Graham Gunderson's Energy conference presentation Most impressive and mysterious

[poynt99]

I've taken the liberty to provide a possible starting point for the simple block diagram.

Let me know of any changes needed.

[Spokane1]

I've taken the liberty to provide a possible starting point for the simple block diagram.

Let me know of any changes needed.

Dear poynt99,

That is a good start. Let me add to it with my Version 1 Block Diagram.

In the actual presentation the input energy was measured at the input of the conversion transformer after the 220 VDC linear power supply and the H-Bridge switcher.

On the back end the oscilloscope made its connections to the capacitor bank while the power analyzer connected directly to the automotive lamp.

Spokane1

[poynt99]

Very nice Spokane1. Obviously you've made a few diagrams in your time. 

One thing that sticks out immediately, is the logic controller feed to the output driver chips (the stage before the output storage caps). I would strongly urge Graham to measure the power being delivered by that feed to make sure it isn't adding significant power to the output.

The other alarm bell that is going off for me is the fact that that same output driver stage has 3 power supplies in it. They too could indirectly be contributing power to the load.

There could indeed be significant power being sourced by the logic controller and the power supplies I mentioned above, but determining if a significant portion of that power is getting to the output could be tricky.

If they are sourcing only milliwatts of power, then obviously they are not an issue. However, if the measured power is significant relative to the output power, then further investigation should ensue.

It would be nice to see detailed input power traces on the scope, and how the power was being computed (no. of cycles, MEAN, etc.). The output power measurement could be getting skewed by what I mentioned above, but the input power measurement seems more interesting since it apparently can go to 0W. I've seen that before and it was attributed to parasitic wiring inductance that was not compensated for. Once corrected, all was back to normal.

[poynt99]

Regarding the input power measurement and a potential source of error;

1) how long is the wiring between the h-bridge and the transformer?

2) which end are the current and voltage probes placed, near the h-bridge end or the transformer end?

[Spokane1]

Regarding the input power measurement and a potential source of error;

1) how long is the wiring between the h-bridge and the transformer?

2) which end are the current and voltage probes placed, near the h-bridge end or the transformer end?

Dear poynt99,

I can take a shot at these two questions. The Oscilloscope current probe is connected right at the Swinging Choke and Capacitor array resonant tank, right were the Litz wire from the Conversion Transformer Primary lead connects. The Scope differential clamps connect across the capacitor array. So does the voltage connection for the input power analyzer. It is not to clear to me just where the Input Power Analyzer gets its current information from without studying the photo more closely. Reiyuki has a nice photo of the top of the capacitor array where you can see all these connections. I don't have that photo at home yet so I couldn't attach it this evening.

So the current and voltage probes for both input instruments are connected closer to the H-Bridge. There is about 8-10" of Primary lead in Litz wire that is of equal length coming from the transformer.

You bring up some good technical questions in your last post that need to be addressed.  Right now I don't know for sure where the H-Bridge and the Synchronous Diode Drivers get their power. Most likely it is from that 5 voltage regulator on the logic controller coming in from those shielded ribbon cables, but it could be from the process itself.

Since the input power is measured after the H-Bridge then what ever energy was added to the input flow from the H-Bridge would be measured as part of the input power to the transformer.

The real question is to understand what is going on in the Synchronous Diode section.  Do you happen to know what the upper limit of leakage power might be from a FET gate to the source-drain current path? I'm sure there is some. Would it be in the data sheet? Better yet, how much energy does it take to run his driver circuit? Lets see 10 mW, maybe 100 mW at the most. The automotive lamp was consuming about 10 watts or 10,000 mW so even if the entire gate driver energy were delivered to the output measurements we would be in error of 100/10,000 or about 1%.

I know this is a cheap way to look at the energy balance situation, but until we learn more about the details of that backend power supply this is what we have to go with for now.

Let me share some thoughts on those driver power supplies:

They are not simple. This is where the advanced electronics is going to be found. It will certainly separate the masters from the armatures. Look at the photos and attempt to start figuring out how that thing works. I know what it does. It provides the three voltages to run the driver IC's -5, +5, and +12. Graham uses negative bias to shut the FETS off and then over drives the gate to +12 volts (maximum is suppose to be +10) to turn it on. The +5 volts runs the logic on the driver IC itself.

Graham does not use gate resistors for his switching FETS. He has thus developed an approach that maximizes the performance of some pretty fast devices.

It appears that the backend network is composed of two power supplies so that each driver IC has its own power source. If I understood Graham correctly those three opto isolators on each end of the board are zener diode optos, they turn on when a certain voltage is reached, these are used as the voltage feedback sensors to operate the PWM's that drive that complex toroid transformer in the center of the circuit board. The idea is to use only as much energy as necessary. I'm sure there are several implementations to achieve the three required voltages. His is a work of art. There are also a number of additional components underneath that
circuit board that no one got a photo of.

I don't think that the method employed to provide power for the driver IC's is going to be a game stopper issue. Who knows maybe someone will discover that a single voltage driver will work as well, better yet a discovery that we don't need the super switching speeds that SiC components deliver (I can dream can't I).

Have a nice weekend,

Spokane1