If a driven array of switches reverses the connection to each and every winding of an ordinary transformer, simultaneously and periodically, then the low frequency response of the array-transformer combination is thereby extended to include DC.
On another forum, the term "DC Transformer" is held to be an oxymoron.
How short can you make an appropriate name for the circuit?
You have a sort of DC to DC converter (if you include all the other stuff such a circuit would need to actually work). That would abbreviate to DDC. I don't think you could get much less than that without introducing excessive ambiguity!
I think that the term "converter" fails to do justice to the transformer-like performance. There are dozens, if not hundreds, of converter circuits, and almost every one of them functions by storing electrical energy from the source, as magnetic energy, and, sometime later, delivering the stored energy, converted back to electrical energy. The transformer reflects voltage or current, in proportion to turns ratio, or impedance, in proportion to the square of turns ratio. It does this very quickly, with only tiny energy storage. The new circuit acts just like a transformer, because 99+% of the time, it is the internal transformer, switch connected.
Hi Newt,
Without seeing your circuit I can't comment further - but it seems like you have a chopping circuit to convert the DC to AC followed by a transformer followed by a second 'dechopping' circuit, i.e. a synchronous rectifier. Am I right?
Yes, that will surely work but you'll need to filter the chopping transients. On the other hand most DC to DC converters are designed to operate with internal storage so that transients on either the input or the output don't get reflected to the other circuit. Have you an application for a DC transformer that requires such a direct transfer function?
Best wishes with your circuit!
Tony,
If you will E-Mail me at [email protected], I'll reply with a .pdf schematic as an attachment. Switching transients are mostly eliminated by the full-bridge nature of the switching. Compared to clocked Pulse Width Modulation, working from a clean DC source, the output ripple is tiny. Fast reflection is key to the performance of this circuit, the same as it is for power frequency distribution, by transformer. Load current spikes produce minimal load voltage dips, because the load is immediately picked up by the near zero impedance of the reflected source. This circuit was invented for a 48V source, driving a 1V load, in data centers. A 1V capacitor to pick up significant energy, is huge, but a capacitor at the 48V source, is reflected as 48^2 times the capacitance.
Newt
Tony,
A use for DC transformers that should interest you, is balancing cells while the cells are connected in series. In a solar panel, it greatly reduces the effects of partial shading. In the Navy's uses of Lead-Acid cells aboard ships, cells can be balanced while being charged or discharged in series strings. To balance lithium ion cells, one would need to know much more about them than I do.
Newt
Hi,
The Transformer is basically an inductance, which doesn't respond to current but rate at which the current is changed. Thus only the rate part of the current is transferred to the other side. Such switching with solid- state devices e.g. thyristors & transistors is implemented in DC- AC inverters, where AC is needed at output.
To use it as DC- DC, another rectifier & filters would be needed at the output.
However, according to the basic transformer equation:
E (rms)= 4.44* freq* no.of.turns* Flux(max);
the output shall be dependent not on the DC level but on Frequency of the switching.
Regards,
Nikhil
Nikhil,
You are mistaken about additional circuitry being needed, A crude model of the circuit is functioning in my lab. It transforms DC power, in accordance with the turns ratio, just as an ordinary transformer does with power frequency AC. Whichever set of switches is at the terminals designated "load", performs synchronous rectification, No filtering is needed, the flat voltage input is referred as flat voltage at the output.
The name for this in electronics is a chopper . It is used to eliminate DC offsets in Op Amp inputs by commutating the input and output with complementary signals. However with inductance transients need to be controlled.
Hi Newt,
Thanks for the circuit! I am sure that it will work as you intended.
However, things to watch for are the number of windings on your driver transformer. Make sure they are are all tightly coupled to the primary; and it might be a good idea to put an electrostatic screen between the input drive secondaries and the output drive secondaries. Alternatively use two transformers. Else you might get unwanted capacitive feedback between input and output causing the system to suddenly burst into oscillation.
Best wishes with the circuit!
Dean
I used to design chopper instrumentation amplifiers forty years ago, I think that DC Transformer is a better name for this circuit.
Ignoring the internals, as a black box with terminals, it does everything for a DC power distribution system, that an ordinary transformer does for AC power.
Fast reflection of currents and impedances, according to transformer theory, works well, because of it's internal transformer.
I would assume the active switches need current sensing for zero valley switching which changes phase due to excitation and load currents being reactive and shifting towards in phase respectively to minimize switching loss.
What power range and efficiency have you achieved?
Dean
The only thing that the switches need, is to be over-driven at the same time. The turn-on is at zero voltage, because of the magnetizing inductance during the switch open time. If the drive is strong, then the driver picks up the load current, at turn-off, through the drain-to-gate capacitance, and the mosfet switch sees a zero current transition.
I have interrupted the testing of the first 150 Watt crude breadboard, after function verification. I am constructing a clean model for 48V to 1V at 100A for data centers.
Newt
that would be interesting. I know another person in EU doing ZVS for similar conversion for some 170 CPU's per board but opting for partial commercial solutions. The efficiency is a great challenge at 1V and the distribution even a greater challenge at 100A