New 4X1 X 4X1's Rig - Problem with audio spikes - Questions - Pictures

[K1JJ]

Wow,
-wish I had a 3 x 41 ft. shack! 
Guess all your stuff is 'hairpin' matched too.

Maybe Frank's talking about a new "alley shack" behind the house? - I dunno...

Steve: Yes, I'd be interested in seeing the MOSFET schematic. I'm not sure how much grid swing is required of the 4X1's in triode-connected service.  I've never seen anything published.  But certainly it can't require more than 833A's.

For the cathode driver I put 50V positive on the triode-connected 813 grids to get all the headroom I needed. Without this bias, there was not enuff monkey swing.  I'll get by with this 8 ohm transformer scheme to get on the air, but will have to build up a transformerless driver later - and choose between the SS MOSFET stuff or the 807 direct coupled design. No rush.


Opcom: Two 3-1000Z's X two?  Those tubes are like gold - do you actually have some? I'd rather be using 3-1000Z's for modulators, if they were common, but I think the triode-connected 4X1's are pretty close in performance from tests.

T

[WA1GFZ]

The south wall of the house is 41 feet wide. I covered 3 feet with flooring with 27 feet to the north yet to do.

[W1RKW]

Frank, How are the knee's holding up?

[steve_qix]

[
Steve: Yes, I'd be interested in seeing the MOSFET schematic. I'm not sure how much grid swing is required of the 4X1's in triode-connected service.  I've never seen anything published.  But certainly it can't require more than 833A's.

Ok, triode connected tubes... how are you connecting them?  Grids together?  Control grid at AF ground and drive the screen only?  There are a number of possibilities.

Let me know 

[KC2IFR]

Steve,
Will u PLEASE stop trying to shove this SS stuff down the throats of the toob guys.....u are polluting the toob gene pool....

ONLY KIDDING guys.........

Bill

[K1JJ]

Ok, triode connected tubes... how are you connecting them?  Grids together?  Control grid at AF ground and drive the screen only?  There are a number of possibilities.

Let me know  

Yep, as I always connect 813's or 4X1's:

Screen and control grid are tied together and driven together.  The bias is supplied by putting diodes in series with the filament center tap. No bias is needed at 3500V, so the fil CT goes directly to ground.

I'm driving the tubes now with a solid state 8 ohm amp/backwards xfmr and from the amp LEDs lighting up, it looks like maybe 30 watts of audio are needed to drive the p-p pair of 4X1's. Then again, this is in the low RF power tap and need maybe 60w of audio later.  But I cud make some careful voltage and power measurements if you need.

T

[WA1GFZ]

Bob,
First two courses were by hand on my knees so Sunday I was pretty crapped out.
Taught my son to operate the nailer by Saturday so that made it easier. White oak is tuff stuff and wider width made it worse.
Drilling and blasting the rig would have been more fun...but just before I left I sanded it and put down a coat of varnish. Looks rally nice. Now I know why old ships were made of white oak.

Tom,
I would think a triode connected audio 4-1000A would take the same grid swing as a GG linear needs cathode swing. Say it needs 100 watts of cathode drive  then E^2/50=100 so E^2= 5000 so E = 70.7 VRMS  70 X 2.8=198 V P-P

[KC2IFR]

Tom,
Let me know when your gonna be on 75 with the rig.........I would love to hear it and record it also.
Thanks,
Bill

[AB2EZ]

Tom

In an application like this, the backward connected audio output transformer may be best modeled as a voltage step-up transformer:

The solid state audio amplifier behaves as a voltage source (unless you place a very low impedance across its output). I.e., in bridged mode,  its output voltage swing will be (essentially) the same whether you put 8 ohms or an open circuit across the output terminals.

As an approximation (below) I will assume that the equivalent circuit (in bridged mode) is an ideal voltage source in series with a resistor whose value is "less than 8 ohms". However, one could probably assume that the equivalent source resistance is much less than 8 ohms, provided one does not draw too much current from the amplifier.

Since the nominal load is 8 ohms, and the nominal peak output is 120 watts (in bridged mode), the nominal peak output voltage (red-to-red output terminal in bridged mode) is the square root of 120 watts x 8 ohms = 30 volts (independent of the load, as long as the load is 8 ohms or more). The associated, maximum, peak-to-peak output voltage is 60 volts.

Note that the B+ and B- supplies are ~ 18 volts... so this is consistent. Also note that when specifying the peak output power of the audio amplifier... the manufacturer is defining it as the absolute peak output power (not averaged over a cycle of a sine wave audio signal).

If (for example) the audio output transformer has a turns ratio of 16:1 (i.e., the output winding has been optimized/designed for driving an 8 ohm load, and the input winding is optimized for being driven by a source with a 2048 ohm source impedance... so the transformer is specified as: "8 ohms on the output winding" and "2048 ohms on the input winding"), then in this mode of operation (backward connected), the output of the transformer will look like the following (as an equivalent source of audio)

Peak output voltage = 30 volts x 16 = 480 volts (960 volts peak-to-peak)
Effective source impedance= less than 8 ohms x 16 x 16 = less than 2048 ohms

If you don't need such a large voltage to drive the grids of the 4-1000's, you can reduce the effective source impedance at the output of the audio output transformer (and also improve its frequency response at low frequencies) by using only one of the outputs of the solid state amplifier. Each output (i.e. stereo mode rather than bridged mode) has a source impedance that is less than 4 ohms ... so the effective source impedance at the output of the transformer would be less than 1024 ohms (in this example). This will reduce the peak output voltage to 240 volts (480 volts peak-to-peak)

Even better: If the transformer has a 16 ohm winding, use that (with the amplifier in bridged mode). You will reduce the peak output voltage to .707 x 480 volts = 340 volts (680 volts peak-to-peak);  and you will reduce the effective source impedance at the output of the transformer to less than 0.5 x 2048 ohms = less than 1024 ohms.  

If you drive the grids of 4-1000's positive, with respect to the cathodes, then the solid state amp will deliver power. Otherwise, it will not. The nice thing about the solid state amp is that it is a voltage source... and its output voltage will not be affected (significantly) by the changing load impedance presented by the grids of the 4-1000's (as seen through the backward connected audio output transformer).

Best regards
Stu

[KD6VXI]

Even better: If the transformer has a 16 ohm winding, use that (with the amplifier in bridged mode). You will reduce the peak output voltage to .707 x 480 volts = 340 volts (680 volts peak-to-peak);  and you will reduce the effective source impedance at the output of the transformer to less than 0.5 x 2048 ohms = less than 1024 ohms.  
Best regards
Stu

Back in my "destroy the ears" days (read, car audio installer), we also found that by bridging our amps to mono, and driving a higher impedance, we could get the same power, but THD dropped like a rock.

Not that we have to have .005 THD for an AM transmitter, but it's NICE to have clean audio chains

--Shane

[K1JJ]

Thanks for the info, Stu -

Here's the impedances and specs of the audio xfmr I'm using.

It's a 62 watt 20hz-20,000 xfmr.

The windings on the secondary are 7,000 ohms CT.

The primary has one large winding with 80 ohms CT and 265 ohms CT.

I am feeding amp between the 80 ohm and CT  - and across the 7,000 ohms CT to the grids.  This is the highest step up I could find on this transformer.

The 80 ohm to CT is less than 80 ohms cuz it uses 1/2 the winding.  The overall ratio seems to work well, though it's not a true 8 ohm to whatever transformer.

The transformer is about 10' away from the 8 ohm amp output. The transformer is about 4' away from the 4X1 grids. The tubes idle at about 100ma each at 3KV - I plan to raise this to 3500V later on. I think the idle current will be fine with zero bias.
Any suggestions?   I don't want to buy another transformer at this point since this one works FB.

Bill, yes it wud be nice to hear a tape of the rig. Shouldn't be too long now. I'm trying to get minor details worked out now like NFB, etc. I tried Stu's amp last night and it basically solved the problem. Sounds good and now the un-key spike is about 40ma - hardly noticable. I'm going to work on that some more too. There is absolutely no sparking in the gaps at all and the rig is very stable overall for both RF and audio.  The other amp was just too close to the RF field - now the new amp is 10' away and loves it.

T

[WA1GFZ]

Stu,
I'm thinking it would be a good idea to use a little resistance in series with the 8 ohm secondary since it is further away from the transformer core. When flipped around as a  primary, I bet the z is lower than 8 ohms. Triode connected 4-1000A at 3500 volts should cut off between -20 to -30 bias so the tubes should load the transformer pretty well.

[WA1GFZ]

Tom,
80 ohms you might take a look at the wavefrom. It may want a shunt load  depending on the SS amp. 16 ohms in bridge may be happy to drive it.

[K1JJ]

Tom,
80 ohms you might take a look at the wavefrom. It may want a shunt load  depending on the SS amp. 16 ohms in bridge may be happy to drive it.

Frank, it's not 80 ohms - I'm feeding 1/2 of the 80 ohm winding single ended = 20 ohms. It's an 80 ohm, CT winding and I'm feeding 80 to CT.

What does that make the impedance of the input and the overall ratio to 7K?

T

[AB2EZ]

Tom
Et al.

As I mentioned, thinking about impedances may be more confusing than helpful... in this application.

If you are using half of the 80 ohm winding, then that is equivalent to using a 80 ohm /4 = 20 ohm winding... in the sense that: if you put a 7k ohm load on the 7k ohm "output" winding, the resulting input impedance (looking into the input winding) will be 20 ohms.

But, if you just think of the turns ratio (which is equivalent to voltage the step-up ratio)...

Using half of the "80 ohm" winding, and all of the "7000 ohm" winding... the step-up ratio is 2 x the square root of [7000 / 80] = 2 x 9.35 = 18.7

This is a good number. I.e., the peak output across the output winding (not from one side of the secondary to CT) will be (for this amplifier, in bridged mode): 30 volts x 18.7 = 561 volts. The peak voltage from one side of the secondary to CT will be 561 volts / 2 = 280 volts.

The output of the transformer will have an equivalent source resistance (from one side of the secondary to CT) of less than 8 ohms x 9.35 x 9.35 = 700 ohms... and that is surely be low enough to serve this application. [The grid-to-grid equivalent source resistance of the output of the transformer will be 2800 ohms... but I think it is less confusing to think of the grid to CT equivalent source resistance]


If the transformer were used as an output transformer: driving an 80 ohm load with 62 watts of peak power, then the peak current (not the rms current) would be the square root of [62/80] amps =  0.775 amps. Saturation effects will come in when the product of the current and the number of turns exceeds some critical value. With only half of the 80 ohm winding being used, you need to put twice as much current (to produce 62 watts of output), but you have half as many turns... so the peak power you can run stays the same (from the perspective of saturation effects).

As you are using it (as an interstage coupling transformer... driving the grids of your modulator's output tubes), I suspect that you will not cause excessive heating (resistive losses) of the 80 ohm winding (or the half of it you are using) because the average power will be much lower than the peak power. However, you should be careful testing the rig at 100% modulation with a sine wave applied for a long period of time.

By the way... some time ago I posted the .wav file of a test waveform that puts very little stress on the components (low duty cycle); but still modulates to up to 200% on positive peaks. It consists of a short positive-going pulse, followed by a short negative-going pulse (having half the amplitude and twice the width of the positive-going pulse), followed by a period of 0 volts. It repeats every 200 milliseconds (if I recall correctly), and has a duty cycle of about 10%.

Stu  

  

[W1RKW]

Why wouldn't K1JJ's solid state amp be considered a current source at 8Ω, even in this application?

[AB2EZ]

Bob

The way the amp is designed (using feedback), it delivers a specified voltage into any load... provided the current delivered to the load stays below a maximum value that is set by the overload control circuitry.

For example, for a given audio input signal, it will deliver the same output voltage whether the load is 8 ohms or 16 ohms or an open circuit. Therefore it is, by definition, a voltage source. Except for the maximum current (overload) constraint, its effective source impedance is essentially zero... but, for simplicity, I have approximated the source impedance as "less than 8 ohms". 

An amplifier designed to work as a current source would deliver twice is much voltage (for the same input signal) if you doubled the output load resistance.

Stu

[K1JJ]

Hi Stu,

Good on the transformer calculations.  I understand - and have the rig running very well now. It took some time to get the NFB right, but I'm real pleased with the response and power of the modulator system now. Your amp is working FB.

OK on the pulse tester. I've been looking for something like that for my linear amp ssb tuning. I don't ever use a carrier to tune up - rather a few "chhh-chhh's" into the mic to keep the audio duty cycle low. I will play your .wav and tape it into a recorder for use on both AM and ssb.  Tnx.

I'll do a search for it now.

I've also heard those "chirper" units that sound like telementry some ssbers use. If seen them sold in a little box. I'd like to build one of those too. Probably a simple 555 wud do it if set up right for narrow pulses.

Later -

T

[KC2IFR]

Quick question Stu,
I thought if using a backward connected transformer, when the grids draw current on positive peaks.....the Z will drop and the transformer output will drop causing some distortion on the audio output of the modulator. Im not trying to split hairs but just wondering. When I went to the cathode follower, because of the low Z of that ckt., it could handle the varying Z changes a lot better.

And Tom......I use my WinRadio SDR receiver for recording...... no tape.......its digital to digital.....

Bill

[K1JJ]

Question:

The recommended bias in class C for a 4X1 is -200V.  I have a fixed -90V and an additional grid leak of any value I choose.

I tried a 5K leak resistor that gave me about -315 total grid voltage. This was at 40 ma of grid current.  (Deep class C)

Then I tried 10K which gave me about  -600 V grid bias.  (Into class D?)

Both seemed to modulate well.

I'm curious - Does the grid of the tube HAVE to run 40ma? What if I ran it at 20ma with the 10K resistor giving about -300V? Is there some reason the tube needs 40ma?

The efficiency seems to be very high now in class D. I did the math, but it seems too high to even say, so I won't ... :-)

BTW, I just looked up the max grid rating and it's -500 V, so guess I'll stick wid the 5K and 315V for now.


Bill: The last recordings you sent me were very good - I was pleased the way I sounded for once...

T

[WA1GFZ]

Tom as duty cycle goes down when you bias further into class c the efficiency goes up, but you need to maintain grid drive to saturate the tube when it is kicked on.
A third harmonic tuned circuit in the screen will drive it even higher. I read somewhere 85% is possible
A backwards transformer will have a good portion of the core air since it is sitting on top of the other winding. This will reduce the inductance of the winding. Sounds like the 20 ohm winding worked well at 8 ohm source but may end up a bit low at 16 ohms bridge. You will have to look at it to be sure.

[Opcom]

Opcom: Two 3-1000Z's X two?  Those tubes are like gold - do you actually have some? I'd rather be using 3-1000Z's for modulators, if they were common, but I think the triode-connected 4X1's are pretty close in performance from tests.

T

I think 99% for sure there are four. In return for fixing a computer, I was given three 13.5MHz 1500W RF generators. One was said to work and has the remote control, all had the big tube and associated parts. 208V filament transformers though. I also have a spare tube. Amperex, very badazz looking. The tubes, the two I have pulled anyway, appear to be like new, because these generators were under maintenance and the tubes were replaced well before lifetime. Cheaper to replace a 3-1000Z than to put a semiconductor line down. I parted one of the not-working generators, the other 2 are left intact.

I know they are costly little beasts. Last I saw was $650-$1500 depending. I was shocked the guy only asked me to fix his computer in exchange for them and I tell him, well, you know those tues are worth quite a bit, are you sure you want to trade all this for just fixing your computer? - I discovered the issue was that it was his XYL computer, and had somehow been corrupted by viruses and worms from "Asian dating sites" so he said, and that this had so far been successfully blamed on his teenage son, and that he needed the computer totally sanitized asap to be presented back to the XYL in proper and non-filth-laden working condition.

What I hate is when you shop for items like this and it says "call". What that tells me is that they do not have it and will act as a broker to get it for you, or will decide the price after you call. I just lose trust there and go to the next vendor.. Does anyone else feel the same way about dealers 'hiding' the prices from prospective customers, or is there a good (honest and customer-beneficial) reason for it?

On the other hand I thought the 3-1000's would be good in case I ran out of 4-1000's for the Tucker xmtr. Which is not likely to happen.

Well back to it I do have 4 of them with all the trimmings and I am not likely to use the set.

[WA1GFZ]

A very nice triode....

[K1JJ]

Yep, I wud pop those 3X1's in as modulators in a second.  Use a 4X1 (tetrode) as the final and you're golden.
Not a bad deal for fixing up his computer...

Well, after some more extensive testing and mods tonight, I'm ready to put this baby on the air maybe Thurs or Fri.  All systems seem acceptable.

I've decided to go sailing with Yaz tomorrow. It's gonna rain in the morn, but then clear up with high winds and 60F temps. Probably the last sailing of the season. Been working too hard and Yaz deserves a good outing. I like to land a lot and he jumps off, sniffs around, gets back on, we raise the sail and off we go to the next destination.

I hear Derb on tonight at 45-50 over. Nice sig Derb!  
Update:  I just gave Derb a call and made him #1 contact on the new 4X1 rig. Tron was #2.  It's nice to hear ya back on OM!  

T

[AB2EZ]

Bill
Et al.

A lot of our intuition (including mine) is based on assumptions that are outdated. In the past... when we designed an audio system to drive a given load... we designed it so that the output impedance of the audio system was matched to the load impedance. This produced the best possible efficiency.

Now, we can buy a 600 watt audio amplifier (if we wish) with essentially a 4 ohm output impedance (or less). If we hook that to a 20:1 step up transformer, we can drive a load whose impedance varies all over the place... as long as it stays above 4 x 20 x 20 ohms = 1600 ohms.

The audio amplifier doesn't care if the load is infinity or 1600 ohms.

The "price" we pay is (perhaps) relatively low efficiency. My amplifier may be consuming 50 watts of average AC power from my 120 VAC outlet... and the average output power may be 1 watt (or less).

Tom is using a step up transformer that he has on hand.

In my shack, I am using a $10.00 (~current price for a new unit) toroidal power transformer from Antek to externally plate modulate my Ranger in Heising configuration. The Antek transformer has a pair of "120 volt" primary windings and a pair of "12 volt" secondary windings. Thus the step up ratio can be set to anything from 1:20 (12 volts to 240 volts) to 1:5 (24 volts to 120 volts). I'm using a 1:20 ratio to drive the Ranger at full output power: 30 volts from the audio amplifier produces 600 volts of output from the Antek transformer. The bandwidth is flat over the relevant frequency range (20 Hz - 10kHz).

Since the output impedance of my 150 watt per channel audio amplifier is (much) less than 4 ohms, the effective output impedance of the toroidal transformer is (much) less than 4 ohms x 20 x 20 = 1600 ohms.

If I were using this transformer to drive the grids of a modulator, I would use a step up ratio of 10:1. In that case, the effective output impedance of the transformer would be (much) less than 4 ohms x 10 x 10 = 400 ohms.

A source with a (much) less than 400 ohm output impedance would not "care" if the load varied from infinity to any value greater than several hundred ohms.

Best regards
Stu