The AM Forum

I finally got around to playing with the homebrew amp (in my avatar)... but it won't amplify. It seems to make a pretty good frequency doubler though.  What am I doing so wrong???

Brief circuit description: four 4-125A in parallel. Intended to run in AB1 on 75 and 40m. Input is to the grids with a 1:4 bifilar-wound toroid and a 200 ohm noninductive swamping resistor. Individual bias adjustment pots for each tube from a regulated -150 volt supply. Screen voltage is 450v shunt-regulated by a 4-65A, plate voltage adjustable (variac) up to 3.5 kv. I bought a commercially made plate choke and feed the B+ through it. Output is the usual pi-network except that the inductor is three stacked T200-2 (red) cores insulated with Kapton tape and wound with #12.

There is drive to the grids, easily 100 v p-p with my SB-102 supplying the "carrier" on 7200. The pi-net does resonate at 7.2 MHz, checked with a signal generator and scope, and the inductance of the toroid is correct on my Tek 130 LC meter. Total idling current initially set to 100 ma, increases to 180-200 ma with increasing drive. But the relative power output on the SWR bridge's meter is the same, or actually less, than the input!  >:(

Watching the output (into a dummy load!) on my spectrum analyzer is also bad news - not only is there less energy out at 7 MHz than is going in, there is almost (a few db down) as much 14 MHz...  :o

so obviously something is really screwed up, but I can't figure out what it is... any thoughts?
???

thanks
Charles

Charles,

What level of plate voltage and grid bias are you actually using?  In straight AB-1 with 450 on the screen I assume you are running pretty high plate voltage and the 25 mil per tube idling current also points in this direction.  With your 1 to 4 step up, even with 200 ohms swamping resistance your drive should be extremely low.

Can you actually dip the plate current?  What is your screen current showing during this process?

I originally intended to use this amp with my screen-modulated ARC-5 transmitter on 40m, which is only about 10W carrier.

The plates are running at 3.2 kv (about 90% on the variac). Not sure what the actual grid bias is, will have to measure.

Could the tubes themselves be zorched? The filaments are lit and they're drawing current and showing a bit of color on the plates. Wondering if I burned up the control grids...

Now that you mention it, I don't think there is much of a dip in plate current at all (although I was tuning for max output, not min plate current). Also the screen current seems to be almost zero. I need to check and make sure the metering circuit is actually working, too.

Will take a look tomorrow when rested. I don't like working around 3 kv supplies with 56 uf capacitors at all, but especially when I have a cold and feel tired!

I am not familiar with toroids in a tube rig as the L in the pi net at those frequencies but it sure sounds interesting. Could the Q be low for some reason? Beware the screen current that you don't fry them.

What is the cap values in the pi network? Also how much inductance are you getting with that inductor. Sounds like the tank is set up wrong. You need about 250 - 300pf in the plate tune and 1500 to 2000 uf in the load cap. 8 to 10 uh inductor on 75.
If it makes power on 40 the tubes are fine.

Any chance for some pics sometimes that helps others help you.


73

Jack

Based upon no/shallow dip and no screen current it is probably loaded way too heavily.  Try reducing the loading and try again.  Screen current is the best indicator of resonance in tetrodes and the screen current will peak at this point. 

Tuning on the second harmonic will do the same thing if the stage is acting like a doubler

Also....seems unlikely that all 4 tubes would be bad, but it is possible.  Have you tried just one or two at a time?  If they are in parallel, this should be possible.  Just plug-in one or two and try it.  Of course, input and output capacitance would be lower, but, hopefully you will still be in tuning range.  Also plate current would be proportionally less.  Is the plate coupling cap from the tubes to the Pi net OK?

Just some thoughts...

73,  Jack, W9GT

I have often used toroids in a pi net - It sounds to me like the pi - net is not correct. How did you calculate it and what are the values. Did you measure the toroid inductance?

Pat
N4LTA

Sounds like insufficient drive for starters.

Hopefully you have a grid meter and its seperate from the screen meter.

Increase drive while watching the RF meter and doing a 2 handed Tune and Load tweak for max. Continue increasing drive until you start seeing some screen current.

Once you start getting the tank to function at the designed Rp then the 2nd harmonic will be down where it belongs.

Ive used toroids for decades as both the L in a Pi as well as the L in a Pi-L and they work fine as Alpha and others have shown since the early 70's up into the 2500W area. I havent gone beyond about 1200W yet.

If the tubes are showing idle current and color they are working. How well needs to be determined but they should be OK on the low bands at least.

Carl
KM1H

Good point Pat!  There are a number of easy online calculators such as this:  http://www.qsl.net/wa2whv/radiocalcs.shtml

If you have an MFJ analyzer or similar you can use it to measure backwards.  With power  removed, connect the MFJ to the output connector and set it for the middle of the frequency range for the band selected.  Connect a non-inductive resistor equal to the calculated plate load from the common tube anode connection to ground.  Leave all other connections intact so that the tube interelectrode capacitance is still present.

See if adjusting your tune and load controls will result in a proper match indication on the MFJ.

REMEMBER to disconnect the resistor and analyzer before your next power up test!

   Charles,

If the toroid on the input side has a 1:4 turns ratio, doesn't that mean 1:16 impedance ratio?

When I make broad band transformers like that, I use my Autek RF-1 to evaluate the part across the frequency of interest to see how it behaves. This way I can quickly see if I need a different core, or possibly more turns.

What is the input SWR look like?

Jim
WD5JKO

Wow... thanks for all the input, everyone  ;)

Re: the pi-network, the plate tuning cap is air-variable 300 pf, the  inductor measures at 8, 11 and 16 uh (3 taps) on my Tek 130 LC meter (140 Khz), loading is a 1000 pf vacuum variable with switched additional doorknob caps up to 2000 pf. I got these values straight out of my collection of ARRL Handbooks...

With the power off, coupling a signal generator to the plates through a 2k resistor does show resonance at 75 or 40m depending on inductor tap. It can't tune to 20m because the minimum capacitance with four tubes in parallel and other strays is above that point.

I will take a look at the grid drive more closely. The transformer is bifilar (1:2 turns ratio) thus 1:4 impedance ratio, with the 200 ohm swamping resistor it's close to 50 ohm and the input SWR is very low too.

How much swing at the grids should I be looking for? I thought after looking at the curves for a 4-125A that 100 v p-p would be plenty. That's about 30 v RMS, or 18 watts into the 200 ohm resistor, and I've gone well above that from the SB-102 driver without getting any screen current either.

Sounds like the no (or very low) screen current is telling us something... if I remember right, it should be about 4.5 ma per tube when properly driven/loaded?

Do you have a good Bypass on each heater lead and each screen lead?
I would think 100 volts would drive the tubes well. Also look at your plate coupling cap. If it is too low a value or bad you won't couple power.

I think you have too much inductance. I would think 10 uh- 12 uh on 75 should be enough.  I think the Q is too low so the second harmonic isn't well attenuated.
I assume you are running AB not class C.

Hi Charles

A class AB1 amplifier doesn't need power to drive it,it needs voltage ,since there's no grid flowing in it.
So the driver power is dissipated  in the swamping resistor,and grid resistor of the Final tube,
You wrote that You can easily drive 100 v peak to peak the Final Rf tube but got no" output."

Maybe,just a maybe You forgot to put an RFC between the G1(control grid) and the Bias supply ,

So the Rf drive (voltage)  goes directly to G1 which is also connected to the bias supply,that works as a smoothing supply.
So the bias changed because of it ,that makes the plate current also change?  not because amply ing the Rf drive .


Gito

Yes, the heaters and screens are bypassed, short leads too.
Plate coupling cap is three 680 pf Russian ceramic doorknobs (about 1" diameter) in parallel.


Agreed, that is why I designed it that way (to drive with a low-power exciter of about 10 watts).



I wish it were that simple a problem ;) - but I did put rf chokes in each grid lead. The 100 v p-p was measured directly at the grid socket pin (with power off, though), so I know the wiring is ok.

More later when I get some time to look at it this afternoon.

What value are you using for the resonant plate load.  I am assuming you are planning on using the 4 tubes in parallel to run around 1 KW input at around 333 mils at 3 KV loaded.  If so, in AB1 I believe you are looking at around 5,500 ohms.  

Unless I made a calculation error (quite possible) you are looking at 85 pf for tune less the stray plus tube capacitance, 300 pf for load, and an inductance tap at 21.  These are for the 75 meter range.

The values you are using are correct for a 2K load but you need to be running either close to 2000 watts (too much for those tubes in AB1) or greatly reduce your plate voltage to hit that value.

are you sure the loading vacuum cap is good?

What tank Q did you use for those values?... I believe I need a fair bit higher than Q=10 in order to "make it work" on 40m, with the large stray + 4 tube minimum. So there will be some increased circulating losses but it's a compromise either way.


Yes... checks out on the LC meter, and the output power  tunes through a very broad maximum as the loading cap's shaft is turned.

(http://i42.tinypic.com/2vlp4c5.jpg)
(http://i44.tinypic.com/swtlxk.jpg)

Please excuse the crappy appearance, I have been cutting and hacking connections all over the place  ::)

That is obvious but one way to see what is going on is to try to drive it into grid current. That should be possible no matter what is on the plate side. Without a grid meter its guesswork. If the tubes arent taking drive it aint gonna work.

Either that input toroid is connected wrong or something else is keeping the drive from exceeding the bias. Try just a 50-100 Ohm non inductive resistor or even the 200 if the rig will load into it.

With only 3pf output C per tube that amp will even run on 10M once the strays are minimized.

Carl
KM1H

Those are for 75 meters, the values I calculated for 40 meters are 47 pf tune, 200 pf load, 11.2 uh inductance.  This is with a Q of 12 which ought to be fine for a class AB1 amp. 

I have a 4-125 Eimac spec that calls for 600 volts on the screen with -96 volts of bias and 96 volts of drive (at 2500V). So you have to drive the grid up to zero.

I would say double the C and 1/2 the L for 40M

OK, some measurements and things I have learned:

First, the huge 14 MHz spur was an artifact because I did not have a direct connection to the spectrum analyzer, but a short length of wire antenna. Now that I've run the analyzer input through a large resistor directly to the dummy load connection, it looks clean. The 2nd harmonic only comes up above the noise floor when whistling or otherwise overdriving the exciter/amp  ;D

Idling: Ep 2.8 kv, Ip 70 ma
Driven at 7.2 Mhz: Ep 2.6 kv, Ip 130 ma dipped, (160 ma off-resonance), Is 2 ma
On the grids: 120 v pp drive, bias -70v (i.e. swing from -20 to -140 Eg)

The output power does tune through a broad peak while adjusting the loading cap.
Plate dip is "normal", not too sharp, not super-shallow.

The power output is much better with the plate supply all the way up, 3.6 kv and 100 ma idling.

The input toroid is doing its job correctly - the output voltage (to the 200 ohm swamper and the grids) is exactly twice the voltage at its input. However, that signal is attenuated because... I'd forgotten that at some point I'd installed a t-pad on the input (20 ohms in each tee arm, 50 to ground) which is a 7.5 db attenuator. I had been planning to drive it from my Heath SSB rigs in that other mode  :-X

I have found that there's plenty of power gain even with the 7.5 db attenuator and I can drive it hard (up to 200 ma plate current) on either 75 or 40 meters. Spectrum analyzer looks clean, too. But although the tank will resonate on two different inductor taps, the "wrong" one really kills the output power. It's considerably more critical than I thought. Guess I do need to study pi-net design more carefully!

It also seems that I have picked the wrong operating point? This amp originally was intended for use (and still may be) on AM, driven from my PW ARC-5. Somehow I had picked 450v for screen, possibly because my plate transformer gives 3600v when directly connected to the ac line...

Charles,

It sounds like you are on the right track now.

Going with  3.6 Kv. on the plates is well within 4-125 specs but you end up with a very high plate load resistance which changes your pi net calculations. 

How much power do you plan to extract from the amp? 

Well... as much STRAP as I can, and still stay under 500 watts plate dissipation!  ;D

Although my Thordarson plate transformer says 900 VA on the nameplate, so I can't run 1+ KW input continuously if I want it to last. Can't afford Peter Dahl iron  ::)

since the rig runs better at higher voltages I suspect the pi network is set up for a high plate Z. You are at a low q point if the dip is shallow. The tubes will love the extra voltage. You will get more power if you reduce the inductance a bit.
130 ma resting it getting up there for dissipation at 3600V. I ran my old 4-1000A GG at 110 ma. You might want to blow some air on the tubes and pressurize the chassis to keep the seals cool.   

Actually set to 100 ma at 3600v for now, but still 90w per tube.

The sockets are mounted above the chassis on 3/4" standoffs so I can't pressurize the chassis with a blower. A big muffin fan is blowing across them, (hopefully) keeping the seals as well as the envelopes within ratings!

I designed this amp about 6 years ago and can't find the schematic (if I ever drew one), so the details are a bit hazy, but since I could make the 3600 volt plate supply with junkbox parts (good for 250 ma continuous), that determined the plate load resistance ;)

It IS 3600/(2*0.250) = 7200 ohms, right?  ???

I'm looking at an Eimac Data sheet (for 4 tubes)
Class C 3000, 600, -150, 1500 watts out
AB1 2500, 600, -96, 660 Wo
AB2  2500, 350, -43, 800 Wo
3500V you could get a little more if you keep it cool.
These AB1, AB2 specs are for audio service so RF you might get get a bit more at low duty cycle of SSB.

Get an old copy of the ARRL SSB handbook. It had a grid driven linear using 4-125s, 4-250s or 4-400s
Build a modulator and kick butt with it on AM.

Building a modulator is one too many projects and I don't have the parts lying around anyhow... How about I just use this one as a linear on AM or SSB?  ;D

After all, my original reason for building this amp was to amplify the 10 watt (carrier) PW'er up to 300 watts or so. A power gain of 16 dB should make me audible at least some of the time...  8)

I just need to find the right operating point and I should be able to make smoke...

Most of the spec sheets list AB1 and AB2 for two tube push pull.  The single tube AB1 RF data I found specifies a maximum of 200 watts output per tube.  This is with 3,000 plate, 510 screen, -95 bias, 30 Ip idle, 105 Ip max, and 6 mils screen current.  The 200 watts output is peak and although it does not specify I believe this is not including output circuit loss so your realistic RF output will be a little less. 

Running AM linear is pretty rough service for the tubes so although the output figures lead you to believe you could run up to 200 watts carrier (net of output tank losses) you would be beating up the tubes pretty badly.  To run at this power I would want to use proper air system sockets with chimneys and a properly sized blower.  Around 15- watts carrier/600 peak would be a more reasonable figure for AM linear and at this level you could probably get away with a pair of fans-one to blow through the sockets and another to blow air past the envelope and plate seal.

Tube parameters for that level are 2000 volts, 615 screen, -105 bias, 40 Ip idle, 135 Ip peak, 14 ma screen.  But if it were me I would use the 3,000 volt parameters and reduce drive to the proper level for a reasonable output.  Then if you decide to also use it with your SB-102 or similar you could run it as a pretty high power SSB rig.  It would also be easy to change the bias if you ever wanted to run it as a conservative 1 KW CW/RTTY amp.

How about screen modulation? That should be easy to do. AM linear would limit you to 150 to 200 watts carrier and make a lot of heat. Heck you could modulate the 4-65 voltage regulator.

To put it in perspective you have the equivalent of a single 3-500Z. That means about a 200W carrier at most with plenty of air.

Thats a lot of work for only 13dB of gain. A rework with a pair of cheap 4-400's would give almost another 3dB with the same drive if your PS is up to it.

Screen modulation makes a lot more sense as Frank suggested at minimal added cost or complexity.

Carl
KM1H

 

True, but the reason I decided to use parallel 4-125A's is one of the great ham homebrew traditions - they are dirt cheap! Especially compared to 3-500's or 4-400's.

Anyway it seems to be working better since I increased to full 3.6 kv on the plates and figured out better inductor tap settings. Today's a rainy day and I have a bad cold, so while stuck inside I built a soft-start for the plate transformer.

Now if someone would come back to a CQ (40m... daytime...) I can test it out   8)

It works, on SSB anyway  ;D

No ugly spurs unless badly overdriven. No parasitics on the spectrum analyzer either (despite not sinking the sockets below the chassis for more grid-plate isolation). 200 ohms swamping resistance appears to be keeping things under control.

Just had a nice long chat with a couple of guys in IL and OH... the large muffin fan is quite audible in my D-104, which they originally thought was 60 or 120 Hz ripple. 56 uf of filter capacitance is plenty  :-\

Aside from that, it does provide the expected 2 S-units of improvement  8)
Now maybe I'll try it out with the screen-mod ARC-5!

4-400's aint expensive, about $20 NIB at most hamfests around here. I was given a NIB pair in 1964 USAF boxes last month....the guy liberated them when he was stationed on Guam back then ::)  Might have to build a more contemporary amp one of these days and quit messing with WW2 and earlier glass.

However if anybody has a nice 6C21 to go with the NIB one here I may return to the dark side ::) I like my tubes in pairs at least.

Carl

that's a lot cheaper than I had been seeing them... of course I haven't looked lately. No hamfests that fit my schedule or within a reasonable distance here... can't make it to Dayton this year  :(

For $20 NIB I'd definitely buy a pair  8)

Just searched ebag and the 4-125's seem to be gold plated these days!

I think I paid $20-25 for a pair of NIB 125's at Nearfest a few years ago. Put them in my 2M FM amp for contests; the old ones from the amps FAA days were real tired and I didnt loaf them either for 25 years. :-X

I'd miswired something in the relay circuitry so I had to flip the standby/operate switch instead of keying from the SB-102... so today I went to fix it.  While doing so I broke one lead of the bias-cutoff relay off flush with the epoxy. Don't have a spare 24 VDC relay either, naturally.  >:( At least I got to use it once before I broke it!
There's a major kwermtest started this weekend anyway, so I won't be operating very much.

After searching my previous posts on this subject I realized I started building this amp in 2004... where on earth did the last six years go??  :o

I ask myself that a lot.

I unsure about something. Was this amp working and crapped out, or is this a HB in initial testing? 

Assuming the latter...
At what frequency are you measuring your plate inductors? Could they have significant shunt C between the windings at RF? 

Yes, homebrew in initial testing.
My Tek 130 measures at 125-140 KHz.
Not sure how to calculate stray capacitance on a single-layer toroid coil... what would you consider "significant"?

I would try to measure the inductance at the frequency which they will be used. Maybe with a signal generator and known approximate value R in series. Even a rice box at low power might suffice. You could measure the drop across the R and XL and determine the L.  Try it at different frequencies, if there's a considerable change (>10%) in calculated L over an octave I'd say you might want to consider a different inductor design.
unds too like
I sounds like your just not achieving any reasonable Q at resonance at RF frequency.