value of C1 and the value of L (after calculating } ,than we can find a dip by turning the C2 .

Than fine tuning to have the rated plate current is by changing the value of the L and dipping again with C2, (C1 is fixed = the plate tube capacitance


gito.n

It is all so possible to save the complete page, so people with a dial up and postal pigeon backed up internet don't have stay online when using the calculator

73 Frits

Thanks for adding that calculator, John.     It is a good one cuz it took several days to find a working web-based Pi-L.   Tnx to W1FVB.

Carl:  Yes, I have Excel here and use it for number analysis sometimes. But I prefer the simple web-based calculators. The last Excel one I used for the Pi-L had frozen cells. I didn't have time to get into and have no patience in lieu of these simple web-based ones. You know what I mean... :-)


Gito: OK on C2 tuning the circuit. I guess the problem is if we want to use the amplifier across 1 mhz of the 10M band, it will require some more adjustment than what the C2 alone will provide. The reason is that loading of an amplifer is important to get it as clean as possible. Once C1 and L1 are resonant, C2 can be increased to provide a better IMD figure.

What I have decided to do is to vary the inductance of L1 thru a tunable panel-mounted slug. This way total C1 is just the tube's internal capacitance and outside stray. That is close to about 45 pf which appears to be the maximum I want to use to get a Q=12 or so.  Otherwise with more C1 the Q soars and the tank circulating currents are too high for good efficiency.

Anyway, I will run some tests today with the slug. I'm not sure if I need a ferrite rod ot a steel rod will do- or if it needs to float vs: grounded to the chassis . The coil is 3" X 3" and the slug will be only about 1/2" diameter with the ability to pull all the way in or out.

Any ideas?

T

A brass rod will DECREASE the inductance, if you overshot your mark on the coil.

this bring to mind a nuclear reactor's moderator rods  :o

I'm curious, which tube(s) have 45 pF output capacitance ?

I figger 12 * 3.75pf is about 45 pf....  ;D

T

gota have headroom

OK OK... so it's not my amp. 

How 'bout this one?    ;D

T

Hi

A pi section network,has 3 variable that's  C1 (resonance capacitor) L and C2 the loading capacitor,usually  we make it to 2 variable circuit with it's L fixed.

These three components makes the circuit resonance,usually tuning is done by opening,closing the C2 (load capacitance} then  after that dipping the plate current by the C1 (resonance) ,this tuning is done ,till the plate current at resonance is what the Tube needs.

We cannot only make the C1 and L resonance than after that  changing the setting the C2 ,since these 3 components depends on each other.

So the C2 setting depends of the setting of C1 and L.to make it resonance.

In Your case C1 (the tube and stray capacitance )is fixed,but C1 value is enough to make these pi section Q.

Since the C resonance is fixed ,we used the other variable component,L and C2
 You are right by making the L variable,but after that ,we must still re tune C2 for a dip in the plate current.

Gito.N


 

I dont find those useful for amp output tanks and prefer the GM3SEK spreadsheet which is real easy the second time thru it. It allows you to make all sorts of changes and see the results. There are several other good links and lots of other info on Ians site.

http://www.ifwtech.co.uk/g3sek/

Carl

Tom you could make this configuration work if you put a little inductance ahead of C1.
This maks a L-pi network. The tube Z is transformed up to the value you need to get c1 to resonate. Then the pi section transforms the new value back down to 50 ohms.
Look at the old 813 single band amps. This trick was used for the 10 meter single band amp.  I did the same thing in my old 4-1000A linear.

Frank,

I might give it a try.  It wud mechanically easier to use than the tuned L1.

If the normal L1 is about .7uh and C1+tube+stray total is about 50pf, what value wud you estimate I shud try for this first coil between the plate and C1?


BTW, right now during testing I'm getting RF into the FT-1000D and audio gear. I don't have the linear shielded yet. Have you had experience with unshielded  vs: shielded tests on 28mhz like this?    I expect to build a cabinet for it later on... no Plexi this time.. ;D

T

Tom did a quick simulation
plate Load 2K 50pf

first inductor .1uh
C1 30pf
main inductor .4uh
C2 load 300pf

Looks like the first L wants to be about 1/4 main inductor. If you go too high the second harmonic comes way up.

Simulated 40pf at the tube and inductor goes up
.1uh, .5uh
c1=30pf
c2=300pf

I will try those values tmw, Frank. It wud be nice to get a C1 variable working again.

Tnx, OM.

T

If you are going the extra coil route you could try simply tapping the existing coil.

If not the slug on a lead screw tunes a lot like having a C1. Just pretend.

I'm still not sure why you insist on paralleling all those 50C5s.

Hi

Attaching a picture comparing a Pi-section and a pi-l program,works on 30 mHz. it's just an example .an amplifier works on 30 mHz ,with an output of 18 watt and peak rf voltage and the anode is 200 volt .
Compare the C1 value in the Pi network and the Pi-l net work,
also the harmonic reduction of the Pi network.I think it's high enough

Gito.n

'cause he wants it to be like a "Dave Made" He wants to be the boss with
all of that hot sauce"   ;D  ;D

Tom,
Running L pi generates a sharp peak around the fourth harmonic or so. This can be adjusted away from operating frequency. If you run 1K ohm load then a pi should work. I see about .5 uh inductor with C1=30pf, C2 300 pf. (I used 40pf for tube)So I think your initial inductor value was too high. I would try smaller value and see if the Q gets too high. I would only use L pi if your Q is too high but be careful to check your harmonics as it will creat a sharp peak that may line up with a harmonic if not careful.
The values I gave you yesterday is a good place to start.
Tune up at the low end of the band and check efficiency then tune up at the high and again check efficiency. If it works better at the low end you can add L. if it works better at the high end you can reduce L

I have yet to find a tube that doesnt work well if you simply center tap the 10M coil with C1 as it gives you an automatic 4X step-up in C requirement.

There is no need for fancy footwork and calculations, just pick the load value and K factor, forget about tube C and build what the program spits out.

Carl

Thanks for all the suggestions.

I have the amp back apart and will try a few.

Carl, your idea of simply tapping the the main L1 coil with C1 is essentially an L -pi network, right?   Think I will try that since it is easiest to set up and adjust - plus I will have C1 variable again.

Will try it later tonight.

T

Carl,
Thr L pi does produce a transient peak in the attenuation well above the operating frequency so you just need to keep it away from any harmonics of the operating frquency.  Be careful with a strap off the C1 cap any L could produce a VHF series resonant circuit. Better to do it with two inductors with a common point at C1.
I think a pi will work at 1kohm load with 40 pf at the tube plate.

Frank,

I did try a straight pi-network already and found it had too much C1. (stray, tube and min C1)   C1 wanted to be all the way out. It's a tiny 30pf cap with a min of about 5pf IIRC.

When I disconnected C1 and tuned L1, the power came up quite a bit.

I'll try the tap and see if I can get away with it using a fat copper strap.

T

Hi Carl,

I assume this spreadsheet is for a pi filter before the final L, where the C1 is tuning, C2 is loading, and the L helps extend the impedance matching?

Thanks

Phil - AC0OB

Hi

Look it this way,what we are thinking is The tube as a source ,than according the pi section theory C1 is the resonating capacitor,

how do we load the tube first set the loading capacitor maybe to 300pf ,make a dip by C 1 ,we got a certain plate current.if it's to low open C2 ,than re dip by C1 the plate current,do it again till the plate current dip is the plate current the tube needs.
If we can't get the plate current dip as needed shorten the L a little ,than use the tuning procedure again ,till we get the plate current dip as needed.

So this has a 3 variable.

After calculating the values of  the transmitter and knowing the output impedance of the tube ( R1.) than the reactance of C1 is   XC1 = R1 : Q
Q =12 ,knowing XC1 we can calculate the value of C1.

So looking at this the C1 of the tube is fixed ,to have the operating condition that we want.

Now If we We think the other way ,

now we used the 50 ohm  output as the source and the tube impedance (R 1) as the load ,now C2 become the resonance C  and C1 the load C .
 Since C1 is fixed, than to know the circuit is resonant  we used C2 to dip ,right?. if we find the plate current is to low, shorten the L than re dip C2.

A pi section works both ways.

Gito.n

Gito has a point. Put a 1k resistor from the plate to ground and then drive the 50 ohm output with a generator. Monitor the resistor with a scope for a peak. make sure you use a low c probe to not add C to the plate circuit.
I've done that trick with a sweep generator.

Tnx for the info, Gito.

Yes, Frank, that's the first thing I did, was to run the MFJ-259B backwards into the 50 ohm pi-network output. I was able to find the proper settings.  Though, the Q was too high with C1 in there and I found with less C the power out was better. Thus, I'm gonnna try the L - pi network next by tapping C1 off the coil.

Unfortunately, I keyed the amp into the MFJ-259B the other day and now its on it's way back to MFJ for repair of the PIN diodes... sigh.

So, I'll use the FT-1000D and swr meter instead.  I'll hook up the scope and 1K load as you suggested.

I have the amp apart on the bench now and added a few things including a handwound single layer cathode choke. I think that commercial choke had too much inter-C anyway. 

It also needs a good metal cabinet and shielding since I'm taking out the station audio gear badly with RF during testing. This 28mhz stuff is a different animal compared to 40M and below.

T

Hoe long did the MFJ fly before it landed.

Tom is so mean to his toys...Toy story 4, Tom's toys revolt
When you had the mfj connected did you simulate the tube operating z with a resistor between the plate and ground? Running without the resistor the Q would seem very high.
Do you know which pin diode is used in the mfj? The ERB amps are loaded with pin diodes and there are a couple 3379 low frequency pins near the BB amp module.

Dave,

I must admit the MFJ took it like a man. A little screaming and burnt flesh, but other than that, he would have made Mr. Jue proud.


Frank, I did use a resistor across the tube.  In fact tonight I used a 1200 ohm globar and ran the backwards 50 ohm pi network test tonight. I found by tapping C1 one turn into the coil I get a good range of tuning for both C1 and C2.  I then tried a 700 ohm globar and saw a similar tuning range adjustment capability.  I was using the FT-1000D swr meter, not a scope probe. I measured my probe and found the capacitance was too high to use.   I think this may work FB when I fire it up tmw.

The input tuning was off too, so put in a 35 ohm resistor and got that matching 1:1 using a mini pi-network at the cathode.

Other than RF in the audio, I'm hoping for some good results tmw.

T

Hi Folks,

Just read this thread and here is the simple solution that I use with excellent results. 

Ready?.....

Use a Q of 20 or 25 or what ever it winds up being so that you have a reasonable tuning range with your C1 cap.  Take a turn off the Pi net inductor.  I've run a Pi Network on 2 meters (8877)with a Q of almost 60.  The inductor was a 1 1/2" wide copper strap about 4" long.  Effficiency was the same as my 8877 amplifier on HF.  Same deal with my 8877 amp on 6 meters.  I used a neutralizing cap in both cases for C1.

Yes, there is very slightly more loss due to the higher tank current but at a Q of 25 the current is only twice what it is with Q of 12.  No biggie.  Plus, you will get much better harmonic supression. 

On HF many years ago I experimented with Q in the pi net and could not measure any change in efficiency with Q's up to 30.  The Inductor did get warmer requiring larger wire.

So....take a turn or two off the inductor and make it from larger tubing to keep the loss low.  I use 1/2' soft copper tubing for tank coils on 10.

Chuck

Yo Chuck -

Tnx for the info.  Yes, I can see how that would work.  Let the Q be what it is and use a proportionally larger material in L1 to compensate for the increased current.

I remember you talking about a test you ran with your 8877 amp using the plug-in coils. IIRC, you put in the 75M coil, took an output power measurement and then retuned the amp on 75M using the 40M coil. The results were very similar, though the Q was much higher with the 40M coil. Both coils were made of copper tubing, which helps, of course.

For this monoband 10M amp, I am using 3/8" copper tubing, about 3 turns, 3" diameter, 5" long.  I figger the inductance is about .5 - .6uh the way it is stretched out.   

I should have some new results tomorrow. As you suggested, I'll be willing to try reducing L1 if need be - for a higher Q.


T

Hi Tom,

I've seen that guidance for a Q of 12 over the years in all the handbooks and other references.  They also warn against higher Q's but in reality the increased loss is minimal and easily dealt with by going to larger diameter conductor which will eliminate the added loss.  For a single ended linear amplifier, a higher Q in the tank will help with IMD also.  Eimac mentions this in one of their publications on the use of their power tubes in linear service.  As I recall, Q's up to 25 helped but with diminishing returns after that. Q's under 10 saw a clear increase in IMD.

The tank coil in my 10M amp is 2 turns of 1/2" tubing about 2 1/2" in diameter, 4 " long.  It's about .3 uH.   I can just get the amp to resonate with the 20M coil (.65 uH) on 10M and there is no measureable change it efficiency.

Chuck

hi

Attaching three different setup of an amplifier with different phase angle,that has a relation with the Q value of the circuit.with a high intrinsic Q of the coil( inductive reactanse/coil resistance).


Gito.n

So if Chuck is running .3uh then as I suspected you are way too high at .7uh.
The tapped inductor works just be careful where the second resonant peak lands or you will generate harmonics.

I spent some time on the 10M amp today.  Finally got the thing working stable and putting out some soup.

The problems were parasitics, low power output and C1 not tuning well.  I smoked a bank of parasitic resistors.

Here's what I found:

1) I originally had a 350 pf plate coupling cap. I thought that was plenty of capacitance because it's about 17 ohms at 29mhz.  But when I replaced it with a .005 ufd, the amp started tuning much better. Maybe the 350 pf was resonanting with some lead length and causing the parasitic.  A .005uf cap shows about 2 ohms at 29 mhz.  

2)  As Chuck suggested, I decreased L1  from 3 turns 2 turns.  This made a big difference and let me use the standard pi-network with C1.  Now C1 resonates at about 20pf and gives a nice tuning range.  Obviously I had too much inductance if I wanted to have a tunable C1.

3)  The original C1 vacuum cap was gassy and glowed blue inside when in operation. This caused power to drop way off when it ignited. I replaced it with another one and it runs FB.


At this point all seems to work FB except I still have some 1-2kc acoustical audio feedback when I run the amp. Not sure where it is coming from. I can disconect the mike and it goes away even with the audio turned up.  There is also a slight bit of RF getting into the audio when I run the amp more than 1/3 power.

I don't know about the acoustical audio problem, but when the rig gets shielded it should help the RF feedback.  Right now it's sitting on top of a rack cabinet with no top or sides.

I may be on 10M tmw trying it out. I'm limping along at about 1/3 power due to the remaining feedback problems , so be nice!   ;)

Thanks to Frank/GFZ  and Dave/VW for their email suggestions and the guys here on the board who posted info.

T

You might want to shield it before you go to high power. There is a lot of radiated energy coming off an open rig like that.
The nice thing about a low z tube is you can tolerate a lot more C1. Just need to have a coil that can handle the current.
You might try a snubber off the cathode with say 100 ohms in series with 50 pf door knob. This will load any high Q VHF signals.

Yep, I expect to take a trip down to the scrap yard in Wallingford next week to pick up some aluminum sheet to build a cabinet.

As you suggested before, I added a 1K in series with a .01uf cap from the cathode to ground.  Next time it's apart I'll add the 100 ohm and 50pf.

After a long carrier, I felt the parts inside and see L1 is room temp warm, the plate choke is cool as well as other parts in the tank.  The problems with the Bird meter went away once I got the tank tuning right. Must have been some VHF parasitics before.  Even the security system stopped talking back.

All quiet on the western front now.  10M is not an easy band to master with homebrew gear.

T

The trick is to find the right snubber value to deQ the feedback that causes oscillations. Might even try lower cap values

Hi

From the RSGB book

Instability in a phone transmitter can often be caused, particularly when working on the higher bands,by the existence of RF fieids or current in the vicinity of the high gain a.f amplifiers.the first voltage-amplifying stage being the most susceptible.Rectification of the modulated rf voltage takes place .and resulting a.f voltage is fed back into the transmitter trough the modulator.At some frequency the phase will probably be such as to produce positive feed back,and the whole system will oscilate at an audio frequency.
Short,direct and screen wiring between modulator,transmitter and power circuits ,in conjunction with the liberal use of r.f chokes and bypass capacitors of 0.001 - 0.01 uf at each end of the power line is generally sufficient to effect a cure.
 The effect of an RF choke choke may be achieved very simply by slipping a small ferrite bead over the grid lead of a valve as close as possible to the grid pin on the valve holder.


Some audio first stage AF amplifier use an RF choke and a bypass C at the control grid of the tube

Gito.N

Thanks for the post, Gito.  That sounds exactly what's happening - the stray RF is getting rectified in the first stage after the mic and then producing positive audio feedback.

Next week I will build a metal cabinet for the amplifier and if that's not enough, I will start heavily bypassing the in and out control leads.  A few ferrites will be used too.

I'm confident I can tame this thang... ;)


Question:  Twice I burned up a few parasitic resistors, but only one out of a parallel bunch of five.   I have five 100 ohm 1 watt resistors in parallel across a 3" long copper strap loop. (20 ohms)  Only one of the resistors chars, while the others stay cool. What is causing this? Do they all get hot but one breaks down faster and drops in value producing a runaway situation? 

Right now I have five in there - one is charcol while the other four look OK and continue to operate like this.   Wish I had a 20W, 20 ohm  glowbar.


T

Update:

I added some ferrites to the microphone cable and audio preamp leads.  The audio rectification feedback is almost gone completely. I can run the power up and run 100% modulation now.

The problem still remains with the security system not liking 29 mhz.  It has a voice synthesizer and says someone is trying to break into the "playroom" where ever that is.... ;D   The new cabinet shielding should fix that next week.

10M was quiet today so just did some testing into the antenna and went black out.

T

No it wasn't. Chuck and I had a nice conversation via backscatter.

Chuck worked a few Euros. So did I.

Pete CWA was working them one after another......

Maybe you were listening on the wrong antenna : )

Are you saying I got sh@@t in my ears and can't hear squat?  

Cool on the Euros. Can't wait to join in and give Pascal a few 59's.

Naw, today I got up late and tuned in at about 1PM, plus the antennas are fixed "Armstrong" west now.  At 14 degrees F out there tonight, I think they will stay there for a while....

T

Don't stick your tounge on the tower when you go out there.  ;D

Not much has been said about the voltage ratings. I always think 2X is right, but I've had arcs during 'mistakes'.

If the plate voltage is 1000 and the DC is blocked as it should be, what should the rating be on the input capacitor for maximum reliability?

What happens to the peak voltage when there is very light loading or when the load is forgotten as might rarely happen?

Does it become many times the plate voltage? (thinking of how resonance can really amplify a voltage, and about the flyback circuit in a TV where B+ is 400V and the plate pulse is 5KV)

Does it make sense to use a spark gap across the tuning capacitor for protection when higher plate voltages like 3000V-5000V are used? In one transmitter the B+ is 7KV peak with modulation and a 20KV cap is in use. For an upcoming project I have a 7500V cap, but the B+ will be 4500V.

Better to ask than start ruining vacuum capacitors.