The AM Forum
December 11, 2024, 07:52:30 PM *
Welcome, Guest. Please login or register.

Login with username, password and session length
 
   Home   Help Calendar Links Staff List Gallery Login Register  
Pages: [1] 2   Go Down
  Print  
Author Topic: Looking for a board from Rod's PWM design  (Read 22719 times)
0 Members and 1 Guest are viewing this topic.
w9jsw
Two shots of Whisky
Contributing
Member
*
Offline Offline

Posts: 599



« on: December 04, 2023, 08:49:28 AM »

Does anyone have a board of this? If not, I will make some.

John



* ClassD_E modul.jpg (118.05 KB, 1000x583 - viewed 282 times.)
Logged
w9jsw
Two shots of Whisky
Contributing
Member
*
Offline Offline

Posts: 599



« Reply #1 on: December 04, 2023, 02:27:46 PM »

I already have a board from Nigel's design. Made them quite a while ago. Which one is preferred? I like this one due to lower parts count. LPF seems more straightforward.

John
Logged
w9jsw
Two shots of Whisky
Contributing
Member
*
Offline Offline

Posts: 599



« Reply #2 on: December 05, 2023, 11:23:04 AM »

Got tired of waiting... Grin

Actually, i really like making boards.


* XRay.png (86.66 KB, 1022x772 - viewed 197 times.)
Logged
vk3alk
Member

Offline Offline

Posts: 264


« Reply #3 on: December 05, 2023, 04:38:10 PM »

Hi John..

Your board looks great ...
Would suggest squeezing a .1uF capacitor where C10 is .... maybe moving that 10uF back up the line...
Also would change C9 to a 100uF low ESR type...
There is so much squarewave switching going on at that point an ordinary electrolytic might not last the distance over time...
That area is really crowded and the 100uF capacitor is larger in size then a 10uF one....

Wayne
Logged
w9jsw
Two shots of Whisky
Contributing
Member
*
Offline Offline

Posts: 599



« Reply #4 on: December 06, 2023, 12:38:18 PM »

Thanks Wayne!

I recall some discussions on diodes in the charge pump area. Are the 1N914s adequate here?

Made the changes you suggested. Also added another 0.1uF at the other end of the 15V rail by U1.

This a good selection for low ESR? 100V rating enough?

Is there any special requirements for C11? I thought so, so provided a large footprint for a 2220 Mica one. Ceramic would be much cheaper.

https://www.mouser.com/ProductDetail/Panasonic/EEU-FR2A101B?qs=u16ybLDytRZBZOsu6DbX5Q%3D%3D

John


* XRay.png (88.09 KB, 1016x762 - viewed 151 times.)
Logged
vk3alk
Member

Offline Offline

Posts: 264


« Reply #5 on: December 06, 2023, 11:56:16 PM »

Hi John...

Hopefully I'm not going to bug you too much but its a good discussion anyway....
The area around C9 is really crowded so maybe using a 10uF low ESR is better for artwork design...
I use a 100uF in all my PWM and the leads of my capacitors are thicker too so soldering them in on thin tracks is a bit of a pain....
A good place for C10 10uF is over at pin 9 there would be more room around there....and the holes over at pin 3 place the .1uF....

1N914 are fine across resistors R12 and R13....

You also need an energy return path from the PWM filter between L1 and L2 is OK and the Power supply or the Drain of the upper FET...say a MUR160 .... they are smaller then the DSE130-6...
Just like Steve uses on his PWMs....

I checked one of my transmitters that uses this type of PWM and have used a green cap for C11 but maybe a poly would be better choice....
Noticed that the resistor in series with this capacitor is showing signs of over heating and needs to be replaced....so when doing your artwork just extend that ground lead a little and have 2 holes so that a 1 watt can be installed ...

Also noticed at the output of the UCC IC pin4 which is 5 volts I think not sure....
I know Steve in one of his designs had a switch enabling the output to be either 5 or 12 volts...
Well that shows it going to the High side input of the 2110 while the low side is going to be at 12 volts controlled by Q1...
Never done that before so not sure whether the 2110 will like 5volts on one and 12volts on the other...
Just guessing would say its fine .....
Since your preparing your artwork would use the circuit that have uploaded etc:

Thats enough....


Wayne


* PWM DRIVER INPUT.jpg (62.66 KB, 640x480 - viewed 193 times.)
Logged
w9jsw
Two shots of Whisky
Contributing
Member
*
Offline Offline

Posts: 599



« Reply #6 on: December 07, 2023, 07:45:41 AM »

OK, Got it. Thanks so much.

John

* KQ6F PWM Schematic V1.0.pdf (85.68 KB - downloaded 101 times.)

* XRay.png (92.77 KB, 1026x771 - viewed 151 times.)
Logged
ua3ral
Member

Offline Offline

Posts: 15


« Reply #7 on: December 07, 2023, 02:06:43 PM »

C8 - Huh??
Logged

Yuriy
Когда Маркони собрал радио он услышал голос А. Попова!
vk3alk
Member

Offline Offline

Posts: 264


« Reply #8 on: December 07, 2023, 04:14:08 PM »

Yes C8 ?

Missed that ......
Its good to have others looking at things too thanks....
Have no idea why thats there so would omitt that capacitor and cannot read the value as well....


Wayne


Logged
vk3alk
Member

Offline Offline

Posts: 264


« Reply #9 on: December 07, 2023, 04:39:05 PM »

Just one other thing too...

The top FET looks like a Cree device and the lower one a IRFP240....
This PWM requires deadtime to survive and lives really on the point of death....
You could play around with R12 and R13 but mixing FETs with different Gate parameters is probably not the way to go....
Perhaps you can redraw the circuit again and make them say IRFP260 ....

Deadtime is not a long period of time....not sure exactly what it is but just say 100nS....
This might sound a little silly..... light travels 13inchs appox in one nS so at night when you take your torch outside to see whats making noises in your tree light has travelled 100 feet... Grin

Thats probably enough from me...


Wayne
Logged
w9jsw
Two shots of Whisky
Contributing
Member
*
Offline Offline

Posts: 599



« Reply #10 on: December 08, 2023, 07:20:50 AM »

Removed C8 and made the FETs the same IRFP260N.

W1DAN has requested to add a cap and R to the SS line. The data sheet suggests a .015uF cap and no R. What say ye?

Going to do another schematic using this board as a component, showing the rest of the circuitry to make it a full implementation. That way I can show the MUR160 diode from the LPF to the power supply.

Really appreciate the help, guys. This is all new to me.

John
Logged
kn4vxc
Member

Offline Offline

Posts: 2


« Reply #11 on: December 08, 2023, 02:32:08 PM »

Hi John...

Hopefully....

You also need an energy return path from the PWM filter between L1 and L2 is OK and the Power supply or the Drain of the upper FET...say a MUR160 .... they are smaller then the DSE130-6...
Just like Steve uses on his PWMs....

Wayne


Wayne, can you say more about what you mean when you say energy return path? What energy? Are you saying this is accomplished by a diode between the drain of the lower FET (output) and the drain of the upper FET (power supply)? Or something else?

This is all new to me too.

Leo

Logged
vk3alk
Member

Offline Offline

Posts: 264


« Reply #12 on: December 08, 2023, 04:54:03 PM »

Sorry John what do you mean by the SS line ?
Logged
W1DAN
Contributing
Member
*
Offline Offline

Posts: 905



« Reply #13 on: December 08, 2023, 09:54:33 PM »

Folks:

SS is the chip's Soft-Start pin. I am not requesting it but rather offering it as an interesting option. Been many years since I used my old Class E rig, but I would have used the UCC25701 (before the UCC25702) datasheet value for the Soft-Start cap. Might have also added a resistor across the cap. Either way, the result was a growth of the PWM pulses to ramp up the filtered DC output feeding the RF deck when going into transmit mode.

On this relatively early Class E transmitter (I believe was the third SS rig on the air in New England), I drove it with an FT1000 at about 20 watts carrier for a couple of hundred watts out on a 6-FET RF deck on 75M. The T/R switch was a DPDT microswitch where one half closed the transmit line of the FT1000 and the other enabled the UCC25701. The DC supply stayed on all the time AC was supplied. When going into transmit the carrier took about a 1/4 to 1/2 second to ramp up. I also used the RF amp's gate to drain capacity to couple the antenna signal to the receiver with about a 10dB loss at 75M, thus no T/R relay.

Dan
W1DAN
Logged
vk3alk
Member

Offline Offline

Posts: 264


« Reply #14 on: December 09, 2023, 05:04:59 AM »

Ohhh I see what you mean...

SS pin 14 on the UCC35701/2 meaning Soft Start

Without making this a long post I use a TL598 as a Voltage switching regulator and its pin4 is called Dead Time or DTC...
I use a RC time constant that sets the charge rate at 250mS.....just like a soft start...
Basically what this means is my output voltage to the PWM input rises from zero to whatever I have set it...
Looks like the UCC has this facility built in on pin 14...which is a nice feature... Cheesy

I was hoping people wouldn't ask me about the UCC because I do not use it as a PWM generator...
I make my own Generator using a HC4060 crystal locked and pick off 250Khz and then into a Triangle wave generator into a LM311 comparator...

Ok so the datasheet has a .015uF from pin 14 to gnd....
This must set a time constant internally that controlls the Duty Cycle from zero to max on startup...
You could experiment with different capacitor values and with your Nice Digital CRO see how long it takes ... or leave it open like what Steve QIX does in his circuit on his WEB site....
If you leave it open on startup I would say ... it would go to maximum Duty Cycle controlled internally straight away...

Would be an fun experiment and wouldn't take long to do....

Maybe there is something else more that only Steve knows about when he designed his PWM...
The UCC is a motor controller and maybe when used in a Audio Circuit its best to leave it as Steve does " open "... Smiley

Thats it I think....


Wayne








Logged
w9jsw
Two shots of Whisky
Contributing
Member
*
Offline Offline

Posts: 599



« Reply #15 on: December 09, 2023, 06:30:03 AM »

I have added a space for the SS cap for those who want it. If you need a R as well, you can piggyback it.

I think the board is done.

I still plan to use the Cree part to keep the implementation true to Rod's original design. He did that for a reason, and it worked fine.

Please note that the PWM will work with any PA topology, provided the LPF is designed to accommodate its load impedance, not just a H-bridge as in Rod's case. My board is also an h-bridge. 

The LPF in Rod's rig is a Chebyshev design to work into the load Z of his PA which is 3.9 ohms.  You will need to recalculate this based on the load Z of whatever board you mate it with. Mine is a higher Z and so the LPF values will be different. Tonnes has a good writeup on how to design the filters.

Also, on the topic of the use of a MUR160 diode from the LPF to the B+ supply.  Rod used a sequencer circuit for enabling the PWM and RF drive to the PA that eliminates the need for a recovery diode.

John
Logged
w9jsw
Two shots of Whisky
Contributing
Member
*
Offline Offline

Posts: 599



« Reply #16 on: December 10, 2023, 09:05:22 PM »

The board switching frequency is 100khz. The assumed PWM board output impedance is 0.5ohms. Rod's PA was 3.9 ohms. Here is a graph of Rod's output filter. He added the 0.05uF bridge cap to cause the deep null at the switching frequency.

I don't know much about this yet. I ran the plots with 2 different PA impedances - 8 ohms and 20 ohms. 8 ohms was still good and passable, I think. 20 ohms was not, IMO.

WA1QIX describes filter design in one of his PWM papers. He says he uses the same input and output impedances in his calculations, and it works fine. So, I surmise a low impedance source driving a filter designed with a higher impedance must not cause too much of a problem? Anyone have an opinion? I plan to discuss this with Steve sometime on a QSO.

Boards ordered.

Did some comparisons. BOM cost for this board is around $70. BOM cost for the Nigel board I made a while ago is around $38. (Second board pic)


* Plot of Rods Filter.pdf (113.71 KB - downloaded 93 times.)

* XRay.png (93.16 KB, 1030x770 - viewed 131 times.)

* PWM-pic.JPG (58.8 KB, 451x336 - viewed 146 times.)
Logged
vk3alk
Member

Offline Offline

Posts: 264


« Reply #17 on: December 11, 2023, 04:12:37 PM »

Hi John..

100Khz sampling rate is a little low...
Steve runs at 160Khz....
I would copy his component values and change on your circuit design...
My sampling rate is 250Khz and you will notice a big different when running music through 100 and then 160/250....
As far as the output filter goes would leave that for others to design their own as people will use different loading in their TX modules...

Wayne
Logged
w9jsw
Two shots of Whisky
Contributing
Member
*
Offline Offline

Posts: 599



« Reply #18 on: December 12, 2023, 08:55:20 AM »

I will play with the switching frequency when I get the boards.

Wayne, how do you design LPFs? I ran elsie using a larger PA impedance and I see huge numbers for the inductors. 334uH and 236uH. Are those reasonable values to build and will I have core saturation issues?

The reason I used 20.6 is that my current H-Bridge board runs at 93V @ 4.5 amps to give me 375W out at 89% efficient. 20.6ohm impedance.

See the elsie plots attached.

See next post for schematic...

The first plot is using an input impedance of 0.5ohms. I get a -10DB transmission level.

Running the plot with the same impedance on the input and output, it looks ok.

Adding 0.01uF across the second inductor gives me a very nice null at 100khz.

* Plot of 20.6 Filter.pdf (114.14 KB - downloaded 85 times.)
* Plot of 20.6 Filter same impedance.pdf (114.02 KB - downloaded 113 times.)
* Plot of 20.6 Filter same impedance with null.pdf (113.73 KB - downloaded 101 times.)
Logged
w9jsw
Two shots of Whisky
Contributing
Member
*
Offline Offline

Posts: 599



« Reply #19 on: December 12, 2023, 08:57:10 AM »

Here is the schematic from the prior post.


* schematic-20.6.png (26.06 KB, 1088x352 - viewed 161 times.)
Logged
vk3alk
Member

Offline Offline

Posts: 264


« Reply #20 on: December 12, 2023, 05:25:57 PM »

Ok on playing around on your switching frequencies when you get your boards back..

On your H Bridge what FETs are you using...
Also the drive inductors and turns ratios and the output transformer etc:
20 ohm is too high and your seeing the problems that come along with it....
On the modules I use looking into 50 Ohms changing the turns ratios on the output transformer I can go down to 10 ohms and below that the efficiency drops away....
For me there was quite a bit of playing around to be honest... until I stumbled ( and I mean stumbled ) on the right combination of FETs and output transformer....
I don't us binocular type cores for the output..
There very limited in the turns ratios like 1:1 1:2 1:3 etc....
But if you use Toroid Cores than all kinds of turns ratios are possible like 1:1.8 or 1:2.2 etc: fine turning things....
If you want use 2 modules together in an adder circuit where each module has to look into 25 ohms forget about using binocular cores...

I use SVC filer from Tonnes WEB site and find it great...
Its free of course and got a bit guilty using it all the time so bought his meter program....


Wayne



Logged
w9jsw
Two shots of Whisky
Contributing
Member
*
Offline Offline

Posts: 599



« Reply #21 on: December 13, 2023, 07:17:09 AM »

The FETs are Wolfspeed C3M0280090A. Output transformer is a T200 red toroid with a 9:13 turns ratio.

I am driving it with my hermes lite at 4W thru a quad of BN-43-202 binoculars with a 9:1 turns ratio on the gates. I am awaiting a board that I did the layout on that will provide a DDS based 2 phase signal generator. I will then redo the input side. I guess the plan should be to rewind the output transformer to try to get around 5-6 amps at 48V to get the impedance below 10 ohms. My bench supply gives fixed voltages of 24/48/96 so that is why I am using those voltages.

This first board was originally built to be modulated by a transformer driven by an audio amp. I may still go that way and just use the PWM board on a second transmitter. Or I could easily modify it to be like Rod's PA that is the companion to this board.

I have been wanting to try your h-bridge for 40M using the 150V MosFets. i would run that at 35V carrier, 70V max. I estimate that impedance to be around 8 ohms for about 200W of carrier.

Logged
vk3alk
Member

Offline Offline

Posts: 264


« Reply #22 on: December 13, 2023, 05:13:32 PM »

I don't know how far I can help...

Those CREE things didn't go well for me and put them in the too hard basket.......
Although they work well using Push Pull....

Your output cores windings seem a good start but the inputs to me are far off.....
Would start by using 4:5 .... and go by the waveform on the gates....
Toriodal cores may give you more room etc: 50 size is OK...


Wayne

Logged
w9jsw
Two shots of Whisky
Contributing
Member
*
Offline Offline

Posts: 599



« Reply #23 on: December 23, 2023, 06:56:36 AM »

Don W9BHI found some errors in my schematic. Thus, the boards I had made are scrap. I will hold off on sending for another run till after the holidays. Attached are the corrected files.

To be clear, the error was mine. Rod's schematic is correct and has been working for him for a number of years. I made an error in R4/R6/R7 connections when I transcribed his schematic into Kicad.

Thanks, Don, for letting me know about the error!

* KQ6F PWM Schematic V1.1.pdf (88.61 KB - downloaded 94 times.)

* XRay.png (91.49 KB, 1018x770 - viewed 134 times.)
Logged
steve_qix
Contributing
Member
*
Offline Offline

Posts: 2598


Bap!


WWW
« Reply #24 on: December 27, 2023, 04:25:32 PM »

Very interesting !!!

Someone mentioned a return path back to the power supply is needed at the input to the PWM filter.  This is correct.

However, in this particular design, one is there "by accident" (or maybe on purpose, and someone was very clever :-) - I don't know).

Ok, Since this is a high-side modulator, energy is stored up in the first inductor of the PWM filter when the high MOSFET is turned on - the voltage is pulled high.  When the modulator high side MOSFET is turned off, the energy is quickly released.  The voltage would fall WELL below 0V (and there would be considerable ringing), except for the fact that the lower MOSFET contains an intrinsic diode (also called the body diode).  This diode will conduct the energy back to the (negative side) of the power supply.

Just a note: Normally, when MOSFETs are used in the RF amplifier, the LOW side MOSFET in the PWM modulator is not needed.  Why? Because MOSFETs, when driven by an RF source, will conduct, even when there is NO VOLTAGE ACROSS THEM (I call this the class E "gift").  Therefore there will be sufficient current flowing, even with 0 volts across the RF devices, to properly discharge the PWM filter.  With a non-MOSFET RF amplifier (like a tube), an active pull down is required.

I think someone mentioned something else related to a protection diode at the output of the PWM filter, going back to the power supply.  This diode is there in case the RF amplifier suddenly does something funky, like radically change its current draw due to loss of drive, antenna failures, etc.  Such a condition could possibly set up an odd resonance in the filter, and could potentially result in a damaging voltage spike.  The protection diode from the filter output back to the power supply will prevent this from happening.

Good stuff happening here !!!  Interesting discussion.
Logged

High Power, Broadcast Audio and Low Cost?  Check out the class E web site at: http://www.classeradio.org
Pages: [1] 2   Go Up
  Print  
 
Jump to:  

AMfone - Dedicated to Amplitude Modulation on the Amateur Radio Bands
 AMfone © 2001-2015
Powered by SMF 1.1.21 | SMF © 2015, Simple Machines
Page created in 0.06 seconds with 18 queries.