# Push-Pull Output Transformers - Part III, The Final Countdown:

Discussion in 'Amateur Radio Amplifiers' started by KD2NCU, Sep 28, 2017.

1. ### KD2NCUHam MemberQRZ Page

Hey RICHS. I understand EXACTLY how the device works and I can derive the equations from scratch that demonstrate the impedance transformation and did so in an earlier post. In addition, the impedance transformation is derived independently in numerous references that I sited as well as an article that KG7SWP cited in an earlier post. It's really not difficult at all though. The transistors see twice the current that the final transformer primary winding sees and half the voltage swing that the final transformer primary sees during their active cycle. Half the voltage, twice the current, that is a 1:4 impedance transformation. This is also very clearly stated in several of the references that I provided wherein the author clearly states that the bifilar feed coil in the push pull output being designed provides a 1:4 impedance transformation. In addition, there is tons of literature out there that already fully explains how "simple transmission line transformers provide a 4:1 impedance change". Where have you been? I would like to see YOU produce a true technical analysis and derivation that shows the device to produce choking action or keeping RF out of the DC supply. THAT would be interesting. You're already involved dude! You're floating out indefensible gibberish, I'm calling you on it, too late to run and hide. Let's see you put that 8 years of college to work and defend some of this with actual technical analysis instead of throwing out platitudes and important sounding words like "phased". Come on, bring it!

2. ### KD2NCUHam MemberQRZ Page

Hey WA7PRC: Yes, again. Here's why.
In my first post I asked some questions about how a particular device worked.
I got a few good answers that made sense and I got some additional answers that just didn't seem to agree with physics.
I did a bunch of independent analysis of the device on my own and came to a certain set of conclusions that didn't agree with what some were saying about the device.
I provided this analysis in a second post and politely asked people to review it and let me know if they agreed and why or why not and if they had any additional information that would answer some of the apparent contradictions.

I don't think anyone actually read the analysis in detail or tried to understand it and certainly NOBODY addressed it on its technical merits and NOBODY offered any information or references that would back up some of the claims about this device that I think are false.

So in Part III, I have gone after fully independent analysis, references, citations, circuits, etc. and consulted experts that have written and published articles on the topic. And guess what, these INDEPENDENT sources have confirmed my suspicions. So it's no longer my analysis, its the analysis of others and I just happen to have come to the same conclusions on my own prior to finding these confirming references.

So this last posting is no longer my opinion or analysis. It is published analysis from experts in the public domain. I don't think anyone on the blog has actually read this posting either. They just continue to use circle logic, presume that the device is a choke, and then use the presumption that it's a choke to defend why it keeps RF out of the DC supply. Nobody has offered anything that in any way explains, proves, or derives analytically how the device can act as a choke or prevent RF from propagating into the supply. Not one shred of defensible technical analysis from anyone has been offered. I'm not surprised. I could tell from the very first posting how this was going to go.

Okay...

4. ### K7JEMHam MemberQRZ Page

The main question you seem to need answered is "Is this device a choke?" And the answer is yes, it is a choke, not a transformer. If you want to prove this to yourself, you could remove the transformer, then replace it with two appropriate chokes, and see if the circuit still works. If it does, then the "transformer" is not needed at all. And the circuit will work fine that way. But now it takes two separate chokes, with two cores and two circuit spaces. But that works, just is more complicated, expensive, and takes more room.

All chokes of this type, those feeding power amplifiers, need locally close connected bypass capacitors, since the choke alone will not filter the AC pulses enough, due to limits of inductive reactance, etc. But once you provide those caps, you have a very clean DC at the input, and wildly swinging AC at the other end, with very little heat. The only way you can have that is if you have an inductor with very little DC resistance, yet high reactance at the RF frequency. That is pretty much the definition of "RF choke".

5. ### WW1WWHam MemberQRZ Page

You are assuming he knows how to use a meter, a scope, connect a power supply, and other such very complicated things. I could not imagine how that discussion would go if he doesn't understand what a coil of wire does.

7. ### KD2NCUHam MemberQRZ Page

Hey RICHS, don't run away yet, you and I are actually agreeing on something. Wouldn't that be amazing! I'll play nice if you play nice.
I never said there was no end to end inductance. I don't think I've ever even uttered the phrase end to end inductance. I've said there's no inductance experienced by the current from the DC source entering the center tap of the coil.
Although, come to think of it, I have used phrases like "not a choke of any kind" although I was always referring only to the connection to the DC supply. If that's a point of contention and confusion, my apologies. I do contend that the device does not act like a choke of any kind to the current coming into the center tap from the DC supply.
The measurement you made does not measure the inductance seen by the current coming into the center tap from the DC supply.
The measurement you made was an end to end measurement and SHOULD show significant inductance otherwise, as you said, it would short out the final transformer primary. And the reason it does is because the flux produced during THIS measurement is constructive and does not cancel.
You measured across the entire coil end to end. The current from your inductance meter goes into or out of both phasing dots simultaneously so the flux from the two coils is constructive. Agree? Said another way, an end to end current will experience positive mutual inductance with a coupling factor close to 1.

However, THAT inductance is not the inductance that the current coming from the DC supply into the CENTER TAP experiences. You would have to make a different measurement than the measurement you made.

We don’t have to argue whether I’m right about the two currents being equal and opposite half sine pulses just yet. Let’s argue about that later. However, wouldn’t you agree, RICHS that if that WERE the case, then all of the following statements would be true?

1.The flux from THESE two currents tends to cancel one another. (One current goes into a phasing dot and the other goes out of the phasing dot.)
2.The instantaneous DC supply current Is is equal to I1 + I2 ALWAYS. (This will ALWAYS be true, regardless of any other assumptions made.)
3.If I1 and I2 are half sine pulses as assumed above, the current Is from the DC source will also be a half sine pulse twice the magnitude of the two currents, not a flat DC.
4.Since the flux from the two individual currents is cancelling, the two individual currents are not experiencing any inductive impedance.
5.Since the current from the DC supply is exactly twice the two individual currents and the same shape, there is obviously no attenuation, inductance, or any type of filtering being applied to the current from the DC supply by this device.
6.Since the current from the DC supply is experiencing no inductance, the device is not acting as a choke of any kind TO THE CURRENT COMING FROM THE DC SOURCE.

To test the inductance presented to the center tap current, still assuming equal and opposite I1 and I2, you would have to jumper the two remaining ends of the coil together as shown and measure the inductance between the center tap and one end as shown below.

Now would you not EXPECT this measurement to show zero inductance except for imperfect coupling and measurement lead effects? Because the current from the meter splits and 1/2 goes into a phasing dot and the other half goes out of a phasing dot, thus producing (close to) zero flux. Said another way, we have negative mutual inductance with a coupling factor of close to 1.
So if you agree with this stream of logic, based temporarily on the assumptions about I1 and I2 being equal, we could save arguing about whether the two winding currents are really equal and opposite half sine pulses for another day.

File size:
32 KB
Views:
65
File size:
36.3 KB
Views:
77
8. ### K7JEMHam MemberQRZ Page

I am trying to figure this out. If it is not a choke, but is a transformer, you should not be able to replace that device with two chokes and get proper operation, since the transformer is now removed. You seem to be saying that the device is working as a transformer, yet we all know that it can be replaced by two chokes that are not physically close together, hence no transformer action.

Once you explain how this is possible, we can go on to the next point. But apparently, no transformer "action" is actually taking place here.

WW1WW likes this.
9. ### WA1GFZHam MemberQRZ Page

All,
The DC feed choke is not balanced. How can it be a source for feedback if it is balanced. I proved it in simulation. I also eliminated it with careful transformer design by integrating the DC feed and output transformation in one transformer core.
I found the choke limited broadband performance and effects second harmonic levels. Helge covers it in his book in the transformer section. This is not new information. I have seen this trick in Harris amps as well as Amplifier Research BB amps. Check out QEX Sept/Oct 2015. I show how I modified a 1200 watt MRI amp to cover 160 through 6 meters. I had limited space on the board and the DC feed choke degraded performance. I struggled with this for a long time and determined the DC feed choke added reactance to the drain circuit that did more harm than good. The stock board running at 80MHz didn't use chokes. Today, I'm running 4 combined 1200 watt modified MRI boards 160 through 6 meters. The transformers run stone cold. I've had it on the air almost 2 years but too chicken to try it full power so far. Frank WA1GFZ

10. ### KD2NCUHam MemberQRZ Page

Good morning K7JEM.
One rather obvious explanation would be that the designer didn’t know (or refused to believe?) that the bifilar coil provides a 1:4 impedance transformation and didn’t take that into account in his original circuit design which resulted in a poor initial match. Replacing the coil with two chokes would result in a better match by getting rid of the extra transformation. The internet has several stories of people replacing the coil with two chokes and getting more power. Now why would that be? When Jim Klitzing, W6PQL, changed one of his circuits replacing the coil with two chokes, I contacted him and asked him several questions about why and then several more questions about whether he thought the bifilar coil provided a 1:4 impedance transformation and whether it provided any choking action on the DC coming from the power supply. I have to give the guy a lot of credit for being very honest. He came right out and said, “I’m not qualified to answer those questions” rather than parroting a bunch of junk science. He then proceeded to tell me he made the change because he experimented and found that he got more output power using the chokes. What might explain that? But he made other changes to the circuit at the same time including removing all degenerative feedback and making several other changes to the input circuit so who knows the real reason he got more output.

Some other obvious explanations why you can replace the coil with chokes might be;
1. Additional components were changed at the same time which compensated for the removal of the impedance transformation.
2. Or, you’re now operating with a poor impedance match between the transistor and the output transformer and you just don’t know it because the entire system is tolerant or you don’t have a way to measure the quality of the impedance match.
4. Your initial impedance match was off by a factor of 2, you took out the 1:4 coil and now your impedance match is off by a factor of 2 in the other direction, so …
5. And so on.

It COULD also mean that there was no impedance transformation, but the mere fact of replacing the coil with chokes doesn’t prove anything.

Now let’s review a couple of other cases where you link cause and effect that have nothing to do with each other and/or use circle logic based on assumptions.

“If you think about it, the "transformer" is acting as a choke, only.” You don’t know this. You are assuming this.

“There is no AC present on the input winding, it is pure DC.” I think you mean on the center tap where the DC supply is connected? You don’t know this, you are assuming this. Who says there is no AC? Where did you get this? What are you basing this on?

“You can't transform a DC voltage on one side to an AC voltage on the other. Your diagrams of the transformer show AC on both sides, when used as a transformer.” Nobody said anything about transforming DC to AC. I said that there will be a full wave rectified sinewave of current drawn from the DC connection into the center tap of the transformer. Do you disagree? If so, what do you think a current probe or other device would say the current looks like as it’s coming out of the supply lead just before it enters the center tap (after any other local bypass capacitors and chokes)?

Yes, I contend that there is RF current flowing in and out of all three terminals of the feed coil. Sounds like you disagree, so tell me what you think the voltages and currents ARE doing.

You said, “But once you provide those [local] caps, you have a very clean DC at the input, and wildly swinging AC at the other end, with very little heat. The only way you can have that is if you have an inductor with very little DC resistance, yet high reactance at the RF frequency. That is pretty much the definition of "RF choke". You are linking cause and effect here that have nothing to do with each other. You are assuming that the cause of the minimal variation in voltage is due to the presence of inductive impedance. You have a fairly clean DC VOLTAGE. But tell me what you think the CURRENT looks like at the connection to the center tap after the local caps & such? Now, having a fairly clean DC VOLTAGE at the center tap does not prove the presence of inductance in any way and, and in fact, it’s NOT the reason you have very little AC VOLTAGE at the center tap.

The reason you have very little AC VOLTAGE variation at the center tap is that there is a symmetric voltage between the collectors with respect to the DC supply voltage. IE; with no signal, both collectors start at 12 volts. As a drive signal begins, one collector goes up from 12 volts and the other collector goes down from 12 volts and the center stays at 12 volts. If you were to connect two resistors in series across the collectors and measure the voltage at the center connection of the resistors with respect to common, you would see 12 volts DC with very little AC variation with respect to ground. You would see very little AC variation with two caps or two inductors, as well. This is called a “virtual AC ground” and the effect is very prevalent in symmetric balanced systems such as the push-pull. I’d send you links and a complete derivation that proves the center tap SHOULD have very little AC variation on it with or without local caps, without choking action by the coil, etc., but I’m pretty sure you’d brush this aside like you have everything else.

Go checkout the web page at http://ludens.cl/Electron/mosfetamps/amps.html

He lays this all out for you starting at Figure 4 including all the voltages and currents that show the impedance transformation. You copied and pasted “choke” from this article but apparently ignored the information about the voltages and currents and the resultant impedance transformation.

So, having very little AC VOLTAGE variation at the center tap is not proof of inductive impedance. It is a simple consequence of the symmetry of the circuit, and actually, a consequence of the transformer action of the feed coil windings producing equal and opposite voltages with respect to the center tap. You have assigned incorrect cause and effect between two effects that have nothing to do with each other.

Why not stop bringing up these red herrings and attempting these silly little end runs and just tell us how you think the device DOES work? You’ve said everything presented so far is WRONG. So why don’t you go ahead and tell us what’s RIGHT. I don’t mean reasoning like, ““If you think about it, the "transformer" is acting as a choke, only.” That’s all hand waving blather based mostly on unsupported and invalid assumptions. I mean tell me during operation what all the voltages and currents are doing and what the flux looks like in the core of the device, etc. I mean use Faraday’s law, Lenz’s law, the right hand rule, calculus, and circuit analysis (current and voltage equations), and the physics of coupled coils and tell me how it works and what all the voltage and current waveforms DO look like. From that, we should be able to determine whether it’s acting as a choke to the DC input, and whether or not it is providing a 1:4 impedance transformation, shouldn’t we? Or find a paper or derivation or reference that lays all this out and proves your point. I put together exactly such a derivation for all to see and review and the web page that KG7SWP referenced lays it all out as well including the impedance transformation. Why not man up and put out your complete derivation and explanation for review.

Maybe start a step at a time. Tell me what you think the current looks like coming from the DC supply going into the center tap of the bifilar coil, for example. IE; draw me a graph of what you think you would see if you connected a current probe or other device around the wire just before it goes into the center tap.