The AM Forum
March 28, 2024, 07:10:19 AM *
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] 3   Go Down
  Print  
Author Topic: Nigel's output transformer  (Read 33486 times)
0 Members and 1 Guest are viewing this topic.
M0VRF
Member

Offline Offline

Posts: 172


« Reply #25 on: February 03, 2020, 02:00:50 PM »

15W for a driver wow!

I use less than a watt and a single chip.

Sine wave drive is not for class E, you want 'very' squarewave.

Anyway if it works....

Regards.

J.

.
Logged
vk3alk
Member

Offline Offline

Posts: 238


« Reply #26 on: February 03, 2020, 07:18:58 PM »

Actually I have not thought of Sinewave drive now for years....
We are lucky today with all the great FETs available and new driver chips....

In the time ago when people used 1st generation FETs like the IRFP450 with large Gate Capacitance and Gate source resistance
the only way to charge and discharge the gate was with power say from a Amateur transceiver....

My first 40M TX was a Class E Sinewave driven and went well.....
To get the correct drive level and Duty Cycle all I had to do was increase / decrease the output power from the Rig...
It was very easy to do for me anyway......

In a way I Sinewave drive my 40M H Bridge Transmitter...
Because my H Bridge uses toroidal cores to drive the output FETs I have no harmonics left and have lack of drive...
To overcome this I resonant the input and which gives me a sinewave.....
It works out well for my IXDD chip because after adjusting the coil by compressing and expanding etc: I had to reduce the voltage to the chip down to 8 volts and only draws 350mA...
Its very happy  Smiley

Wayne
Logged
YB1AHY
Member

Offline Offline

Posts: 34


me


WWW
« Reply #27 on: February 04, 2020, 01:55:15 AM »

Hi Wayne


My transmitter was build in 2007, that is my first in class E still operate till now without fail.

I am interested in H bridge also, i was build one with no success. mosfets and output transformer creating excessive heat. I am interested on your design, i saw your picture of H bridge using small gate transformer. if i zoomed it in, it looks like you are using IXDD driver? My question is do you connect directly the output of IXDD to primary of transformer or there is capacitor in between?

I want to build H bridge again soon, i have bunch of used STW20NM50FC from friend


BR

agus
Logged
vk3alk
Member

Offline Offline

Posts: 238


« Reply #28 on: February 04, 2020, 03:14:05 AM »

Hi Agus...

What FETs do you use in your 2007 Class E ?....

I still have my 75M Class E TX....
Do not use it any more but it was reliable and never let me down....
It was a copy of Steves from his WEB site.....using 11N90s etc:.....
Because it was so reliable I cannot take it apart so its found a home at the back of my shack... Grin

STW20NM50 are a fine FET but are no longer manufactured unfortunately.....and are usable in H Bridge.....

Yes there is a capacitor from the IXDD to the primary of the 4 cores driving each FET...
Its a DC blocking capacitor thats all .... a 1uF Polyester ....otherwise the IXDD would have a direct path to ground and blowup  Embarrassed


Wayne





Logged
M0VRF
Member

Offline Offline

Posts: 172


« Reply #29 on: February 04, 2020, 05:11:41 AM »

40nC Qg, no surprise you need loads of power to drive them!

Plenty of devices under 10nC, no need for so much power.

I don't see the point in using old obsolete devices when there are far better spec, cheaper and more easily available devices around.

J.
Logged
YB1AHY
Member

Offline Offline

Posts: 34


me


WWW
« Reply #30 on: February 04, 2020, 07:15:28 PM »

Hi Agus...

What FETs do you use in your 2007 Class E ?....


Hi Wayne, and John


I m using IRFP460 configured as a current mode push pull class E. i made two module and i combine it, totally four mosfet. i believe that IRFP460 has a narrow opening drain on high band also hard to drive, so i need a lot of power. when i turned on drive quite a while without VDD on mosfet, it was warm to touch act like dummy load Smiley.

Another question on H-bridge, do you blocking dc between bridge then the output transformer or made it direct connection without capacitor?  My experiment was direct connect, what had happened is that transformer was getting hot. In H bridge i used IRFP350LC (gate enhancement). If you have web link to share, i will learn it appreciate for that.

At that time i was using PWM modulator, direct from mains 220 volt, i set carrier to 120 Volt it generated 250 Watt RF.

Thank you

73s agus

   
Logged
vk3alk
Member

Offline Offline

Posts: 238


« Reply #31 on: February 04, 2020, 09:43:05 PM »

Hi Agus...

Yes I DC block one side of the transformer just like K7DYY circuit ....
Those capacitors C16 C17 and C18 ......

The IRFP350LC FET is not one I would try though .... a bit hard to drive probably....


Wayne


* Senior K7DYY.jpg (109.83 KB, 1024x768 - viewed 804 times.)
Logged
km6sn
Contributing
Member
*
Offline Offline

Posts: 91


« Reply #32 on: February 08, 2020, 12:43:25 PM »

I am considering using a transmission line transformer on the output of a Class D deck to convert the 12.5 ohm deck impedance to 50 ohms before the low pass filter.
I know the transformer would have to pass at least 3x the carrier frequency.
Has anyone tried this?? Any actual experience with this?? Any ideas?? Thanks, Rod
Logged
KQ6F
Member

Offline Offline

Posts: 124


« Reply #33 on: February 08, 2020, 03:16:02 PM »

A few years ago I built a current-mode Class D deck using a transmission-line transformer in the output.

http://amfone.net/Amforum/index.php?topic=38429.0

Rod
Logged
N4LTA
Contributing
Member
*
Offline Offline

Posts: 1075


« Reply #34 on: February 08, 2020, 04:19:26 PM »

W1VD uses this on his class d rigs.  Not sure why you think it needs to pass three times the fundamental frequency?  He uses homemade 25 ohm coax and ferrite sleeves. The coax is made with 50 ohm RG 142.

Pat
N4LTA
Logged
N4LTA
Contributing
Member
*
Offline Offline

Posts: 1075


« Reply #35 on: February 08, 2020, 05:27:36 PM »

I am still trying to get a clear picture of Nigel's output transforner in my head.

He appears to have two 1:1 transformers wound on a single core.

Two of the windings are wired in series to form the output winding and two windings are wire to feed the two drains from a common Vdd
point (center tap?)

My first assumption is that this is much like a simple push pull output transformeras in my crude drawing. It does not appear to be connected that way. I have shown it jumpered as if it was.

It does not appeared to be connected as I have shown

Where am I screwing up?

Pat
N4LTA

Pat
N4LTA



* one.jpg (73.56 KB, 640x480 - viewed 689 times.)

* two.jpg (71.31 KB, 640x480 - viewed 704 times.)
Logged
vk3alk
Member

Offline Offline

Posts: 238


« Reply #36 on: February 08, 2020, 06:06:53 PM »

Yes its typical looking VMCD...
Just like Jays earlier stuff as per photo uploaded....
The unlabeled capacitor across the drains is appox 100pF....

Its only my opinion but of all the class's ... found VMCD my most unsuccessful and gave it away quickly....
Stray inductance and high spikes everywhere ... bias through the output transformer causing heat issues....

Is the TX your building up going yet .... have you applied power ?

What I might do is build up a 4 FET CMCD like my 2 FET one and see how that goes....


Wayne

* 100WclassD.pdf (24.29 KB - downloaded 355 times.)
Logged
KQ6F
Member

Offline Offline

Posts: 124


« Reply #37 on: February 08, 2020, 07:39:29 PM »

Wayne, et al -

Before you give up on VMCD you might want to consider something like the attached.  I've been running this successfully for 2 years now without any hiccups.  I call it Class D/E because it combines the two circuit topologies that provides the best of both, at least in my mind.  The stray C's and the drain C's are swamped by the 1000pf shunt caps and produce Class E-like half sinusoid waveforms that are free from the usual Class D spikiness.  But instead of the series-tuned output filter commonly used in Class E (requiring large unwieldy components), the filter here is a modified Butterworth.

One fairly important detail - the two halves of the center-tapped primary should be wound on opposites sides of the core.  This provides normal flux coupling but eliminates capacitive coupling between the two windings that would otherwise distort the waveforms.

The peak voltages on the drains are 4X the supply voltage (see attached) so high voltage MOSFETs are required.  That's the only downside to this configuration that I can see.

Rod


* ClassD_E PA.jpg (114.33 KB, 1000x755 - viewed 862 times.)

* classd_e_002.jpg (206.04 KB, 1000x750 - viewed 692 times.)
Logged
vk3alk
Member

Offline Offline

Posts: 238


« Reply #38 on: February 08, 2020, 08:17:10 PM »

Ohhh thanks very  much for those comments Rod....

I'm not an engineer so only rely on my own experiences.....
I did have a close look at Nigels photo and noticed those capacitors to ground .... I scratched my head wondering whats going on without giving it too much thought really...
You have 15 volts to the IXDDs...maybe thats due the FETs your using requiring a little more drive energy....

Modified Butterworth Filter .... what does that actually mean ?
When I saw the waveforms using CMCD looking Class Eish my main reaction was " well thats less demands on the LPF "....

Pat .... maybe you should build this....

I've made up a board for a 4 FET CMCD but may consider yours Rod instead....only because have never built this type before and would be interesting to do...


Wayne

Logged
N4LTA
Contributing
Member
*
Offline Offline

Posts: 1075


« Reply #39 on: February 08, 2020, 09:38:12 PM »

i have blown up a set of FETs twice so far with my set up. What  my question is  - is that Nigels photos appear connected differently than my connection.

Maybe I am looking at them incorrectly.

I very well may try current mode with my board. I may try it first with a toroidal transformer rather than using a transmission line transformer as I can't get the transmission line transformer on my board. Not enough room.

I am using W1VDs circuit using a 74F74N  to generate the 180 degree out of phase signal. I have laid out a double sided board using Steve's circuit, but the bare board will not arrive until next week and needs to be tested.


"One fairly important detail - the two halves of the center-tapped primary should be wound on opposites sides of the core.  This provides normal flux coupling but eliminates capacitive coupling between the two windings that would otherwise distort the waveforms."

 This may be part of my problem - On my latest transformer I wound a primary wingding with a center tap and did not split the windings on each side as you state and that Nigel's seems to have done.

I am using  Nigel's recommended drivers. They seem to fail when the FETs fail. Changing the SMT parts are a PITA.

Right now - I am not confidant with my signal generation/phase splitting system or my output transformer .

Pat
N4LTA
Logged
km6sn
Contributing
Member
*
Offline Offline

Posts: 91


« Reply #40 on: February 08, 2020, 11:04:03 PM »

Pat, can you post a picture of your output transformer?? Rod
Logged
N4LTA
Contributing
Member
*
Offline Offline

Posts: 1075


« Reply #41 on: February 09, 2020, 10:12:14 AM »

Here is the transformer and a photo of the rf board and a photo of the PWM generator. The modulated Vdd is fed to the center tap at the top of the transformer.

I originally had 470 pF caps to ground. I'll replace them . I will try Rod's circuit with my installed drivers and his output filter. I'll also modify my transformer.Rod, How do you develop you bi phase drive? Do you use a simple flip flop like W1VD or a more complex circuit like Steve with adjustable duty cycle?


In the transformer photos, a blown FET and driver have been removed.  I'll change them this afternoon.


Pat
N4LTA
 


* rf1.jpg (103.38 KB, 640x480 - viewed 674 times.)

* rf2.jpg (126.85 KB, 640x480 - viewed 662 times.)

* rf4.jpg (97.61 KB, 640x480 - viewed 671 times.)
Logged
N4LTA
Contributing
Member
*
Offline Offline

Posts: 1075


« Reply #42 on: February 09, 2020, 10:19:20 AM »

The PWM generator



* gen1.jpg (128.17 KB, 640x480 - viewed 688 times.)
Logged
N4LTA
Contributing
Member
*
Offline Offline

Posts: 1075


« Reply #43 on: February 09, 2020, 03:11:08 PM »

Here is my new output transformer wound per Rod's drawing (I think)

First photo is the primary - 6 turns on each side center tapped at the top. (#14 enamelled wire)

Second photo is with the red #14 secondary 11 turns spaced equally over the primary.(red solid #14 insulated THHN wire.

Modulated Vdd at the top center tap and the other ends to the drains D1 and D2


Pat
N4LTA


* t1.jpg (41.87 KB, 640x480 - viewed 633 times.)

* t2.jpg (103.63 KB, 640x480 - viewed 629 times.)
Logged
km6sn
Contributing
Member
*
Offline Offline

Posts: 91


« Reply #44 on: February 09, 2020, 04:00:44 PM »

Pat,
I am not sure because I have not built and tested one yet. That said,
referencing the original picture in this post:

1. it looks to me like the primary windings are NOT one continuous center
tapped winding.
Look down at the solder solder lug where the DC connects; the primaries approach
from opposite directions. Much like the bare wire secondary, if they were extended UP
through the core, they would  need a "bridge" like the secondary winding.

2. The secondary appears NOT to be one continuous winding; if it were, the wire
  "bridge" would cross over the top of the core. It looks like the secondaries are
   opposed windings.

I hope that is useful in some way.
Regards, Rod
Logged
N4LTA
Contributing
Member
*
Offline Offline

Posts: 1075


« Reply #45 on: February 09, 2020, 05:35:01 PM »

Rod

I don't think they are (Nigels) and that is why I have been asking questions for a couple of weeks -  It appears that his change directions at the tap which I think make the dot end at the top on one winding and the no dot at the top on the other - That makes sense as the center taps should be one dot end and one no dot end.

For the output winding - it makes sense also - I am assuming that if the winding direction changes - one side upper will be a dot and the other upper side is a no dot. If that holds trus- it was an ingenious way to wind and make the hookup easier. I wish Nigel would let us know if that is correct.

I am a electrical power engineer, but certainly no transformer winding person.

I am trying to build 1/2 of your design for starters. I have a board for two FETS on each side, but I need to get something running for testing.

I have rewound my transformer like your circuit (I think) and added the 1000pF 1KV SM caps, replaced the blown FET and driver. Here are some photos. No power has been applied until I get the Butterworth filter built. If you see anything screwed up , please let me know. Thanks for your help.

Pat
N4LTA



I have rewound mt transformer to be like you


* front.jpg (115.31 KB, 640x480 - viewed 632 times.)

* front2.jpg (104.97 KB, 640x480 - viewed 659 times.)

* rear1.jpg (75.54 KB, 640x480 - viewed 649 times.)
Logged
KQ6F
Member

Offline Offline

Posts: 124


« Reply #46 on: February 09, 2020, 06:55:44 PM »

Hi Pat -

We've got two Rods in this post and it sometimes gets confusing.  Suggest for clarity you start calling me Rodney (which is my given name that I've always hated but let's do it anyway).

Your transformer now matches mine.  I'm not certain about Nigel's transformer but I believe he achieves the same result by connecting the two secondaries in opposite phase.

One small detail - I recommend you mount the transformer so that the primary center tap is down and as close to your bypass caps as possible.  It's best to use several bypass caps in parallel to keep the center tap as RF "quiet" as possible.  I used four 0.1uf SMD caps.  Of course their total capacitance should be included in the design of the last stage in the PWM LPF.

I'm suspicious of your 1000pf SMD caps.  The peak voltages across them will get very high, especially on positive modulation peaks, and the resultant currents will be large.  I don't know the current-handling characteristics of the ones you're using but I doubt they are up to the task.  I use 1000v silver micas.

Attached is the schematic of my clock generator.  I use a flip flop to generate the two phases and gate both outputs with an enable signal.  THIS IS VERY IMPORTANT.  You need to shut off both phases when Tx stops.  Otherwise one side of the flip flop will be left high and the driver and FET that follows will be left in ON condition which will rapidly saturate the transformer and blow the FET (and probably the driver chip too).

Hope this helps.

Rodney


* ClassD_E ctrl.jpg (170.58 KB, 1000x755 - viewed 807 times.)
Logged
N4LTA
Contributing
Member
*
Offline Offline

Posts: 1075


« Reply #47 on: February 09, 2020, 07:14:36 PM »

Rodney,

Thanks for the schematic. When I said SM caps I meant Silver mica - I used 1000pF 1KV silver micas. 

I was sort of six of one, 1/2 dozen the other as if I left the CT at the botton - that made for long inductive drain leads - Probably better bypassing is more important than short drains leads as you say.

I'll probably build your driver circuit and test it tomorrow.

Nigel's reversed phase windings made installation easier with shorter leads.

Thanks again

Pat
N4LTA

Logged
KQ6F
Member

Offline Offline

Posts: 124


« Reply #48 on: February 09, 2020, 07:33:06 PM »

Pat -

Sure, if you like Nigel's xfmr better, go with it.  Or better yet, try both and let me know the results.  Heck, I might change mine...

Rodney
Logged
N4LTA
Contributing
Member
*
Offline Offline

Posts: 1075


« Reply #49 on: February 09, 2020, 08:27:26 PM »

I just laid out a small PC board for your phase splitter circuit. I notice that the flip flop clear pin 13 is tied back to the microprocessor. I assume this is a reset signal . Can it be done manually
or exactly what does this line do. Thanks again for all the info.

Pat
N4LTA
Logged
Pages: 1 [2] 3   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 0.064 seconds with 19 queries.