push pull topology?

Welcome and feel free to ask questions about class E, mosfet's, and AM.
You must be registered to post.

push pull topology?

Postby N9NEO » Tue Apr 08, 2003 12:00 pm

Seems the literature I've seen on the net is mostly single ended topology. Why not push pull?

regards,
Bob
User avatar
N9NEO
Class E Enthusiast
 
Posts: 322
Joined: Tue Apr 08, 2003 11:52 am
Location: Uxbridge MA

Push Pull works

Postby steve_qix » Tue Apr 08, 2003 1:58 pm

Hi!

I've also seen some push-pull circuits - class "E - F" - RF amplifiers.

The advantage is a lowered 2nd harmonic output. Disadvantages are a more complex circuit. I also don't believe the efficiency is quite as high - a couple of precent difference from what I've read, although that may be due to the particular implementation, and not the actual topology.

But, it *does* work.

Regards,

Steve
Visit the class E web site at:

http://www.classeradio.com
User avatar
steve_qix
Class E Guru
 
Posts: 434
Joined: Fri Jan 31, 2003 11:03 am
Location: Townsend MA

Postby frank carcia » Mon Apr 14, 2003 3:31 pm

I've been push pull since '96 works fine. A little bigger transformers
fc
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Postby Nathan Sokal » Mon May 26, 2003 8:48 pm

Reply to "push-pull topology?"
Yes, it works OK. I can send you a circuit diagram of 3 usable topologies by postal mail if you wish.

Pro: cancels even harmonics and adds the powers from the two transistors, less a few % power loss in the power-combiner.

Con: Requires adding a power-combiner to combine the outputs of the two transistors, and a second driver circuit to drive the input of the second output transistor.
Nathan O. Sokal
Design Automation, Inc.
4 Tyler Road
Lexington, MA 02420-2404
NathanSokal@compuserve.com
Nathan Sokal
 
Posts: 5
Joined: Mon May 26, 2003 8:26 pm
Location: Lexington, MA

Postby frank carcia » Tue May 27, 2003 7:19 am

Hi Nathan,
I'm the guy at Boxboro who found class E by mistake thinking I built
a resonant converter. I would be interested in seeing your ideas on
push pull. I've always felt push pull was superior to single ended
for all your reasons and it puts a more stable load on the driving source.
My present rig runs 14 IRF840s in push pull parallel but plan to build a new final with the FETs Steve found. I'm considering asecond final in series by by lining up the output transformers and winding the secondary
through both transformers. fc
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Postby Nathan Sokal » Tue May 27, 2003 4:30 pm

Reply to Frank Carcia's message of May 27:

If you give me your postal address, I can send you a sheet with 3 push-pull topologies. They are all identical mathematically, but you might prefer one or another for ease of construction for a particular project.

"More stable load on the driving source" is true if you are driving BJTs, but I don't tsee why single-ended or push-pull would make much difference if you're using MOSFETs. If you see a substantial difference, tell me what and why.

Nathan O. Sokal
NathanSokal@compuserve.com
Nathan O. Sokal
Design Automation, Inc.
4 Tyler Road
Lexington, MA 02420-2404
NathanSokal@compuserve.com
Nathan Sokal
 
Posts: 5
Joined: Mon May 26, 2003 8:26 pm
Location: Lexington, MA

Postby frank carcia » Wed May 28, 2003 7:55 am

Hi Nathan,
I have to comfess that I play a little drive trick with my set up. The drive is transformer coupled. The turn on voltage at the gate is about 4 volts.
I set up my drive so it is not 50 % on but less say 40 to 45 %. This helps the final avoid both phases on at the same time. I found the efficiency
improved when I took advantage of the FET operating point and transformer coupled drive. Looking at the drain I see the a little fly back
action then the other phase takes over. This puts a little step in the drain wave form but couples over to the other phase driving it in the direction it wants to go for the next state. My best measured efficiency
is 90.8 % if I trust my nonNBS test equipment.
Frank Carcia, WA1GFZ
181 Columbia Rd.
Enfield, Ct. 06082
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Postby Nathan Sokal » Thu May 29, 2003 7:40 am

Reply to Frank Carcia's message of May 28, 2003:

OK, Frank. I'll send you by postal mail a printed sheet with three versions of a push-pull topology. They are all equivalent mathematically, but they differ in construction. One of them has the two transistors well-isolated from each other, so a little overlap of conduction by the two transistors has almost no effect.

In fact, even if the two transistors *are* coupled (as in the other 2 topologies), a little overlap of conduction does no harm (like you could have in other kinds of circuits) because the transistors are fed from one or two current sources (the one or two dc-feed inductors, depending on which topology you are using). Because the transistors are being fed by current sources, what they do in case of overlap in the topologies where they *are* coupled: they share the total current between them. Almost no "shoot-through" effect like you can have in other kinds of circuits, because they are fed from one or two current sources.

That's one of the reasons why Class E is found by some people to be more robust than voltage-switching Class D. But if that Class-D circuit is designed properly, it will not have conduction overlap between the two transistors, so the shoot-through problem will not occur. And in current-switching Class D, conduction overlap is no problem. The two transistors simply share the current being injected into the circuit by the dc-feed choke. In the current-switching Class-D topology, a wide conduction *gap* should be avoided.

73,

Nat, WA1HQC
Nathan O. Sokal
Design Automation, Inc.
4 Tyler Road
Lexington, MA 02420-2404
NathanSokal@compuserve.com
Nathan Sokal
 
Posts: 5
Joined: Mon May 26, 2003 8:26 pm
Location: Lexington, MA

Postby frank carcia » Thu May 29, 2003 12:13 pm

Hi Nathan,
My drive is applied through a transformer so there is no overlap to cause shoot through. The center tap of the secondary is connected to the sources of both phases in the final amplifier. My drive swings plus and minus 12 volts but the FETs switch at about 4 volts positive.
This gives me a little less than 50% duty cycle.
My DC is applied through the center tap of the output
transformer. The heat sink temperature was much warmer before I
used transformer coupled drive and had over lap. Also much lower
efficiency. The real reason my heat sink is so large. I didn't understand
my problem until after I machined the heat sink.
I may have been saturating my output magnetics when I had over lap
problems. Maybe the over lap time was too long. I don't know but
the drive transformer fixed the problem. Prior to that time I was using
DC coupled square wave drive.
I will be interested in your push pull schematics to see if I did anything cyazy. fc
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Postby frank carcia » Sun Jun 01, 2003 10:57 am

hi Nathan,
I received your interesting mail. I think you published this in Ham Radio years ago. I felt the class E effects could be reflected back through the output transformer if the drive duty cycle was less than 50%. I only have
a tuned circuit after the broad band output transformer. I started
with square wave drive so it may also help. It will be interesting to see if I can use the same design on 80 meters. My present design has problems
in the pre driver above 2.5 MHz. The MOS drivers I'm using get pretty hot
at 10 VDC at 80 Meters. I'm working on a new transmitter that will use
the Fairchild Fets Steve found. They appear to have the same switching performance as the IRF840s I am presently using but they are rated at 900 volts and much higher power dissipation. I may try direct drive from
a transceiver with a bias on the center tap of the drive transformer to control the drive duty cycle but all this stuff is my head at this point.
My goal is to only have a tuned circuit at the output...KISS.
Steve is doing the same thing with his single ended amplifiers. He is using
a broad band transformer to couple to the output circuit. He also has taken multiple amplifiers and series the transformer seconadry windings to a common tuned circuit. He reflects the effects back through the transformers to multiple stages. Push Pull is just reversing the input and output phase of half the stages. Think along these lines I feel multiple push pull stages could also be series up to get higher power. There are no additional components and still get the advantage of push pull.
My present rig runs 1700 watts peak output power with a single output stage and the FETs never get above 100 degrees F. I run about 85 volts
at full modulation so the new FETs should allow me to double the output power with the increased power supply voltage. fc
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

Push-Pull vs Single-Ended

Postby Todd WD4NGG » Mon Jun 02, 2003 11:54 pm

I had recently written to Nathan Sokal asking his views on push-pull vs single-ended. Nathan had asked me to post his replies here as he will be tied up for about a week. Here are his replies to some of my questions. 73 Todd WD4NGG. >My feeling is, if you have a single-ended Class-E amplifier that is
running 93%
efficiency, it is simple to build and tune, then how is "push-pull" better?


**That's perfectly correct if a single transistor gives you the output
power you want.
But consider the case when you need two transistors to get the output power
you
want. Then you have the choice of connecting the two transistors in
parallel
or in push-pull. [In the parallel connection, you don't actually connect
the two
drains together. To ensure that the two transistors share the L2 current
reasonably
equally, each transistor has its own L1 and L2 (each is double the
inductance of
what you would use if you had a single transistor that could handle all of
the power),
and the two L2s are connected to a single C2 that serves both L2s. The
right-hand
end of that C2 connects to the input port of the low-pass filter. Or if
you need lower
ESR of the C2 capacitor, you can use two separate C2s, each connected at
its left end
to its L2, and connect the right-hand ends of the two C2s together at the
input port
of the low-pass filter. Each of the two C2s is half of the single C2 you
would use if ...]

Now the choice of single-ended or push-pull is a trade-off of one hardware
setup
vs. the other. Push-pull needs to add a power combiner to combine the
outputs of the
two transistors (e.g., a push-pull transformer primary). On the other
hand, the
low-pass filter need not reject 2f nearly as well as is needed in a
single-ended PA,
so the LP filter gets a little simpler. If you were going to use
transformer coupling
anyway, for the single-ended PA, you need add only a second primary winding
to
make it push-pull instead, and you can wind it bifilar with the first
primary winding,
so little extra work is needed in making the transformer.

>Second harmonics can be easily dealt with using a low-pass filter or
second-harmonic trap
I believe.

**Yes, that's correct. The 2f trap costs you one extra small-size
inductor. Trade that off
against a second primary winding, bifilar with the first one.

Another small advantage of push-pull: The RF ripple current in L1 must be
C-bypassed to
ground at the top end of L1. If you use push-pull, the RF currents in the
push and pull L1s
almost cancel each other, so less bypassing is needed.
Todd WD4NGG
Member
 
Posts: 29
Joined: Sat Feb 15, 2003 8:30 am
Location: Hilton Head Island, SC

Postby frank carcia » Tue Jun 03, 2003 11:51 am

Hi Todd,
I guess you got the same mail as I did. Steve single ended and I push pull
have been able to use a broad band transformer to couple multiple devices to a common tuned circuit to get the effects of class E. We have both been doing it for years. This allows the transmitter to easily change frequency. Nathan's configuration makes a lot of sense but more complicated and may require a number of L/C value changes when the operating frequency is changed. I bet there is extra efficiency there. Hams running in the range of 90% can live with the extra power loss compared to the old tube rig. Our goal is to be able to build something simple that works well and can travel the bands easily. Who cares if the FETs don't exactly share power as long as they are derated properly. The FET by design compensates for thermal effects.
This is why I choose to run push pull because the configuration is not really more complicated than single ended if you need to run multiple devices to get the rated power. I found that as long as the tuned circuit
had a little time to fly wheel the final before the next power stroke
no ping would be observed. Nathan uses series inductors to deal with possible overlap so power isn't wasted. My configuration gets hot whan
I have overlap so I just solved the problem in the driver. I bet there is a drive sweet spot in Steve's configuration to get a similar effect with sine wave drive. I may try the same thing with a little negative bias on the gates to offset the duty cycle of the drive. fc
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.

My $0.02

Postby N9NEO » Tue Jun 03, 2003 8:54 pm

Lots of good posts on the subject. Please excuse my ramblings.

My intent with the first post/question was to primarily determine if push-pull had been done before. I had been mulling over in my mind a scheme where I would use two single ended class E topologies driven 180 out, in order to derive a symmetrical waveform. My backgound is power supplies, and not RF. Push-pull to me is a rather obvious step from a flyback when stepping up the power levels. (Or full-bridge from half-bridge if you prefer.)

The question was quickly answered by Steve when he replied, "It *does* work." So the answer for me was a green light to build one. I was looking for a hombrew project for quite some time, and had almost decided on a digital vfo to drive my Knight T150. Heheh. (If anyone has ever operated one of these nifty little rigs I think you may understand my motivation.) Anyway, was tuning up the 75m band one day and came across a nice group of people discussing something called class E rigs. Punched it in to the search engine which brings me to this site, and oh, oh, back to the point...Sorry.

Yes, second harmonics will be mitigated, and drive circuits may be a tad more complex I'm not sure that the efficiency will suffer. My sense of engineering intuition tells me for the same $ the push pull would be at least as efficient, and perhaps a bit more. Steve if you're listening, do you recall the literature which discussed the efficiencies? In any case the design is not so different from the single-ended and should offer a few not unsurmountable difficulties. A little bit of meat with those potatoes as they say.

Hey it's not an off the shelf $10,000 piece of A***n plastic with lotsa knobs and buttons that does jack-shit for the USA economy. And does a fine job in keeping with the spirit of Ham Radio. Experimentation for the advancement of the hobby, and so on.

I say GO FOR IT!

A special thanks to our sponsors who without their support I would not be able to give my $0.02.


73
Bob
User avatar
N9NEO
Class E Enthusiast
 
Posts: 322
Joined: Tue Apr 08, 2003 11:52 am
Location: Uxbridge MA

Postby frank carcia » Wed Jun 04, 2003 11:23 am

Bob,
The short answer is that I have been donig push pull for years and don't
feel it is complicated. I have never done anything solid state single ended
over 1 watt. Go for it! fc
frank carcia
Senior Contributer
 
Posts: 765
Joined: Fri Jan 31, 2003 10:48 am
Location: Enfield Ct. / Niantic Ct.


Return to Class E Discussion

Who is online

Users browsing this forum: Bing [Bot] and 5 guests

cron