Call Search
     

New to Ham Radio?
My Profile

Community
Articles
Forums
News
Reviews
Friends Remembered
Strays
Survey Question

Operating
Contesting
DX Cluster Spots
Propagation

Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net

donate to eham
   Home   Help Search  
Pages: Prev 1 [2] 3 4 Next   Go Down
  Print  
Author Topic: Building 4-1K amp - parasitics and other problems  (Read 8281 times)
W8JI
Member

Posts: 9748


WWW

Ignore
« Reply #15 on: June 30, 2008, 03:46:59 AM »

You don't need a .01uf cap for plate coupling. that's just silly.

Even a 500pf is far more than needed for 160-10 meters.

The problem is the voltage rating, the series inductance of the capacitor and anode path, and the CURRENT rating of the capacitor across HV.

The last worry in the world is the capacitance of the blocking cap. Think about it logically. The anode operating impedance is several thousand ohms. A capacitor with a reactance of a few thousand ohms would barely affect the anode impedance.

If you do the numbers on paper or do an experiment you will see this is true.

I had a buddy who built a 160 meter amplifier with a doorknob of 100 pF for coupling. A friend of mine and I laughed and laughed, but when he changed the capacitor to a 2000pF NOTHING changed except the plate tuning knob moved a little bit.

73 Tom
Logged
KD7YQM
Member

Posts: 71




Ignore
« Reply #16 on: June 30, 2008, 06:30:38 AM »

Hi Tom thanks for your input.

The socket I'm using is the plastic one. The connection from the 2 screen pins is a copper strap 1/2 inch wide and about 1 1/2 inches long grounded directly to the chassis. The grid pin is grounded by a 1 ohm resistor and a .01 ceramic cap.

Tom please share a good design of tunned input that worked for you.

Tom what size and kind of cap would you recommend for the
plate choke bypass? Is further filtering required in this area?

Thanks to everyone for input.
Dennis

Logged
W8JI
Member

Posts: 9748


WWW

Ignore
« Reply #17 on: June 30, 2008, 06:23:49 PM »

<<The socket I'm using is the plastic one. The connection from the 2 screen pins is a copper strap 1/2 inch wide and about 1 1/2 inches long grounded directly to the chassis.>>

Get some thin copper or brass foil. Cut pieces that will fit through the gap in the socket out to the chassis. Clean the top of the clip up on the flat part and tin it with a dot of solder. Tin the bottom of the foil and sweat solder the foil pieces to the flat part of the spring clip with them sticking out through the socket gap by the pins. This will give you a direct screen to chassis connection of an inch or less.

You don't want to run through the pin, holder, terminal, and then back down to the chassis. Too long.

<< The grid pin is grounded by a 1 ohm resistor and a .01 ceramic cap.>> Why? Metering?

<<please share a good design of tunned input that worked for you>>

I always use a pi lowpass. Not of any special design. The tube is about 80 ohms on the lower bands but 20 through ten will have to be cut and try. You need to keep the pi near the cathode so it presents a low impedance on all harmonics of the drive frequency at harmonics.

Sometimes you can get away without it, but if the cathode impedance gets high on harmonics it can ruin IM performance and efficiency.

<<Tom what size and kind of cap would you recommend for the
plate choke bypass? Is further filtering required in this area? >>

Depends on the plate choke impedance. The plate choke and the plate bypass cap forms a low pass filter to keep RF out of the supply. Generally if the plate choke is large enough you only need 1000pF or so. .005uF would certainly be big enough even for 160 for a bypass.

What parasitic choke are you using?? Are you sure you are using low inductance resistors??

73 Tom
Logged
KD7YQM
Member

Posts: 71




Ignore
« Reply #18 on: June 30, 2008, 06:47:52 PM »

Hi Tom
The suppressor is 3 turns of #12 3/4" dia. by 1" long with 3 100 ohm 2 watt metal oxide film resistors inside.

That does sound like a better ground for the screen.
I will try that.

The 1 ohm and .01 on the grid is for metering.

The choke is a B&W #800 and I was using a 500pf doorknob at the base.
I changed the capacitor to a 1000pf high current ceramic.

I also put a large ferrite 43 bead after the cap on the hi voltage lead.

I also put some ferrite beads on the lead between the feedline capacitor and the filament/cathode. Not sure if that was good. Could that attenuate my drive?

Anyway after those changes I haven't seen an oscillation like yesterday.
Maybe those mods helped. I hope they didn't hurt.

I'm not sure on the inductance of those resistors. I know I don't want to use carbon comp anymore. They tend to short out and cause real problems.
Logged
WB2WIK
Member

Posts: 21836




Ignore
« Reply #19 on: June 30, 2008, 07:04:20 PM »

>RE: Building 4-1K amp - parasitics and other probl  Reply  
by KD7YQM on June 30, 2008  Mail this to a friend!  
Hi Tom
The suppressor is 3 turns of #12 3/4" dia. by 1" long with 3 100 ohm 2 watt metal oxide film resistors inside.<

::Are you sure they're non inductive?

>That does sound like a better ground for the screen.
I will try that.

The 1 ohm and .01 on the grid is for metering.<

::IME, the bypass here is pretty critical.  .01 if it's a chip capacitor with no leads should work pretty well, but if it has any lead length, it might be a tuned circuit at 28 MHz.

>I also put a large ferrite 43 bead after the cap on the hi voltage lead.<

::I had to use five or six, over the lead of a Z-50 (Ohmite) RF choke.  Don't know why that was required, but when I did that, everything became very calm and I've never seen the amp become unstable since 1984.

>I also put some ferrite beads on the lead between the feedline capacitor and the filament/cathode. Not sure if that was good. Could that attenuate my drive?<

::What's a feedline capacitor?

>I'm not sure on the inductance of those resistors. I know I don't want to use carbon comp anymore. They tend to short out and cause real problems.<

::They do?  I never heard of that.  Most old fashioned carbon comp resistors have a low breakdown voltage (300V or so), so using them in HV applications wasn't good.  They also drifted with temperature and age.  But haven't we all?  I still have a pile of "Ohmite" brand 220, 270 and 330 Ohm 2W carbon comp resistors stored indoors (dry) and never used, and I keep using them for parasitic suppressors.  

WB2WIK/6
Logged
KD7YQM
Member

Posts: 71




Ignore
« Reply #20 on: June 30, 2008, 07:48:28 PM »

Hi Steve
No I don't know if they are inductive or not. I just know that I have had 2 of the carbon comp resistors fail by shorting almost to 9 ohms or something. That caused an arc and nearly blew my breakers.

No it's not a chip cap on the 1 ohm resistor. It's just a disc ceramic.
What kind of chip cap do you suggest? I do have one with ribbon leads. I don't know what voltage it's good for. Are there going to be any high voltage spikes across that 1 ohm resistor?

By feedline I meant the capacitor in the drive line feeding rf to the cathode.

Logged
K4DPK
Member

Posts: 1128


WWW

Ignore
« Reply #21 on: June 30, 2008, 09:01:16 PM »

Hi Dennis...

Had a similar problem once, and I put a rubber duck antenna on a freq counter situated away from the amp but close enough to see.  Let it take off, looked at the counter.  Said ~40 Mhz.

So I built a little series-resonant 40 Mhz circuit, put it from cathode to ground and stopped the problem.

Your mileage may vary.

Phil C. Sr.
k4dpk
Logged
KD7YQM
Member

Posts: 71




Ignore
« Reply #22 on: July 01, 2008, 05:25:03 AM »

Hello Phil
Yes I was kind of curious as to what frequency this thing was oscillating. This last time it was oscillating it was not quite as violent as when it first showed itself. I probably could have easily measured it.

But I think I may have cured the oscillation with the ferrite beads and better quality bypass cap on the plate choke.
The bead I put on the b+ line right after the bypass cap was a very large bead. It measures 1 inch long by 1/2 dia.
I put about 4 smaller beads in the drive circuit between the .01 mica cap and the tube cathode/filament. What I want to know is if I'm attenuating my drive input to the tube by doing that?

I guess I'm still not sure I have the pi network taps at their most efficient placement.
On some bands like 20m I'm getting output of 1kw easy. The tube plate is showing a little red glow.
On other bands like 40m I am getting my 1kw output but the plate wants to get really red.
On 15m I'm only getting about 500w output and the plate wants to start glowing really quickly.

Now I don't know if this is just because my driver has that much lower output at 15m or if my tap point is still not optimum.
There's a lot more going on in the process of designing a so called simple gg amp than you might think, and I still don't have the process down to a science. ;>)
What fun.
Dennis

Logged
K2XT
Member

Posts: 39




Ignore
« Reply #23 on: July 01, 2008, 07:23:07 AM »

Do we know for sure that the oscillation is a parasitic? I don't like the sound of it taking off at 40 MHz. When you are making these tests do you have shielding in place?  When I first tested my 8877 I could see "grass" on my spectrum analyzer as I tuned the plate tuning.  As the grass (wideband crap) level rose up, suddenly I had dozens of spurs everywhere.  When I put the top cover on the final compartment it cleaned right up.
Rick  K2XT
Logged
WB2WIK
Member

Posts: 21836




Ignore
« Reply #24 on: July 01, 2008, 08:07:55 AM »

>RE: Building 4-1K amp - parasitics and other probl       Reply
by KD7YQM on July 1, 2008    Mail this to a friend!
I guess I'm still not sure I have the pi network taps at their most efficient placement.
On some bands like 20m I'm getting output of 1kw easy. The tube plate is showing a little red glow.
On other bands like 40m I am getting my 1kw output but the plate wants to get really red.<

::My 4-1000 amp with 4kV Ep delivers >1700W output with about 750mA Ip on all bands.

>On 15m I'm only getting about 500w output and the plate wants to start glowing really quickly.<

::Definitely an inappropriate network ratio!

>Now I don't know if this is just because my driver has that much lower output at 15m or if my tap point is still not optimum.<

::Sounds like a bad tank design.  If stuff like chokes and bypass capacitors were dissipating all the extra power, they'd have exploded by now.  All the extra power that you're not putting into the coax is being dissipated by the tube.

Logged
W8JI
Member

Posts: 9748


WWW

Ignore
« Reply #25 on: July 01, 2008, 08:37:48 AM »

First on carbon comp resistors. If you didn't specifically know you were using carbon comps, you almost certainly were NOT using them. The last company making carbon comps was Allen Bradley and they are special order custom parts.

Those shiny brown phenolic cased carbon resistors people like to call carbon comps are virtually never carbon comps. Since about 1970 or so they have mostly been carbon film or carbon spiral resistors. Since 1980 or so the carbon comps have been all but phased out. The only way you can tell is to break one open and look at the core, because the CASE is identical in every way. It is absolutely impossible to look at the resistor externals and tell if it is composition or film.

A suitable substitute is a an Ohmite OX or OY series metal composition resistor. They are actually a much better resistor.

One thing that does work in a suppressor is to tune the suppressor to the grid resonance. If you add a small mica capacitor in series with a long lead resistor you can significantly increase the loading of the cicuit at VHF while decreasing it at HF. I did that in the Ameritron amps.  

<<Yes I was kind of curious as to what frequency this thing was oscillating. This last time it was oscillating it was not quite as violent as when it first showed itself. I probably could have easily measured it.>>

The key to stabilitly is moving the grid resonance as high as possible in all the grid leads. This means VERY short direct chassis connections right through the socket slots.

<<But I think I may have cured the oscillation with the ferrite beads and better quality bypass cap on the plate choke.>>

The bypass cap would never be involved in an MF, HF or VHF parasitic. There is far too much isolating impedance from the plate choke and there is no real path that causes a common problem. The bypass cap is only involved in AM BC band and lower frequency instability, if anything at all. Most commonly it occurs in conventional tetrodes where there can be a screen to anode feedback path for RF at VLF frequencies, although it it rare even in the most likely cases.

You might have done something with the beads. That is possible, although even that is not very likely.

The normal problem is at some frequency the grid is just not grounded any more. The shunting C of the grid combined with the inductance of the grid leads makes the grids float at some frequency. Then if the anode is resonant near or above the grid frequency the tube forms a TPTG oscillator.

There can be other oscillatory modes but they are rare. The anode has a very high impedance so any very small amount of feedback capacitance to the grid will destabize the system. you don't see many tuned-cathode tuned-grid oscillators unless external feedback capacitance is added, but you certainly comnnly can find TPTG oscillators.

By the way the 4-1000A is a nice rugged tube, but it is a terrible tube for stability. If you want a stable tube pick one that has good VHF or UHF performance. As a matter of fact the better the VHF and UHF performance of the tube the easier it is to stabilize at HF.
 
<<The bead I put on the b+ line right after the bypass cap was a very large bead. It measures 1 inch long by 1/2 dia.>>

I'm not sure what you are doing there because the plate choke and bypass cap should virtually divorce that part of the circuit from the tube. Maybe you changed something else by accident.
 
<<I put about 4 smaller beads in the drive circuit between the .01 mica cap and the tube cathode/filament. What I want to know is if I'm attenuating my drive input to the tube by doing that? >>

Depends on the bead mix, but loading the cathode with loss can help. The reason?? The stray capacitance from the grid to the cathode is part of the tuned circuit involving the grid. Ther real problem you have is you have no lowpass tuned input  near the tube. If you had a low-pass low impedance conventional pi, then the beads or a parasitic suppressor would dominate the cathode system.

Right now you have the transmission line and the exciter in the feedback path, not just the beads, so you have a lot of uncontrolled variables.

<<
I guess I'm still not sure I have the pi network taps at their most efficient placement. >>


It is a common but absolutely INCORRECT myth that tap placement critically affects anode efficiency. Get that idea out of your head.

Tap placement can do two things. It can make the network just not be able to match the impedances, in which case the loading capcitor will actually work backwards (less C means more output or no change in output) from too low of a tank Q.

Tank Q has to be more than the square root of impedances, so with a 4000 ohm anode and 50 ohm load the tank Q would have to be greater than sqrt of 4000/50 or 8.9 plus some safety factor.

Beyond that point the tank Q can be any value and the tube will not get any hotter for the same plate input power. It can be 100 or 1000 and the tube won't get hotter for the same input power.

What the tank Q does do is increase network losses. If you are losing 2% (a typical value) of available power in the tank as heat with a Q of 15 and you change tank Q to 30, you lose 4%.

Let's assume the anode efficiency is 70% and the tank loss is 2% with a kilowatt input. You'd have 700W at the anode and 98% of that to the load, for 686W at the load or 68.6% overall efficiency.

Now if we changed the tank operating Q to twice that value with no changes in component ESR, we'd have 70% anode efficiency and 4% tank loss. You'd have 672 watts at the load from DOUBLING the tank losses, and anode heat and anode color would not change a bit.

<<On some bands like 20m I'm getting output of 1kw easy. The tube plate is showing a little red glow.>>

Good. If the anode never glows red the tube will have shorter life.

<<
On other bands like 40m I am getting my 1kw output but the plate wants to get really red.
On 15m I'm only getting about 500w output and the plate wants to start glowing really quickly.>>

Remember the input circuit. The input circuit has a large effect on anode efficiency. If the tank operating Q is greater than some number larger than the square root of impedance ratios, the remaining factor is conduction angle of the tube. If the cathode is allowed to have harmoncs, especially even harmonics generated by the short conduction angle of the tube, the anode current pulses will be rounded off and extended. I've seen amplifiers have as litrtle as 30-40% efficiency due exclusively to cathode impedance at harmonics of the drive frequency.

This is why you need a reasonable Q low pass flywheel at the cathode.

Without that the exciter, the cable length to the exciter, and all the input wiring inside the amp can have a large influence on anode waveform. This directly affects IMD perfromance and efficiency.

While some amps work just fine without a tuned input, it is a matter of dumb luck. With an infinite length input cable the cathode would be terminated in 50 ohms which isn't so bad. If the ouput port of the radio looks like zero ohms on the second harmonic and the coax is 1/2 wave on the second harmonic, the cathode could see a virtual open on the second harmonic and IMD performance and efficiency can tank no matter what you do with the rest of the system.

The shorter the tube conduction angle the greater the effect of the input system on efficiency.

73 Tom
Logged
W8JI
Member

Posts: 9748


WWW

Ignore
« Reply #26 on: July 01, 2008, 08:50:51 AM »

Dennis,

You are right about carbon resistors.

Carbon is a semiconductor and like all semiconductors can fail "shorted". It is very common for carbon resistors to age down in value from heat.
They age up in value from moisture. You just can't win, so metals are better types.

It sounds to me like your suppressor inductor is too small.

The anode circuit is a long path that extends from the anode through the blocking cap and through the tuning cap to chassis and back to the tube.

The suppressor is in series with that path.

The resistor is normally across the suppressor inductor.

In order to increase stability we want the anode path to be dominated by the resistor value at the frequency where the system is unstable. This means the inductive reactance the resistor is across has to be large compared to the path inductance from the tube anode (inside the tube) to the chassis.

This is why you need short WIDE anode leads all the way to the capacitor, and why the capacitor should be mounted solidly to the chassis as close to the tube as possible. You want the inductance of that path, outside the suppressor itself, to be as low as possible.

Then you make the inductance of the suppressor high enough that any resistor across it will control the impedance of the entire anode path.

You ideally want the resistor path to be capacitive so the desired RF stays out and does not heat the resistor, but the undesired path has low imnpedance.

This is why you want a low inductance resistor and a low inpuctance anode path, but the suppressor itself has to be a significant inductance compared to the entire anode path.

73 Tom


Logged
KD7YQM
Member

Posts: 71




Ignore
« Reply #27 on: July 01, 2008, 10:54:36 AM »

Wow this is a chunk of info to digest here;>)

Rick: I don't know the frequency of the oscillation. I think someone else said their amp oscillated at 40. I wish I did. Now it won't start on it's own so I guess I will attempt some 10m drive and see what happens.

Steve: I'm pretty sure I'm not getting enough drive on 15m because the grid current is fairly low (50ma) at full output of my driver (old ft-101b). I'm probably not getting enough drive on other bands or I probably would have more output. I do not think my power supply is capable of 700ma anyway. I think I do need that tuned input circuit. Someone please give me a good example.

Tom: You write an excellent text book.;>)
  It's going to take me a while to digest all of this.
First off what would be an acceptable suppressor inductance and construction for this application?
 
What is an acceptable resistor inductance? A lot of these metal oxide film resistors do have some spiraling in their construction. I don't have a lcr bridge but I couldn't get any of these resistors to measure any inductance on a cheapy meter I have or resonate with a 100pf cap using a grid dipper up to 250mhz.
 
So I guess if my 4 beads in the cathode circuit were absorbing a lot of my drive they probably would have popped by now. Those beads are fairly small.
 
But What ever I did with the new plate choke cap and bead and the cathode beads did something because I can get it to oscillate on it's own again without drive.

I definitely had 2 resistors fail by shorting. They are the phenolic like shinny type resistor. My carbon resistors went up in resistance when they got really hot. I broke them open and it's a phenolic body with a carbon filling with the wire leads at the end. I don't know what else would be called a carbon comp resistor.

My anode circuit is mostly 1/2 inch copper strap except for the 3 turns of #12 in the suppressor. The vacuum variable is mounted to a bracket that is bolted to the chassis as close to the tube as I can get it. I have a glass chimney. The loading variable is also bolted directly to the chassis not to the front panel.

I guess in summary this all boils down to I need a tuned input, probably more drive and a better suppressor not to mention a better power supply.


Logged
KA5N
Member

Posts: 4380




Ignore
« Reply #28 on: July 01, 2008, 01:55:56 PM »

For tuned input circuit all you need to know is that
the driving impedance of a 4-1000 is 105 Ohms.  Go to

http://bwrc.eecs.berkeley.edu/Research/RF/projects/60GHZ/matching/ImpMatch.html

 which is a calculator for various networks.  Input  50 Ohms for source and 105 Ohms for load (use zero for reactance) and input frequency for each band
use Q of 2 or 3 and go down to lowpass pi network and you will get the inductance value and capacitor values for that frequency.  Repeat for all bands. You can use powdered iron cores for the inductors and if you make part of the capacitance variable you can tune the network for best SWR.
Allen
Logged
KD7YQM
Member

Posts: 71




Ignore
« Reply #29 on: July 01, 2008, 03:34:19 PM »

Thanks Allen

Do you know if these pi circuits will be broad enough so you don't have to tune anything?

My driver has a tube output that can tune 25-100 ohms or something like that.
Logged
Pages: Prev 1 [2] 3 4 Next   Go Up
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!