looking for high power regulator

[Rerouter]

like has been said before, for that level of wattage, it would probably be wise to use a switching pre-regulator, then you could just as well get away with an op amp and a big pass transistor like a 2n3055t to stabalise it and remove most of the noise of the switcher, while keeping your wattage down to near nothing,

i say op amp instead of a LT3080 or similar being how you can easily manage a small negative tap for each rail, so can manage down to 0V without issue, just would need a op amp able to supply a few hundred mA, or with a darlington arrangement, just a stock standard quad op amp package, so you could even play with constant current, and making the switcher track the linear output,

though if you want the negative rail trick can be used just as effectivly with any other LDO regulator to pull down its set pin to offset its dropout voltage, and make a minimum load much simpler,

[eevblogfan]

hey

thank you for helping me .

are you talking about 300w ?  or 75W ?   either of them . I need the power supply to be linear not half switching (I saw the "tracing pre regulator" ) .

I also need it to be made out of mosfet's ( I am working on the circuit but from some reason I cant make the simple "linear Dave's power supply" on the LTspice .

I also notice that changing the sort of opamp give's me whole another result .  if so . what opamp should I chose for that purpose ( the simulation only . ) ? 

I think the 75W should be dissipated by 2 IRFP064N's (TO-220) , perhaps each one can Handel about 40W simultaneously .  so two will be able to Handel 75W no problem ,

since I am newbie , I'll ask , what is the difference between "LM" and "LT"( linear tec ) and  lm ?

and I'll ask . why do I need to use transistor (current controlled and not so much efficient ) instead of mosfet (voltage controlled . very efficient, in terms of Vd at full load .  )

thank you in advance !

[T4P]

hey

thank you for helping me .

are you talking about 300w ?  or 75W ?   either of them . I need the power supply to be linear not half switching (I saw the "tracing pre regulator" ) .

I also need it to be made out of mosfet's ( I am working on the circuit but from some reason I cant make the simple "linear Dave's power supply" on the LTspice .

I also notice that changing the sort of opamp give's me whole another result .  if so . what opamp should I chose for that purpose ( the simulation only . ) ? 

I think the 75W should be dissipated by 2 IRFP064N's (TO-220) , perhaps each one can Handel about 40W simultaneously .  so two will be able to Handel 75W no problem ,

since I am newbie , I'll ask , what is the difference between "LM" and "LT"( linear tec ) and  lm ?

and I'll ask . why do I need to use transistor (current controlled and not so much efficient ) instead of mosfet (voltage controlled . very efficient, in terms of Vd at full load .  )

thank you in advance !

ALL Transistors are voltage controlled . http://amasci.com/amateur/transis.html
LM is by national semicon and LT is by Linear Tech .
75W is a shitload of heat to get rid off  , you can have a switcher and then a linear after that, less heat to get rid off , less money spent on a massive heatsink .

[eevblogfan]

hey

thank you ,

I know that they're not really "current controlled"  but with mosfet's you can put only 200uA or so in order to fully open the gate and with  transistors you really can't , ( much higher current ) .

the reason I am not buying a power supply is because all of the costy stuff are here on my desk (IE . power transfromer , and big heat sink, and 2X4.1/5 20,000 count panel meters ,

now I "only" need the circuit design and circuit manufacture and I finish ! (I know , it's much more harder )

now , on the begining of the power supply series , dave sais that he can also do that with mosfet's but he will not explain , would someone explain me how to ?

I know that the "usual" opamp+mosfet/transistor CC either CV circuit is so noisy , please tell me you have some chip either some sort of circuit who can fit here. even a 30W one ( I'll build 3 of them in order to handel the ammount of power needed ,)


thank you in advance !

[T4P]

hey

thank you ,

I know that they're not really "current controlled"  but with mosfet's you can put only 200uA or so in order to fully open the gate and with  transistors you really can't , ( much higher current ) .

the reason I am not buying a power supply is because all of the costy stuff are here on my desk (IE . power transfromer , and big heat sink, and 2X4.1/5 20,000 count panel meters ,

now I "only" need the circuit design and circuit manufacture and I finish ! (I know , it's much more harder )

now , on the begining of the power supply series , dave sais that he can also do that with mosfet's but he will not explain , would someone explain me how to ?

I know that the "usual" opamp+mosfet/transistor CC either CV circuit is so noisy , please tell me you have some chip either some sort of circuit who can fit here. even a 30W one ( I'll build 3 of them in order to handel the ammount of power needed ,)


thank you in advance !

With that sort of stuff , i give the thumbs up to go true linear ! How big is it actually ? 20,000 counts panel meters ! Where did you get it ? You must have paid a bomb if it wasn't free !

[eevblogfan]

hey

sarcasm ? either serious ?

BTW , 20,000 count meter isn't as costy as big 300W either 1KW power transformer , or big nice heat sink ,

thank you in advance !

[Rerouter]

The half linear design is simply so you can get away with small thermal considerations, if the linear only needs to handle 1-5w in place of the full 75w per rail.

as for the MOSFET to regulate with them means operating them in there linear region, just below the rated voltage for rson, however mosfets are not ideal for a supply as they have gate capacitance delaying there response to transients unless driven quite hard

As for op amp regulators being noisy its is partly correct was just throwing up the option any regulator with a set pin can function adequately pulling it below zero with some pissy little milliamp rated winding is all thats required, and allows an output voltage down to zero, with how much voltage just being the net offset voltage of the regulator and the pass element

[Rerouter]

Even though you have the giant heatsink I would personally prefer to not have a piece of metal close to or exceeding 100 degrees in my lab, but if you wish to go linear then by all means sounds like you have the pieces to do it. As a secondary throught if you went with full linear how would you isolate you few thousand uF capacitors from boiling

I don't mean to sound pushy just been jumping through these hoops myself recently

[eevblogfan]

hey

yes . I do have bearskin   ( I indeed get burn regularly and it "stops" been hurt )

BTW , 100C with  0.064C/w heat sink will need about 1250W (with ambient of 20C )

so it's kind of allot power

assuming I do want this 75W power supply be linear , and I do want to get burn , will you help me figure out the circuit needed ?

thank you in advance  !

[T4P]

I can figure out how my largest heatsink is extremely inferior to your heatsink 


I believe all you need is to use BJT's instead , controlled by a LT3080 .
Let the others help you out , i'm not too good at this area yet .

[A Hellene]

eevblogfan, here is a couple of suggestions on how to use the LT3080 to drive a 300++ W power stage. Q2 is a (low saturation voltage) current amplifier that drives two (or more) forced air cooled 150W (or more) power transistors. Of course, to fully utilise the LT3080 capabilities you might need faster power transistors than the 2N3772/2N3055 shown at the drafts.

[...]
Additionally, if your LT3080 drives a power stage instead of the load directly, R8 shuts the power stage off, providing the first milliamperes to the load by the LT3080 itself and after that the power stage takes over; R9 ensures that the LT3080 will never provide more than 40mA, keeping it always cool on the PCB -no need to mount it in the hot heat-sink and have its characteristics drift off.

-George

[Rerouter]

sorry about that, combination of overlooking the junction and iphone "auto-correct" still as for a mosfet, i am really struggling to find a design that responds to wide variations in loads,

however this seems to be one you can play with simulating, http://cdn.head-fi.org/9/91/1000x500px-LL-91977a40_mosfet-voltage-regulator-circuit-diagram.PNG

if uses a voltage doubler arrangement to get enough voltage to properly drive the mosfet, though you could well just use another puny winding to get a higher one,

as for the actual transformer, i assume you know the number of windings in the primary and how to calculate how many you will need on the secondary, if not just post the number on the primary, and the gauge of wire your using, (will be a voltage drop in the windings that needs to be compensated)

to go furthur into that circuit, it would seem that the higher voltage is being fed in and being pulled down for control. with some feedback off the output, then fed into an adjustable voltage reference to keep it accurate and stable,

[eevblogfan]

hey

I cant really tell you what is the winding count , because I didn't really count it . I just winded ><"

well I ran a little experiment , and the result are :

no load

V input : 223.4VRMS (the 87V says)
V output: 4.355VRMS (the 287 says)



now with load

V input : didn't cheked ,
V output : 3.62VRMS (with 50W 12V halogen lamp )
A lamp : 1.93A (the 87V says)

number of windings on the secondary :5 ( 9m'm wide cable)

I hope it will help .

thank you in advance  ! 

[T4P]

You most certainly have to rewind ... 15VAC is the target .

[Rerouter]

(4.355-3.62)x1.93 equals about 1.42ohm of impedence on the winding, almost 1cm thick seems a tad crazy to think of unwinding,

now just at a guess, i am going to guess your going to remove that and wind on your own with thinner gauge wire, if so, count the number of turns, at the moment the ratio is somewhat close to 52:1 primary to secondary, however you will need to bring that close to 13.5:1 to get it able to handle 16V under a 5A load, 16V as you will have some drop out voltage,

as for the guage of wire, i would say something close to 2.5mm CSA, or AWG10, you can get away with less, this is just trying to keep the temperature increase in the windings down, any neagtive or positive biasing windings can be as low as AWG30 as its milliamps,

[eevblogfan]

hey

thank you

the "big transformer" is 2 "E" core out of 2 microwave transformers , and they re held together with a"hand held vise " , so in order to "rewind" I just have to release the vise and then pole the windings out of it  

and the winding will be as easy as needed ,

BTW :

I measured the size of the transformer .

the S of the winding area is (1.4X5X6.5)=45.5CM squared (for each side )
 
the H is 5CM and the W is 1.4CM and the D is 6.5CM (each side )

is that help ?

BTW , I don't have 10AWG winding wire , and I remember someone has told me that as the thickness of the winding wire goes low , so as the loss of the copper goes low , is that true ?

if so , I think I have like 150 meter's long 18AWG or so , is that better one ? I mean, 18AWG can pass about 8A no problem (http://www.geocities.com/CapeCanaveral/Lab/9643/awg.htm)

so that's the wire I was going to wind with , am I wrong ? IE , 10AWG can pass 32A , isn't is crazy ?

oh and I don't know If I am right , the Resistance of 10m 18AWG is 0.214ohm , so the max short circuit current is 16/0.214=74.76A ?

if so , the max current of 10m 10AWG will be : 0.0334ohm , so 16/0.0334=479A .

almost any diode will die with that sort of current , huh ? . well I hope I am wrong with that calculation ( I don't understand transformer's calculations , and I know you've got the inductance and impedance that should be count either ,)

oh and BTW , is the usual schottky rectifier diodes (in ATX power supply ) will  be good for that purpose ( IE , full wave rectifier built by them  )

if so , I have so many , I can build a welder rectifier out of them


thank you in advance !

[Rerouter]

the other stuff i will have to wait till tommorow to have a good think about, as for AWG10 it was soley a reccomendation, being how the nature of bunched up windings mean they get a little hot from any losses,

i wasnt working of theoretical maximum current, just peak capacitive loading, and keeping your temp and by such losses a little lower, AWG18 still feasable.

as for diodes, aim for high current, the large value cap will on start up try sucking down all it can get, that was why i aimed for such high currents, up to 50A for ~ 1/15th of a second on switch on,

still i will say knowing exactly how many turns on that output instead would make it much easier to spit out a value for you to wind to, but by tommorow i might have remebered something to work off the measurements,

[eevblogfan]

hey

based on the ratio , I calculated the number of turns on the primary coil . it was 260turns,

now , the coil is "super heavy " heat shield, IE , very thick isolation . besides the usual Emile isolation . it has some yellow ( I don't know what is that stuff , but it keep leaving good with nearly 300C )

and he is about 1mm thick (18AWG)

so , I'll ask , the heavy shielding causes loss in terms of room inside the core , I'll ask if I can wind another primary , this time out of my 18AWG coil . how many windings should I do ? (I haven't measured the bias current taken while without any load , but I assume it's allot ,

less winding then 260 , will rise the given voltage achieved by one winding on the secondary coil ,while more then that will lower it , assuming the lowest voltage will be 203VRMS , and the highest will be 234V. at 50Hz ,
 

thank you in advance  !

[T4P]

You mean "quiescent current" don't you ?

[eevblogfan]

hey

I was tier when wrote this , please forgive me for any silly mistake , I ment the current that the transformer takes from the wall while ther's no load ,

thank you in advance !

[eevblogfan]

hey

I ran a simulation :



and the results :



can someone tell me where I am mistaking ?

thank you and sorry for double post .

[A Hellene]

hey

I ran a simulation :
[...]
and the results :
[...]

can someone tell me where I am mistaking ?

thank you and sorry for double post .

Hello eevblogfan,

If you are addressing me, the author of the orignal design you have modified and posted above, I expect that you have already taken into account my numerous remarks about:

1. The very long overload/saturation recovery times the vast majority of the Zero-Drift op-amps suffer from. For example, the LTC2050 input overload recovery time is 2ms, which makes the device too slow to participate in such a fast (>100kHz * 2..10) feedback loop.

2. The response time and the stability of the LT6105 current sense amplifier, especially at that high amplification gains (20 V/V), where noise becomes an issue.

3. The original design being a draft, which means that there are several oversights or omissions. For example, the LT3080's operating current of a few milliampers is directly fed into the load, bypassing the current monitor circuitry.

Finally, that I posted the design as a draft, in order to suggest a couple of possible solutions for a low voltage op-amp that drives a high voltage load, and not as a design complete and ready to be implemented.

This is hardly a finished design; it is just a proof-of-concept model.


-George

[eevblogfan]

thank you Georg

I assumed he isn't finished yet , I assumed the LTspice suffer from few disadvantages either . so I post the circuit as I thought it will fit my own needs . I haven't read all of the topic , yea I missed up here ....

anyway, I presume I'll make a new drift ,

here drift :


and here result :


BTW . I wind 20 windings and the ACvrms was 17.4V or so , the DCv was 22V all of this without a load ,

now with 5.4A load , I had about 17VDC !

but the strange thing is , way the transformer take 3.9~4Aac no load . and 4.3~4.45Aac with 93W load ! ? ! ?

it's like 892W no load and 992W with 92W load ! ? ! ?


edit:

with overkill config , I literally "kill" the transformer ,

I'll go and rewind the primary againe ..

what is the number of winding I need to do in order to achieve about >/= to 1A of current (no load )

the coil will be made out of the same 18AWG usual emile wier ,

the data I have is :

with about 260 windings I get 3.9~4A (no load )

and with another coil I added ( another 220 windings but 10 ohm in series , IE total of 13 ohm , it draw 0.3A and it was very quiet , no noisy buzzing anymore )

in order to achieve 4X15V taps and end up with quiet transformer , since I dont know transformer's theory , I am asking you to tell me way is the number of winding (I dont know what difficulties are hidden after this question , so please consider that )




Wat's wrong now ? ( I really don't know ) 

thank you in advance ! 

[eevblogfan]

hey

I just try some formula I thought about ,

mains V /R (coil resistanse) = /(mains freq) times crest factor

so

220/1.45= 151.72

151.72/50=3.34

3.34*1.3=3.9448 (about the same current the transformer draw )

is that formula real ? is this ok ?

thank you in advance !

[eevblogfan]

hey

I found and "mark" this power supply schema :



It's Agilent E361XA  30W power supply ,

I would like to ask you people , how can I modefie this power supply for my needs ( E.g higher power pass transistors+upgrading the driver circuit , or so ... )

I am sorry for my 3'rd post in raw ,

thank you in advance !