A Clean Canvas - 24 Pill Class E Rig Construction

[K1JJ]

Correct     Tom, you can test the modulator drivers, etc. by simply grounding the output of your PWM filter (or the output of the modulator board).  Obviously, no hi voltage can be applied to the modulator !!!!  This will bring the charge pump diode into conduction, and the power supply on the PWM output board will function properly.
There's a nice feature of using the charge pump approach - if there is no load on the modulator, it doesn't produce any output!

Yes, I now see how it functions.

I grounded the floating ground and both PDM power boards worked FB.  Nice square waves at the FET gates with almost no overshoot. I'll wait until I put the transmitter together to test the drains/outputs, etc.  TNX!

Well, I just cleared the bench and put on the 24 pill RF deck for testing. This is the last component before wiring it all together and then getting the many T/R and sequencer functions working right.  Then time to breathe life into Rico Suave.

T

[KF1Z]

"Clear!"


 

[K1JJ]

Hooked up the 24 pill RF deck today and fired it up. I saw good pulse drive thru the 414's into the final fets. I put 15V on the final MOSFETS and saw output into a 4 ohm resistive load.  However, my outside RF connections were clip leads - long and JS for the input and output so I saw signs of parasitics.   Otherwise everything worked.  This is good.

I'll do the final tests on the RF deck when it's properly hooked into the tank circuit and has shielded input cables, etc.

So, everything has been tested in component form and works FB as far as I could test.  This includes the HV supply, PDM generator, PDM power boards, shutdown board, VFO and driver board, RF deck, smaller power supplies, sequencer, metering, T/R switching and tank circuit.

Now on to mounting, integrating and final troubleshooting the various components to make Rico Suave a real live rig. 

T

[W1IA]

Hooked up the 24 pill RF deck today and fired it up. I saw good pulse drive thru the 414's into the final fets. I put 15V on the final MOSFETS and saw output into a 4 ohm resistive load.  However, my outside RF connections were clip leads - long and JS for the input and output so I saw signs of parasitics.   Otherwise everything worked.  This is good.

I'll do the final tests on the RF deck when it's properly hooked into the tank circuit and has shielded input cables, etc.

So, everything has been tested in component form and works FB as far as I could test.  This includes the HV supply, PDM generator, PDM power boards, shutdown board, VFO and driver board, RF deck, smaller power supplies, sequencer, metering, T/R switching and tank circuit.

Now on to mounting, integrating and final troubleshooting the various components to make Rico Suave a real live rig. 

T

Houston....WE HAVE LIFT-OFF!!

[K1JJ]

heheheh... no lift-off yet, Tina.   The major components are still scattered all over the shack needing to be mounted in the rig and wired up together. It may be another two weeks before first light.



Threading the wires thru the cores - output transformer layout:

Here's the PHASING diagram to hook up the output transformers to the fets for a 4 module (24 fet) rig.  The two left modules are fed at 0 degrees and the two right modules are fed at 180 degres out of phase. This phasing arrangement can be used with either a digital gate drive or analog gate drive configuration.

The drives marked "0 degrees" or "180 degrees" go to either the 414 driver chips or gates of the 11N90 MOSFETS.  

In my case, the four primary wires marked "Drains" are #10 wire (one loop for each of four modules) and the one secondary wire marked "tank" is #6 wire, (one wire threaded thru to combine all four modules) -  although these are not critical wire diameters.


This physical phasing layout info is not clearly shown anywhere, so print it out if you plan on using this general rig plan.

T




Attached Below:

1)  Diagram of output phasing

2)  Picture of the actual output transformers: (BTW, in this picture, two of the the output transformers are wired incorrectly. Refer to the diagram for the correct wiring...  )

The meter polarities are also wired backwards in this picture. That's what initial testing is all about.

[KF1Z]

I would have swapped #3 and #2

To help balance.

[K1JJ]

I would have swapped #3 and #2
To help balance.

Bruce,

It's simple enuff to change it.  Please elaborate about the effects of balance.

I did it this way to keep the drive areas segregated on each side of the heatsink - no other reason.

What might happen if I left it the way it is?

T

[WA1GFZ]

The secondary lead is so long I don't think it is worth the trouble.

[KF1Z]

Oh I;m not real sure Tom...

Just all the diagrams I've seen do it that way...

0deg, 180deg, 0deg, 180deg  etc...

your "0 deg" phased modules are 2 transformers away from ground end...

Thats the only real reasoning...

Frank or Steve would be the ones to ask for sure if it makes any difference at all.


I've found that usually if there is any unbalance between phases, you can almost guess that the module who's transformer is closer to ground is the one drawing slightly more current.(or was it the other way around?)

[steve_qix]

It doesn't matter about the phasing.  I do it - 0 deg, 0 deg, 180 deg, 180 deg in my transmitter to keep the leads shorter.

Works just fine 

[K1JJ]

Here's the latest shot of Rico Suave almost wired up and ready to do some initial testing. All the components have been individually tested, but working together is another thang.

All these wires are DC/bypassed, so length is not a problem. I will tie wrap and neaten everything up later, once everything is mounted inside the cabinets.  

Pictured are the VFO/Driver, shutdown board, sequencer, 12V supply,  +- 18V supply, HV supply, PDM generator, PDM power board and RF deck.  (PDM filter not pictured)

I will later be able to easily pull everything out for servicing, just as pictured. I am not bolting down any modules. They will be bonded electrically.

T



There's a lot to building a 24 pill class E PDM rig! Don't let anyone tell you it's easy the first or second time. It takes lots of time and concentration. Despite having the nice PC boards, it's no Heathkit...   But well worth the effort.

[K1JJ]

While we were taking E-rig pictures, Fabio started mumbling about getting some attention. Notice his slight green hue.

Included is Fabios new SS MOSFET audio driver designed by Frank/GFZ.  It's now safely mounted under Plexiglas.

Also see Fabio's 3-diode circuit high level limiter I've been fussing with lately.

T


Pic 1:  Look closely for the ghost reflection of the GFZ driver board.

[KA2QFX]

Is that 10 FETs in the PDM modulator output?! What are you planning on modulating after this is built, a second bank of 24 and combine them?  That's quite a few amps of modulator you got there Rico.

[K1JJ]

Is that 10 FETs in the PDM modulator output?! What are you planning on modulating after this is built, a second bank of 24 and combine them?  That's quite a few amps of modulator you got there Rico.

Yep, Mark, if I can find more room for a second PDM filter and squeeze a second RF deck in there, Rico has the chance of becoming a 48 piller.  It's been in the plans since the beginning. We'll have to see how the rig runs at 24 pills first.  There's already a 40 piller out there and I've heard  of plans for even bigger stuff in the wind, so gotta keep up with the E-Jones, caw mawn...

T

[K1JJ]

The gods must have planned some fun for me tonight. Talk about rotten luck and life not being fair sometimes...

Rico Suave is almost all wired and ready for full RF testing. All the infrastructure and outside whirl connections are ready to fire him up. The modules are still spread out, but easily placed into the rack intact, once tested.

Anyway, I had the power off and was snaking my RF deck cables to the VFO/Driver digital board to connect up. This was one of the last items before firing it up for RF.  Suddenly I heard a big snap and smelled smoke. Yaz got up and wanted to leave the room.  The exposed ends of my RF deck driver input cables had brushed against the 240AC terminal strip on the back of the rig. I had forgotten to unplug the rig.  I was Yallowfied!

At first I figgered it was nothing, but then realized the two cables went directly to the twelve  RF 414 driver chips. I did an inspection and found six of the chips were burnt and broken up with soot on the heatsink. Who knows what else got wasted? The many bypass caps, the twenty four MOSFETs fed by these driver chips?  I don't mind an RF crapout, but to lose them from a space-out like this?

After a couple of hours of tight work testing all the solid state devices I found just one half of the RF deck was affected and the other half was OK.  Fortunately the MOSFETS were still OK, but assorted components with the six driver chips had to be replaced.

It was really rotten luck cuz I had already run the cables and decided to reroute them again to make it prettier.  The only good thing I gained is the confidence that the rig is servicable when these things occur. Usually RF decks can be a real pain to work on.

I fired the system up and found good driver RF at the MOSFET gates. Tmw I might try some HV on the finals and see what happens.


The moral:  Double check to be sure there are no exposed voltage points (including 120/240) when you think everything is shut off.... sigh.   I plan to put a plate around that terminal strip tmw. I already added 18V transorbs to the RF deck input. That may have saved it in the first place.

T

[The Slab Bacon]

<<snip>> "There's a lot to building a 24 pill class E PDM rig! Don't let anyone tell you it's easy the first or second time. It takes lots of time and concentration. Despite having the nice PC boards, it's no Heathkit...   But well worth the effort." <<snip>>


Not to mention the $10,000.00 worth of crimp on wire ends and hook up wire!!   

[KA2QFX]

Tom,
First, really sorry to here about the lab accident.  I know from whence you speak and feel your pain. It especially smarts when you know it was avoidable. UGGHH.  Still happens to me now and then, but I'll refrain from publicizing such events. All success and sunshine here at QFX labs!  Yep,... wanna buy some swamp land with that?


I noticed you are using 32 beads in your output. I went around a while with Steve on this and settled on 64 at a minimum for 24 pills. You might want to check with Steve again since he did those calculations.   If 32 is the right number maybe I'll consider 48 pills! 
Later,
Mark

[W1IA]

The gods must have planned some fun for me tonight. Talk about rotten luck and life not being fair sometimes...

Rico Suave is almost all wired and ready for full RF testing. All the infrastructure and outside whirl connections are ready to fire him up. The modules are still spread out, but easily placed into the rack intact, once tested.

Anyway, I had the power off and was snaking my RF deck cables to the VFO/Driver digital board to connect up. This was one of the last items before firing it up for RF.  Suddenly I heard a big snap and smelled smoke. Yaz got up and wanted to leave the room.  The exposed ends of my RF deck driver input cables had brushed against the 240AC terminal strip on the back of the rig. I had forgotten to unplug the rig.  I was Yallowfied!

At first I figgered it was nothing, but then realized the two cables went directly to the twelve  RF 414 driver chips. I did an inspection and found six of the chips were burnt and broken up with soot on the heatsink. Who knows what else got wasted? The many bypass caps, the twenty four MOSFETs fed by these driver chips?  I don't mind an RF crapout, but to lose them from a space-out like this?

After a couple of hours of tight work testing all the solid state devices I found just one half of the RF deck was affected and the other half was OK.  Fortunately the MOSFETS were still OK, but assorted components with the six driver chips had to be replaced.

It was really rotten luck cuz I had already run the cables and decided to reroute them again to make it prettier.  The only good thing I gained is the confidence that the rig is servicable when these things occur. Usually RF decks can be a real pain to work on.

I fired the system up and found good driver RF at the MOSFET gates. Tmw I might try some HV on the finals and see what happens.


The moral:  Double check to be sure there are no exposed voltage points (including 120/240) when you think everything is shut off.... sigh.   I plan to put a plate around that terminal strip tmw. I already added 18V transorbs to the RF deck input. That may have saved it in the first place.

T

What a drag Tom I know from mistakes. Try not to get too discourage and move on. I have had to learn patience with the current project. I tend to be compulsive and want my project finished; but the slow pace is due to the detail work. I get frustrated cause I want to get MM on the air. Mark has the patience of a saint when it comes to the detail work. I finished the wiring on the triode and driver board yesterday. Everything in the pressurized tube assembly is finished. I'm waiting on meters and a few part to finish the tank, band switch etc. The biggest part of the project is milling the supports for the vac caps for C1 and C2; and the gear and chain assembly for the sub band switch (driver band switch). These will be made out of 3/8 aluminum plate. We have critical spacing due to the fact the parts are so large and have to maintain 1 inch spacing to keep B+ from jumping to ground etc. What seemed like a large chassis quickly becomes crowded once you stuff it with part.
Let me know when Rico comes on the air so the farmers in CT can wrap foil arund there sheep.

[WA1GFZ]

you  want to shoot for 150Gauss per bead for low loss.
I used 16 type A material (ui=600) in my push pull 22 fet final.
They might get luke warm after it sits at 48V/1200 watts for 5 minutes on 75m so 8 per stack sounds good for me.
My 160 meter rig running at 80 volts I have 10 beads type a material for a push pull final. I've never felt them warm up running 300 watts carrier and it will do almost 200% positive peak. This rig has been on the air since 1996.
16 beads per stack would support a high voltage. I think it is a bit much.

[W1AEX]

The exposed ends of my RF deck driver input cables had brushed against the 240AC terminal strip on the back of the rig. I had forgotten to unplug the rig. 

Ouch! Glad you and Yaz were uninjured. Don't want to see you on the local action news!!!!

[WD8BIL]

I just love "wireless" communications!!

[K1JJ]

Thanks for the get-well greeting cards, guys...

Thirty-two cores appears to be enuff, Mark. That's what I ran with the other 20 pill rig six years ago - no heat after an OB xmission.


Yep, this stuff can be real fragile to mistakes. It will run forever once running, but cannot take random shorts or external voltages inserted into it's inner workings...  The tube stuff will usually survive.

I try to make everything servicable. This means running longer harness wires that will permit the boards to be flipped over easily when in-circuit.  I also cut square 16" X 16" holes (with access doors) on each side of the cabinet for inner access. I can actually replace FETS on the RF deck without pulling the rig apart.  I'm tired of building with things cramped and impossible to get at. Some day we old buzzards will not like the twisted positions needed to get at this stuff.

Might make some more progress over the next few days if all goes well.

T

[K1JJ]

Hopefully there are still guys following this old thread and thinking of building up a big class E rig. We're getting near to the end of this project with testing to follow. Hope you all have enjoyed riding along.

Well, had a few more problems today.  One of the shielded cables shorted that goes to the input of the PDM modulator. I had heated the inner conductor too much and it broke thru the insulation.  This caused my PDM 414 driver to short out. I went through two drivers before finding the problem.

Got the whole system to key up smoothly with the sequencer. There are about eight relays that kick in. The big 12V coil primary relay arrived today and worked beautifully.  All levels from the VFO on up are within specs.

We're now at the point where there is gate drive to both the RF deck and PDM modulator. All systems appear to be working except for the high-current shutdown part of the shutdown board - I'm awaiting a new chip from Steve for that.

So, tomorrow I hope to put the HV on and see what I have. It's been a slow process with lots of minor issues along the way to get to this point.

Here's some shots of the rig cobbled together ready for a test. The cables to each module are intentionally long so I can pull each module completely out of the cabinet to work on. It still needs covers and dressing up. I will test and perfect everything first before tie wrapping and making the wires look good. I expect to be pulling modules in and out for a long time yet during testing and troubleshooting.


Notice the side cabinet access for the RF deck. This will have a door later on.  Also notice the two layers on top - the control circuitry on top and the PDM modulator below.  The RF deck sits in the bottom of the cabinet. There is room to add another 24 RF pills next to the first RF deck.  The bottom of the rack holds only the RF deck and tank circuit, so there is room to play in there.

T

[K1JJ]

Here's one of the cooler shots. Notice the schematic and glasses in the background. Just like a how-to magazine...

[steve_qix]

Hmmmmmmm.. I just noticed a couple of things that you may want to consider fixing before attempting normal operations.

There appears to be a "driver bus" rather than individual coax cables running to each IXDD414.  This did not work in my own transmitter.  It works OK with 4 MOSFETs and 2 IXDD414s per side, but when I went to 6 MOSFETs per module with 3 IXDD414s, there were parasitics caused by the long, unshielded driver input bus.

I corrected the problem by running an individual coax cable to each driver, with termination (300 ohms in this case - 6 devices per phase).  I would recommend fixing this.  Also, do ground the shields at the driver coax cable common point - where they all come together.  It should resemble this:


Second - hey, what's all that extra wire connecting the output transformers???    This will surely introduce inductance that's not supposed to be there.  The primary and secondary wires should all be exactly the same length for each transformer, and be as short as possible.  Similar to this:


The transformers started out like this:


The shunt capacitors are way over at one end of the drain bus.  They should be in the middle, or use 2 capacitors of 1/2 the total value and put one at each end.  For 160 meters, you can add additional shunt capacitance at the end, as long as this is not the total capacitance of the shunt capacitor.

The entire transmitter, with all the construction steps is shown here:
http://www.classeradio.com/24_fet.htm