The Cliplead 400 transmitter lives!

[WA1GFZ]

Gee no wonder my hair is turning gray

[steve_qix]

I have created a preliminary schematic of the ClipLead 400 transmitter's modulator output section (all of this is new).  The remainder of the modulator (PWM generator, etc.) is from existing schematics (and boards) and is already documented at the class E web site as part of the Pulse Width Modulator and Power Supply.

The ClipLead 400 modulator takes a standard PWM signal from the existing PWM generator board.  This signal is then fed to a phase splitter, drivers, and ultimately to the output MOSFETs operating in push-pull.  All of this circuitry is included in the new schematic.

This schematic is subject to change, etc.  Don't build any of this without contacting me first !

Comments welcome 

Click on the link below to view the schematic.
http://www.classeradio.com/off_line_pwm_output.pdf

Regards,

Steve

[W1DAN]

Steve:

SAWEET!!

And sounds good too!

Dan
W1DAN

[Ralph W3GL]

Steve...

PUSH PULL TRANSFORMERLESS PWM "OUPTUT" SECTION

Really?

[WA1GFZ]

Something doesn't look right in your clamp circuit at the transformer primary

[steve_qix]

Something doesn't look right in your clamp circuit at the transformer primary

Hi Frank,

I included the circuit snippet and explanation of the clamping function in general, so everyone can see what we're talking about..  The purpose of the clamp circuit is to hold the line on spikes which result from leakage inductance and other stray elements in the circuit.  It is possible for there to be very short and very high spikes during the turn-off cycle of the MOSFETs.  In order to control these spikes, clamping and other circuits are employed.



In this case, D800 and D801 conduct energy into C807, charging the capacitor.  The "normal" voltage will be twice the power supply voltage.  High, but otherwise low energy spikes will have very little effect on the voltage across C807, however the diodes, working into C807 will effectively clamp the maximum drain voltage at twice the power supply voltage.

The voltage is sent back to the power supply through R805.  A small amount of power is lost in R805, however we are looking at only a few watts.  If one wanted to conserve the energy, it could be utilized in some way, using a buck switcher or other switching regulator.  In the interest of simplicity, I chose to drain off the few watts.

Anyway, what did you see as odd in the circuit?  It definitely works, although there may be other ways to accomplish the clamping function.  Let me know !! 

Thanks !

Steve

[WA1GFZ]

you are limiting the flyback energy in the transformer by ac coupling it to ground through the 3 uF cap. Yes it will peak charge but I think you are wasting power. Any spikes are going directly to ground. You may have excessive leakage inductance in the transformer and lead length and this bypass is the fix but usually this energy is dumped back into the supply not ground.

[steve_qix]

you are limiting the flyback energy in the transformer by ac coupling it to ground through the 3 uF cap. Yes it will peak charge but I think you are wasting power. Any spikes are going directly to ground. You may have excessive leakage inductance in the transformer and lead length and this bypass is the fix but usually this energy is dumped back into the supply not ground.

Interesting idea !  I'll try an experiment by moving the 3uF capacitor's "ground" connection to the power supply and not ground.  I'll let you know if I see any change in efficiency.  Everything is metered, so I'll be able to tell right away.   Let you know 

Regards,

Steve

[W1VD]

Steve

Look forward to hearing it on the air! Which material are you using for the 48613 core?

[steve_qix]

Steve

Look forward to hearing it on the air! Which material are you using for the 48613 core?

P material.  That's right, I have to specify that on the print.  The number isn't enough!   

Also, I did the experiment - moving the snubbing capacitor's return from ground to the power supply.  I was unable to measure any difference in power input, output, voltage or current.  It there were a change, I couldn't read it.

I didn't necessarily expect much because the modulator is VERY efficient, and nothing gets warm that one would not expect to get warm.  It appears as if the greatest loss (and it's not much loss) occurs within the switching MOSFETs due to the R D-S on (I'm only using 2 MOSFETs per side - 4 would really do it), and the diode drop loss in the switching diodes that are part of the bridge and damper.  Again, small losses, but I'm trying for 95%     That's where conducting the snubbing energy efficiently to the power supply using a switcher would make a difference of a 1 or 2 percent - but now that's starting to get rediculous    (but isn't that the whole point?).

Anyway - I'm so pleased with how this is working, that I am contemplating laying out a PC board and making a kit of parts, like I have for the other modulators. Thing thing about this one is, it can be built very lightweight, and that's attractive as we AMers start pushing old-buzzard-ness.  And, of course, these units can be combined for greater output if one were to need it for whatever reason, and the weight would be very reasonable.

Regards,

Steve

[WA1GFZ]

Steve,
I have a real good app note on snubber design somewhere on my work computer. I will send you a copy when I find it. One other thing is R805. A wire wound resistor has a lot of inductance and ineffective in eliminating high frequency transients. 5K and 3 uf is a 15 ms time constant on a 130 KHz switcher is very high.
P material is cool stuff for power supplies.

[steve_qix]

Steve,
I have a real good app note on snubber design somewhere on my work computer. I will send you a copy when I find it. One other thing is R805. A wire wound resistor has a lot of inductance and ineffective in eliminating high frequency transients. 5K and 3 uf is a 15 ms time constant on a 130 KHz switcher is very high.
P material is cool stuff for power supplies.

I'd like to check out the app note if you can find it !  I figure the 3uF is pretty much a short circuit at 150kHz (.35 ohms) as compared to the 5K load.  The voltage is pure DC at that point (checked it with the 'scope)  - no 150k stuff showing up.  I've talked with a few power supply engineers and they tell me they usually have to use snubbers with this sort of application.  But, in this case the transformer is pretty good, so I don't have a lot of energy to "snub out", so to speak.

If I weren't so concerned about modulation linearity at the extremes of the pulse widths (1% - full off, to 49% which is full on), I wouldn't worry so much about the snubbers (the voltage doesn't go that far out of line), but I find that if I do not snub these tiny pulses out, they affect the modulation at the extremes - not a lot, but it's there!!  Anyway, not a big deal but definitely worth getting as perfect as possible (are there degrees of "perfection") 

That P material is GREAT, and amazingy enough, not expensive.

Oh, on another note - I've tried some experiments with toroidal inductors in the filter and air wound inductors.  The air wound inductors dissipate slightly more heat in the windings due to DC resistance (about 2 watt for both inductors wound with air), but the core in the toroidal filter dissipates about the same amount of power due to magnetic stress - so it seems to be a wash.

Now, I do believe the air wound inductors work better, and are more stable over the wide current variations one gets with modulators for AM transmitters, and the air wound inductors are MUCH MUCH less expensive.  Each core in the filter costs about $22.00 - I'm using the very best cores - HI FLUX cores (check out the CH777060).

The downside of the air wound inductors is the size.  They are a lot larger, and cannot be mounted against metal as can toroidal inductors.

Regards,

Steve

[WA1GFZ]

Yes you want to use a snubber but you just want to quench the high frequency overshoot and ripple on the drain pulse. I just worked on a switcher running at 300 KHz and due to wrong snubber design it was making a pile of crud up to about 80 MHz. We had to burn about 1/2 watt to get rid of it.
The problem was layout and transformer design. Snubber design is all about the little things you can't control like leakage inductance and interconnects.
I'm sure 15 ms TC hanging off 130 khz is well integrated. WW resistor is also acting like a RF choke. 

[steve_qix]

1/2 watt ain't bad :-)   If you find that snubber document, I'd like to have a look at it - definitely!

I may add 2 more MOSFETs to the output, to slightly increase the efficiency, and to bring up the power.  I think a good design value for this project is around 550 watts (or thereabouts) carrier power (output from the modulator), to yield 500 watts output from the RF amplifier.  So, I would like to design the modulator/PS to do up to 600 watts at full output.

Then, one just needs to choose a filter to choose the actual operational carrier power.  My particular filter gives me 450 watts output, but I have other filters which will let me run 500 watts and 400 watts (DC input) at carrier.

500 watts makes a good "building block" power level, either for standalone or for combinations of 500 watt building blocks.

Fun!

Regards,

Steve

[WA1GFZ]

Steve,
The 1/2 watt was in a 100 watt supply so you may dissipate more unless your transformer design is perfect and very short connections. Secondary side needs to be just as good. Well read the app about 10 times and notice the time constant you need. I think we had 4700 or 10,000 pf and 2kohms in our case.

[steve_qix]

Hi - thanks for the comments, etc. so far.

I am about to release Rev D of all of the Pulse Width modulator output section boards, and the PWM generator board.  I am currently out of PWM generator boards, so this is a good time to make minor changes.

On the PWM generator, the main functional change I'm looking at so far is a 50 ohm, low impedance 12V output for the pulse width signal, in addition to the medium impedance TTL output that's part of the current rev.  This low impedance output will allow the PWM signal to be cleanly transmitted over many feet of coaxial cable, in the event the PWM generator is not located right next to the PWM output (which is often the case).  The low impedance output will be the output used with all of the rev D pwm output boards.  The signal is converted to standard TTL at the output board.

I will probably make changes to the current output board - heavier etch runs, and possibly add another MOSFET to the board for higher current capability.

The new PWM modulator output section will get a new board, as none currently exists.

I am also going to create a "sequencer" board, which will implement complete transmit/receive control for antenna relays, receivers, VFOs, etc.  The board will implement a "first on, last off" type of system - the first thing switched (like the antenna and VFO and driver) will be the last thing turned off.  This will give an orderly transition from transmit to receive, without any glitches or "overlaps".

Any other suggestions, etc. would be appreciated.

Thanks and Regards,

Steve

[WA1GFZ]

How about the pwm board you replace the 1n34 with an opto to flash a led if you are hitting the negative peak limiter. You might be able to use the led in the otpo in place of the 1n34. I use a led in the 160 meter rig and 75 meter rig modulators. That is a modification I plan to make.
Output board I'm not sure you need so many "damper" diodes.
The pwm output chip Z is low but the voltage swing is high. I wonder if you could just reduce the resistor divider values?? I was thinking of a load resistor at the output boards with a lower series resistor off the pwm. This would save adding another ic or transistor. fc   

[W1VD]

1) Balanced / unbalanced line level audio input. Maybe eliminate phase select switch...most folks probably already handle this before the transmitter.

2) Can the pdm chip drive more than one output board? If so, maybe some extra pads. Thinking of the possibility of multiple output boards with multiple filters powering multiple TX modules.

50 ohm between pdm chip and output board(s) would be a good idea.

 

[WA1GFZ]

I agree with Jay on the switch. Heck a field of jumpers would be fine.
Jay look up the chip it has a fairly low output Z but the voltage swing is up around 12 volts. I'm going to drive two boards and just put a load resistor at each output board and a lower series in the voltage divider. I have about 1 foot of shielded wire to each output board. I'll load it and increase current until the waveform is clean. I would think the opto input z high so the load would just match the shielded wire z for a clean pulse. another option would be to use an 74AC00 and invert twice to get a strapping 5 volt low Z pulse.
This would allow 3 outputs. AC gates are 25 ma source / sink and faster than s logic using less power.