A Clean Canvas - 24 Pill Class E Rig Construction

[WA1GFZ]

Tom,
I have also had problems as I told you with parallel drivers going into weird oscillations as they ping -pong against each other. I had a LF oscillation in my 160 meter rig when I first built it. A long cable feeding the drive needs a good load on it. The shielded wire you are using looks like about 20 ohms impedance so you may get away with parallel drive if you load each end of each bus with a low value resistor. You need to take a careful look at the drive to make sure it is stable before you add drain voltage or you will be replacing parts. Steve's method of multiple drive cables reduces the drive impedance but adds complication.  Your predriver can handle some current so you might get lucky adding load resistors to clean up the pulse. A small series resistor at the pre drive end of each cable will reduce reflections at the expense of a bit less drive voltage. 

[K1JJ]

OK - Thanks for the suggestions, guys.

It's close to firing up, so this is good info. I put a simulated 1.75 resistive load on the input tank and tuned it to give MFJ analyzer 50 ohms on the output, so that shud be a close start.

Yes, I was aware of the bypass caps being on the end, Steve. It was a matter of servicability and crowding in that caused me to do it. But guess I'll move them to the center now, since I got caught.

The same with the feed cables to the driver chips - I wanted to simpify things, but maybe I'll try doping it more as Frank suggested.    Frank, what value series predrive resistor?  10 ohms or something like that?  Or since it's on the table, I cud always run extra cables and do BOTH.

As for the longer primary xfmr leads - they needed to be longer cuz of the physical position of the transformers. In the old E rig, I did it like yours and servicing was a nightmare cuz the transformers were right over the fets. I'm hoping the extra 2-3" will be forgiven on 75M....  

Do you think firing it up with the PS at 1/2 voltage (~ 70V total, 25V on Mosfets) will prevent fet damage if I have parasitics or shud I go even lower by adding a test Variac?


** I'll do the mods and take a picture and post it before firing up.

T

[WA1GFZ]

Tom,
Adding any L in a low Z circuit adds ringing so be aware of that.
Make sure the gate drive is clean before you put any voltage on the drains. Also make sure the the signal on the input of the drivers is nice and square. I would set the PDM generator output to a low voltage and work your way up to carrier. I would modulate at each output voltage to make sure everything is stable.

[steve_qix]

Hi Tom,

There are some simple things you can do to improve the situation without changing the basic layout.  First, I would really make changes to the output transformers.  This should be a simple thing, but those long lead will cause problems.  You may not even be able to balance the modules correctly.

So, some suggestions.  The RF output doesn't have to come from the center of the bus.  Because you are dealing primarily with fundamental frequency (and not lots of harmonics like the shunt capacitors handle), it is not as crutial to take the output from the direct center.

Given this, and knowing your layout, you should be able to bring the transformer primary within 1/2 inch of the drain bus and connect it there.  The bypass capacitor should be similar - maybe 1/2 inch of transformer primary and short leads on the bypass caps.  Come of the ends - just make the leads SHORTer.

The secondary should be dressed to be as short as possible between transformers.  At the ends, it's not as critical.  You don't want a lot of phase shift along the line between the modules.

Anyway, some ideas.

Regards,

Steve

[K1JJ]

OK, Steve and Frank -

Yes, I wondered about using the end of the bus for the pri connection - will do and make them shorter.

Maybe I can tighen up the sec between transformers somewhat too.

I'll make all the changes mentioned in the last few posts and post a close-up picture later.

I can't fire this thing up until Monday when I get that LM-239 chip anyway. I'm also adding the 100A and 200V digital meters today. That shud finish the construction - at least until I get beat up during testing time...

Thanks.

T

[K1JJ]

Steve and Frank -

Here's the mods I made.  There are separate cables per driver, the bypass caps are near the middle and I shortened up the xfmr primary connections to the end of the bus.

I can see now how the predriver input bus was facing the output busses of both the driver and gates of the fets, creating a possible feedback loop. Now the connections are smaller and shielded better.

I didn't add any resistance in series with the pre-driver cables yet. If I have parasitics, that's the first thing I'll do.

Watsa?

T

[WA1GFZ]

Tom,
I would have run the cables back to the predriver chip. Now you have a bunch of 20 ohm cables hanging off a single cable. Big Z bump. Maybe you could add 2 driver chips where the terminal strip is located if there is a degraded pulse at the joint.

[K1JJ]

Tom,
I would have run the cables back to the predriver chip. Now you have a bunch of 20 ohm cables hanging off a single cable. Big Z bump. Maybe you could add 2 driver chips where the terminal strip is located if there is a degraded pulse at the joint.

Frank,

The VFO output uses a 414 driver chip and has four foot long cables going to the rig. After doing the mods, I see no difference in the RF drive waveform. The waveform at the predriver chips in the rig and at the gates of the MOSFETs looks very good, like a class E waveform. I see little signs of deformities, so guess we're all set.

Tina said his E rig burned up tonight becuz of an RF parasitic... so guess it's something to be careful of...

I will take your suggestion of slowly ramping up the voltage of the PDM pulse to test the finals at a low power level.

T

[WA1GFZ]

You want to drive the gates with a square wave which will degrade a bit when you add drain voltage miller effects.  Long leads cause ringing there is no hiding from that. Remember what I said about the source leads. Everything matters.

[K1JJ]

You want to drive the gates with a square wave which will degrade a bit when you add drain voltage miller effects.  Long leads cause ringing there is no hiding from that. Remember what I said about the source leads. Everything matters.

Hmmm.... I looked at the 414 output from the VFO when disconnected and it's a nice squarewave. But when I hook up the 4' cable going to the rig's input the waveform looks half way between a sinewave and squarewave. (like a class E waveform)  

Steve mentioned he runs up to 20'? cables from the VFO driver output to the rig. What does yours look like at the rig, Steve?


My old 20 pill E rig using analog drive waveform looked like a standard call E waveform, cuz that's what the driver was.

So, how much will the efficiency change by my using this E-waveform  as compared to a squarewave? Will it make it more prone to parasitics cuz the on time is higher?

I can't see how adding a 414 at the rig input (after the 4' cable) will help cuz I already have one driving the finals 6" away and that helps clean up the waveform very little into the MOSFET gates. ie, It's still basically a class E waveform at the RF final gates.

I'll try some load resistors at the input and see if that helps clean it up. Hopefully there is plenty of current capability in the 414 driver.  I have little heatsinks on them now.

T

[WA1GFZ]

The harder (more square) you drive the gate the better it is.

[W1IA]

Tina said his E rig burned up tonight becuz of an RF parasitic... so guess it's something to be careful of... Wink

After effecting a boatload of repairs and upgrades in the smaller rigs; inside and out (antenna repairs), Murphy bit me in the arse I lost a couple of fets in the big rig. I have been meaning to upgrade to digi drive and install Steve's overload protection circuit. I am populating a new heat sink and tearing down the old project for a complete overhaul. As if I don't have enough projects already.

[WA1GFZ]

It's all about layout

[K1JJ]

Brent,

Will the shutdown board protect against parasitics?  I doubt it. Except for no VFO drive, it just senses DC current and has no idea what freq it's at. Maybe it wud be wise to add in the swr input. That might do it.

But, if there is no load at the parasitic freq, big voltages can occur....  what do you say, Frank?

Yes, hopefully the digital drive will make the whole lash-up more stable. I'm about to find out here myself.

I'll try the load resistors later today and see if it gets better.

T

The HossTraders Manchester, NH WX has finally cleared out a hi-hi FB sunny gap for us. Thur, Fri always looked good. Sat was questionalble, but looks good now!

T

[W1IA]

Brent,

Will the shutdown board protect against parasitics?  I doubt it. Except for no VFO drive, it just senses DC current and has no idea what freq it's at. Maybe it wud be wise to add in the swr input. That might do it.

Byt, if there is no load at the parasitic freq, big voltages can occur....  what do you say, Frank?

Yes, hopefully the digital drive will make the whole lash-up more stable. I'm about to find out here myself.

I'll try the load resistors later today and see if it gets better.

T

The HossTraders Manchester, NH WX has finally cleared out a hi-hi FB sunny gap for us. Thur, Fri always looked good. Sat was questionalble, but looks good now!

T

No Tom, but at least the pulse train will shut down faster than the old overload circuit....the digi gate drive is the biggest benefit. The old 2 fet analog driver stuff was always unpredictable. I am adding shunt resistors on the 40 meter deck and running cables like yourself to each driver chip. Since the small deck is 10 fet, I will use 250 ohm terminations per cable.

I am trying to track down and odd problem with the 8 fet 160/80 deck (also digi drive. I noticed a slow change (maybe thermal) in current draw between the modules as the deck runs for a bit. Starts out 4 amps per module and then creeps to 4.5 and 3.5 per module? I have 1000pf atc chip caps as shunt in the center of each bus. I changed out the doorknobs to ATC's to see if it was the problem. I am measuring about 375 volt peaks on the drains.
Any Ideas Frank...Steve?

Brent

[K1JJ]

No Tom, but at least the pulse train will shut down faster than the old overload circuit....the digi gate drive is the biggest benefit. The old 2 fet analog driver stuff was always unpredictable. I am adding shunt resistors on the 40 meter deck and running cables like yourself to each driver chip. Since the small deck is 10 fet, I will use 250 ohm terminations per cable.

I am trying to track down and odd problem with the 8 fet 160/80 deck (also digi drive. I noticed a slow change (maybe thermal) in current draw between the modules as the deck runs for a bit. Starts out 4 amps per module and then creeps to 4.5 and 3.5 per module? I have 1000pf atc chip caps as shunt in the center of each bus. I changed out the doorknobs to ATC's to see if it was the problem. I am measuring about 375 volt peaks on the drains.
Any Ideas Frank...Steve?

Brent

Just make shorter transmissions...    

When you blew out the RF module, what did the waveform look like on the scope - did it have the normal hairly fuzz of parasitics look? And how many seconds did it run like that before blowing? 

I tried mine the other day without the latest mods, but with a 2 ohm load on the RF deck and saw parasitics.  I used only 12VDC, so no damage. Hopefully everything will be tame when I test it soon. I just calibrated the shutdown board, so getting close. I decided to add swr sensing. I have an old Bird slug set up that I can keep in reverse and sample some voltage for the board.

T

[W1IA]

When you blew out the RF module, what did the waveform look like on the scope - did it have the normal hairly fuzz of parasitics look? And how many seconds did it run like that before blowing?

Wasn't watching Tom. Just the typical shotgun effect and a blue flash. I can't really confirm the parasitic issue, but its a likely due to uneven pep drive voltages of the past. Was just talking with Al W1VTP and BANG!
Since Mark is building similar rig...I will construct rf deck when he does his. In the mean time I still got Baby E to keep me on the band.
B

[WA1GFZ]

Just pulled up one of my class e final simulations and added .1uh in series with the drain lead of 1 phase. The drain signal went from nice and clean to high frequency ring on both turn on and turn off. This adds a lot of extra dissipation in the device die and reflects this crud back at the gate. If this crud is generated at the switch point the device oscillates around the switch point making it unhappy. Driving the gate harder helps overcome this problem but the real fix is layout reducing leakage inductance and strays. 
So if your layout looks like it fell off the back of a cow I wish you luck

[K1JJ]

Heck, no problem.  If it blows up I'll use the parts to build a GFZ Joystick....


In the meantime, I'll pull out the output transformers again and make them more compact with less wire.

T

[K1JJ]

More improvements:

I stacked the cores tightly and was able to take about 7" off the secondary and about 2" off each primary winding.  The only extra lead lengths now are the 2"  primary connections to each drain bus. The bypass cap leads are about 1/2" to each capacitor.

I also added some .02 disc caps to the transformer bypass caps for good measure.

That's about all I can do to shorten up the output transformer.

Got the digital blue 100A  and 200V meters working. Look pretty.

Still need to build a high swr sense circuit.

T

[The Slab Bacon]

Just FWIW guys,
                       I designed and built a stripline IPA for my 220 repeater some years back. (I used an NPN bipolar device, dont remember the number) I was popping the tops off of pills right and left. I found that it would go into a parasitic oscilation whenever and only if it felt like it. I bypassed the 12v feed at the end of the stripline six ways from sunday and it would still do it occasionally.  I found that I had to bypass it for many different frequencies. I wound up using a 100pF, a .01uF,
a 220uF electrolytic, and are you ready for this..............a 1N5400 series diode.

The diode is what really did the trick. The diode really did the trick, it just tamed it down and stabilized it so well that it has been now running for close to 15 years without a problem. I guess slicing off the negative half cycle prevents it from going into full oscillation. Whatever it did, it now works FB.

I wonder if something like that would work for youse guys??


                                                         The Slab Bacon

[WA1GFZ]

Tom,
I'm concerned with the drain lead to the transformer could be as high as .050uh. Also you could flip the phase of the secondary to get shorter leads to the tank circuit if that helps. The joy stick layout is a bit anal. There are other ways. to get shorter leads.

[K1JJ]

Slab,

Good ideas. I made note of using extra varied caps and the diode. If I have problems I will add those items to the two supplies in the RF deck.

Frank:  My other 20 piller had a similar layout and worked well for nine months until I pushed the drain voltage up too high...

About those 2" pri leads... if I have problems, I will replace the wires with thick brass bus bar right to the transformers to reduce L.

T

[K1JJ]

Frank,


Please take a look at this schematic of the shutdown board. I want to drive the shutdown board with an external Bird rev slug for high swr protection. To the extreme left shows the  swr sensor inputs marked  "optional external inputs."  This is a balanced input to a LM-239 chip.  My Bird slug in the reverse swr mode will put out maybe 0.3 volts max positive voltage. I tried hooking it to these inputs many ways, but the high swr .3V will not trigger the stage no matter where I adjust the "external threshold" pot on the LM-239 stage. If I put 5 volts to the  input,  it triggers OK.

Shud I use an external transistor or op amp to amplify the 0.3V DC bird signal? The Bird output is unbalanced.  Since it needs DC to work, this is more complicated.  Can you come up wid a simple circuit to interface?  TNX.

http://www.classeradio.com/efficiency_meter_rev_d.pdf

T

[WA1GFZ]

Take the bird positive and connect it to the negative input. Then take the bird negative and connect it to the positive input. Then adjust the pot for the trip point you want. If that doesn't work you will need to change some resistor values around the comparator.