The new Easy Class H Board kicks BUTT!

[steve_qix]

Well, I got the new Easy class H modulator boards back from ExpressPCB.  It took me about 1.5 hours to stuff the board (including gathering all the parts), and everything came up the first time.

The on-board implementation (4 modulator MOSFETs) will support RF amplifiers up to 5 Amperes at 30 volts carrier out.  The board could be used with higher voltages (up to 45 volts DC at carrier at 4 amperes).

Here is the board mounted on a 12 x 6 heat sink, and there is more than enough room to built the entire RF amplifier on the same heat sink (which is what I'm going to do), creating a compact, 150 watt (input - about 135 watts out) transmiter.

I do have extra boards if anyone is interested.



Here is the latest schematic.  It is also available as a .PDF at http://www.classeradio.com/easy_class_h_rev_b.pdf

[n2ry]

NICE !!!!!!

[steve_qix]

The .pdf of the modulator (located at: http://www.classeradio.com/easy_class_h_rev_b.pdf) is MUCH easier to read.  Schematics converted to .jpg files don't do real well unless they're fairly small 

[KD6VXI]

The .pdf of the modulator (located at: http://www.classeradio.com/easy_class_h_rev_b.pdf) is MUCH easier to read.  Schematics converted to .jpg files don't do real well unless they're fairly small 

What's the cost on the boards, and assuming one had to source the parts, a ballpark.

This looks COOL.



--Shane

[K6IC]

Looks great Steve !
May be interested in a board.
Thanks   Vic

[WA1GFZ]

Steve,
Do you have a schematic error at the current sense? It sounded good tonight. fc

[steve_qix]

Steve,
Do you have a schematic error at the current sense? It sounded good tonight. fc

Hi Frank,

I don't believe there's an error, but please let me know what you see, so if there is something unusual, I can correct it !!

[KF1Z]

I see what you see Frank....

But it's not an error....

Steve,

I believe Frank ist talking about how the hall-effect sensor APPEARS to be in-between the two IRFP260n's.

But, in fact the fets are still in parallel., And the current from just one of the IRFP260n  is being monitored.

At least, I believe that's the idea.   

[WA1GFZ]

So if the FET not in the monitor circuit fails nothing will happen. I would think the Hall could be configured to handle both FETs
I do like the source follower. My design used a big pull up resistor.
I solved the crossover distortion the same way as you did with the top gate above bottom by a few volts. Loss stays the same just transfers it to the other FET.
I think it sounds cleaner than the PDM but can't say why. Low frequency response is very good. We had a lot of slop bucket so had to limit BW

[n2ry]

Steve,

I would be interested in a board also

Rich
n2ry

[steve_qix]

So if the FET not in the monitor circuit fails nothing will happen. I would think the Hall could be configured to handle both FETs
I do like the source follower. My design used a big pull up resistor.
I solved the crossover distortion the same way as you did with the top gate above bottom by a few volts. Loss stays the same just transfers it to the other FET.
I think it sounds cleaner than the PDM but can't say why. Low frequency response is very good. We had a lot of slop bucket so had to limit BW

Hi Frank,

The overload detection methodology gets real interesting    In this case, the output of one MOSFET is measured to allow the builder to configure from one to many MOSFETs without having to change any values in the overload circuit itself.  

In the unlikely event of a modulator MOSFET short circuit, the board-based monitor circuit would not sense this as an overload, as the output current and voltage would track (which is what this circuit measures).  A secondary circuit located in the power supply is used to shutdown the modulator in the event of such a problem.  But, it would be possible to monitor both MOSFETs on the board and such a change would be quite simple.

It is very interesting comparing the PWM to the class H modulator.  The biggest difference I see is the lack of filtering in the class H modulator.  All of the PWM systems have a multi-pole aliasing filter, and of course the output filter.  All these add phase shift, and of course they compromise the high frequency response which is what the filters are supposed to do.  There are also no ripples in the frequency response of the class H modulator.

These differences are audible in headphones, although they are quite subtle.

On the design - that source follower after the voltage amplifier really solved a lot of problems!  Due to the variable nature of the load represented by the gates of the output MOSFETs (the capacitance varies, and the reverse transfer capacitance couples the output to the input), as the frequency is increased, the effects of the capacitance on a pullup resistor only voltage amplifier became quite pronounced !  And, the effect become increasingly worse as the voltage rose, and the current through the pull up resistor become lower and lower.  I observed this myself during the design phase.  So, I added the source follower and got rid of all those problems.

Further enhancements in this design come by running the voltage amplifier and source follower MOSFETs to the -12V supply.  This gives an effective increase in the voltage swing available to the gates of the modulator MOSFETs of 12 more volts.  This additional positive swing is recovered across ZD5 (bypassed by a 10uF capacitor), and effectively allows the gates to be driven sufficiently positive with respect to the source to allow the output MOSFETs to achieve saturation at the full supply voltage.   So, the maximum positive peaks are realized and none of the supply voltage is wasted due to the threshold voltage of the MOSFETs.

The negative peak limiter is an improvement over previous designs in that the limiter supplies very little current - leaving the output MOSFETs to carry the load.  The limiter is also adjustable.  Previous designs used a low voltage power supply and a diode to directly supply the RF amplifier.

Anyway - a few design ramblings     I had fun doing the design !

Sorry for the old-buzzard post.

Regards,

Steve

[KA1ZGC]

Schematics converted to .jpg files don't do real well unless they're fairly small 

Schematics are best exported to PNG format. JPG's compression algorithms are geared towards photographs, with vivid colors and gradients thereof. Sharp lines and small characters get blurred and pixellated by JPEG processing.

Or, as the great philosopher Solo once said: "I don't think the Empire had wookies in mind when they designed her, Chewie". 

I don't think there's a single web browser or graphics viewer/editor in use today that doesn't support PNG.

[W2XR]

So if the FET not in the monitor circuit fails nothing will happen. I would think the Hall could be configured to handle both FETs
I do like the source follower. My design used a big pull up resistor.
I solved the crossover distortion the same way as you did with the top gate above bottom by a few volts. Loss stays the same just transfers it to the other FET.
I think it sounds cleaner than the PDM but can't say why. Low frequency response is very good. We had a lot of slop bucket so had to limit BW

Hi Frank,

The overload detection methodology gets real interesting    In this case, the output of one MOSFET is measured to allow the builder to configure from one to many MOSFETs without having to change any values in the overload circuit itself.  

In the unlikely event of a modulator MOSFET short circuit, the board-based monitor circuit would not sense this as an overload, as the output current and voltage would track (which is what this circuit measures).  A secondary circuit located in the power supply is used to shutdown the modulator in the event of such a problem.  But, it would be possible to monitor both MOSFETs on the board and such a change would be quite simple.

It is very interesting comparing the PWM to the class H modulator.  The biggest difference I see is the lack of filtering in the class H modulator.  All of the PWM systems have a multi-pole aliasing filter, and of course the output filter.  All these add phase shift, and of course they compromise the high frequency response which is what the filters are supposed to do.  There are also no ripples in the frequency response of the class H modulator.

These differences are audible in headphones, although they are quite subtle.

On the design - that source follower after the voltage amplifier really solved a lot of problems!  Due to the variable nature of the load represented by the gates of the output MOSFETs (the capacitance varies, and the reverse transfer capacitance couples the output to the input), as the frequency is increased, the effects of the capacitance on a pullup resistor only voltage amplifier became quite pronounced !  And, the effect become increasingly worse as the voltage rose, and the current through the pull up resistor become lower and lower.  I observed this myself during the design phase.  So, I added the source follower and got rid of all those problems.

Further enhancements in this design come by running the voltage amplifier and source follower MOSFETs to the -12V supply.  This gives an effective increase in the voltage swing available to the gates of the modulator MOSFETs of 12 more volts.  This additional positive swing is recovered across ZD5 (bypassed by a 10uF capacitor), and effectively allows the gates to be driven sufficiently positive with respect to the source to allow the output MOSFETs to achieve saturation at the full supply voltage.   So, the maximum positive peaks are realized and none of the supply voltage is wasted due to the threshold voltage of the MOSFETs.

The negative peak limiter is an improvement over previous designs in that the limiter supplies very little current - leaving the output MOSFETs to carry the load.  The limiter is also adjustable.  Previous designs used a low voltage power supply and a diode to directly supply the RF amplifier.

Anyway - a few design ramblings     I had fun doing the design !

Sorry for the old-buzzard post.

Regards,

Steve

Hi Steve,

Superb work as usual!

Now on to the WA1QIX FET audio driver circuit! You have a few guys foaming at the mouth ready to use that thang, myself certainly included.

Seriously, as I indicated to you on-air this past Saturday, I am really eager to install this device within my rig.

73,

Bruce

[WA1GFZ]

I got around the high frequency response problem with dc coupled NFB
I was good out to over 20 KHz.
I like the source follower better with peak charged gate

[WA1GFZ]

Bruce the source follower drives the grid with a pull down to bias voltage.
Then just need a second source follower 180 degrees out of phase for the second tube. I think the 11N90 should be up to the job.

[K4QE]

Hey, Steve,

I know I'm kinda late to the party here, but do you still have any of these boards available?

73, Tony

[steve_qix]

No, there are no more class H boards.

It's better to use pulse width modulation anyway, and there are plenty of these boards and parts around.

Regards,

Steve