hi power smps

[jamesrnz]

ahhhhhh.... i been playing with a 115 vac full bridge to make 20 volts at 50 amps to charge the batteries. i am fried.
anyone have a schematics that work?

mine seem to get hot at 10 amps.

my little ones work............

[poobah]

What topology are you using?

[jamesrnz]

wellll.

bridge rect from 115vac through cm filter. then a full bridge with mosfet into a large toriod with 16 t on pri and 4 t on sec to bridge ind/cap filter

the transistors are at home i guess i will have to get more info for the thread they are about 50 amp 600 volt.

jimbo

[N-Channel]

Hmmm,

Let's see. 20V @ 50A = 1kW. From 115VAC, we get ~162V. Assuming your design is good for low line of 90VAC (~127VDC), this is 7.85A. Through a full-bridge, the drain current is 1.56 I(max), or 12.25A. Giving a 20% margin, this is just under 15A.

Total conduction losses through the 4 MOSFETS will be I^2*R, or (15^2)A x say, 0.24W (Rds(on) for each MOSFET) gives 54W per MOSFET power losses, This doesn't take into account the switching losses, which will be a function of the MOSFETs' Rds(on) and the switching frequency. This is at the max of 50A.

At 25A, again at low line, this is (7.35^2)A x .24W = ~13W per MOSFET. Heatsinking will DEFINITELY be necessary.

Instead, I would do a conventional fullwave-doubler input off the AC mains, just like every AT & ATX PSU box does. Then, you will be dealing with ~320Vdc (~270VDC @ low line), and your (V^2)/R losses for each MOSFET (assuming you're using MOSFETs) is cut by three-quarters.

Then go on to either do a half-bridge, or full bridge. At 1000W, max, I would recommend the Full-bridge.

Or, if you really want to get wild, and further reduce your MOSFET drain currents, you could Active PFC the front-end, yielding a semi-regulated ~385V-400VDC. Your 600V units should handle 400V just fine. Then, with those decreased drain currents, you should be able to get your thermal problems under control.

At 1000W, I would strongly recommend PCF'ing to reduce the peak rectified currents that would result from the non-linear current waveforms that result from charging the input bulk capacitors at the peak of each haversine wavecrest. Also, PFC'ing instantly makes it compatible with all world voltages, without the need for a 115/230V switch.

Hope these little bits help you out.

Steve

[jamesrnz]

yes of course , i should have thought of that.

i was looking at the L4981 or mc34262 for the PFC controller.

what are your thoughts on transformers? I am using a 2.4" toroid at 66khz, i belive that the primary is 16 turns made of 8x22ga. and the sec is 4 t of 36x22ga.

with the change in input voltage i will have to in creas the pri t to about 32ish.

i am using a 34025 for my switch controller with voltage control from a opto and current control through a current trans..

lets make a 5 kw for big stuff. \\

i am also working on a 12v to 115 invertor at 1.5 kw.........


thanks
jimbo

[N-Channel]

Wow, James. That's aggressive. (Not in this forum, though)

I use the 33025 (Temp range of -40C to +105C) in current-mode control.

What is the material of the 2.400" toroid? For the #77 material (and this is getting old, as newer materials are supplanting it) at a frequency of 66kHz ( I will assume this is your switching frequency, and not the clock frequency, at 385-400V, by my calculations, your primary turns should be 40.38 (rounded to 40 Turns).

This equation comes from Chryssis' chapter on transformers:

N(pri) = (V(in) *e^8)/(k * f(sw) * B(max) * A(e), where:

N(pri) is the number of primary turns,
V(in) is the input voltage (Nearly constant for a PFC'ed front-end),
k is a constant (4.0 for sq waves, and 4.44 for sinewaves)
B(max) is the max flux density at the frequency
A(e) is the effective cross-sectional area of the core material.

From here, you can calculate the secondary turns.

For your PFC controller, the MC34262 is Critical-Conduction Mode (CrCM), which is good for power levels mainly below 200-300W. It is a variable-frequency control technique that is not best suited for power levels approaching 1kW.

I would look at the UC3854A Continuous Conduction Mode (CCM) PFC Controller. This constant-frequency PFC Chip is more suited to medium- and hi-power applications. Also, since both the PFC and the PWM chip are constant frequency, their oscillators can be synchronized to avoid producing beat-frequencies.

Consulting the Amidon Associates datasheets, for an FT-240-77 core, your max power throughput at 66kHz should be somewhere around 1200W. I don't have the portable sheets in front of me right now, and the main catalogue doesn't have this graph, but I think it is ~1200W. I will check when I get home.

So, for 66kHz, I would think your chosen core should be OK. .

As far as the Opto feedback, and current-transformer, those are both fine.

It sounds like you have already consulted several books on SMPS design. I smell Marty Brown's work here, maybe a little Abraham Pressman or the aforementioned Geo. Chryssis. These three books are my 'Holy Bible' in designing all of my SMPSs.

Methinks you are off to a goods start in the design process.

Steve

[jamesrnz]

n-ch

thanks, i just orderd some samples from ti.

i just got browns book it is pretty good for low med power stuff.

my experience is with thyratrons and microwave tropo scatter power supplies from the old days,

what would be a good forum for insanely hi power stuff?

jimbo

[jamesrnz]

n-ch
yes the 33025 is a better choice, i have a 77 toriod i was thinking about f but the sec turns might be too few...

[jamesrnz]

steve,

tnx so much this forum is valuable when you dont have anyone to bounce stuff of of.............


what do you think of diodes in the drain and source??

jimbo

[N-Channel]

Intgreal Body diodes,

You can use some MUR880Es across each MOSFET, but I don't think it will be necessary, especially for constant-frequency current-mode PWM control.

Is 66kHz your switching or clock frequency?

Also, if you do consider PFC'ing the front-end, you will need to replace the normal standard-recovery rectifiers with ultra-fast rectifiers rated at at least 800V and 15A. The reason for this is now, there are high-frequency pulses being drawn from the AC line: XXkHz (whatever frequency the PFC is running at), modulated at the AC Line frequency of 60Hz.

I checked the Amidon Associates nomograph for the #77 ferrite toroids, and for the FT-240-77 (2.40" Ferrite Toroid, #77 material), at 66kHz, you have a theoretical ceiling of 1.5kW throughput. So, for 1kW, I think you have a satisfactory margin of safety. I would caution against letting the switching frequency get too high, as core losses will go up with higher frequencies. Another caution to take is when winding the primary and secondaries, you must provide for sufficient insulative layering between windings. I use teflon transformer tape, which works rather well. Actually, it's used for plumbing, but it works just as well for electrical applications, and I believe it also has a very high dielectric constant. Available at Homeless Depot or Lowe's.

As for "insanely hi power stuff", I guess right here. But please be aware, as you will be dealing with high AC & DC voltages (up to 400VDC and 240VAC) you MUST MUST MUST exercise EXTREME caution. I know you've probably already heard this, but it is necessary to repeat because of the danger involved.

This forum has strict rules about discussion of dangerous and potentially lethal topics (and reasonably so) . Discussion of this topic (Line-operated SMPSs) is, of course, permitted here, provided that proper cautions are mentioned, the proper precautions are taken, and that the poster(s) asking the questions demonstrate a reasonably good knowledge of what they're doing.

You will find that there are several SMPS gurus here in the forum (EVA, jackinnj, poobah, luks, chas1, just to name a few), and sooner or later they may chime in and give their thoughts. Lots of good advice here.

Steve