Sciencemadness Discussion Board
Not logged in [Login ]
Go To Bottom

Printable Version  
 Pages:  1    3  ..  9
Author: Subject: PC PSU to laboratory PSU
Organikum
resurrected
*****




Posts: 2250
Registered: 12-10-2002
Location: Europe
Member Is Offline

Mood: lonely

[*] posted on 11-5-2005 at 04:51
PC PSU to laboratory PSU


Look here

Hey! Thats a good one isnĀ“t it?

/ORG

[Edited on 22-1-2007 by vulture]




Irgendwas is ja immer
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 12-12-2006 at 09:44


Good idea ........

However , it's only a start towards a true Lab Power Supply
since it doesn't allow for any incremental control of the
voltage or current .

I have been trying to find a way to provide those controls .
The attached circuit is what the effort at designing something
has produced . The design is the product of what has been my interpretation of the parameters in discussions with 12AX7 and Twospoons . It is only a theoretical model at
this point and has not been built or tested , so it remains
something of a work in progress , but hopefully near completion . The schematic I am attaching here because
this seems to be the most pertinent thread where any searches may locate it .

This schematic is like I said a work in progress and a
few changes will be made before it is complete .
A couple of errors have been found already , a typo
of sorts that I need to correct concerning the inputs
on U1 and U2 which are flipped , and should have the same orientation as the other four . And there are a couple of
other things I have caught proofreading the schematic .
So check back as there will be some minor changes ,
and corrections made as this design is completed .

The corrected schematic with regards to U1 and U2 is attached . A few other changes are planned yet .



[Edited on 14-12-2006 by Rosco Bodine]

Attachment: Converter for 550W ATX Computer Power Supply to 50A Laboratory Power Supply.pdf (52kB)
This file has been downloaded 1815 times

View user's profile View All Posts By User
The_Davster
A pnictogen
*******




Posts: 2861
Registered: 18-11-2003
Member Is Offline

Mood: .

[*] posted on 12-12-2006 at 09:52


Yeah, that link is the design I based this:
https://sciencemadness.org/talk/viewthread.php?action=attach...
and more recently this:
https://sciencemadness.org/talk/viewthread.php?action=attach...
off of. Works great. Only reason I had to buid a new one is my older ones contact points rusted very badly. And I killed a few wires pulling >15A through only 2 16 or 18 gauge wires.

[Edited on 12-12-2006 by The_Davster]




View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 12-12-2006 at 10:40


Yeah it has certainly been a motivator to make use
of a PC power supply if possible , once you check the
prices on manufactured power supplies having similar
current capability . Even if I had that kind of money
I would sure rather spend it on something else , if
an acceptable substitute could be improvised for
less than a tenth what the conventional model costs .
It can be fun to try to " beat the system " with a bit
of ingenuity and improvisation :D

Somebody please send that fellow who posted that other page a link to this thread and/or the attached schematic . People who are actively building and
prototyping will likely get to this a long time before
I do ....if I ever get to such things anymore . I'm
getting old and slower by the day sort of like an old
clock winding down . So any of you younger folks
that may have any use for any of this stuff , just
go ahead and make use of it .
View user's profile View All Posts By User
jpsmith123
International Hazard
*****




Posts: 764
Registered: 24-6-2005
Member Is Offline

Mood: No Mood

[*] posted on 12-12-2006 at 18:16


Unless you want to do it as a hobby, why bother trying to build a variable, high-current lab power supply, when you can buy one, new, rather cheaply, on Ebay?

While the cost for most of the electronic parts may be relatively cheap, the hardware costs, e.g., case, panel meters, fan, heatsinks, etc. can add up fast.

I built one several years ago, and I ended up spending more money than I would have for a good used (and for certain brands, new) supply.

Then there's the factors of time and effort; by the time you come up with a design, and then etch, drill, and populate a board; prepare the case, wire it, debug it, etc. (and assuming your design is reasonable in the first place), you're probably talking about several days of labor.

On the other hand, you can get a new Mastech 30V 20A general purpose (switching type) lab power supply for $200.00 on ebay.

For a little more money, you can get 50V 20A or 30V 30A units.

I just got the HY3020E, and although it's no Hewlett Packard or Sorenson, so far it seems hard to beat at a price of $200.00.
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 12-12-2006 at 19:57


Quote:
Originally posted by jpsmith123
Unless you want to do it as a hobby, why bother trying to build a variable, high-current lab power supply, when you can buy one, new, rather cheaply, on Ebay?


Because you can build something having more than twice the
current at the lower voltage you need for a quarter to a third
the cost ......could be one reason .

Quote:

While the cost for most of the electronic parts may be relatively cheap, the hardware costs, e.g., case, panel meters, fan, heatsinks, etc. can add up fast.


Depends on what you can scrounge or what you may
have on hand .

Quote:

I built one several years ago, and I ended up spending more money than I would have for a good used (and for certain brands, new) supply.


What type of power supply was it , what current rating ,
and what sort of range and type of controls did it have ?

Quote:

Then there's the factors of time and effort; by the time you come up with a design, and then etch, drill, and populate a board; prepare the case, wire it, debug it, etc. (and assuming your design is reasonable in the first place), you're probably talking about several days of labor.


I wouldn't even bother to etch a board for something
like this . I'd stick the components on perfboard and
bend the excess lead wires into point to point traces
on the underside using bare wire to fill the gaps
and using jumpers of connection wire where needed there and also topside . Assembly time is more like
hours on something like this , not days .

Quote:

On the other hand, you can get a new Mastech 30V 20A general purpose (switching type) lab power supply for $200.00 on ebay.

For a little more money, you can get 50V 20A or 30V 30A units.

I just got the HY3020E, and although it's no Hewlett Packard or Sorenson, so far it seems hard to beat at a price of $200.00.


For anybody who wants to spend $200 on a 20 Amp supply , that's their privilege ....I won't . And when it gets to higher currents the price goes up substantially .

Actually I watched the auctions on eBay regularly for
awhile and sometimes there are what looked to be some decent bargains , although to me that isn't one of them .
It's like " where's the beef " :D What is the price for a
40 A unit ? Maybe send a copy of my file to Mastech
and ask 'em what they want for a 50 A unit . They really ought to love that .

One advantage of a converter is that the ATX power module can be replaced pretty cheaply if and when it expires . You can buy another one and just plug it in .

The current capacity is also there right where it is needed
in the lower voltages used for electrolysis .

And probably there is indeed the " hobby " factor
involved in something like this ....which if you build
it , you know what is in it and you make it the way you want it . It is sort of like a RC airplane I guess .....
you can pay somebody to build 'em for you I suppose ,
or you could just build it yourself .
View user's profile View All Posts By User
Natures Natrium
Hazard to Others
***




Posts: 163
Registered: 22-12-2004
Member Is Offline

Mood: No Mood

[*] posted on 13-12-2006 at 13:04


Damn, those are some really cool designs you guys have come up with. Something like that is exactly what I need for my project, but when I opened up the schematic I sounded like a backwoods red-neck trying to read chinese. "Durr, whats all the funny squiggly lines for?"

Although I dont see myself attempting this anytime in the near future, I did have a question about it: How do you adjust the voltage/ampreage on the schematic posted?

-NN




\"The man who does not read good books has no advantage over the man who cannot read them.\" - Mark Twain (1835-1910)
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 13-12-2006 at 14:33


Quote:
Originally posted by Natures Natrium
Damn, those are some really cool designs you guys have come up with.


Thanks . I think they are pretty cool too . Even cooler if they work well in reality like they are supposed to work in theory and on paper . You don't really know for sure until you throw the switch:D Then comes the moment of truth .

Quote:

Something like that is exactly what I need for my project, but when I opened up the schematic I sounded like a backwoods red-neck trying to read chinese. "Durr, whats all the funny squiggly lines for?"


Those are symbols for resistors . They are little cylinders
of carbon mixed with various amounts of ceramic so that
they have measured value of resistance to electrical current ,
as if they were wire filaments of a certain length and diameter ....smaller giving greater resistance and larger giving lower resistance . They are something like tiny heating elements having different ratings . The small ones which
most of these are , are about the size of a single grain of rice having a wire about the size and length of a straight pin
coming out each end .

The triangle shaped symbols are amplifiers .

Quote:

Although I dont see myself attempting this anytime in the near future, I did have a question about it: How do you adjust the voltage/ampreage on the schematic posted?

-NN


Look down at the bottom left quadrant .

The two " front panel controls " are like volume knobs ,
you just rotate 'em clockwise to turn up the power
and watch the meters to set it where you want .

You see that little arrow that points against the resistor ,
is a sliding contact , or " wiper " in the control pot that
makes contact at different points along the length of
the resistor inside , and that gives a voltage that varies
accordingly , and changes the signal to the amplifiers ,
which then changes the output .

[Edited on 13-12-2006 by Rosco Bodine]
View user's profile View All Posts By User
12AX7
Post Harlot
*****




Posts: 4803
Registered: 8-3-2005
Location: oscillating
Member Is Offline

Mood: informative

[*] posted on 13-12-2006 at 15:07


Quote:
Originally posted by Natures Natrium
Although I dont see myself attempting this anytime in the near future, I did have a question about it: How do you adjust the voltage/ampreage on the schematic posted?


Unfortunately, you can't. Rosco completely botched the feedback loop, so you aren't going to get a constant voltage output at the very least. If it doesn't explode or oscillate under a variety of conditions first.

Tim




Seven Transistor Labs LLC http://seventransistorlabs.com/
Electronic Design, from Concept to Layout.
Need engineering assistance? Drop me a message!
View user's profile Visit user's homepage View All Posts By User This user has MSN Messenger
Maya
National Hazard
****




Posts: 263
Registered: 3-10-2006
Location: Mercury
Member Is Offline

Mood: molten

[*] posted on 13-12-2006 at 15:14


New to this thread but what I basically pick up w/o reading everything is low variable voltage supply: why not go old school like I always made my supplies?

Variable Transformer + 4 diodes + some large capacitors? Inductors and resistors optional.....
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 13-12-2006 at 15:24


Quote:
Originally posted by 12AX7
Quote:
Originally posted by Natures Natrium
Although I dont see myself attempting this anytime in the near future, I did have a question about it: How do you adjust the voltage/ampreage on the schematic posted?


Unfortunately, you can't. Rosco completely botched the feedback loop, so you aren't going to get a constant voltage output at the very least. If it doesn't explode or oscillate under a variety of conditions first.

Tim


The component numbers and paths are right there for
your specificity in describing where *exactly* you see a problem . So which one of the feedback loops are you
seeing as a problem ....the output stage I presume ,
since all the others are textbook models ? Actually
I think the output stage is a textbook model too ,
it's just probably not in a book you have read yet .

You want to chug back on over to whimsy and sort it
out ....or right here would be fine too .
View user's profile View All Posts By User
Natures Natrium
Hazard to Others
***




Posts: 163
Registered: 22-12-2004
Member Is Offline

Mood: No Mood

[*] posted on 13-12-2006 at 15:27


Well, honestly, I knew those were resistors. I was just trying to convey in a humorous way how I found the total diagram to be overwhelming on first sight. The only circuit I have ever built is the flyback driver (powerlabs design, for use in generating ozone). I did not know however that those triangles were amplifiers. I have heard of mosfets, and I recognize the two flat parallel lines as capicitors, but what does it mean when one of the lines is paranthesis shaped (ie "--|(--" )?

Hmm, either way I would not want to even attempt it until someone else had verified it as a working design, I know too little and it would take far too much time. I suppose for someone like me, spending the money on a commercial unit might makes sense. Even if I spent $200 on a PS, combined with my costs so far, it would still be half the price of a more expense unit and probably put out twice as much O3.

But, I digress from the topic.

-NN

PS Maya, do you have plans for this thing? I very much like simple, and I only need 24V @ ~5-8A.




\"The man who does not read good books has no advantage over the man who cannot read them.\" - Mark Twain (1835-1910)
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 13-12-2006 at 15:30


Quote:
Originally posted by Maya
New to this thread but what I basically pick up w/o reading everything is low variable voltage supply: why not go old school like I always made my supplies?

Variable Transformer + 4 diodes + some large capacitors? Inductors and resistors optional.....


Maya , that's perfectly fine for supplies that are maybe
30 Amps or less .....but we are looking at power supplies for
electrochemistry that are in the >30A range , where
straight linear supplies are very bulky and inefficient ,
as well as hugely expensive .
View user's profile View All Posts By User
jpsmith123
International Hazard
*****




Posts: 764
Registered: 24-6-2005
Member Is Offline

Mood: No Mood

[*] posted on 13-12-2006 at 15:36


Quote:

Unless you want to do it as a hobby, why bother trying to build a variable, high-current lab power supply, when you can buy one, new, rather cheaply, on Ebay?


Quote:

Because you can build something having more than twice the
current at the lower voltage you need for a quarter to a third
the cost ......could be one reason .


I would say the relative "cost" is still debatable at this point.

In any case, from your earlier statements, you seemed to imply that you saw the project as a general purpose lab supply that others might be interested in building, accordingly. For what application do you anticipate needing such a high current, regulated, low voltage supply?

Quote:

While the cost for most of the electronic parts may be relatively cheap, the hardware costs, e.g., case, panel meters, fan, heatsinks, etc. can add up fast.


Quote:

Depends on what you can scrounge or what you may
have on hand .


Well, yes, and if you don't have it on hand, you're looking at spending some potentially serious money.

Quote:

I built one several years ago, and I ended up spending more money than I would have for a good used (and for certain brands, new) supply.


Quote:

What type of power supply was it , what current rating ,
and what sort of range and type of controls did it have ?


It was a 0 to 120 volt DC, 10 amp, switching supply with variable current limiting, in a nice anodized aluminum case, with fairly high quality analog panel meters, and fairly high quality 10 turn potentiometers.

Between the case, meters, heatsink, fan, switches, pots, and miscellaneous hardware, I had about $200.00 into it as I recall. I probably had at least another $100.00 invested in the electronic parts.

Quote:

Then there's the factors of time and effort; by the time you come up with a design, and then etch, drill, and populate a board; prepare the case, wire it, debug it, etc. (and assuming your design is reasonable in the first place), you're probably talking about several days of labor.


Quote:

I wouldn't even bother to etch a board for something
like this . I'd stick the components on perfboard and
bend the excess lead wires into point to point traces
on the underside using bare wire to fill the gaps
and using jumpers of connection wire where needed there and also topside . Assembly time is more like
hours on something like this , not days .


Trying to do that with a circuit like the one you've posted would be a sloppy mess and a major PITA, IMO. Especially if you end up having to fool around with it and make some component changes to get it working to your satisfaction.

In any case, even if your design is good as is, and your "circuit board" comes together the first time with no mistakes, just mounting heat sinks and fans, drilling and punching holes for meters, switches, LEDs, wiring, etc., generally takes much more than "hours". Hell, just figuring out the layout of the parts in the enclosure, or rummaging through online catalogs for necessary parts, can take hours.

I've built enough gadgets to know it almost never takes merely "a few hours".

Quote:

On the other hand, you can get a new Mastech 30V 20A general purpose (switching type) lab power supply for $200.00 on ebay.

For a little more money, you can get 50V 20A or 30V 30A units.

I just got the HY3020E, and although it's no Hewlett Packard or Sorenson, so far it seems hard to beat at a price of $200.00.


Quote:

For anybody who wants to spend $200 on a 20 Amp supply , that's their privilege ....I won't .


You almost certainly will, if your time has any value at all. In fact, how much time have you spent on it already? Don't tell me you haven't already spent a few hours.

Quote:

And when it gets to higher currents the price goes up substantially .


The price is generally more a function of power throughput than current. The 30 amp Mastech is capable of 0 to 30 volts, and you'll pay about $50.00 more for that extra 300 watts.

Quote:

Actually I watched the auctions on eBay regularly for
awhile and sometimes there are what looked to be some decent bargains , although to me that isn't one of them .
It's like " where's the beef " :D What is the price for a
40 A unit ? Maybe send a copy of my file to Mastech
and ask 'em what they want for a 50 A unit . They really ought to love that .


Well, look up the prices of some similarly rated U.S. made power supplies, if you don't think $200.00 is a bargain.

Quote:

One advantage of a converter is that the ATX power module can be replaced pretty cheaply if and when it expires . You can buy another one and just plug it in .


Perhaps; but IIRC, I've never seen one go bad during normal usage, so unless you abuse it, I can't see too much of an advantage there.

Quote:

The current capacity is also there right where it is needed
in the lower voltages used for electrolysis .


Electrolysis at 40 or 50 amps implies a big cell, and if you're trying to make chlorates or perchlorates using MMO or platinized anodes, a very costly cell. Regardless, why not just directly use the 5V output and adjust the electrodes for the current you want to run?

Quote:

And probably there is indeed the " hobby " factor
involved in something like this ....which if you build
it , you know what is in it and you make it the way you want it . It is sort of like a RC airplane I guess .....
you can pay somebody to build 'em for you I suppose ,
or you could just build it yourself .


Of course if you do it for a hobby, most of the issues I brought up will be moot; but if you need a robust, relatively high current general purpose DC power supply, and don't have much experience building something like this, don't have half the parts lying around, and don't want to spend a day or two or three working on it, you're far better off just buying one on ebay, IMO.
View user's profile View All Posts By User
hinz
Hazard to Others
***




Posts: 200
Registered: 29-10-2004
Member Is Offline

Mood: No Mood

[*] posted on 13-12-2006 at 15:54


Anyone an idea whether these PS could be connected paralell, (voltage stays connstant,amperage adds up)? I thought about this several times before, because if you look at the prices at ebay http://search.ebay.com/pc-power-supply_W0QQssPageNameZWLRS, these things are that cheap, you could easily buy 10-20 (as much as your power line holds up to), if you choose those those for 7$ and buy all at the same guy. Then you have a supply with 200-400A @ 5V, perfectly to use in all kinds of electrolysis (purification of alloys like nickel from fake silver cutlery by electrolytic oxidation, chlorate and perchlorate cells etc.) without a long time to wait for the result. This would be far better than rewrap an welding transformer with thicker wire/rods to get a lower voltage furthermore you could replace a died supply cheaply.




[Edited on 13-12-2006 by hinz]
View user's profile View All Posts By User
Natures Natrium
Hazard to Others
***




Posts: 163
Registered: 22-12-2004
Member Is Offline

Mood: No Mood

[*] posted on 13-12-2006 at 16:03


Hmm, I was just reading on a web page about how thier commerical supplies can be hooked up in series (not parallel) in order to increase the voltage which works as long as the max amps for either of the supplies by itself is not exceeded. I wonder, can the same be done with an ATX power supply, or is there too much regulating circuitry inside 'em to allow that to happen?



\"The man who does not read good books has no advantage over the man who cannot read them.\" - Mark Twain (1835-1910)
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 13-12-2006 at 16:27


IIRC , the common ATX supplies are *not* parallelable
for providing extra current .

There are some commercial supplies that have that capability and IIRC they are also very expensive .
View user's profile View All Posts By User
The_Davster
A pnictogen
*******




Posts: 2861
Registered: 18-11-2003
Member Is Offline

Mood: .

[*] posted on 13-12-2006 at 16:32


If you two check out the powersupplies thread in Miscelaneous, around page 2 or so the hooking up of PC PSs in series and parallel was mentioned. The reasoning I forget and had no idea what it meant at the time, but the consensus was they could work in series, but not parallel..



View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 13-12-2006 at 17:33


While you cannot directly parallel the naked ATX's ......

It is an idea that has been in the back of my mind ,
that there is however a way to parallel the outputs
of the circuit I have shown .....but it would require
an additional control stage , and frankly I haven't
really occupied myself with contemplation of how
best to go about it .

I can outline what I think would be required ,
but admittedly at this point I haven't studied this
possibility thoroughly .

The ground rails of the two stages would be tied together . The output current of the first unit would be voltage sensed at point A on R27 , and used as the reference for an error amplifier whose other input
would voltage sense the corresponding point A on
the slaves R27 , and the output of that error amplifier
would be amplified to drive the MOSFET array on the slave , to produce the identical current output as the master .

Or it might be better to do the current sensing on R19
on the master , and then compare it with the signal from
R19 on the slave array , and drive the slaves MOSFETS to
match current with the masters into their shared load .
I haven't given a lot of thought to how best to go about this and there are different ways it could be done .

Now the hitch .....the positive rails have to be connected
with nose to nose diodes , and the regulated negative
rails have be connected with back to back diodes ,
and the load connects to the nodes between the diodes ,
which results in an output voltage drop across those
diodes .....which would be unavoidable , unless
there is some other way of accomplishing the paralleling
of ATX's which I just don't see . By such a scheme
you could I believe parallel a bunch of ATX's and force
their current sharing .....but at the cost of two diodes
voltage drop right off the top .

However it probably would be about the cheapest way
anyone would ever find for daisychaining a bunch of
cheap PS's into a massive current capability array .

[Edited on 14-12-2006 by Rosco Bodine]
View user's profile View All Posts By User
12AX7
Post Harlot
*****




Posts: 4803
Registered: 8-3-2005
Location: oscillating
Member Is Offline

Mood: informative

[*] posted on 13-12-2006 at 22:16


Ya the only way to parallel stuff is with a resistive or constant-current characteristic. If you parallel two voltage sources with low output impedance (approaching zero, as an ideal regulated supply is, individually), the voltage difference inevitably will not be zero (and can not be zeroed for all temperatures, loads, etc.), leading to one power supply powering the other. Since each supply has diodes, current won't backflow, but the higher voltage one will hog all the current.

You can add resistors to force current sharing, but you lose regulation in proportion to the current sharing you gain. The circuit I designed (and which Rosco has tried, I emphasize tried, to redraw) takes this to the extreme by paralleling current sources (or sinks, as the case may be) and controlling them with an error amplifier to stabilize the voltage. Both voltage and current can be set, so that at low loads (i.e. lower than set current/relatively high resistance), voltage equals the set value so long as current is lower than the limit. Conversely, at high loads (voltage pulled lower than setting/relatively low resistance load), current reaches a maximum of the set current, no matter what the voltage is pulled down to. What you get is a square operating point, if you graph V vs. I for all load values from 0 to infinity ohms.

Tim




Seven Transistor Labs LLC http://seventransistorlabs.com/
Electronic Design, from Concept to Layout.
Need engineering assistance? Drop me a message!
View user's profile Visit user's homepage View All Posts By User This user has MSN Messenger
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 13-12-2006 at 22:47


@12AX7 ,

Thanks for hollering about the feedback polarity ,
I have been staring at for two hours and realized
I have the inputs flipped on U1 and U2 .

I need to straighten that out . A negative regulator
works backwards , and I have to think backwards and upside down to keep it straight :D Compared with the
divider values on a positive regulator , the voltages
on the sensing divider between the regulated negative rail and the positive rail *rise* towards the fixed positive rail , as the regulated output voltage is *descreasing* , so the error amplifier is non-inverted instead of inverted as in the usual positive regulator .

I have the polarities on U1 and U2 right where they ought to be for a positive regulator , but upside down
for what they should be on this negative regulator :P

I'll flip 'em where they should be , and I probably have
the sense leads for U1 and U2 transposed also on R21,
I'll check this , and the LED's labeling also .

Also I was thinking about gain limiting U2 to something
from around 40 up to 100 ...to decrease its sensitivity .
I am not sure exactly what gain would be good , as
I have seen regulation values common anwhere from
plus or minus 5% , down to 0.01% .....and knowing the
switching bandwidth for the MOSFET will be limiting ,
I am not sure what sort of gain and accuracy on the regulation is even possible , but am guessing about
that gain for the error amplifier .

The corrected schematic with regards to the feedback loops
on U1 and U2 is substituted for the earlier file above .


The feedback and control loops look okay to me now ,
but the voltage reference selector needs to be configured differently , and range referenced with regards to zeroing the negative voltage output with respect to the positive rail .

The way I have it now is actually the way I had set it up earlier for a positive regulator , referenced to ground , a similar brain fart as was the reversed inputs on the error amps . It is like riding a bicycle backwards because of the referencing to the positive rail being exactly backwards to the usual perspective . Some of my old textbooks written
by British authors followed that convention and completely
designed circuits referencing everything to the positive rail ,
and it drove me up the wall then just like is occurring now .
These circuits which are inverted to the more common arrangement are difficult for me to follow , but this one is getting there .

The reference voltage select is connected wrong . The reference voltages are correct and the values are right ,
but the bottom of the control pot actually needs to connect to the cathode of the LMV431 and the top of the control pot
needs to connect to the positive ATX rail attached to R18 .

I'll check it out later , redo the connection points , and see if the referencing works out then for zero up adjustability like I was intending .


The moral of this story is not to work on schematics when you need sleep . I just was looking at what I was doing
earlier and I think the original problem is not only about the
inputs on U1 and U2 being flipped or perhaps the reference voltage selector being wrong , but the problem is the sensing
divider string R20 , R21 , R22 is between the wrong rails
and that is what is inverting the feedback signal , and causing the excursion above common mode input limits as
well , even after I flip the inputs on U1 and U2 to correct the signal polarity .

I am going to move the voltage sensing divider down
so that its top is at the regulated negative rail and its bottom is at actual ground and see if that helps .

Yeah .....that looks like it was the problem , had the sense divider between the wrong rails . I am still working on it
and need to check the feedback signal range against the common mode input limits on U1 and U2 , and recheck
the reference voltage selector ......it will take awhile .
I need to take a break from this and do some other things for awhile , get rested and come back to it to finish the
design . I see now why you weren't in any hurry to transpose and translate the positive regulator configuration
to the negative regulator configuration :D , it becomes
complicated to keep the inverted signals properly formatted ,
because of the "ass backwards" polarities and signal direction changes which are absolutely counter intuitive ,
or opposite to the more usual positive regulator circuits which
seem more straightforward because of familiarity . Looking at the configuration for a negative regulator , it is like
having to " retrain your brain " to see everything inverted
from the more familiar scenario .

I'll post a corrected schematic later that should make more sense . I'll rethink the feedback loop after I take a break ,
and put it right .


[Edited on 14-12-2006 by Rosco Bodine]
View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

biggrin.gif posted on 16-12-2006 at 01:55
Eureka !


Nothing like a couple of differential amplifiers for sorting out
the load sensing feedback from a rising and falling negative regulated rail , and referencing that feedback properly to the actual ground . Those differential amps are unity gain ground referencing amps which simply translate the absolute voltage measured across the sensing divider , and output the same voltage only referenced to actual ground , which is
an intelligent signal for the error amplifier . Presto we got load regulation from servo locked feedback , load regulation
whether the postive rail fluctuates , or the load fluctuates ....no matter which it is actively tracked and regulated .

That's what was the missing link to make this negative regulator , current sink scheme work .....so that everything
can be self powered from the +12V rail , and need no
separate power supply .

The attached schematic is the design which has distilled from all the back to drawing board scenarios which I think were
leading directly to this .


Anyway , I think this is the right combination of components
and in the right arrangement to do the job and have a
good chance of providing high performance at reasonable
cost .

If this is as good as I think it probably is , then it was worth the effort . And if this ATX converter was so very simple to implement , it would already be in an application note , or on a project page and we wouldn't have to be sweating the details of how to go about it . All we would needed to have done is google it and go right to the information . From what
little I have been able to find , if anyone has already designed and built a converter of this sort , then it is something not very well known. It just could be that
we are breaking new ground here on this one and it is one for the books .

Check this out ,
I think I have this ATX converter hack defined .
The numbers on it check out for me anyway .

And of course let me know if you spot anything I missed .

BTW from the +3.3 , +5 , and +12V supply there should be
full current available from 0V up to -0.15V of the respective
rails capability , half of that drop being from the current
sensing shunts . That is 0 to 3.15V , 0 to 4.85V , and
0 to 11.85V .

That is pretty damn low dropout voltage for a regulator :D

Update : I cleaned up a couple of things on the previous
schematic and this should be the final design attached now .
I appreciate everyones patience with these revisions that have occurred as the final details are pinned down , proofread and edited . This design is to a point now where I believe ( and sincerely hope:D ) it is complete ....unless
some specific required revisions are pointed out by others .
It never hurts to have more than one set of eyes on one
of these things .

[Edited on 18-12-2006 by Rosco Bodine]

Attachment: Converter for 550W ATX Power Supply to 50A Laboratory Power Supply.pdf (54kB)
This file has been downloaded 1263 times

View user's profile View All Posts By User
Rosco Bodine
Banned





Posts: 6370
Registered: 29-9-2004
Member Is Offline

Mood: analytical

[*] posted on 21-12-2006 at 12:08


Upon reviewing the component selection I recognized that
a quad op amp could be conveniently used as the controller stage for this regulator , at a reduced cost for the construction and simplifying the connections . So an amended schematic is attached reflecting this change . The functional structure of the circuit and the way it works is unchanged , but the quad op amp is an improvement to the earlier specification . The TLV2374IN quad op amp also has a better headroom with its 16V operating voltage rating for this application .

Biasing currents have been increased to provide better signal
strength and noise immunity .

[Edited on 22-12-2006 by Rosco Bodine]

Attachment: Converter for 550W ATX Power Supply to 50A Laboratory Power Supply.pdf (58kB)
This file has been downloaded 1844 times

View user's profile View All Posts By User
Misanthropy
Hazard to Self
**




Posts: 69
Registered: 24-3-2005
Member Is Offline

Mood: Variable

[*] posted on 21-1-2007 at 17:40


So, I'm butchering a 100W ATX PS with a purple +5vSB line rated at 0.8 A.

Does anyone foresee any problems with me paralleling it with the regular +5v (10A) bus for a little extra umph?




View user's profile View All Posts By User
vulture
Forum Gatekeeper
*****




Posts: 3330
Registered: 25-5-2002
Location: France
Member Is Offline

Mood: No Mood

[*] posted on 22-1-2007 at 10:07


I've changed the thread title to PSU instead of PS, as the last abbreviation usually stands for PolyStyrene in chemistry...;)



One shouldn't accept or resort to the mutilation of science to appease the mentally impaired.
View user's profile View All Posts By User
 Pages:  1    3  ..  9

  Go To Top