Shop OBEX P1 Docs P2 Docs Learn Events
Quick 18A @ 2 volt power supply needed — Parallax Forums

Quick 18A @ 2 volt power supply needed

HulkHulk Posts: 68
edited 2009-12-19 15:01 in General Discussion
I could design this power supply with a 2n3055 if it could handle the current, but I'm seeing 15 amps as a max and that's at a die temp of 25 degrees C (which will never happen). I have a bunch of hexfets (IRFZ42) and an ATX power supply.· The ATX can supply around 20 Amps on its 3.3 volt (orange) line and 25 amps on its 5 Volt line.
·
My problem is my lack of knowledge of FET power supplies.
·
I·have a friend that teaches jewelry design and needs a power supply where you set the voltage somewhere between .5 and·2 volts and immerse a nonconductive thing like a flower seed pod in a solution of some copper compound and the copper builds a thick layer on the object and it looks really cool.· The key·to the process is a proven method of depositing a sprayed silver conductive layer on the, otherwise, nonconductive nature thing.
·
So, my stamp friends, can you show me how to use wiz bang hex FETs to design a voltage-controlled power supply using my 3.3 or 5.0 outputs from ATX power supply that will supply a max current of 18 amps·to a plating solution?
·
BTW, I've gotten past the tricks to keeping the computer supply active so I don't need that help.
·
I will be grateful for any help you can provide.



▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


·

Comments

  • kwinnkwinn Posts: 8,697
    edited 2009-12-17 04:28
    What you want is a design where you control the pulse width on the high voltage input side of a high frequency transformer so you only need to rectify and filter the low voltage output.
  • HulkHulk Posts: 68
    edited 2009-12-17 14:11
    kwinn, Thanks for your response!
    I know that's the best and most efficient electronic solution, but the bulk of that function is handled by a single chip dedicated to that function on the input to the high frequency transformer. I'm looking for a simpler series regulator circuit to be added after the stable circuit that's already there. I'm basing my request on a circuit like an emitter follower where you supply a variable voltage to the base of a power transustor like the 3055 and you see that voltage (minus about .6) at the emitter. I'm sure there's a simple hex fet solution, but I havent found it yet.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


    ·
  • kwinnkwinn Posts: 8,697
    edited 2009-12-17 21:32
    This should give you what you want. Voltage is adjustable from 0 to about 2.4V. The fet has to be on a good heat sink and you may need more than one in parallel to meet the current requirement.
  • HulkHulk Posts: 68
    edited 2009-12-18 00:26
    Thanks kwinn!
    That sure looks like a source follower. I didn't know it was so similar to the emitter follower that I was familiar with.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


    ·
  • kwinnkwinn Posts: 8,697
    edited 2009-12-18 03:29
    It's not quite a source follower, but close. The gate voltage needs to be a bit higher than the desired output voltage so you may need to increase the value of the trim pot and possibly use the +12V supply for the LM117. Depends on the fets you use. The nice thing about using fets in parallel is the Rsd is fairly consistent for fets of the same type so no resistors are needed to split the current.

    BTW most industrial electroplating I have seen is done with a constant current source to control the current per unit area of the item to be plated so a couple of 3055's in parallel with .01 ohm emitter resistors might have been a better way to go.
  • HulkHulk Posts: 68
    edited 2009-12-18 03:44
    kwinn,
    The lady that wants to do this gave me an article from a trade publication. It states some empirical study is necessary to get good results and bases their control on the source voltage rather than the current/area. I, like you, would have expected controlling the current at a constant level based on square inches covered, but that's not the way the article is written.
    I've done some platting and some anodizing so could be considered as "experienced" (certainly not an expert), but I will stick with her source until I can demonstrate that current control works better.
    Thanks for your help and advice.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


    ·
  • kwinnkwinn Posts: 8,697
    edited 2009-12-18 03:48
    After taking a look at the data sheet for the fet you specified I am sure you will need to connect the LM117 to 12V and use a 1K or 2K pot to get a high enough gate voltage. It needs 10V on the gate to be fully on (Rds = .035 ohms). When using this in it's linear region the thing to watch for is power dissipation. With 5V in and .5V out at 18A power dissipation is 81W so you will need a pretty good size heat sink.
  • HollyMinkowskiHollyMinkowski Posts: 1,398
    edited 2009-12-19 01:17
    You can alter an old and unwanted PC pwr supply to give lots of amps
    at different voltages...there are plans on the web, just google.

    I did it once to get 13.8v from the 12v line...it involved rewinding the
    small transformer inside and took about an hour.

    Perhaps a similar project could drop the 5v line down to 2v?
  • kwinnkwinn Posts: 8,697
    edited 2009-12-19 02:13
    Holly, it seems he wants a variable supply between 0.5V and 2.0V. Much simpler to do with a linear regulator even though efficiency is poor.
  • VIRANDVIRAND Posts: 656
    edited 2009-12-19 08:14
    You said you have a 3.3V supply. If it is 20A then two silicon rectifiers (OR one bridge rectifier) in series with it
    will drop the voltage closer to 2 volts. It is probably easier to find a high current bridge rectifier than diodes rated
    over 20 Amps.
  • HulkHulk Posts: 68
    edited 2009-12-19 15:01
    Thanks for all of your suggestions. It will have to be variable so I think I'll use the 5 volt (25A capable) from the ATX supply and, per kwinn, use the 12v line to supply gate drive with an lm317 that I have l laying around.
    I have one of those cheap panel meters that reads 0 to 2 volts and thought I would read the drop across the ground lead using a second (Kelvin) lead on the ground supply lead; this to measure current. I'm shooting for .001 ohm which should yield .020 on the el cheapo meter (@20 amps) and I can wire it to put the decimal point in the right place. Given .00118 ohms per foot of 10ga wire, i figure I can use a pair of parallel leads 20.3 inches long (and a third Kelvin lead) to read th IR drop to get the current. Does that sound like a good plan?
    Naturally, I can switch the + meter lead to the + power supply lead to get the supplied voltage (and move the decimal point at the same time).

    BTW, this is a different kind of plating process than usual (for me anyway). You can plate an object like a flower pod by first spraying it with a conductive paint. Then the copper in the solution plates itself onto the seed pod. I've seen some successful results from equipment that costs about 800 bucks so this will be a good deal for her jewelery school if she/we can pull it off. When the part is fully covered, you heat it and burn out the original organic material and this leaves a very sturdy hunk of copper.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


    ·
Sign In or Register to comment.