CCM boost inductor design

Yes, a good idea, but requiring an unusual piece of insulation equipment, with tough calibration issues. Probably the only way, for less than 1% or 0.5% loss.

Yes, I thought of that, and it'd probably have to be my workaround, but leaving one uncertain of the efficacy of the results. But it's the ideal way to measure 380V dc converters. After thinking about it, and considering the basic instrumentation weakness, I decided to purchase a new-in-box WT310E on eBay. It's actually a rather impressive instrument. This was a personal, rather than Institute, purchase. Now that I'm finishing THE BOOK, time for some fun.

One could put the circuit in a die-cast aluminum box, along with a power resistor. Thermocouples to the box and ambient. It wouldn't be hard to measure the box theta using the resistor, then figure out how much power the booster is dissipating.

Tedious, but probably pretty accurate.

This is actually pretty good:

It was just an example, which corresponds nicely to the numbers we have for our PFC converters

I cannot share more details which would be a breach of my NDA

Cheers

Klaus

How are you going to measure your 99% performance?

For 99%, that is difficult

We normally have a very accurate simulation model for the entire converter

We then do individual check on key components to check the model

On the measurement side, we have a special instrument bought at very high cost just for that measurement (I can check the model number next week)

Often the difficult parts are after the rectifier which can be fed with DC to do an extra check (DC measurement is a lot easier)

We also do thermal measurement of each component verified afterwards with a complete dummy load dissipation test where we match all temperature measure ring points

Cheers

Klaus

piglet

Exactly. The best PFC efficiency I know of is 99.3% according to the author (claimed to be obtained using a calorimetric meaurement). Totally bridgeless, with some insanely low switching frequencies, just barely above the audio range.

You just can't get 99% with the old-school boost approach.

Best regards, Piotr

Nobody does active rectification? Common at lower voltages. I think the HV transmission-line guys can get well over 99%.

I heard 5% is lost in transmission. That's why some long distance HV transmissions use DC. But that's not a reason to ignore conversion losses.

If you are shipping 5GW, then 1% is 5e9*0.01 = 50,000,000W. A very good reason to reduce losses. Where are you going to put the heat?

Correct , that is why I explicit wrote ?PFC stage?, also th at comparison since the OP was talking about coil losses

Cheers

Klaus

Like Winfield wrote, they operate with higher than 99% efficiency for the converter

Quite impressive

Cheers

Klaus

It doesn't pay off to stop halfway. In order to have losses low enough to justify the added complexity you need a sufficiently low R_DS_ON high voltage MOSFET. That means an expensive superjunction unit or, preferably, a SiC/GaN device. The SiC ones are particularly price competitive. But then, having a classic full-bridge arrangement of the MOSFETs and the sensing circuitry you discover that you can shave off the downstream boost stage entirely if you just wiggle the gates in a bit smarter way than a classic SR would do. This is the totem-pole bridgeless boost configuration. One example (fig. 5 and everything that follows):

Indeed, would be required at the low side of the converter anyway.

Best regards, Piotr

Synchronous rectification would be great for the bridge rectifier

We have been toying with that a long time, problem is complexity, price and protection

Protection is difficult since you are right at the ac terminals with no dampening components except for the cm coil, most active components spontaneously turn on when subjected to high dV/dt

Cheers

Klaus

On a sunny day (Sun, 4 Aug 2019 09:42:38 +0200) it happened Piotr Wyderski wrote in :

Very nice solution!

Yes, I realize that. I was just using 1% for illustration. So for 99.9% efficiency, the losses are 5e9*0.001 = 5,000,000W, which is still a large amount of heat to get rid of.

The obvious question is how do they do it?

And can they get higher efficiency?

I'm questioning the litzwire size, when I check the charts the recommended wire for 150kHz is #40 wire, 0.0031 inches or 0.07874 mm. It think you are saying 0.4 mm which is wasting copper. If your limit to get the turns needed is 1.2mm, then a litz wire using the recommended #40 wire with 108 strands is 1.14 mm in diameter. See chart here,

Full page gauge/frequency chart.

Mikek

or

DC transmission is actually mostly to reduce corona losses, which depend on the peak voltage. You can put twice the power through a given line.

Cheers

Phil Hobbs

Hydro Quebec uses AC to ship power at 735 KV, and DC to ship at 450 KV. So there are other reasons beside corona to decide on DC.

The AC losses range from 4.5 to 8%, varying due to temperature and operating situations. I don't know if temperature affects corona.

For DC, you still have to convert it at the transmitting and receiving ends. With twice the power, it is crucial to reduce the conversion losses.

Another reason for DC is to connect grids that operate at different frequencies or phases.