The AM Forum

Boy, have I been busy! It's been a while but progress does continue with the 8-FET transmitter. I decided to make a series of YouTube videos as I go along. It's not really a tutorial - Steve's website has all the detailed instructions, which I do cite. But, there are literally no YouTube videos out there to sort of chronicle building one of these beasts, so here's my stab.

Video 1: Intro to the project and talk about the phase splitter / duty cycle adjust board

https://youtu.be/j4F0KA-qlE8 (https://youtu.be/j4F0KA-qlE8)

Video 2: Demonstration of the phase splitter and talk about next video for crystal oscillator as signal source

https://www.youtube.com/watch?v=SpoEd438Xsg (https://www.youtube.com/watch?v=SpoEd438Xsg)


Jon

Nice work, Jon!  Looking forward to the upcoming videos.

73, Tony

Great idea Jon, to chronicle the build with videos, I enjoyed the videos, looking forward to following along with your build. Good job. Keep up the good work.
Donnie/AG5UM

Thanks Jon, nicely done. Your camera person is doing a great job also.

Thanks all! There aren't many builders doing class E stuff in America. I'm guessing 20 people. I enjoy this sort of thing and may be the only current project out there. Sit back and I'll try to keep you entertained.

Kuddos to my wife with camera duties.

Jon

I am actually almost finished with the oscillator for the 3rd video. It's shown in two pictures attached to the first module. Come to find out, 7740 KHz crystals are common enough on eBay and are twice the frequency for 3870 KHz. I'm waiting on the remaining parts to arrive to finish.

The first picture shows the stand I built for the transmitter itself. I welded up a mount beneath the transmitter that'll hold 4 meters (IXDD bus voltage, drain voltage, Iphase1 and Iphase2). The cabinet on the left will hold the power supply and analog modulator. On top is my antenna tuner with all components mounted at 90 degrees from each other.

Jon

I finished the oscillator plug-in module and it works correctly with the phase splitter. The first photo shows one of the output square waves along with the clock input. The second picture shows both output square waves.

I'll post video when my wife has a chance to record it.

Jon

Here's the third video in the series:

https://youtu.be/iWRlX0hVlrw (https://youtu.be/iWRlX0hVlrw)

Jon

Here's where things stand now and the 4th video is coming. I've changed the design a little bit on the modulator shelf to also incorporate a negative peak alarm, a negative peak keep-alive circuit, and a pre-modulator overcurrent trip. I was trying to come up with ways to trip on overcurrent but they all weren't very practical. So, I ended up borrowing from Steve's a little bit, but simplifying it for my application. Sorry Steve, but you WAY over complicate things. Aside from the RF deck stuff. That's pretty much how it has to be.

Some parts from DigiKey for the overcurrent trip are on backorder and I want to finish that before I do the video.

Until then, keep busy, partake in some indulgences, and keep the BS below a 7.

Tootles!

Jon

Getting ready to assemble the analog modulator. Here's the layout. My finger is pointing to the simplified overcurrent trip circuit. The power supply is all done.

I'm going to try using epitaxial diodes in the negative peak limiter. They'll be stacked for easy mounting as shown with a TO-220 style resistor on top.

I'll shoot the video after my wife gets back from Vegas this week.

Jon

Here's video #4: Power supply and analog modulator with overcurrent protection

https://youtu.be/MRetD5NT8no (https://youtu.be/MRetD5NT8no)

Jon

For those with attention deficit disorder, this video may be to your liking:

https://www.youtube.com/watch?v=tfZLS6WqKMU (https://www.youtube.com/watch?v=tfZLS6WqKMU)

Jon

I nearly have the RF deck completed, and as usual the plans have changed. I'm now going to use 6 FETs in push-pull for 250 to 300 watts carrier. Last iteration I was going to use 4-FETs only and built the RF deck transformer holder to accommodate only 8 cores. With more power I need more cores (to be on the safe side with no heating), so I had to build the transformer upward rather than outward. Here's the result:

I made it into a 4x4 brick with still only using two cores stacked. I went all fancy and put 1/4" copper tubing in the centers of the cores for the FET windings. I think the brick adds a little style to the look rather than using 16 cores horizontally in a 4x4 arrangement.

Jon

Here's everything so far. Modulator and power supply mounted in the cabinet with tuner on top. Raspberry Pi and touchscreen mounted on the side of the cabinet to be the efficiency calculator (programming half done). I learned (learning) to code Python.

Scope to keep track of drain waveforms and I'm also going to mount a couple blue LED strips 1/2" wide behind blue stained glass pieces on the RF deck to add some bling. Above the analog meters is a 20 segment LED bar graph tied to power output. Bling, bling.

Jon

All looks like its going to work Jon  :)


Wayne

Most importantly the blue LED strips. We just finished a Monet house party (Avon for the 2020's) with a houseful of young women. They're not interested in tank coils and gate capacitance. They like the bright shiny things.

My job for these parties is to be the caged bear in my shack. My wife parades the women up and says, 'This is my husbands office!' They go, ooh and ahh, and I make sure they have something to look at.

I don't verbalize it, but in my head I'm also saying ooh and ahh.  The blind leading the blind.

Jon

Viva Las Vegas!

The LED power output bar graph below will modulate with blue LED's as well.  Now to actually finish the RF part of it (ha). Everything is mounted, I just need to do all the parts on the heat sink and run the copper on the deck.

Jon

The heat sink assembly is complete. This rig is getting close to completion and I'll post a video of the testing.

After all that I still have to finish writing the code for the Raspberry Pi efficiency calculator. That'll come in handy for tweaking the last few ounces of efficiency out of this. I'm hoping for the best as I made this tighter than any previous transmitter I built.

Jon

** and yes, 680pf shunts, close enough.

Ok, I give, looking for advice. Steve, wake up!  :)

So, I'm testing this 6-FET rig and all is well except when I apply power to the drains (a low 13.8 volts in this case, low for initial testing) I hear sizzling bacon. There is RF output but the waveforms are crazy and I don't leave it on long due to the cooking sound.

Here's my setup, all home-brew just for a quick snapshot:


Crystal oscillator ---> Steve's phase splitter duty cycle adjust schematic with IXDD614's on output ---> Short 12 SMA coax going to RF deck ---> Traditional E RF deck


At the gates without drain voltage applied, I am getting the proper 40%-ish duty cycle, 180 degrees out of phase, all that jazz. I apply drain voltage and it sounds like parasitics, but here's what I noticed: With RF coming from the phase splitter/duty cycle adjust module I am getting nearly 12 volts on the IXDD supply rail on the RF deck. When I turn off the crystal oscillator, the voltage goes away.

So, RF from the phase splitter module is getting coupled onto the IXDD supply rail. I've isolated everything and the only possible path is through the 6 IXDD614's on the heat sink assembly itself.

I took a resistance check for each phase of the IXDD's input to the 12 volt supply rail and I'm getting about 7k in one direction and infinite in the other direction. Apparently 7k is enough resistance to couple RF energy in one direction to cause a problem.

I've never had this problem with other E decks and I don't have a spare IXDD 614 to experiment with. Any suggestions as to what would cause this?

Jon

** Ok, my bad. I posted incorrect information. The problem of the induced voltage is still there without applied RF and with the oscillator off. I simply apply power to the phase splitter/duty cycle adjust board and the problem shows up.

I'm going to drive the RF deck with a completely isolated DDS VFO and see if the parasitic problem is there or not. It may be and I'll need to figure that out, but at least I'll be able to rule out this weird applied DC issue. One thing I did not do is put a 'grid snubbing resistor' inline with the to the IXDD inputs. We've talked about this extensively years ago but I didn't see the benefit and thought it was just a joke. However the way I've built this one, I could easily just add one to each phase without a problem to give it a try.

And actually, I used a morphed version of Steve's VFO. The duty cycle adjust portion is verbatim but I'm driving it with a JK flip-flop that is always enabled. I didn't incorporate a switched enable like his but rather switch DC on and off completely to the whole module during TX/RX transitions. Also, since the JK is always enabled, one of the Q outputs will always be 'high.' I'm not sure how this would all work out in the end with the logic on the duty cycle adjust portion. It probably wouldn't matter but to think through all that logic will make my brain blow up.

Not to mention I'm surrounded by mosquitoes.  ::)

Anyway, I've got it. I'll putter some more and get it.

Jon

So, yeah... after thinking about the problem it is entirely a parasitic.

I've attached a photo showing the how I currently have the IXDD drive from a common bus fed from an SMA connector and how I will modify it. Steve's method of parasitic suppression allows the full drive voltage to be applied to the input of the IXDD. My 150 ohm resistors are too buried to dig out so I'll need to leave them in place. I'll get rid of the common bus and feed each input via its own 150 ohm resistor. To bring total impedance down to 50 ohms I'll need to add a 100 ohm resistor at the SMA connector.

This method will cut the applied drive voltage in half as it is seen at the input, but with about 14 volts peak it won't be a problem. In fact, keeping the original 150 ohm resistors in place and bogging down the system even more may help to bury the parasitic.

The other photo shows you how I have it currently. You can barely see the SMA connector and it has a small gauge wire coming from it and is attached to the common drive bus wire. That thin wire was probably the recipe for disaster.

I'll let you know how it works out.

Jon

Hi Jon....

Ok on the parasitics and stuff.....
Have never had any parasitics before that I know of anyway .... maybe I'm just lucky  :)

Your output stage is made really well and looks great...
Couldn't see any RF bypass capacitors there on the IXDDs but sure that you have them soldered in...
With only 14 volts applied to the drains you would think all would be good at that power level....

Have attached a photo of how I run tests on things...
I have anything attached to a sheet of Aluminium acting as my chassis ground for want of a word...
If you look at the photo you will see my test oscillator and Steves Duty Cycle control board .....
Not sure how well you have everything bonded together or whether your relying on your coax for common grounding etc:
On your Duty Cycle adj board are those blue capacitors disk cermanics acting as low impedance to ground ?

Maybe I'm barking up the wrong tree but I am careful with my grounding....
I probably would not have those yellow wires as links too on the DC board and are not sure how good the gnd is on the underneath side etc:

Just some thoughts  ;D
If I was close by would drop over and have a look etc:

Wayne

Hey Wayne,

Could also be grounding.

I have those same blue capacitors, 6 in total, on the 12 volt IXDD bus.

I’ve never had a parasitic issue either and I’ve built 4-ish E rigs.

It’s something simple. I’ll figure it out without killing it, I hope.

Jon

The fact remains however that when I apply DC power to the phase splitter module, it also 'turns on' the rail voltage for all the IXDD614's (which is not good). Very likely that could be the cause of the parasitic since, as you stated, I did build my heat sink assembly very well.  ;)

Anyway, as a quick check, I'll do the DDS VFO drive thing to see if the problem goes away. If it does I could just put .47uF's in the line going between the phase splitter module and the coax leading to the heat sink assembly. I'd just need to readjust the duty cycle, probably.

But first, M-F 9 to 5, then meet my-main-man-Dennis in ye ole' Arizona, then test.

Jon

Yes put capacitors in line as DC blocking .....  :)



Wayne

That seemed to be the answer in large part. The DC is no longer feeding the IXDD power bus but parasitics still creeped in until I adjusted the duty cycle down more. Applying 13.8 vdc on the drains yielded 12 watts output with class E looking-ish waveforms. Once I raise voltage by testing at 22 volts, things should look proper.

Before that though I'm going to bog down the IXDD inputs like I mentioned before. I want to see if things get less touchy for parasitics by doing so. I've never had an E transmitter as sensitive as this one, so I want to correct the issue before I continue on with voltage levels that will destroy devices.

I made a short video showing this initial testing:

https://www.youtube.com/watch?v=R5_Laqb4RcE (https://www.youtube.com/watch?v=R5_Laqb4RcE) 

Jon

Hi Jon...

If I remember correctly ... Steve mentioned the design didn't run all that well at low power levels and needed to be run at say 25-30 watts minimum for the low power tests etc:
Still find it hard to believe you have parasitics though  :o

Can you post a picture ?

Run it with slightly more power and test again....


Wayne

Definitely. Yes, it doesn’t run well at low power, as I’ve experienced with every transmitter. I just always do this first test to get settings within the ballpark. I’ll run it at 22 volts, probably this weekend.

The force is telling me that the thin wire from the SMA connectors isn’t good.

Jon

Pulse confirmed.

Here's the first full power test at 300 watts carrier with clean drain waveforms. This was the first tune up. I haven't tweaked duty cycles yet. Also, drain voltage is 53.4 volts at 300 watts, which is way too high.

I also found out that the overcurrent trip circuit works wonderfully! It was trip-tripping away all through the initial tuneup testing (shows you how careful I was this time). Anyway, FETs are safe.

Jon

In Class E, the nominal peak drain voltage is 3.8 times the DC applied voltage. In your scope plots, if I squint, I see 170v peak, and that would be 170/53.4 or ~ 3.2X

Maybe some output adjustment would help.

Also what is the peak gate drive voltage?  One of the scope plots shows the drain saturation voltage a bit higher than the other. I'd verify the probe compensation first, and try to use the spring ground on the probe instead of a leaded ground clip.

Looks like fun Jon!

Jim
Wd5JKO

Thanks!

I need to do a little more welding to it, but pretty soon it should be belching carrier on 3870!

Jon

Ok, so check this out.

The drain voltage on my power supply is 54.5 volts at full power output of 300 watts. It's way too high for maximum efficiency. As I said before in the previous message that it's hovering around 80%. That's not a completely accurate figure because I'm currently using an analog wattmeter to make my RF power measurement. To make the final bragging rights RF power measurement, I'm going to rectify the RF voltage coming off the antenna jack and used a formula to accurately calculate the power down to the knats ass. This way, the efficiency calculation will be completely accurate as I'm using a digital voltmeter to take that measurement as well as drain currents as measured from a shunt.

So, now we get to the question as to how I'm going to reduce the drain voltage from 54 volts DC down to where it should be around 44 volts DC. That's quite a large step (an 18.5% reduction to be exact). There are many possibilities to do this. I won't drone on about the 100 ways to skin a cat, but rather expound upon the solution that I'm choosing to do this.

Here's my idea:

I'm going to run a toroidal transformer in series with the primary winding of the unregulated power transformer. I'm not going to place a load on the output of the this 'power wasting' transformer but rather just use this series winding as a reactive resistor to AC. As you all know, and I don't even need to say, the low voltage winding on a large current power transformer has negligible DC resistance. However, with AC signals and the iron ferrite in the core gives a reactive resistance which will absorb, if you will, some of the AC voltage and thusly reduce the applied voltage to the primary of the power transformer. The primary voltage still uses the same step-down ratio as before and now, with the primary voltage being lower, the secondary voltage will also be lower.

Voila!!

We are talking class E nirvana!

What do you guys think? Will this work? Has anybody tried anything like this before? Inquiring minds want to know!

Regards, I will wait for your reply...

Jon

The idea of reducing the voltage with another transformer is valid/common but doesn’t work how you described it. You need to hook the primary to the ac line and the secondary in series with the main transformer. Connected in one direction it will boost the voltage of the main transformers output, the other way it will reduce or buck it, hence the term boost/buck transformer. There’s plenty of diagrams and links out there, they even sell packaged boost/buck transformers for this.

Ed

Why do you want to run a lower voltage?

Surely just run a higher Z to bring the current down?

Max efficiency is I2R, lots of volts and less I would be better, would it not?

JB.

Hi Jon....

Great to see your TX coming along  :)

Since your power supply outputs 55 volts I would probably leave it and run the Transmitter at that level.....
The 11N90s drain to source resistance is appox 1R which is rather high and Steve parallels many together to lower that down so to reduce loss causing heat etc: particularly when he runs his power levels  ;)
I think your using 3 FETs either side which combines to appox .3R generating appox 8 watts of heat at standing carrier and with modulation would increase at lot more.... using 45 volts that would increase to appox 12 watts...
All this adds up but I know you like using big heatsinks !!!
At 55 Volts there is still plenty of headroom available  :) :)

All this is just resistive loss let alone your efficiency % .....

You do what you reckons best though....


Wayne

I agree with Ed. Bucking is an efficient, cheap, and good-regulation way to do what you want.

Thanks for all the input. I still haven't gotten around to more tweaking yet as I've been busy. It'll go on the air as-is, but as Steve has said before, "Test, test, test!"

Quite honestly though, I'm tickled poop-less that my version of an over-current trip works wonderfully. I only wish I had made a schematic though. I've already forgotten most of it.

Anyway, more to come!

Jon

** I just did the math for adding the second transformer inline (both primary and secondaries), in phase obviously. With 120 volts on the primary and two 35 volt windings in series on the secondary, this will bring me nicely to 44 to 45 volts.

Wham-bam, thank you Ed.

Hi Jon,

Hmmmmmmmmmmmm. .  Not knowing exactly what your circuit looks like, I can't be sure, but definitely you should be able to DC couple from the phase splitter to the transmitter.  In fact, this is a good thing.

When the input is not present, are you still getting output from the 2 phase converter ?

Do you have 300 ohm resistors in series with the MOSFET driver's input pins (one for each driver) ?

I assume the driver outputs are NOT paralleled.  If so, that will cause many issues.

Anyway, it sounds as if things are coming along, and that's great !!

Regards,  Steve

Hey Steve,

I did reconfigure the drain power supply to use just the toroidal transformer only, and quite honestly I'm getting the cleanest class E waveforms I've ever made. I did change the IXDD driver inputs from being parallel to individual feeds via resistors and that may have helped. I'm still getting a carrier drain voltage of about 54 volts and I'll just need to live with that as I'm not using PWM. It's acceptable.

I don't quite understand how the DC coupling to the IXDD's was causing such a problem, but using DC blocking caps fixed the problem.

I had a problem coupling audio to the Heising circuit using an old Radio Shack PA amplifier. The circuit connected directly to the Heising capacitor. I did incorporate a time delay with a large parallel 200 ohm capacitor which allowed the cap to charge to drain voltage before connection to the PA amp. This protects the audio amp from power spikes and has worked well on previous designs.

However, I was not able to push audio onto the RF deck. Not having a schematic of the Radio Shack amp, I'm guessing that maybe it was a class A amp and wasn't able to deal with the voltage on the speaker outputs. Measuring AC voltage on the speaker outputs did show drive, but very little.

To fix this, I decided to go tried-and-true with a class D Crown PA amp. Additionally, I put a large power toroidal transformer in between the PA amp and the Heising circuit for further isolation. This has worked well on previous designs as well. The impedance ratio between the 8 ohm audio output and the modulation impedance was close to 1:1, so I used the 120 volt winding for the PA amp and the 100 volt winding for the modulator deck.

I did this a couple days ago and haven't really had a chance to test it yet at this point. I'll probably get to it this weekend.

Jon

I know everyone is dying to know, (not!) but I figured out my modulation problem. The 12 volts from the sequencer isn't keying the PA amp cut-in relay. It's a simple wiring issue. I'll test more soon. Very busy, just trying to keep everyone a breast.

Jon

I'm happy to report that I put it on the air for the first time tonight at 300 watts carrier. It sounds good! I was able to hear myself in the headphones from sampled RF audio.

I stayed under 100% modulation and decided to go a little higher when the overcurrent trip shut down the power supply. I also blew a Transzorb on one phase. No biggie, it's just $2 and everything was protected. I should've implemented this stuff on my E transmitters back in Phoenix.

Anyway, running the drains at 54 volts DC carrier leaves me a little less headroom on voice peaks. I apparently exceeded 540 volts. So next go around I'll run it at 200 watts carrier and maybe just run it at 100% modulation.

Jon

Quick and dirty efficiency calculator using a Raspberry Pi.

https://www.youtube.com/watch?v=lDBehtJrPXU (https://www.youtube.com/watch?v=lDBehtJrPXU)

Jon