A Clean Canvas - 24 Pill Class E Rig Construction

[K1JJ]

Nice filter layout Tom. (Nice to see layouts that don't look like they fell off the back of a cow.)
Keep an eye on temperature rise with the caps you might need more lower values in parallel.

Thanks, Frank.

Well, actually, you're seeing the results of your advice over the years.  That last 4X1 PDM filter system I built is laid out like this one. I'm simply using your information on layout and general practices....

Those orange caps should be OK. They are the same ones I used with the last 20 pill MOSFET rig I built six years ago, just different configs. They didn't get warm back then.

You will notice in the third filter picture I have the diode (and filter cap) that dumps spikes into the power supply  - it's mounted near the output, ready for action.  I might even mount a trap there to further attenuate the residual 140kc switching frequencies. I still have the old series trap lying around. We'll see if it's needed first.

T
 

[K1JJ]

I added the last transorbs, RF bypass caps, input RF drive connections and assorted stuff. The 24 pill RF deck is officially ready to fire up and test. However, I will hold off until the rest of the safety infrastructure is wired up and the rig is a complete package.  Then we just gots to flip the switch and play the Rico Suave tune...

Looking at all the work required for just 24 pills, I'm not sure I wanna do this again to make it 48...

I hope this series of posts is giving others the encouragement needed to do their own class E rig project!

T

[WA1GFZ]

Tom,
The diode on the output of the filter. Put a cap to ground right at the cathode. Reason. The wire back to the power supply will have some inductance. This will limit transient current. A cap will provide a nice place to dump any HF transient.
Come up slow and measure every gate. I'd measure every gate before I put any drain voltage on the final. Then feed it a few volts and see how the gate waveform changes. It is all about that wiggle as the gate goes through 5 volts or so. Also keep an eye on the driver  voltage and make sure there is no high frequency crap on the 12 volt rail. Good luck

[K1JJ]

Tom,
The diode on the output of the filter. Put a cap to ground right at the cathode. Reason. The wire back to the power supply will have some inductance. This will limit transient current. A cap will provide a nice place to dump any HF transient.
Come up slow and measure every gate. I'd measure every gate before I put any drain voltage on the final. Then feed it a few volts and see how the gate waveform changes. It is all about that wiggle as the gate goes through 5 volts or so. Also keep an eye on the driver  voltage and make sure there is no high frequency crap on the 12 volt rail. Good luck

Yep, there's already a 0.5 at 2KV cap on the diode's cathode to gnd. Did you mean to make the lead shorter? It's about 2" long now.  Refer to the 3rd picture of the PDM filter.

OK on the fire-up precautions. It will be a week or two yet cuz I wanna wire everything up first and have protection, etc. Still have a ways to go before I start the final assembly.   

T

[WA1GFZ]

2 inches is better than 2 feet. Also the Damper diodes need a local bypass an OD in parallel with one of those  680uf/500 volters would be nice if it will fit (if there is any distance back to the supply).

[K1JJ]

2 inches is better than 2 feet. Also the Damper diodes need a local bypass an OD in parallel with one of those  680uf/500 volters would be nice if it will fit (if there is any distance back to the supply).

Damper diodes?  You don't mean the diodes on the output of the PDM modulator?

Or do you mean the single diode that dumps the residual PDM filter spikes into the power supply as mentioned before?  I have 0.5 uf to gnd at the diode cathode. Are you saying to put a big 580 uf cap there too?  It does makes sense since the PS is about 3' away in cable length.

T

[WA1GFZ]

Tom,
I'm a dope, diodes go to ground in a floating modulator. But it would be a good idea to have a cap at the drains of the modulator FETs. A 680uf would be a nice low Z source. Otherwise you have the inductance of the wire back to the supply limiting current during short pulses. I think there is a cap on the output board but not that big.

[K1JJ]

Tom,
I'm a dope, diodes go to ground in a floating modulator. But it would be a good idea to have a cap at the drains of the modulator FETs. A 680uf would be a nice low Z source. Otherwise you have the inductance of the wire back to the supply limiting current during short pulses. I think there is a cap on the output board but not that big.

Yep, there are a pair of 560 uf caps - one at each end of the modulator FET drain bus already.  Steve had that covered.


Well, now all I have left is to build the phase splitter / driver board and then start integrating the whole system together. That in itself is a big project considering the many supplies, shutdown board, interconnections, digital meters hook up, etc.

Got my blue-digit freq counter for the VFO yesterday, so got all the parts to finish the job.

T

[K1JJ]

One of the guys building a class E rig wanted to know the capacitor values of the PDM filter.  Bear in mind this filter is designed for 24 pills and has an output load designed for 1.75 ohms.  If using a different load impedance (more or less MOSFETS) then the filter values will change.




(I cannot post the SwitcherCad schematic cuz it's in .asc format, not allowed here)


Six Element Butterworth infinite impedance low pass filter:

The values are:  L1= 28.8 uh    L2= 28.8 uh    L3= 14.1 uh

C1=10.7uf    C2=7.3 uf    C3 = 1.6 uh.

Remember that the C3 value is in parallel with the input RF deck bypass caps. These are 0.15 uf X 4 caps = 0.6 uf.  So, use a 1uf cap for the filter C3 position and allow the other 0.6uf to come from the RF bypass caps at the foot of each output transformer.

Use the picture I posted in this thread to see how the filter is wired and laid out.

T

[K1JJ]

Here's the completed and working VFO. I added a digital display and it works FB.  

The top and side covers of the VFO have been removed for pics. Easy to service and seems stable in all regards. Ready to drive the E-rig with digital pulsed RF.

The second set of pictures shows the VFO output, the VFO buffer output and finally the digital board's output.  I don't like the overshoot that's occuring on the output of the digital board. Anyone know what may be causing this?  I had the same overshoot with no leads attached to the board's output, just terminated with 100 ohms - and I had a short lead feeding the input.   It looks the same at the earlier clock chips too.

This VFO was not an easy thing to get working. I struggled trying to get enuff VFO analog output for the digital board for almost a day.

T

[K1JJ]

1) VFO out

2) VFO Buffer out as in goes into the digital board.

3) Overshoot/ringing of the final digital output at 12v p-p.  What could cause this?  The carrier sounds clean in the receiver and is stable.

I just tried the digital board with an external sine wave signal generator. The digital board output still has overshoot.  Maybe the sinewave drive needs to be more square/pulse-looking to generate a clean, square output.

T

[K1JJ]

VFO and splitter/ digital driver schematic-

http://www.classeradio.com/vfo_2_band.pdf

[KF1Z]

That is truly nasty... !

Tom... please send me the schematic Steve sent you to build the board!

Thanks...

[K1JJ]

Bruce,

Here's the "Two-Band VFO" schematic I scanned for you.

I tried an HP external sine generator and the digital board output looked the same. So it's not the VFO / buffer.  Does the sinewave have to be squared off to drive the board properly or can a pure sine do it?  I notice Steve talks about adjusting to get a sine saturation waveform.

T

[KF1Z]

VFO and splitter/ digital driver schematic-

http://www.classeradio.com/vfo_2_band.pdf

I wondered WHY you posted this schematic link......

Because this link does NOT go with the board!

The one you scanned does...

[KF1Z]

Bruce,

Here's the "Two-Band VFO" schematic I scanned for you.

I tried an HP external sine generator and the digital board output looked the same. So it's not the VFO / buffer.  Does the sinewave have to be squared off to drive the board properly or can a pure sine do it?  I notice Steve talks about adjusting to get a sine saturation waveform.

T

Pure sine wave up to the base of Q2.

At the emmiter of Q2, it is clipped...

Not at all what your scope pattern shows....


Your last two pictures should look very much like the two pics at the bottom of this page...

http://classeradio.com/vfo.htm

[KF1Z]

But no... the sine does not have to be squared off....

You basically just need about a 3 volt positive pulse to clock the flip-flop.

Anything negative will be wasted to ground via a resistor.

Sumpin not right in the set-up Tom...

[steve_qix]

Hi - here is the correct schematic, and a link to the same 

The oscillator section should produce about 3 volts (or a bit more) - sufficient for a logic 1, and enough to drive the first flip-flop.

Link: http://www.classeradio.com/vfo_2_band_rev_c.pdf  The old schematic on the class E web site had also been replaced, since we have the new one.  The old one is completely valid, but the new schematic goes with the boards that I made.  Note: the next board rev will contain *minor* revisions, so a new schematic will be available for that one.

Picture:

[steve_qix]

1) VFO out

2) VFO Buffer out as in goes into the digital board.

3) Overshoot/ringing of the final digital output at 12v p-p.  What could cause this?  The carrier sounds clean in the receiver and is stable.

I just tried the digital board with an external sine wave signal generator. The digital board output still has overshoot.  Maybe the sinewave drive needs to be more square/pulse-looking to generate a clean, square output.

T

The input will not affect the output waveform unless the input is causing the flip-flop to unlatch due to ringing.  It is hard to know if the waveform REALLY looks like that, or if this is a measurement anomaly.  I say this because all fast edges shown in your pictures appear to have ringing.  Can you look at any fast edges cleanly?  check the output of the 74AC112.  What does it look like?

Is that measurement right at the board itself?

I have found that the SLIGHTEST inductance in the measurement system (the probes, grounding, etc) will cause ringing where none is actually present.  I have build a number of these VFOs, and the output is a clean square wave.  This is what I get:


Ignore the high duty cycle.  It should be more like 30 or 35%.  The important place to measure the duty cycle is at the gates of the RF amplifier MOSFETs.  It should be about 40% to 45% THERE, which means it will be smaller at the VFO because every IXDD414 adds a bit of storage time!!

I use a Tektronix 475A with 100mHz Tektronix probes.

As an aside, I have some 20 foot runs of coax (and some much shorter) running from the VFO to various transmitters.   Even the output at the far end is quite good, so it should be excellent at the input side.  I put a selector switch on my VFO, allowing me to select between up to 5 different RF amplifiers.  I got tired of disconnecting cables when changing transmitters     That's a nice feature to have!

Sorry you had trouble getting the oscillator section of your VFO to produce enough output (reference from a previous reply to this thread).  Your experience is not typical    Did you finally use 2n3904 transistors (as specified in the design) or continue with the 2n2222?  Both should work, but the 2n2222 has a higher input C, and appears to be a little slower, so some values might need adjustment.

Regards,

Steve

[WBear2GCR]

fwiw, and I am not sure it is an important point, but apparently there is significant difference over time in the 2N2222 transistor!
The early metal can ones apparently are best, and have better specs than any of the later plastic paks. The PN2222 is allegedly actually a different transistor physically (inside, the way it is made), as are subsequent ones marked 2N2222...

                        _-_-bear

[steve_qix]

fwiw, and I am not sure it is an important point, but apparently there is significant difference over time in the 2N2222 transistor!
The early metal can ones apparently are best, and have better specs than any of the later plastic paks. The PN2222 is allegedly actually a different transistor physically (inside, the way it is made), as are subsequent ones marked 2N2222...

                        _-_-bear

That's my understanding also.  I don't use the 2n2222 myself, but Tom did (at least at the start) in his VFO oscillator.  I like the 2n3904 for a number of reasons, but particularly because there is a good compliment - the 2n3906 - which I use all the time in conjuction with the 2n3904.  They are also really cheap and plentiful 

[K1JJ]

Thanks for all the replies.

Steve, I will try some different scope probes and even try a direct, short wire to the scope just to make sure it's not adding to the waveform ringing. It COULD be the probe and I'll see some improvement which will indicate this. I see the same waveform throughout the whole board.  I did notice this false ringing effect when measuring the PDM waveform of my old 4X1 PDM pulse. It was all in the test setup and the pulse was actually square. We'll see later today.

I am using a 2N2222 for the VFO osc and a 2n3904 as the buffer.  The VFO sig is pretty good at 12v p-p, but the buffer drops way off.  I'm out of 904's but will try one for the vfo osc as soon as I get one.  Right now I am getting barely 2.2V out of the buffer to the digital driver board - the bare minimum to keep the digital board and relay circuit happy.  But the real problem with the ringing is hopefuly the test set up.

Otherwise, the VFO is working FB.

I'm testing the shutdown board and power supplies now. Actually, the whole 24 pill rig is to the point where I could JS clip lead it all together and fire it up, but I decided to test it all one component at a time.  The PDM modulator and RF deck need to be tested next.

I'll post with results later.

T

[K1JJ]

What a difference a probe makes.

I tried two different additional probes. The first pic uses a Tek 6027 UHF probe and the second uses a Tek HV probe.  The difference from the original waveform pics yesterday is big. Looks like that other Tek probe has inductance.

The third pic shows the VFO buffer output.


It appears the duty cycle of the digital board output is around 30% or so?


Pic 1:  Digital board output probe 1

Pic 2: Dig board output HV probe

Pic 3: VFO buffer output to Dig Board probe 1


BTW, a straight wire probe (no compensation) was the worst.


I still need a little more buffer output to play with the VFO waveform pot. Right now it's on the hairy edge of just enuff voltage to drive the digital board. Any waveform compression and the digital board drops out.  Maybe using a 2N3904 in the VFO osc will help - when I get one later.


Looks like everything was FB all the time.


T

[WQ9E]

So how do these probes look on the square wave calibrator output?  Make sure the compensation cap is adjusted properly in the probe or neither will give you a true picture.

If you are using a Tektronix P-6015A (or similar) the compensation setup requires several steps.

To precisely set the compensation, use delayed sweep to expand the critical points on the square wave calibrator signal.

Of course you want to make sure your built in scope calibrator is doing its thing first so take a look at it with an X1 probe.

[WA1GFZ]

Tom,
Why do you have the duty cycle set so low? The two pots on the board should allow you to get closer then 50%