class C linear amplifier

[K1JJ]

Wow, that's pretty significant, Clark.

So you are getting more power output for a given plate current (more efficiency)  and a lower temperature for the same power output with the class C mode.  (again, more efficiency)   And it stays clean on the air.


Now the next logical test is to try it with an audio tone and see if the same trend holds up under modulation. The efficiency normally goes up from carrier to modulation for a standard linear amp, so let's see if it follows this pattern in "class C"  AM linear.


BTW, I used to own a T-Bolt linear myself. But can't remember how it gets placed into class C mode... is there a switch that selects  the regulator tube bias?

If these tests work out, I will certainly be placing my linears in a more heavily biased class for AM in the future. I use a big string of diodes in the cathode for bias with a rotary switch to go from AB1/2 to C.    Later I will run some IMD tests to see what effect there is. Apparantly it is clean, but there's usually no free lunches...


T

[ke7trp]

Let me know exactly how to test it and I will.  I would guess you want a 1K audio tone injected at some % of mod for each mode?  Heat, current and power measured?

Carl, A popular mod for many years for the SB220 is to change out the Bias diode so lower the Class for AM only use.  Shane will jump in here I am sure. He told me about that modification. 

The Tbolt has a CW/Tune and Linear mode switch Tom.  In CW you are class C with your choice of RES or Tuned input. 

My Tbolt is factory except I used the new high curent hockey puck DC filter caps and double size Screen supply caps.  I also plugged in VR150s in all screen reg tubes to go from 450 to 600 screen.  I then put one of the VR125's I took out and put it in the Bias regulator spot which biased the amp down in curent.  I posted about it in my Tbolt thread.  I think johnson intended you to do the tube swap and shipped them the factory way to keep it "legal".
 
This is why alot of guys down talk the Tbolt.  In factory form they are so deballed they only do 150 to 200 watts Am out of a 125 lb amp!

In CW mode you must have the Blocking Bias contacts switched by a relay from the exciter.. Otherwise, Its FULL plate current and damage to amp.  Mine is switched from the ranger so when I unkey the ranger, the rig is in cutoff.

What all this means to me is that I can load this thing up higher in the Class C mode and still have less heat.


C

[K1JJ]

Let me know exactly how to test it and I will.  I would guess you want a 1K audio tone injected at some % of mod for each mode?  Heat, current and power measured?

Yes, that's right.  Do it just like the last test but use as high a percentage of modulation as you dare.  100% for 3 minutes would be ideal.


OK on the regulator tubes. Yep, that is what it takes.

A tetrode amp can be a super clean stage if the grid draws zero current and the screen is electronically regulated and draws it's reccomended current under a given loading.   Some guys run them unregulated (both grid and screen) and that will create a mess.

I'm crazy about the prospect of building the cleanest amp chain I can.  Just finishing up a new 4CX-350FJ  linear amp right now to drive my 8877 amp. It is possibly the cleanest tube ever made for linear service, being at -45db 3rd according to Eimac.  The "J" is almost 15db cleaner than a standard 4CX-350F. I built an electronically regulated screen supply (+- 100mV)  and also working on a sensitive trip safety circuit for the screen. It's easy to make mistakes and pop a tube like this.  In fact, I like the trip circuit so much, I may build one for all my amps and place it in the grid circuit to shut down the amp when overdriven, mistuned, etc.

I'll post some pics of the amp and this trip circuit in another thread next week.

Looking forward to your next tests, Clark.

[kb3ouk]

I'm seriously thinking of converting my 2000B to class C now. On that amp, I think it might actually be pretty simple, since it appears that they are using a 300 to 350 ohm cathode bias resistor to cut the tubes off on standby, then shorting it out on transmit. But they also have 37.5 ohms from each grid to ground, so don't know if that would need to be changed or removed or not.

And Tom, I was going to mention to test the IMD of AB vs C too see which was cleaner, but I see you already thought of that.

[K1JJ]

I'm seriously thinking of converting my 2000B to class C now. On that amp, I think it might actually be pretty simple, since it appears that they are using a 300 to 350 ohm cathode bias resistor to cut the tubes off on standby, then shorting it out on transmit. But they also have 37.5 ohms from each grid to ground, so don't know if that would need to be changed or removed or not.

And Tom, I was going to mention to test the IMD of AB vs C too see which was cleaner, but I see you already thought of that.

Shelby,

The PTT keying resistor shud stay as is. Also, those 37.5 ohm resistors in the grid are for neutralization. But I usually just strap the grids directly to ground with copper strap.  There is some contraversy of the value of using those grid resistors vs: a direct path to gnd.

To bias the stage into class C, make up a board of about 50  1A 1KV diodes  in series. You may need more or less, depending on plate voltage, tubes, how far into class C you want, etc.  Use whatever diodes you have that will handle the current.  You can use a rotary switch to select various points of idle by shorting some out.  Place it in series with the cathode filament center tap of the fil transformer.  Point the arrows down - the diode cathodes go towards ground and the anode points towards the tube cathode pin.  This will give you a very well regulated grid bias for whatever class you wish. 

This works for both tetrodes or power grid triodes, etc.  This makes the tube's cathode slightly positive in relation to the grid, which is actually a negative grid bias. Used is most GG amps these days.

Some use zeners, but the diodes offer a finer control method.

T

[kb3ouk]

Ok, that's what I wanted to know. Of course, at the rate I'm going with things, it might be a while before I even attempt to tear that thing apart again, seeing that I had it apart twice within an hour last night (second time due to screwing up what I was trying to fix the first time I had it apart) because I might have to do some wire tracing before I start doing mods, I think a previous owner did some mods and of course there's no documentation.

[ke7trp]

Ok.  Time for bed. Not feeling to hot.   Tomorrow, I will drag out the SDR radio/scope and AF gen.  Then take some tests.  I dont know about 100% for 5 minutes on an old Tbolt.  But maybe 60% will do.  I am intested to find out if there is much gain when modulated.

I already noticed that there is a Peak on the Grid drive in Class C.  If you vary the exciter there is a point where current goes up and then with more drive, the current goes down.  However, There is one spot where the output is the greatest.   

Just like my 2x4 Class C transmitter.  Which I now realized, I could run as an amp if I wanted to  

c

[KD6VXI]

Clark, you saw this years ago..  Look at that SB220 I did when I came out to help with the IT Security years ago.  LOTS of peak power, and great efficiency.

Most people won't believe what the SB220 can do, AM, when modified.  1600 PEP with modified input, output and bias.  New BIG C caps in the power supply.  This from a 300 watt carrier.

Another (needed) move is to add a large amount of C to the end of the bias string.  Keeps the dynamic bias source stable.  

I went over this on the telephone to Jim, VE7RF a few years ago.  You can search the yahoo amps archive and find a large writeup about bias, AM and SSB, and how surprised some people get.  He's done large amounts of bias on 3x3 and 3x6s and tested them.

Then, it's not impossible to understand some of those numbers thrown around on CB.  Relays click another half volt on the filaments on TX, and they are biased deep class C, then driven like dog snot to overcome it.  They take it to an extreme, and it's not clean, but done correctly, it can and does work.

--Shane
KD6VXI

[ke7trp]

Yeah. I remember that Shane.  Gotta be 20 years ago. 

kb3ouk-  How about an EBS-1 board for your hunter?  Its triodes.  You can mod the board to have adjustable bias and set it class C to Class A.

Probably work FB for the SB220 and other amps. 

C

[kb3ouk]

I might try that sometime, for now I'll try the diode string in the center tap idea, plus I'm in the process of gathering parts to build a driver for the hunter, that way I can quit dealing with the stupid FT-901 audio issues. Basically use the 901 for an RF source and receiver, then have a 25 to 30 watt modulated amp that feeds the linear. Which that thing can use all the help it can get as far as eliminating heat is concerned. That's what I was working on last night, ripping out the old fan that somebody put in to replace the stock fans, then I just placed a quiet 12v fan on top of the cabinet to pull air out the top and that thing has never ran cooler. It is cooler running with the little fan than it was with the big noisy thing that was in it when I got it, and if I rebias it, it should run even cooler.

[VE3LYX]

I have been watching this thread with great interest. Often wondered about this idea. Somewhere way back in my fadded memory I remember there was a way to use a diode to provide feedback to the grid to improve a class C for AM but I was too young then to pay attention.
I was told often too class C has too much distorion but then SSB came along. Too much distortion ? How bad could class C AM be? Too much distortion or a small percentage compared to none? Then would the ear hear it? The ear that can only sense a change on signal strength of 1 DB . When you think about what 1 db is it makes one wonder. So during operation the signal goes around the corner a bit of the linear portion of the slope. Would it be discernable? In small doses probably not. How much would be too much. That would probably surprise us. For those playing Wolfman Jack with their AM transmitter (broadcast quality) probably they are not interested. I personally like the old antique carbon mic sound on AM and use one on my favourite homebrew. Minor distortion if mostly unperceptable would not bother me. I cant hear well anyway. ;>). Very interesting thread. Thanks to all. I run my small linear right up to the edge.Now I may try leting it sneak around the corner a bit and see what happens.
Don Ve3LYx

[W3RSW]

Back to reading today.
Kb3ouk, your illustration was simply an in/out projection based on a perfect slope, m, of 45 degrees.  It was not real tube characteristics.  Perhaps the tail of a pentode's constant current characteristics, but certainly not real.   Thank you for your class C vs. B idealization.

Still wondering about commercial practice all these years missing "the discovery of the century."  -Not counting recent development of class D and E, of course.  Yes, engineering practice can get stuck it ruts, designs can get ossified,etc.  and yes, a ham's definition of distortion free might not correspond with commercial AM's, so I'm wondering why class ab/C linear amplification hasn't been used, AT ALL.

[W2VW]

Back to reading today.
Kb3ouk, your illustration was simply an in/out projection based on a perfect slope, m, of 45 degrees.  It was not real tube characteristics.  Perhaps the tail of a pentode's constant current characteristics, but certainly not real.   Thank you for your class C vs. B idealization.

Still wondering about commercial practice all these years missing "the discovery of the century."  -Not counting recent development of class D and E, of course.  Yes, engineering practice can get stuck it ruts, designs can get ossified,etc.  and yes, a ham's definition of distortion free might not correspond with commercial AM's, so I'm wondering why class ab/C linear amplification hasn't been used, AT ALL.

Once again the amplifier is only biased class C without drive.

Nothing new here. It was about 10 years ago when I first noticed this. Mentioned it to W3DUQ and he was familiar with the concept.

Same amplifier used on ssb without resetting bias gave the expected bad signal reports.

Not too many full carrier M.W. or H.F. high power linear applications out there today.

Rewind to the days when AM M.W. broadcasters used linear finals before high level modulation was popular.

Maybe some info from 90 years ago.

Like anything else in amateur radio some seem to find a way to take the concept beyond its capability.

If one looks around long enough on the bands one can get any signal report they desire.

[W3RSW]

I understood it perfectly.  I and many of your readers understand that the AM carrier of a previous stage sets a pedestal bias voltage sufficient to allow linear amplification of a superimposed audio signal in what once was a class C stage is now at resting a class AB final stage.
I excepted class E and D..
It is a very neat circuit application and due to bias being applied thusly is no longer class C, but now AB within limits.

Further in your remarks, How does linear amplification of lower level stages into linear final as used in BC service some time ago have anything to do with it?   ...except trying to fluff through an argument.
Why do you feel that nobody "Gets it "except the annointed.?

My question was exactly historically that which you posited again.  What happened 90 years ago or whenever that it was missed if it is more efficient than C plate modulated?

I am digging into " electronic Designers Handbook, " Landee, et. al. , and other references to see if there's any mention.  

Sorry for all the "I's" but certain situations tend to draw them out.  

[W3RSW]

Another thought. Perhaps 100 percent modulation being a limit didn't historically satisfy those seeking 125 percent plus.

[W2VW]

Further in your remarks, How does linear amplification of lower level stages into linear final as used in BC service some time ago have anything to do with it?   ...except trying to fluff through an argument.
Why do you feel that nobody "Gets it "except the annointed.?

My question was exactly historically that which you posited again.  What happened 90 years ago or whenever that it was missed if it is more efficient than C plate modulated?

Rick, you were the person asking why this revelation did not appear on your radar. Your q:
"So why Hasn't this mode been the commercial standard all these years?
Relative efficiencies"?

I offered a possible answer.

Who said this is more efficient than class C plate modulated service? It ay be marginally more efficient than B or AB.

Some are trying to take this 5 steps further referencing CB amplifiers. That is a far cry from what works on AM on the ham bands.

As far as I'm concerned this is not much more than a curiosity.

[K1JJ]

heheheh...  It's starting to get somewhat confusing to follow here.

I agree that this is just a curiousity - that needs further testing.

The test results seen by Clark are worth pursuing. Let's see what he shows when putting an audio tone through at full modulation.


What I don't fully understand is -   is this the result of simply running the amplifier in a lower class (class B vs: class AB) or is something else going on here?    What if the normally class AB linear is "being pushed" (or at least acting like efficiency-wise) being closer to pure class A when running the constant AM carrier? Poorer efficiency.  So by increasing the bias maybe we are simply bringing the operating point back down where it belongs - back to class AB or B again.  

But if lower class, then the tradeoff would certainly be poorer IMD between nearer class A vs: AB or B, which I plan to test myself once I get caught up on projects here.


Let's focus on Clark's results last night and continue with tests:


"Class C mode:
250 watts carrier on Bird43
300MA plate current


104F output at rest.  This was stable.
After 5 minutes 140F.  
It hit 140 and stayed there at 3 min point

Class AB (linear mode)
250 watts carrier on Bird43
400MA plate current
After 5 minutes air temp was 170F.  Rising steady... Was going up when I let off.

In Class C,  The 4-400s have very little color.  Slight red spot the size of a dime or less.

In Class Ab,  The 4-400s are Red from top to bottom and white/yellow in middle

A HUGE decrease in Temp and current.  In Class C, the plate current swings UPward.  My Friend called to say that the Class C mode is Crystal clear, loud and is Clean on the air. "


T

[KD6VXI]

Some are trying to take this 5 steps further referencing CB amplifiers. That is a far cry from what works on AM on the ham bands.

As far as I'm concerned this is not much more than a curiosity.

Not true at all.  I only pointed out that the claims made by some of the mfgs of those amps are always poopoo'ed as impossible, but then on the amateur bands they call the technology something else, and it's WOW, LOOK AT THIS!!!!!

You may think of this as a curiosity, but other people might actually want to learn what, if anything different, is going on.

I think Tom hits it on the head pretty well later, where we may be using our amplifiers in class A during carrier periods, resulting in the poor efficiency.

I've done some experiments with audio derived bias in both transistor and tube circuits.  It DOES work.  It also works with EBS designed to change the class of operation (although you need to set the EBS to act upon modulation peaks, rather than a carrier).  You run pure class C during the carrier, GREATLY increasing efficiency, and during any modulation (as detected at the microphone level, not RF in my circuit), the bias slides up to AB / B.

Saves old unobtanium based tubes a lot of Pdiss averaged over time.

Clark, about 15 years ago.  1998 maybe?  It was near Y2K, I remember that much.  Man, that is close to 20.  Getting old?


--Shane
KD6VXI

[K1JJ]

I've done some experiments with audio derived bias in both transistor and tube circuits.  It DOES work.  It also works with EBS designed to change the class of operation (although you need to set the EBS to act upon modulation peaks, rather than a carrier).  You run pure class C during the carrier, GREATLY increasing efficiency, and during any modulation (as detected at the microphone level, not RF in my circuit), the bias slides up to AB / B.

--Shane
KD6VXI

This EBS  (electronic bias system) stuff is very interesting to me. It was first written about in the 70's but may be a better design now. I've been reading about it and thinking of adding it to some of my GG linears.  I think it gets a bad rap cuz some circuits on ssb linear operation go from hard class C cutoff to class AB and generate splatter on the first word.  Some have too slow of an attack time. In addition, I think maybe an audio derived trigger using a downward expander trigger might be a better source -  instead of an RF derived trigger.  (100mW, etc)

That said, maybe a Schottky diode can give 20uS attack as the latest EBS-1 advertises. Also going from 50 ma idle to 150ma idle is still in the linear portion and shud not generate the crud of full cut-off to class AB with the first word.

The Ameritron EBS-1 circuit (a W8JI design?) - where does it connect to the cathode of the tube for bias - at the D3 / D4 area?  This can be modified for any GG amp, right?   I have an 8877 I'd like to try it on.  If it works cleanly (with mods) on ssb, I will try the carrier idea where it is running in a lower class with carrier only to idle higher under modulation. That's an interesting idea if it can be done audio transparently for gating - and done cleanly IMD-wise.

T

[W7TFO]

The class-shifting of the final is beginning to sound a lot like a Doherty circuit, albeit with one (or parallel) tube(s) and no phasing network.

It may indeed work, but TANSTAAFL.

73DG

[kb3ouk]

What I don't fully understand is -   is this the result of simply running the amplifier in a lower class (class B vs: class AB) or is something else going on here?    What if the normally class AB linear is "being pushed" (or at least acting like efficiency-wise) being closer to pure class A when running the constant AM carrier? Poorer efficiency.  So by increasing the bias maybe we are simply bringing the operating point back down where it belongs - back to class AB or B again.  

I was thinking the same thing last night, and I think that's really what is happening. When the amp is biased into AB, it would stay in AB for ssb since there is no carrier, but with AM the carrier is actually pushing it into class A which is why the efficiency and power output sucks. On AM, if the resting bias is in class C, then the RF carrier voltage is pushing it into class AB or possibly even class B. So the limit on how deep you can bias it into class C would have to be based on how much bias and drive you have so that on key up the tube is running in at least class B or lower. If the tube would be in class C on transmit, then that will be when things start to distort. This brings up another point, if that's what is happening in a linear amplifier when it is being driven, that the bias voltage is actually lower than we think, then the same thing could possibly be happening in a high level modulated final, which makes sense why with some designs you can gain efficiency by putting heavy bias on the tubes. It could be possible that in a plate modulated final, if the bias isn't deep enough into class C, that the tube might actually be running closer to class B when it is being driven. Which would then prove something else that is against common ham knowledge that you cannot modulate a final that is in class B or lower.

[W3RSW]

My apologies to those for the previous.

[ke7trp]

I spoke with Robert this morning on 15 meter AM.  He said this was Class B.  That makes sense to me.  I think when modulated and the Grid drives forward, then you  are in Class AB2 again.   Slept for a few hours. I am going to Drag the gear out to take measurements.

C

[K1JJ]

I was thinking the same thing last night, and I think that's really what is happening. When the amp is biased into AB, it would stay in AB for ssb since there is no carrier, but with AM the carrier is actually pushing it into class A which is why the efficiency and power output sucks. On AM, if the resting bias is in class C, then the RF carrier voltage is pushing it into class AB or possibly even class B. So the limit on how deep you can bias it into class C would have to be based on how much bias and drive you have so that on key up the tube is running in at least class B or lower. If the tube would be in class C on transmit, then that will be when things start to distort. This brings up another point, if that's what is happening in a linear amplifier when it is being driven, that the bias voltage is actually lower than we think... 

Yep, I think that's it, Shelby.

And think about it....  there is really no practical limit to how much bias we can run...even biased into class E, as long as we have enuff RF drive to push the amp up out of the class C  into B - enuff to make a normal carrier. It's a diminishing return at some point.   So if we have enuff carrier, then we will always be above class C to make the linear operate linearly.  If there is no idle current (class C) this means we don't have enuff carrier to work with on AM. It's almost a self adjusting mechanism...

So, the bottom line is when running AM linear, we may need to add more bias to bring the amplifier back to where it normally runs on ssb - in class AB or B rather than close to class A.  As a reward, there is less heat per given power output.   The IMD will probably not be a significant factor since we are talking about the small difference between AB and A.  Getting close to class B may begin to make a slightly bigger difference in IMD, however.

The thing to do is make power efficiency and IMD measurements while adjusting RF drive and bias looking for the compromised sweet spot between efficiency and best IMD.

T

[ke7trp]

I was wrong..  The modulated plate current is MUCh higher in AB1/2 then in Class C.


"Class C mode:
1000 watts modulated on Bird43
350MA plate current


104F output at rest.  This was stable.
After 3 minutes 145F. 
It hit 145 and stayed there at 3 min point
Seems to never run over 145F.

Class AB (linear mode)
1000 watts modulated on Bird43
450MA plate current
After 3 minutes air temp was 190F.  Rising steady... Was going up when I let off.


I cant see any difference on the Spec An. Its clean either way. 

For these tests I used the RESISTiVE input selector.  They suggest this for use with Ranger exciter

Things to note:

In class C mode, The Exciter output is MUCh higher.  I have to load the Ranger to 130MA (factory spec) to get the perfect Grid drive

In Class AB1 linear mode, I can only load the ranger to about 60 MA or I over drive the Grids on the Tbolt.

I do have a Z communications 50 ohm attenuator for use with the ranger and Tbolt but do not normaly use it. 

I decided to try it with the Tuned input Mode on the Tbolt.  I got a GREAT increase in signal quality on the spec An!  That tuned input REALLY cleans things up!  Its showing Second harmonic at 35 DB with RES input and 50DB with the Tuned input.  WOW.

So Tom, Can you explain why in Class AB1 carrier and Class AB2 modulated uses an extra 100 MA of plate current?  I tuned this thing to perfection in both modes.. No matter what I do, linear mode, Uses another 100 MA at 2400 volts DC!!!!

C