1800W 50V Power supply

[sourcecharge]

So this is my first post here, and I just would like to give a little background on my tech capabilities.

I have extensive research in inductor design and production and multi-phase resonant circuits.

Recently I have been trying to make a new shuttle for my homemade toroid winder as the one I have is not perfectly straight,

so I wanted to use my 3d printer for a lost PLA process that I will use for other things too...

So I'm using a 1800W ZVS Induction Heater...

I'm using a 1 1/4'' iron nipple with cap and I was able to melt aluminum in it...

The problem I had was that I had to use all of my power supplies to give the heater enough power...

I had two options:

Buy a power supply:

Almost $300 for a 50V 20A ps to put in parallel with one I already have giving 50V and 40A, or I could get one for almost $600 for 50V and 50A.....

These Power supplies still wouldn't power the 48V fans that are on the main power of the heater if the load brought the main power below 48V....

So I would have to have a 3rd power supply to make sure the fans had a constant voltage

Using 2 or 3 power supplies just is not realistic....

2nd option, make my own power supply:

So I decided that I wanted to make my own power supply because I already had two T650-26 toroids laying around...

these are massive powered iron toroids...

funny thing is that at 6k Guass, the permeability of this material increases by 3x, so the 119mH  14AWG primary is actually about 390mH...

this seems to be the case for the 2 secondaries too...

Anyways, long story short, I put the fans on their own circuit so that there would be a constant voltage for the Fans vs the power side...

and I used an old computer case with a computer power supply to supply 12v...

Here is my design of a regulated power supply using some mosfets and zeners that I had laying around...

(N ch: IRL530n P ch: IRF9530 Zeners: 1N4740 bridge is 60A600V microsemi chase mount bridge)

The power side (top) has at least 12 zeners and 12 P channel mosfets in parallel to handle the current.

The 23.9mF cap is a 100V while the 182mF cap is actually 2 caps at 63V each

The power supply is variable between the zener addition using a 1Meg pot in parallel.

computer power supply is on it's own switch on the same input mains power from the 240AC input..

I got a 15A fuse on the primary and a 40 amp fuse on the output of the secondary.

Note that a lot of these values are really only because of what I had laying around...

I have tested this circuit already and it seem to work so far...

It will probably do about 49V and 40A but at a regulated 50V it seems to be 30A easy..

Hope I did well!!

Let me know what you think!

The only real question I have about this is why is it soo inefficient?

[TiN]

11 amp MOSFETs rated for 88W for 36A PSU?
Something feels off.. 

[sourcecharge]

11 amp MOSFETs rated for 88W for 36A PSU?
Something feels off.. 

I assume you are referring to the IRF9530 p ch mosfet.

I put 12 of these in parallel to handle the current....

These are the ones I used:

https://www.infineon.com/dgdl/irf9530n.pdf?fileId=5546d462533600a401535611b4a71dbf

[TiN]

I actually looked at IRL530. Also MOSFETs cannot be simply paralleled due to risk of runaway.

[sourcecharge]

I actually looked at IRL530. Also MOSFETs cannot be simply paralleled due to risk of runaway.

Well, remember, I'm only using stuff that I had laying around so there are components that are probably way overkill....

The mosfets in parallel may have a problem, I don't know,

I actually haven't fully tested it at full load yet, only at 40V and 20A due to the lack of a resistor that is less than 2 ohms that could handle the power...

They are all on the same massive heatsink, right next to each other...

I am using the tab of each TO-220 as the drain and taking a point in the middle of the heatsink as the output of the "module" as what I am calling this thing..

The regulation circuits work though....haven't seen this done with mosfets too much...

thought maybe someone might be interested...

[fourtytwo42]

The schematic looks sane but what tool are you using as some of the symbols are unfamilier ?

Are you asking why a linear power supply is inefficient ?

[sourcecharge]

The schematic looks sane but what tool are you using as some of the symbols are unfamilier ?

Are you asking why a linear power supply is inefficient ?

I use B2spice..

which symbols are you referring to?


Lastly, basically ya, why is a linear power supply so inefficient?

[fourtytwo42]

The horizontal oblong boxes with an arrow or +/- at the ends, are they measurement points or something ?

Poor efficiency is because the regulating device is carrying the full load current and dropping the voltage between the bridge rectified and smoothed transformer output and the output voltage. I dont know what your figures are but supposing the unregulated voltage is average 65V and the output is 50V@30A the regulating device will dissipate 15x30 = 450Watts!! Another way of expressing this is to say the efficiency is only 77%.
Of course as the voltage drop across the regulator increases the efficiency worsens.
OTH a simple buck switcher could be expected to deliver an efficiency of at least 85% across a broad range of voltage drop.

[sourcecharge]

The horizontal oblong boxes with an arrow or +/- at the ends, are they measurement points or something ?

Poor efficiency is because the regulating device is carrying the full load current and dropping the voltage between the bridge rectified and smoothed transformer output and the output voltage. I dont know what your figures are but supposing the unregulated voltage is average 65V and the output is 50V@30A the regulating device will dissipate 15x30 = 450Watts!! Another way of expressing this is to say the efficiency is only 77%.
Of course as the voltage drop across the regulator increases the efficiency worsens.
OTH a simple buck switcher could be expected to deliver an efficiency of at least 85% across a broad range of voltage drop.

Yep, the horizontal oblong boxes with an arrow (current) or +/- (voltage) at the ends are current and voltage meters....

Sorry..

But ya, I don't know.....is that the real reason why it's so inefficient?

Because, it's inefficient without the regulation...like only 50% or so..

[fourtytwo42]

I guess you are measuring the AC input power, then maybe your transformers have a huge magnetization current ? Try measuring there load with secondaries disconnected ?

[sourcecharge]

I guess you are measuring the AC input power, then maybe your transformers have a huge magnetization current ? Try measuring there load with secondaries disconnected ?

In no load condition, (with the secondaries disconnected from any load), the primary pulls 1.65 A...(measured with current clamp)

And that's from the 240AC mains...

IDK, maybe some type of SMPS would be more worth doing but it would require alot more time and parts...

[fourtytwo42]

So I decided that I wanted to make my own power supply because I already had two T650-26 toroids laying around...

these are massive powered iron toroids...

What an earth persuaded you these toroids were suitable for 50hZ power transformer ??

I should think your primary inductance is far to low for 50hZ, I have never heard of anybody trying to use a powdered iron core as a mains transformer............I had naturally assumed the "massive toroids" you were talking about were normal silicon steel strip cores (my fault for not paying attention).

[sourcecharge]

So I decided that I wanted to make my own power supply because I already had two T650-26 toroids laying around...

these are massive powered iron toroids...

What an earth persuaded you these toroids were suitable for 50hZ power transformer ??

I should think your primary inductance is far to low for 50hZ, I have never heard of anybody trying to use a powdered iron core as a mains transformer............I had naturally assumed the "massive toroids" you were talking about were normal silicon steel strip cores (my fault for not paying attention).

I didn't have any of those type of cores laying around?

I stack two of the T650-26 together which gave a 868 AL, where normal ferrite or stripped silicon steel probalbly have ALs of like 4000 to 20000....

So ya, that might be a problem...

[sourcecharge]

Well, btw, it was a calculated ...

B(max) = V(rms) x 10^8 / (sqr(2) x pi x N x F x A(e))

which gave about 6000 guass,

This actually is not saturating the core...soo why not?

Other than the low efficiency, it actually should work...

Am I wrong?

[NiHaoMike]

Just get a "telecom rectifier".

[mikeselectricstuff]

Just get a "telecom rectifier".

I was about to post this. Get a used telecom rectifier from a decommissioned telecom server farm is the cheapest way to get high power 48V power supplies.

This. 48V is not commonly used outside of telecom so extremely high quality surplus PSUs go for peanuts

[sourcecharge]

Well, if this thing doesn't work, I might have to either redesign it or look for another solution...

How much do those used power supplies cost and where can I find them?

[NiHaoMike]

Well, if this thing doesn't work, I might have to either redesign it or look for another solution...

How much do those used power supplies cost and where can I find them?

I got mine for $15 from Ebay. I have yet to actually use it for anything that needs very much current, most likely what I end up using it in would involve some sort of motor.
https://www.eevblog.com/forum/reviews/teardown-fuji-48v-40a-psu/

This. 48V is not commonly used outside of telecom so extremely high quality surplus PSUs go for peanuts

A "common" hobbyist use is large audio amplifiers.

[sourcecharge]

Well, if this thing doesn't work, I might have to either redesign it or look for another solution...

How much do those used power supplies cost and where can I find them?

I got mine for $15 from Ebay. I have yet to actually use it for anything that needs very much current, most likely what I end up using it in would involve some sort of motor.
https://www.eevblog.com/forum/reviews/teardown-fuji-48v-40a-psu/

This. 48V is not commonly used outside of telecom so extremely high quality surplus PSUs go for peanuts

A "common" hobbyist use is large audio amplifiers.

wow 48V and 40A for 15 bucks!

You just can't beat that...

[sourcecharge]

You should try out induction heating if you already haven't....

it's fun and you can make a whole bunch of stuff now using a 3d printer...

I'm sure you know about it, but just in case you don't, check out, "lost pla"

So using a 3d printer is no long about just plastic, you can make metal objects designed from a 3d cad program, I use Freecad..

just search for 1800W ZVS on ebay to find a good heater for cheap...(<$100)



I did a search for "telcom rectifier' in ebay but didn't give me the same luck that you had, maybe in the future I might find a good deal...



2nd test of my DIY PS didn't go so well, a 40+ yr old br-84d bridge  burnt up with only the fan as the load, but the power side worked perfect at 40V and 20A...

I'm guessing the bridge was bad, because the bridge is rated for 2A, and I'm only pulling 0.7A at 63V peak through the transformer secondary...

It's rated at 50A surge current and I'm only charging <10mF up to 48V..

I'll figure it out though...

Thanks for the ideas!

[NiHaoMike]

Good 1800W induction cookers start at $50 or so. (I have a $20 1300W induction cooker that I plan to mod into a precision temperature controller, but I don't have the time to try reverse engineering the control ASIC quite yet...) That said, I don't think induction is the best way to melt aluminum. Rinoa Super-Genius had great luck melting down aluminum with carbon rod arcs.

[sourcecharge]

Good 1800W induction cookers start at $50 or so. (I have a $20 1300W induction cooker that I plan to mod into a precision temperature controller, but I don't have the time to try reverse engineering the control ASIC quite yet...) That said, I don't think induction is the best way to melt aluminum. Rinoa Super-Genius had great luck melting down aluminum with carbon rod arcs.

I'm using Iron pipes with screw on caps as crucibles and melting the aluminum inside. 

Works great, and very efficient with the royer ZVS type drivers...80% - 90%

My work coil is about 11.5 uH before the pipe is in the coil, and about 13 uH when the pipe is in the coil.....measured at 10khz...


I was looking into a good blast furnace, but it seemed messy..


How's that carbon rod arcing work to melt, let's say 300cm^3 worth of aluminum?

[T3sl4co1l]

Didja check the core loss at that frequency and flux density?  I think you will be surprised...

0.3H at 60Hz is also 113 ohms or 1A at 120V, or 2.5A at 50Hz 240V.  That's a lot of magnetizing current.

Tim

[sourcecharge]

Didja check the core loss at that frequency and flux density?  I think you will be surprised...

0.3H at 60Hz is also 113 ohms or 1A at 120V, or 2.5A at 50Hz 240V.  That's a lot of magnetizing current.

Tim

It's actually about 0.39H so with the 240 AC mains, its about 1.65A(rms) measured with the current clamp...

If I calculated the R(ac) from the 0.39H value the resistance would be way too high to even get 10 amps out of it, so it seems that R(ac) is only probably calculated by the uncoupled inductance....and since I'm using a toroid, the coupling factor is about 0.9996...

Therefore core loss is very low...

IDK, it seems like the 50% efficiency is probably due to the charging and discharging of the capacitors or the loss from the bridge....

Guess that's the price to pay for DC...IDK

[sourcecharge]

Did my first test with a small iron crucible, 48V and 36A....

(at least the panel meters on the unit said 36A but simulations show about 40A as that panel meter is not very acurate)

It held for about 3 to 4 mins, and the iron crucible was up to about 700F, then it started to decrease in power...


Then, something started to smoke and I think the breaker tripped...

So the mosfets held up, but the 60A 600V bridge had a rectifier fail, and simulations show that it would have pulled 30+ amps on the primary with a dead short of one of the rectifiers on the full bridge.

My bad, I didn't manage it's heat well enough...

Anyways, cheap new bridges on the way...

[sourcecharge]

Well, it was worse than I thought, the secondary burned through and possibly shorted to the primary.

So, since this is no longer simply a matter of doing it easy and cheap, I went and bought 10 lbs of magnet wire.

5 lbs 14 AWG for the primary and 5 lbs 10 AWG for the secondary, both 200C rated.

The current capacity of both wires should be way higher than needed but this time I'm not taking any chances along with a new 100A 1000V bridge...

I think this actually could withstand about 3kw, but Im limited with the heater as only 1800W.

So, ya I was looking at those rectifiers and one was like 160 bucks and another was like 68 bucks, but I'm just not sure what's good or if they can be used exactly like a normal PS or if they need some type of battery bank or capacitors...

Since there is a lack of interest in DIY High powered PS with voltage regulation, this will be my last update unless someone asks.

Thx for all your guys' input 

[T3sl4co1l]

Yeah, that sounds quite beefy.  As long as it's not copper clad aluminum, which may cause some problems (~half current density, making connections).  But even that's not so bad, for what you're doing, really.

Rectifiers suddenly rise in price at 50A.  Amazingly, 50A just barely fits inside a regular 10-35A size bridge module, very cheap.  After that is just industrial modules, starting at maybe $30 or so, and kind of just going up and up (no, you don't get much of what you pay for, in this case).  So, the $68 part is probably okay, can probably find something cheaper and still good enough, definitely no need to gouge yourself on the $160 part...

You don't actually need a filter cap (which will need to be quite large indeed to do its job), if you don't mind pulsating output.  However, note that you must keep gate drive powered up at all times.  Use a separate 12V supply, find the bias resistors and hack them to a separate supply.

Tim

[dmills]

I suspect by Rectifier the OP was talking about a telecom "Rectifier" module (Which everyone else calls a power supply), basically an industrial grade switchmode supply usually in some sort of hot plug form factor.

I suspect the reason the OP is getting so little traction in terms of building his own is because a 2kW 48V supply with power factor and all the trimmings on ebay is cheap enough that rolling your own is just not that interesting to most here.

See for example Ebay.com 152935516200 which is 100A@48-54V for $40.

I don't know exactly what you would need to jumper to what on that one to get it going, but it is usually something that can be figured out. 

73 Dan.

[T3sl4co1l]

I suspect by Rectifier the OP was talking about a telecom "Rectifier" module (Which everyone else calls a power supply), basically an industrial grade switchmode supply usually in some sort of hot plug form factor.

Oh, yeah, we haven't talked about those in a little while.

In case anyone is wondering, that's a fine value for a power converter.

Anything going for less than $1/W is a good deal.  That computer PSUs, "rectifiers", inverter microwaves, induction cooktops and more, can sell for $0.1/W or less, is between stupendous and ludicrous.

Tim

[dmills]

Yea you are probably paying cents on the dollar compared to the original pricing.

The Eltek Twinpacks are good value at the moment on US ebay, just a few percent less efficient then the latest generation but that makes them worth replacing in a 24/7/365 application where energy costs dominate, $50 for 52V @ 50A, gets my vote!

Be careful out there, it is only 50V or so, but there is PLENTY of energy there to do damage and some of these are positive ground (Telecom standard for corrosion reasons). 

One of the RC hobby groups has the thread that goes into much detail on what to jumper to what to get these things to come on.

73, Dan.