A NEW KW TUBE Transmitter AM PDM BUILD - the beginning...

[steve_qix]

Interesting !  If you do go the PWM route, analog compensation is absolutely necessary in order for the modulator to operate in a linear fashion.

Graphically, what happens is this:



With small pulses (such as over about 85% negative modulation), the rise time of the waveform lengthens, and gets worse and worse as the pulse gets smaller (closer to 100% negative modulation).  This is due to energy stored in the filter, which is not fully depleted at low current.  The result is that the modulating waveform flattens out, this flattening increasing as 100% negative modulation is attempted, and then the modulator output suddenly goes to 0 (100% negative modulation).

At the other end (150% positive modulation), the modulator tube, being a tube, has a somewhat nonlinear transfer characteristic, particularly at higher currents.  The result of this is a non-linear modulation characteristic at high levels of positive modulation.

So, how much of a perfectionist do you want to be ?  You can certainly get away with no analog compensation, but there will be distortion.  This is amateur radio.  Most people probably won't be able to identify what they are hearing, and it will sound "fine", so to speak.

An interesting example of this is Rob W1AEX.  He always sounds really good.  However, one day I noticed a distinct improvement in his audio.  It was striking, and Rob always sounded "fine" (good, actually).  Now he sounded Excellent !  He got a new SDR, and it was better than his previous SDR.  But, without the comparison, I didn't think anything of it.  Rob sounded good.  Now he sounds even better !

I would go the perfectionist route    I figure if something is going to take that much time, and cost money, etc., might as well go all the way and make things as perfect as possible.

That said, with cathode switching of the modulator tube, the idea would be to construct the modulator driver such that you vary how hard the switching MOSFET turns on, depending on the modulating signal.  Set the cathode MOSFET such that, at carrier, the cathode MOSFET is only turning on about 30% or 40%.  Set the bias of the modulator so the tube turns on hard at this level (getting close to full grid dissipation).   Make sure you have enough swing at the cathode such that the modulator turns off completely when the cathode MOSFET is off.

Conceptually, this sounds easy.  In practice, with cathode switching, it's an interesting problem.  It is easier with grid drive of the modulator, because you just put a source follow in series with the driver MOSFET to vary the voltage fed to the driver MOSFET under modulation.  I'll think about a similar solution for cathode switching.

[K1JJ]

OK Steve -

Yep, it's tough to sleep knowing a rig can be improved further...

I have no problem going with grid drive for the PDM modulator since we are in the design phase.   It sounds easier to control and tweak.

My 6LF6 PDM modulators (cathode driven, no analog comp) required heavy grid and screen currents to best modulate the 4D32s. It was to the point of exceeding the data sheet numbers to find the sweet spots.

If you can a draw a grid drive circuit with analog drive using an 11N90, I will try it out and test it.

BTW, do you have a preference for choosing the lineup of 3-500Zs and 4-400As for PDM or RF finals?  It seems that tetrodes are used a lot as PDM modulators. I see you used a single 4X1 tetrode as a modulator and 450TL triodes as RF finals, thus my thinking this way.

Thanks.

T

[DMOD]

Well, it's time to end the old thread about building a pissweaker 25w rig.  The end result was a pair of 813s series modulating a single 813. Works great and all finished.
http://amfone.net/Amforum/index.php?topic=46165.0

Time to move on and build something else...


After thinking and talking it over with a few guys, I decided to build a big AM PDM TUBE rig. It is a challenge that scares me a little, so this is good.  I was toying with the idea of doing something like Rick/KHK's big series modulated rig, but I already have one now, though small, but wanted to swing the other way for a high efficiency rig.

Haven't decided on the tube type yet, but it will be at least a pair of 3-500Zs PDM'd by a pair of 4-400As.

Check out the first steps in the pics below.  Shown is the PDM generator, an old 2006 QIX board I have used for both MOSFET and PDM tube rigs in the past. It didn't work when I first fired it up, but after a few hours I have it putting out a nice stable squarewave with good modulation from 1 Hz to about 10 KHz where the artifact LP R/C  filter rolls it off. It also has a very nice low level negative peak limiter (actually clipper) which I feel is one the the best implementations out there. It limits those negative peaks well as an insurance policy. Just a touch at -95% is a good setting.. I use a 10-turn pot to set it.

I added a pot so I can adjust the switching freq from 30 KHz to over 300 KHz.  But 70 KHz is a popular switching freq used by the tube BC PDM pros, so that's where I set it as shown on the freq counter.  

Also, an important part of the rig is the big integration PDM output filter using L-C-L-C-L-C.   The inductors pictured are homemade; wound on 5 gallon plastic cans - and have worked with my old 4-1000A rig PDM'd by a 4X1. They are about 34 mH and the caps are around .001 @ 10 KV.  The filter values will have to be tweaked based on the final plate impedance.

Just cleaned up a very strapping 35" high X  19" wide X 31" deep cabinet and sprayed it Baby Blue. It will have a side window viewing and easy ACCESS of about 2' X 2' using Plexiglas OR sheet aluminum, for testing or permanent use.  It will have three levels inside, one for the PDM deck and one for the PDM filter and PS, fil xfmrs, etc. and the RF deck on top.  At least that's the plan for now.  For safety and stability, I am working hard to fit EVERYTHING into this one cabinet.  This and the tube choice is my next design challenge.


The finals will be in regular class C, maybe 80% efficiency at best.  The PDM tube modulator will probably hit 88% efficiency or even better, based on experience.  I have a pre-driver chip in the PDM generator putting out 10 V p-p with good current. A 11N90 MOSFET or two in the cathode of the PDM modulator tubes will drive it well.  No need for the GFZ driver board this time and no audio transformers.

Have built a few of these rigs in the past using quad 4D32s X 6LF6s, 4-1000As and other big tubes. It is an advanced project and can be dangerous due to the floating supplies, meters, etc.  But the results are really amazing when you key it up and wonder where all the heat has gone...  It will generate audio "shark fins"  on the scope just like the big MOSFET class E rigs. I would compare it very closely to my recently built series modulator rig's performance...  FB triangles and square waves.

In no rush, so it will be several months before completion I would guess.

A journey starts with the first step -  and here we go.

T

Steve did a nice design job on this PDM kit board.  (The 10 turn pot on the front is for precise negative peak limiting adjustments on the fly.)

Hi Tom,

Here is potential, "conceptual" switching modulator circuit using parallel 572B's. I have not simulated the circuit in-total as this is just an idea for consideration.

Phil - AC0OB

[K1JJ]

Very good, Phil!   I appreciate the  circuit since there is nothing else I have to work with at this point.   It is a start.  Is this an analog compensation circuit? ( I don't think so)

I'm curious about the 75V p-p that gets dropped to 15 V p-p thru the 9:1 xfmr...    

I may decide to build the PDM rig with an empty grid drive module -  and by the time I'm finished there will be a working analog compensation circuit available. I am convinced about the need to linearize the modulator this time around.


T

[DMOD]

Very good, Phil!   I appreciate the  circuit since there is nothing else I have to work with at this point.   It is a start.  Is this an analog compensation circuit? ( I don't think so)

I'm curious about the 75V p-p that gets dropped to 15 V p-p thru the 9:1 xfmr...    

I may decide to build the PDM rig with an empty grid drive module -  and by the time I'm finished there will be a working analog compensation circuit available. I am convinced about the need to linearize the modulator this time around.

T

I had an error on the schematic for the pulse transformer as it is a 3:1 turns ratio, not what was previously shown. (9:1 impedance ratio, Primary-to-Secondary).

Gate Drive Transformers provide electrical pulses and are primarily used as pulse transformers to drive the gates in switch-mode power systems. These pulses control timing in the circuit by turning semiconductors such as MOSFETs and IGBTs on and off.

Gate Drives, which are often toroidal, are also used for impedance matching and voltage isolation.

Here is what they look like, about 1.25" X 1.25"
https://www.schaffner.com/product-storage/datasheets/it-series-single/

The goal was to provide voltage isolation while switching on and off the MOSFET.

To reduce any distortion we need to find a switching transistor with the highest speed switching specs, lowest inductance for the filters, and the lowest filter capacitance to allow a fast a rise time as possible.

Someone here at one time had some software for designing these filter circuits for ClassE and as I recall it had an optimization subroutine for speeding up the rise times of the waveforms.


Phil

[DMOD]

Very good, Phil!   I appreciate the  circuit since there is nothing else I have to work with at this point.   It is a start.  Is this an analog compensation circuit? ( I don't think so)

I'm curious about the 75V p-p that gets dropped to 15 V p-p thru the 9:1 xfmr...    

I may decide to build the PDM rig with an empty grid drive module -  and by the time I'm finished there will be a working analog compensation circuit available. I am convinced about the need to linearize the modulator this time around.


T

Here is another idea since we require a 50V supply so why not use it to drive the pulse transformer?

Your PDM generator (Vs = 15V) would feed a 15V Gate driver which pulls up M2's gate to saturate M2. M2's Source pulls up to +45V generating a 100 mA current pulse through the pulse transformer which generates a 15V pulse train on its secondary. The secondary's 15V positive pulses in turn saturates M1 to pull the grids of the 572B's up from -200V up to about 45V to saturate the 572B's.

Since the 572B's are in parallel, the total plate resistance is 1/2 that of a single tube, meaning that the cathode of the 3-500Z will be brought closer to ground than if only one tube were switching it.

K6IC reminded me of Nigel's post regarding filter designs here on AMPhone:

http://amfone.net/Amforum/index.php?topic=38161.0

There may be others as well.

Phil - AC0OB

[Opcom]

Kinda late to the blower vibration/telegraphing, but this mount was made from materials and worked very well for me. I hear the air right now but there is no vibration transmitted to the rack It's a dual 6FT rack 35" deep with door-type sides. It seems to have eliminated the mechanically transmitted 60Hz type hum and I can't feel or hear any vibration. It's a 1/2 HP 3450RPM dayton ball bearing motor . The foam rubber is kinda squishy and jiggly. Closed cell, not the kind of foam that eventually rots and leaves a sticky residue -that would be too weak to support the blower. Maybe this is too complicated but I never made a mount before. Don't know what to call the foam rubber exactly but its compliance is likely responsible for isolating well.

[w8khk]

Patrick, thank you for sharing the details of your blower mount.  Great that you got rid of the annoying hum and vibration.

I was thinking of doing something similar, and I have decided to make the blower housing and plenum from fireproof fiberglass, enclosed in a metal chassis, shielding the entire modulator tube assembly.

Right now I have the lower power series modulator spread out on the bench, pine-board style, with 810s and 8000s, looking at plate curves and distortion measurements.

[steve_qix]

I have found through experience that using a transformer to drive the gate(s) of the driver MOSFET (high side drive) only works if it is GUARANTEED that the duty cycle will not get too close to 100%.  Once 100% duty cycle is reached, nothing passes through the transformer.

I use transformer coupling for some PWM circuits, but only those where the duty cycle is carefully controlled to never exceed 90% or thereabouts.

The proposed PWM generator does not do this.

However, there are other ways to drive a source follower (high side drive) that are DC coupled, and a 100% duty cycle signal (full on) can be passed safely.

I am not criticizing the circuit - just pointing out something that not everyone may have noticed.

This circuit might be a starting point. 



The concept is sound ( I have built this exact circuit it is works well ).  Analog compensation is included, and the high side driver is DC coupled.  Notice the transformer is used to facilitate an "active pullup" for the gate of the output MOSFET.  Since a resistor is used as a pull-up, it is necessary to "help things along" to ensure a fast rise time at the gate of the output MOSFET.

It is not necessary to build it exactly.  For instance, the MOSFETs can be FQA11N90s, and other op amps can be used.

[K1JJ]

Tnx for the info, Steve.   There is a lot there compared to a single 11N90 in the cathode.  I will study it some more.


Different subject:  AIR BLOWERS - Check out this one.  Would it push enuff air thru a 3CX-3000A7?   The noise level is only 32 DBA  which is the sound level of a quiet room. It also has a muffler option.  It has a 6" inlet and outlet.  It will put out 402 CFM which is 3 times the needed air for a 100 CFM 3cx-3000A7 tube.  It says it can handle restricted airflow environments, but how about tube fins? I dunno.  8 speeds controlled with PWM.

I need to reduce the air noise in my linear or it cannot be used on AM.  SSB is OK cuz it can be masked thru a noise gate, etc.

Any thoughts on this for $109?  The reviews are great, but they are not using it to cool a tube, rather for ventilation.

https://www.acinfinity.com/hvac-home-ventilation/inline-duct-fan-systems/cloudline-s6-quiet-inline-duct-fan-system-with-speed-controller-6-inch/?gclid=EAIaIQobChMIjbXZ4pTo7QIVxcqGCh1EqwB3EAQYASABEgJ30fD_BwE#product-description


Here is a 4" version for $89.     Airflow: 205 CFM | Noise: 28 dBA | Bearings: Dual Ball:

https://www.acinfinity.com/hvac-home-ventilation/inline-duct-fan-systems/cloudline-s4-quiet-inline-duct-fan-system-with-speed-controller-4-inch/

Any thoughts?

T

[W1ITT]

Tom...  In regards to that air mover, I suspect that "restricted airflow" means different things to an HVAC guy than it would to someone who is  pushing high velocity air cooling a finned tube.   Fluid dynamics change with velocity.  The Dayton line of blowers all have impellers that are designed to shovel air into a back pressure situation.  And of course they shovel noise to a certain extent.  For noise abatement, I like the Dayton slow rolling blowers...  1500 rpm or so...such as their 2C647D.  The noise is less and of a better spectral mix, not as tiring, and it carries the sound of power and confidence needed in so many on-air situations.
I'd also be wary of the PWM motor controller in that duct mover.  It may well require filtering so as not to trash your receive environment.
Perhaps coolies down cellar, pulling a rope through a hole in the floor to activate a bellows to move air through the transmitter would be sufficient.
73 de Norm  W1ITT

[K1JJ]

Thanks for the dual info, Norm.  

Yes, the PWM controlled blower cud make noise and the blower will probably sag with a restrictive load.

I use a blower now (similar to the Dayton 2C647D with new bearings) that uses a Variac and can run very slowly, but still, on AM it is audibly offensive -  though OK on SSB.

The best air system I've ever used on this rig is a blower located on the outside of the house, pushing in air. It is amazingly quiet that way. I shud do it again. The big amp really needs to be serving in AM service in addition to SSB.


T

[Opcom]

long-winded on blowers:


I looked at that consumer duct type one, and nothing makes me think it will cool a 3CX3000.
https://www.acinfinity.com/hvac-home-ventilation/inline-duct-fan-systems/cloudline-s6-quiet-inline-duct-fan-system-with-speed-controller-6-inch/?gclid=EAIaIQobChMIjbXZ4pTo7QIVxcqGCh1EqwB3EAQYASABEgJ30fD_BwE#product-description

Marketing babble instead of a pressure-volume graph:
"mixed flow design using a stator blade and dual hydrodynamic wind circles"

gimme more of them hydrodynamic wind circles.

That kind of fan (according to it's internals picture), a type of vane-axial fan, is designed to be quiet and efficient when moving high volume air with low back pressure like <1". At some point, when the pressure rises, the 205 CFM dwindles to very little, and long before the pressure drop through the duct and 3CX3000 is surmounted.

The larger version of 405CFM is probably good for exhausting the whole rack while a high speed forward-curved-blade blower cools the 3CX3000 inside. Low speed backward-curved-blade centrifugal fans also work well for high volume low pressure and are also quiet.

EBM Papst was mentioned earlier and I downloaded their catalogs. Lots of learning in there. Then went looking at different fan designs.
(note these are all 'fans' but centrifugal ones are popularly called blowers.
https://www.rs-online.com/designspark/fan-types-why-choose-a-forward-curved-centrifugal-fan
https://www.rs-online.com/designspark/fan-types-why-choose-a-backward-curved-centrufugal-fan
https://www.rs-online.com/designspark/fan-types-why-choose-an-axial-fan
don't confuse tubeaxial with vaneaxial, they perform differently. They are not like centrifugal fans though.
https://www.tcf.com/wp-content/uploads/2018/06/Fan-Performance-Characteristics-of-Axial-Fans-FE-2300.pdf
and
http://www.dynamic-air.com/fan-selection-guide.pdf


A great fan for the 3CX3000 is a centrifugal fan (blower) with forward curved blades and RPM about 3450 which is suitable for a reasonable size wheel. That will make the most air flow through the radiator at the needed pressure with the least power and noise.

The forward-curved blade, high RPM, and wheel diameter is what makes the pressure. The wheel width and horsepower makes the volume.

What this whole noise situation is up against is that a blower with a high enough speed and large enough wheel (and fraction of horsepower) to make the pressure at the desired volume of 3cx3000 is going to be noisy. Noise reduction techniques must therefore be used to abate unwanted noise.

Also keep in mind the blower pressure must overcome the resistance of any ducting because the only thing that matters is CFM through the tube.

The table below from Eimac is simply a measurement across the tube, probably in the standard socket.

Therefore a manometer in the tube socket plenum can be used with the chart below to tell if the blower is doing enough. I have a bunch of nice 0-5" Magnehelic gauges pulled from STMicro when they started ripping the guts out of the building. These are wonderful and a lot less messy than a water manometer.
I just used one last week to discover the negative inlet pressure and positive outlet pressure across a large unknown blower used with a kitchen fume hood in my house. LOL Now I know what $450 Dayton blower to order and get it right the first time.

Eimac: The 3CX3000 @4KW dissipation:
sea level: 127CFM @ 1.40" H2O base to anode flow
10,000 FT: 185CFM @ 2.55" H2O base to anode flow
------------------------------------------------
sea level: 178CFM @ 2.50" H2O anode to base flow
10,000 FT: 260CFM @ 4.50" H2O anode to base flow

Whew! I'm winded after all of that hot air!

[W1ITT]

For efficient air movement, I believe that the method using six hydrocoptal marzal vanes works well, assuming you can get them a a decent price.
de Norm W1ITT

[w8khk]

For efficient air movement, I believe that the method using six hydrocoptal marzal vanes works well, assuming you can get them a a decent price.
de Norm W1ITT

You could part-out a retro-encabulator to get a half dozen, but I have not seen them on the bay for quite some time.  Shipping might be a bit expensive.  I wonder if any members here have attempted to reverse engineer the device?

Specifications here:  https://youtu.be/RXJKdh1KZ0w

[W3SLK]

Rick said:

You could part-out a turbo-encabulator to get a couple, but I have not seen them on the bay for quite some time.

Those are the swerving hydrocoptal marzal vanes that I use with my retro encabulator! I keep them right next to my interossitor!

[w8khk]

Rick said:

You could part-out a turbo-encabulator to get a couple, but I have not seen them on the bay for quite some time.

Those are the swerving hydrocoptal marzal vanes that I use with my retro encabulator! I keep them right next to my interossitor!

But I accidently discovered that if you cross-couple a flux capacitor to a static inductor with alternate reversed phase polarities, the resulting superior conductor renders the marzal vanes totally unnecessary. Flux capacitors are plentiful, but just try and find a static inductor!

[Opcom]

Well that's good because you'd need a deplenerator to operate marzal vanes in the manner proposed, and they are hard to find.

[K1JJ]

Tnx for the elaborate blower info, Pat.  

I decided to take two approaches for the big linear amp. First I ordered a Dayton 2C647D blower for $63.  It's a 1500 RPM @ 120VAC motor on a Variac. (Haven't received it yet)

Norm, your first post convinced me:
"For efficient air movement, I believe that the method using six hydrocoptal marzal vanes works well, assuming you can get them a a decent price."

Second, I mounted the existing blower on the outside house wall in the cold sitting on a steel frame. The 4" smooth fiberglass duct is about 36" long -  from the blower outlet to the amplifier tube, so not bad.  The dB noise level in the shack has dropped about -15 to -20DB. It's pretty quiet now - except for the small whine of the old blower bearings.  The new Dayton shud fix that with all white noise I hope.


This linear amp is the only one in the shack that has an external anode -  so needs brisk  air flow. All the other rigs in the shack are glass finals with small internal blowers or muffin fans and I can usually keep them Variac'd down and quiet due to the heat radiation and convection cooling.

With the cold air, I can reduce the speed down to 1/2 of normal and still keep the tube cool, so helps noise even further. I use a digital temp sensor above the final tube with a warning alert. During the late fall to spring the rig will run cool and at low noise. I don't run it during the summertime anyway.

I plan to get a 4" shutoff valve to keep the cold and moisture out when not in use.


* Still designing up the PDM rig. Hard decisions.  I got a generous offer (an offer I can't refuse) for some old BC transmitter PWM coils which will help immensely to shorten the usual days spent winding coils. I'll talk more about this later.  I've been taking the time to integrate the shack and catch up on the little modifications needed after building a few rigs. Once done I can focus more on the PDM rig build.

T

[K1JJ]

I received three really BEAUTIFUL FB commercial PDM coils from Pat / KD5OEI / Opcom.   He sent them to me for no charge. Much appreciated Pat!

There is a lot of stuff that gets given free to others on this BB.

I've always wanted a set of BC PDM coils to build around. These are small and compact and out of a Gates 5KW PDM transmitter.  The ones I've wound in the past were huge and I've had one arc over. These coils are the size of a beer can. Compare them to my older 5 gallon pail coils!

Looking at these coils, the first coil had a flashover in the past and I had to remove one of the pi-sections. It measures about 16 mH total now and is the input coil without ferrite.   The second coil with a ferrite adjustable core measures 26 mH and the third coil with a ferrite core measures 43 uH.  They are all the same size and same number of turns, so those ferrite cores really make it a widely adjustable filter. Once I figger out the tubes to be used and plate impedance, these coils can be dialed in with the right capacitors.  Usually it's a L-C-L-C-L-C  lowpass filter at 70 KHz with a rolloff of about 10 KHz.. It will have to be modeled out based on the tubes and the value of filter caps I have in the junk box.

Interesting that the two ferrite core coils have a 100 pF cap across the coils.

I coated some of the damaged turns with nail polish. That arc over affected the adjacent coils somewhat. Shud be OK. Otherwise I could always add some more external L if needed.  This shud make a FB PDM rig if I can get by the analog compensation obstacle.

This is like getting modulation iron for an analog rig, but it's "digital" instead.   A good bath and they will look vely pletty.

T

[K1JJ]

More:

[K9EID]

Hi Tom
Yes, let's get out the soldering irons and why not breadboard?  When I did the Pine Board Project back in 2019, never realized how many folks - girls and boys- would 'dig in'.  Hundreds followed after it hit the cover of QST .  The beautiful diagrams of W4IQN (retired GM graphic artist) and having Antique Electronics add the parts list in their catalog.    It ended up using the full coil  on 160, tapped for 75 and 40.  Bandswitched those  along with the three crystals.  Even added a receiver RF out and mute. 

I designed the mic preamp with two band equalization and the audio was always reported excellent.  The final piece was an 807 final to get a good 40 watts out.  All of the many great diagrams are on our site. www.heilsound/pineboard. about 50 pages, diagrams and pictures.

Fun project and will look forward to your new project, Tom.  Always enjoy your work.

Dr. Bob Heil, K9EID

[Opcom]

www.heilsound.com/pineboard

[K1JJ]

Hi Bob,

Cool on the pine board projects.

Next, you might build a tube VFO for your xtal transmitter.  

T

[K9EID]

Thanks for yoru note... yep- let's do that!,,,... fun project I do believe.   Presently I have Blessed to have Wayne WA4FTY 'borrow' my Haegerty digital boards for a month or so and when I got the boards back - this is what I received.   What a builder !  Check his WRZ page where he built the only Harvey Wells receiver and matching 4-400 final. 

Stay in touch.  Love to see your 40 watt tube transmitter.

73
Dr. Bob Heil, K9EID