Full Bridge 90A SMPS...

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've been reading and studying and I want to build a 90A supply for ham use. If I am in the wrong place please let me know. I aquired an EE65-32-27 core and am currently searching for a bobbin supplier for the FE0300 bobbin. I'm not an engineer, so any help would be greatly appreciated. I have plenty of parts and would be willing to give a complete set to someone for helping me design this supply. My specs are 117V AC input at 60hz, 13.8V output at 90A. I need help with the wire size, number and method of turns, coil design, and some basic layout help for the PFC/transistor configuration. The simplest design with the least parts would be the best. If anyone is interested in helping, I will upload my parts list and ideas. My idea for PFC/PWM transistors are Fairchild FCH47N60 MOSFETs. Will those work?

Thanks, Hawkfeather
 
I think that your goals are far too ambitious for a first SMPS project. You are willing to get 90A at 13.8V with some efficiency and to use a PFC input stage to avoid extreme voltage drops and distortion on mains line, but you can't figure out how to calculate suitable magnet wires for a transformer.

Furthermore, you have already chosen a E65 core with no particular criteria, without knowing that a E42 or E55 may be enough for your needs (some calculations may reveal this). And you have also chosen a MOSFET with very high current rating without considering that such devices are very hard to drive and suffer from high capacitances, and that a typical 110V PFC stage outputs something like 230V DC, so 5A are just enough to get over 1KW output from a full-bridge PWM stage.

Selecting parts randomly is certainly not the most efficient way of doing things. Try to do some calculations first, learn exactly how each stage works, look for formulas in application notes, figure out the minimum and maximum values of the currents and voltages involved in each power device in each stage. Then, selecting devices (and even magnet wires) will be much easier.

PD. Trust me, then again, I've been able to develop this from scratch with very little money through rational design: http://www.diyaudio.com/forums/showthread.php?s=&threadid=60969
 
So Eva... from what you are saying, the 47A MOSFETs are overkill for a 90A or better SMPS. Got that. I will take your advice and study more as I am really eager to learn and am very interested in electronic design. Like I said, I'm no engineer, but common sense and listening/reading skills go a long way. Thank you for your advice. I think my topology choice is right on though. Just need to tweek my component studies a bit and start out with each section, step-by-step. The Rds (on) for the FCH47N60 is .07 Ohm. But the 47A rating I will have to change by choosing another MOSFET. In a full bridge topology, with four drive transisors, 15A would suffice for up to 3.4kW. Am I correct in this? Say I want to use the EE65 core or a 55 size. Then I would be able to go higher than the 90A? Like I said. I have a humongous parts bin, I just need to learn how to utilize them best!
Thanks again, Hawk
 
hawkfeather said:
So Eva... from what you are saying, the 47A MOSFETs are overkill for a 90A or better SMPS. Got that. I will take your advice and study more as I am really eager to learn and am very interested in electronic design. Like I said, I'm no engineer, but common sense and listening/reading skills go a long way. Thank you for your advice. I think my topology choice is right on though. Just need to tweek my component studies a bit and start out with each section, step-by-step. The Rds (on) for the FCH47N60 is .07 Ohm. But the 47A rating I will have to change by choosing another MOSFET. In a full bridge topology, with four drive transisors, 15A would suffice for up to 3.4kW. Am I correct in this? Say I want to use the EE65 core or a 55 size. Then I would be able to go higher than the 90A? Like I said. I have a humongous parts bin, I just need to learn how to utilize them best!
Thanks again, Hawk

If you have couple of hours free time daily you should be able to read TI's APT's and ixys appnotes and online seminars in 2 months :)

First thing first tough, I suppose you have isolation transformer and 100Mhz scope at your use? I see supprising amount of people trying to build swithcers without scope, its #¤!"# waste of time.

1kW-range offline SMPS is not the thing to start with, build some small 12v push-pull pwr supply for starters at first. (altough i know one guy who started smps-stuff at 5kW)
 
Hawkfeather:

15 A transistors won't suffice for a 1KW SMPS (unless you parallel them).


you would have about 7 A rms current on the primary, and at 50% duty cycle (100% by some) the peak current through each device can reach 28A.

You can't use the old 60Hz transformer math here. Because of very high peak-to-average current ratios, SMPS's follow a math that often seems counter-intuitive to some.

Driving such MOSFETS can require driver circuitry with several Amps capability.


Funberry
 
funberry:

At such power levels, I assumed that everybody was thinking about a fixed frequency continuous mode PFC and a (full bridge) transformer coupled buck converter with low current ripple.

In such a PFC, with low line conditions (76V AC input), the switches are required to conduct only approx. 20A peak (including some unavoidable inductor ripple current), and RMS current is substantially lower than that (approx 15A) since duty cycle approaches 50% when the current amplitude of the PFC is at its maximum value. The magnitude of the required currents with 110V AC input is somewhar lower (13A peak ana probably approx. 8A rms)

Also, in such a buck converter, the supply rail is pseudo-regulated to approx 250V thanks to the PFC, so little duty cycle margin is required, 80% nominal may suffice. For a 15V output (13.8V plus voltage drops), a 13:1 transformer would work fine, yielding a primary peak current of 90/13=7A plus inductor ripple (let's say 7+1=8A), and also a RMS current of approx 7A due to the high duty cycle.

So I'm quite curious concerning where did you got that '28A peak' figure.

Also, concerning 50/60Hz transformer-rectifier-capacitor PSUs, they always work with approx 30% duty cycle. So mains RMS current, transformer RMS current, and diode RMS current are *always* three times higher than the average output current, and everything gets *nine* times hotter than it would get when powering a resistive load.

Actually it's quite easy to overheat and blow a 1KVA toroid if 1KW of rectified and filtered DC is drawn continuously from it, because there are as much losses as there would be with a 3KW resistive (non-rectified) load. Fortunately, audio amplifiers draw in average only 10..20% of its rated power when driven with music signals at full output swing...
 
Hawkfeather, is there any compelling reason why you want to have a PFC stage in this PS?

(forgive me anyone who likes PFC, because I resent it)

I have built SMPS with a boost converter pre-regulator, yes, to reduce ripple and do line regulation ( and reduce size of filter capacitors), but that's different from PFC, because it made no effort to reduce harmonics on the line.

I think recent regulatory constraints about line harmonics are oppressive (ok, flame me for having this opinion).

But European regulatory agencies are making it much harder for smaller companies to market electronic products, reducing competition in the process.
And all the nonsense about creepage and clearance, is forcing products to have larger, heavier, magnetics, higher leakage inductance, and are less optimized.

Meanwhile. the Chinese are stamping products with the CE logo that don't even comply.

A vendor needs to listen to the needs of the customer, and supply what the customer needs. Power utilities are using technology that hasn't been updated in over a century. The burden of tailoring the product to the need is, conceptually, on the vendor.
 
telewatt said:
Hawkfeather, is there any compelling reason why you want to have a PFC stage in this PS?

(forgive me anyone who likes PFC, because I resent it)

I have built SMPS with a boost converter pre-regulator, yes, to reduce ripple and do line regulation ( and reduce size of filter capacitors), but that's different from PFC, because it made no effort to reduce harmonics on the line.

I think recent regulatory constraints about line harmonics are oppressive (ok, flame me for having this opinion).

But European regulatory agencies are making it much harder for smaller companies to market electronic products, reducing competition in the process.
And all the nonsense about creepage and clearance, is forcing products to have larger, heavier, magnetics, higher leakage inductance, and are less optimized.

Meanwhile. the Chinese are stamping products with the CE logo that don't even comply.

A vendor needs to listen to the needs of the customer, and supply what the customer needs. Power utilities are using technology that hasn't been updated in over a century. The burden of tailoring the product to the need is, conceptually, on the vendor.

Personally i dont give a sh1t about PFC in my projects. For hobbyist it just doesnt matter, one non-pfc 2kw power supply made by rare hobbyist is not going to change the big picture anyways. For larger-scale effects there is more justification for it as nowadays almost everything works with smps.

With creepages and clearances I dont agree with you, there is good reason for them and I happen to like myself... who cares if others get fried.

Why dont you also just stamp CE-markings on your equipment?

BTW, does anyone know how EMC-regulations are working in practice? At least in here they make random tests on imported chinese **** on safety basis but i believe that nobody actually cares about EMC? Reveal/burn your chinese competitor if they dont comply with emc-regulations?
 
It's actually quite funny to draw 2KW of rectified power from mains in a pulsed way. This causes the lights of the entire flat or building to dim in a more than noticeable way, and also causes a standard 15A plug (and wire) to get somewhat hot. Furthermore, mains waveform gets so clipped that it starts lookng more like a square or trapezoidal wave _/¯\_/¯\_/¯\. This obviously causes the rectified voltage to drop a lot, it's not uncommon to get 320V DC with no load and 275V DC at full load (with +-20V ripple or so).

On the other hand, when 2KW are drawn through a PFC, all these effects are minimized or disappear. (Yes, I have the prototypes and I have done these tests). Unfortunately, a 2KW PFC adds something like 50% more complxity, weight, space used and parts to a 2KW SMPS.

PD. Drawing 2KW of rectified power from a 110V mains line is probably going to be even funnier...
 
Eva said:
It's actually quite funny to draw 2KW of rectified power from mains in a pulsed way. This causes the lights of the entire flat or building to dim in a more than noticeable way, and also causes a standard 15A plug (and wire) to get somewhat hot. Furthermore, mains waveform gets so clipped that it starts lookng more like a square or trapezoidal wave _/¯\_/¯\_/¯\. This obviously causes the rectified voltage to drop a lot, it's not uncommon to get 320V DC with no load and 275V DC at full load (with +-20V ripple or so).

On the other hand, when 2KW are drawn through a PFC, all these effects are minimized or disappear. (Yes, I have the prototypes and I have done these tests). Unfortunately, a 2KW PFC adds something like 50% more complxity, weight, space used and parts to a 2KW SMPS.

PD. Drawing 2KW of rectified power from a 110V mains line is probably going to be even funnier...
Yeah, for sure waveform looks ugly :D
At 2.4kW I measured 298vdc on filtering caps (8000uF)
RMS current is about 16A at this point..
4.5-5kW around 270v, not sure as 16A slow-blow lasts apprx 10 seconds.
Your numbers sound reasonable also, I guess i have quite short cablings.

2-3.5kW PFC stage is next gonna-be project as I need to max out 10A fuses. :eek:


110v without PFC at these power levels, just forget about it. :)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.