Double vs single phase class D/E driver

[AMLOVER]

Hi to all,

After some searching around net I couldn't find any information about this subject.

Is there any benefits to drive multiple class D/E bridges/pallets in series with both phases (A) in comparison with only one (B)?

What would you suggest between A and B comparison? What are the pos or neg in each topology?

I have used till now only type A (both phases) but I would like to know what if I use only the positive phase to drive few pallets/bridges in series?

Stefano

[AMLOVER]

Is there any reason that my previous post is disappeared?
Thank you.

[Pete, WA2CWA]

Is there any reason that my previous post is disappeared?
Thank you.

You were posting the same topic in multiple forums.
I'll keep this one under the Class E forum and delete the other one.

[AMLOVER]

I should delete the wrong one but didn't know the procedure.
Thank you.

[Pete, WA2CWA]

I should delete the wrong one but didn't know the procedure.
Thank you.

It's already done. You don't have to do anything.

[vk3alk]

Hi Stefano....

To me they would both work as they are adder circuits....
B would be easier to build as it is just one phase.....

Do you connect 4 modules together with your configurations .....


Wayne

[AMLOVER]

Hi Wayne,

Four modules are connected in series. Each module however consists two bridges and so eight bridges are connected definitely in series.
With the right carrier voltage and current can get closed to 6.25 Ohm from each bridge and with 1:1 output transformers the final result is very closed to 50 Ohm plus some leakage inductance which is cancelled in the input of the output low pass filter. Somehow like in the picture.
I am thinking to drive eight sigle bridges but with the same phase, it seems to me also much easier for some reasons but I had some doubts. I appreciate your positive opinion and give it a try. A plan is attached.

Thank you for supporting us in this kind of projects.
Stefano

[vk3alk]

Wooww there is a lot of work there.....very good...
Method B would be easier but since this TX is working maybe best to leave as is....
Perhaps in your next transmitter try then....

I have a replacement for the IPP530N50 FET which will post to the Forum soon...
You could give them a go on say 40M .....


Wayne

[AMLOVER]

Attached is a better schematic with the old (two phase drivers) vs the new (single phase driver) designs.
I am wondering if with the single phase the total combined output misses some balance or decreases the efficiency.
Steve combines in his E class amplifier few modules driven with one phase but he also combines other pallets with the opposite phase.
What would be the disadvantage if he wouldn't add the opposite phase driven pallets other then getting less output power?
I'm in source for VFO DDS with TFT screen, advices or a source to provide are welcome.

Stefano



 
 

[AMLOVER]

Correct schematic

[vk3alk]

Hi Stefano...

I understand what you are doing.....
For the moment forget about Steve circuit...

You have 8 modules .... each one is a transmitter in its own right producing say 90% efficiency and looking into 6.25R...
From your squarewave generator with a 50:50 DC apply that to each modules input ( single Phase )....
Then you just feed a wire through all the transformers grounding one end and the other to your LPF etc: .... ( new design )..

You have already done this with your OLD system but complicated it by using different phases....

As you already know its an adder to increase your output power level...


Hope this helps...


Wayne

[AMLOVER]

Hi Wayne,

It is exactly as you described it, adders. Of course your support helps me to decide the final design.
From the DDS square wave output, 8 same length cables will feed the single phase to the inputs of 8 drivers (IXDD614CI).
PCBs and supplies on hand. Looking only for DDS, power/swr meter with swr and temperature protection.
No final decision though for mosfets, I'll expect your suggestion.
Thank you.

Stefano

[vk3alk]

I have never added 8 modules only 2 but the same principal exists I would say......

Can I ask why your using so many modules and what output carrier power you are hoping for....
Also can you supply the circuit diagram and what FETs they are....as well as the band of operation...
I'm just a little concerned about the distances between your modules for the secondary winding but I suppose time will tell...
Have you tested the modules and are you happy with their proformance etc:


Wayne

[WBear2GCR]

Assuming that 1/2 the drive is out of phase by exactly 180deg with the other, and then sent to separate modules,
and wired as Push-Pull, then there is significant reduction in even order harmonics...

                       _-_-bear

[AMLOVER]

Wayne,

Some people including me as getting older don't like high voltages. With 50V switching supply, 40% duty cycle (20V carrier level) you can achieve 150% assymetry modulation with 8 modules in series for about 500W carrier. It works like harm without ventilation.
The schematic is like the one in the attachment multiplied by any number.


Bear,

Hmmmm that's why almost everywhere are used both phases. It is the push pull even harmonic reduction benefit.
A 7th order Butterworth LPF instead of the usual 5th order could also help however in this part.


 

[vk3alk]

Hi Stefano...

It's true what Bear states about Push Pull but the circuit you provided is exactly what I do for 40M...and other bands...
Two separate H-Bridge modules feed in Single Phase with the outputs both having to see 25R connected as an adder to 50R...
So if you wanted to connect 3 other Groups of 2 totalling 8 you just continue the secondary along....
But be careful when feeding the wire not to get the phasing mixed up as cancellation will occur etc:..
You would have to change the turns ratio of the output transformer to 1:1 so that each H Bridge will see 6.25R though...
For 500Watts appox just one group of 2 should get close if all going well...
Your output transformer .... if those 2 cores are type 43 will get hot when modulation is applied...

Wayne

[AMLOVER]

Hi Wayne,

With only 2 modules in series you can get 260W carrier, 150% modulation peaks and this only when high voltage (150V) supply is used.
150V x 40% = 60V carr. level. P=Vcar^2/Rload and so 3600V/25Ohm=144W x 90% = 130W carrier per module/bridge.
With 50V supply at least 6 or better 8 modules are required.

Ferrites for modulation and rf output is a special case for this kind of projects.

Ferrites should fulfill few different conditions. Material adequate for the frequency, voltage/turn withstanding, amount of beads to achieve the adequate inductance and density/temperature issues.
My procedure is to first multiply the Rload X 8-10 times to find the required reactance for the frequency in use.
From V=sqr(PxRload) to calculate the voltage and multiplied by sqr2 to find the voltage peak value with modulation.
Per example FB-1020s can stand 10-12V/turn before they develop 20C temp increasement.
If the peak voltage per example is 60V then with a single turn at least 5-6 FB-1020s must be used.

With all those in mind, material and beads number in use are decided.

There are two ways to make it work, to use the right amount of beads or to use less amount but well ventilated.
Right amount of beads is preferable and so the next dilemma is how many turns to the right amount of beads.
One turn is always preferable but in the lower frequencies material 61 which is almost perfect has very low μ' =125 and for few uH big number of beads in series are needed.
Material 43 has however higher μ'=850 and with 1 or 2 turns to 1 or 2 of them we can get the required inductance BUT....
1 or 2 beads type 43 are not enough for the required voltage which is depended to power and Rload.
More power to higher Rload means higher voltage which means more beads are in need before they become hot.

My solution in this complicated case is the series/parallel connected ferrites.

Let say per example we need 6uH, each 1020-43 offers 3uH and 2 in series offer 6uH but 2 are not enough for the power in use and get hot.
If instead of 2 we use 4 x 1020-43 in series then we get twice the inductance we need (12uH).
In parallel connection with another 4 x 1020-43 (12uH), the inductance becomes what we need (6uH) but the nr of beads is 4x (8 beads instead of 2).
Almost more then what is needed for voltage and temperature withstanding.
In this case the smaller size (FB-7351) or the even smaller size (FB-4852) could be used and the total final space needed is not far from the 2 bigger ones..
With the right ferrite inside diameter for the wire in use, the leakage inductance which is more important on higher frequencies can be also minimized.
In the attachment are 3 series/parallel to another 3 just for example.
This helps if ventilation must be avoided.

Stefano

  

[vk3alk]

How interesting Stefano....

Sounds like you will have no problems getting this to work....
I have never been very good with output transformers.
My method is wiggling and giggling 
Trying different material or stacking until I get things to behave to a satisfactory level...

I must say that even at 40M to only get 130watts from a single module would disappoint me...
The attached photo shows an experimental 40M TX.
I was trying to get a H Bridge to work at 7Mhz....see the little coil at the input....
It did work and I used it probably for 2 years...
Ran the TX at 180 watts output with no problems except those 4 FB1020-43 cores got hot and required a fan ...

The next photo shows my current setup.
Two FB1020-61 cores and they run cooler ...
Under testing with modulation each module ran very well at 200 watts output....

But thats enough about me...
Will printout your reply and read while having a cappuccino coffee this morning....


Wayne

[vk3alk]

Hi Stefano

I had a nice cappuccino coffee and read your last posting and " I get it ".....
Series / Parallel winding of output transformer ......
I run higher voltages then you around 50 volts carrier level ....
Have attached another photo showing a different Transformer using FB114-61 cores..
Its trifilar-wound with 6 twisted turns ..... one pair going to the FETs and the other 2 connected in series for the output...
Its works very well but at higher power levels still gets hot....
But its only in parallel as you say or is it series .....
Have plenty of FT114-43 cores so will try to play around with series / parallel conbinations a little etc...

Another interesting thing is the resonanting coil for 40M...
You can get away with not using that coil too ....
But it depends on the FETs you use .... you need a low Gate capacitance....
I am using more turns on my FT50-75 input cores then you.... the secondary has 6 turns....
The inductance combined with the gate capacitance enables adjustment of the drive level although you cannot reach the peak...

Thats enough for now...

Ohh just one other thing....
Have never been able to achieve good efficiency under a 10R load...... it drops off quickly ...
Your loading needs to be 6.25R .....


Wayne

[AMLOVER]

Hi Wayne,

I hope to keep you good company meanwhile you enjoy one more morning cappuccino.

Efficiency shouldn't normaly related to Rload. Ohm's law doesn't make distinctions.

In audio most of the amplifiers are usually designed to see 4 Ohms Rload with high efficiency, look their tiny coolers without ventilation when they make 1000W/4Ohm easily. They can make it even better to 2 Ohms. If there is ventilation it is mostly to cool their power supplies.

So the efficiency to Rload factor in our case has probably to relate with the frequency in use.

If I understand well, Nautel in its NX50 (50kW) transmitter decreases each module's (2500W) Rload from 8.28Ω το 0.23Ω by using 6:1 turns ratio output transformer and then connects in series 0.23Ω x 20modules to get 4.6Ω before LPF output filter.

In this model they combine in series 20 modules to get 50kW output power with very good efficiency but in low frequencies - mw band.
A BC station very closed to my location transmits with NX100 with the same topology but 40 modules.

Typical values for NX50

1. B+ = 400 VDC
2. PAV = 163 VDC
3. MOD Depth = 145%
4. Number of RF modules = 20
5. Filter input Impedance = 4.17 Ω<25° or 4.60 Ω//+j9.87 Ω
6. Primary Turns = 6
7. PA impedance = 8.28 Ω//+j17.77 Ω
8. PA Power = 146 𝑉rms^2/8.28 Ω = 2574 watts
9. 2574 watts x 20= 50 kW

As the frequency goes higher some mosfets can't deal with fast switching high currents, some others can make it better and so they can be more efficient in lower also Rloads.

Nautel uses mosfets from APT, they seem to have higher Ciss but with the right drive to charge fast their gate capacitance (to around 40ns) they switch efficiently in very low Rloads.

I think that mosfets suffer on higher frequencies because they are built to work as switchers in supplies and later in audio industry, to very low or just low frequencies.

Nautel and few others managed to 'pull' them to work higher in the mw/sw bands.
The hams try even higher to 40 and lately to 20m.

http://www3.nautel.com/pub/Post%20NAB%202024/Design%20Considerations%20%20for%20AM%20Transmitters.pdf

As it has to do with the inductance in the input transformer for me is always preferable more ferrites then wire as it is also in the output transformer (single turn to fewer ferrites). Series/parallel though

It is better 2-3 stucked FB50-75 or a long bead instead of more turns to a single ferrite. This is also the law for quality in audio output transformers, better more iron with high gauss and less wire turns.

In my new project instead of ferrites, 40mm long 12.5mm diameter ferrite rods will be tested as input transformers.

https://gr.mouser.com/datasheet/2/150/FRPCS02034_1-2548765.pdf

Stefano