Power Supply / Scavenging Components

[fridgid]

Goal: mains powered 24V DC @ 8-9A power supply

The voltage quality doesn't need to be high as it will be going into motor control chips which will regulate it or control boards which will also be able to regulate it.

Background:
I'm a senior mechanical eng undergrad student and I also have a computer science degree
I've built several robotics projects for classes before, but they've always provided a power supply or we've used batteries. I basically only have motors so far, no controllers at all (well i guess i have a small DAC board).
My current project will need to power 3 high torque stepper motors (Lin Engineering 5718L-01S) and a control board. Eventually this will be powered via a battery, but being a long term project, I'd like to create a simple 24v (ish) psu and start playing with the power draw before I commit to a battery (as weight is a large factor).

Side Question:
How do you define the power requirements of a bipolar stepper motor? Worst case 2 of the phases will be on, or I guess it depends on the controller used. I will need some degree of microstepping. Would it be false to assume that there will never be 2 phases on at any time with a controller using microstepping? If this is true, my original power calc is off. I was counting worst case of all 3 on with 2 phases active each (not sure if steppers produce additional current when being back driven that would cause a spike).

Motor Specs:
5718L-01S
Bipolar (4 leads), 1.4A / phase, 4.5 ohm / phase, inductance 15 mH/phase
http://www.linengineering.com/LinE/contents/stepmotors/5718.aspx


Assuming my power calcs are correct, I've considered many ways to approach this problem but not sure what's best. I'm not particularly adept at finding datasheets for scavenged components. Seems like an art-form really.

Ideas:
1. buy a purpose built one
           these seem quite expensive when i have access to a very large quantity of printer/copier/computer parts

2. series 2x 12V atx power supplys
           all the designs seem to use a floating ground which seems pretty dodgy

3. modify atx power supply to output 24v
           most of the info i can find on this is poorly documented at best. it seems doable if you can swap out select components (mostly resistors) and upgrade a bunch of caps, though all ive found simply disable the overcurrent protection which im hesitant to do (or upgrade it, which sounds like a nightmare to calculate since you will most certainly not have the original schem)

4. build one
           A professor of mine recommended using a 24 Volt transformer with a simple prebuilt rectifier/filter caps from canakits.
(http://www.canakit.com/4a-universal-ac-dc-converter-power-supply-kit-ck354-uk354.html) (putting two of them in parallel to give me the 8A)  (i have what i believe to be a 24v ac transformer, but i can't find the datasheet and as such the max amps. i guess for that matter i have the entire power supply board from a rather large hp 2200 printer which i figured i might be able to just use that damn thing, but again, no spec sheet or max amps)

First post so be gentle, I've been lurking for a couple of weeks, but I guess it's time to jump in.
Any help/advice would be appreciated. Thanks in advance.

[billhowl]

How about this one "MW 24V DC14.6A 350W Regulated Switching Power Supply" US$62.90 FREE Standard Shipping.
http://www.ebay.com/itm/MW-24V-DC14-6A-350W-Regulated-Switching-Power-Supply-Meanwell-NES-350-24-/120771610467?pt=BI_Electrical_Equipment_Tools&hash=item1c1e8c8763#ht_4128wt_957

[Soeren]

Hi,

Eventually this will be powered via a battery, but being a long term project, I'd like to create a simple 24v (ish) psu and start playing with the power draw before I commit to a battery (as weight is a large factor).

If it's just for short term testing, you might be able to borrow two car batteries.

Would it be false to assume that there will never be 2 phases on at any time with a controller using microstepping?

Yes - microstepping or not.

Ideas:
1. buy a purpose built one
           these seem quite expensive when i have access to a very large quantity of printer/copier/computer parts

You wouldn't happen to have a large photo copier to strip?
(They're usually 24V)

2. series 2x 12V atx power supplys
           all the designs seem to use a floating ground which seems pretty dodgy

Dodgy? how?
Problem is, that they regulate on the 5V side (And doesn't have that much current to give at 12V without some rebuilding).

3. modify atx power supply to output 24v
           most of the info i can find on this is poorly documented at best. it seems doable if you can swap out select components (mostly resistors) and upgrade a bunch of caps, though all ive found simply disable the overcurrent protection which im hesitant to do (or upgrade it, which sounds like a nightmare to calculate since you will most certainly not have the original schem)

If you have never worked with SMPSUs, that would not be the best option (to put it mildly).

4. build one
           A professor of mine recommended using a 24 Volt transformer with a simple prebuilt rectifier/filter caps from canakits.
(http://www.canakit.com/4a-universal-ac-dc-converter-power-supply-kit-ck354-uk354.html) (putting two of them in parallel to give me the 8A)  (i have what i believe to be a 24v ac transformer, but i can't find the datasheet and as such the max amps. i guess for that matter i have the entire power supply board from a rather large hp 2200 printer which i figured i might be able to just use that damn thing, but again, no spec sheet or max amps)

First, a 24V transformer will give you around 32V DC when peak rectified.
You'll need an 18V (AC/RMS) transformer for getting 24V DC
If you want to regulate down to 24V, shoot for 27..28V rectified DC.

You don't need a PCB for a simple rectifier. I usually solder 4 diodes and a capacitor directly to the transformers terminals and then wires from that to whatever further circuitry.

Your HP supply can be measured to find the max current.

How much does your 24V transformer weigh and what's its dimensions?
Those numbers will give you a rough estimate of what poer you can expect it to put out.

[fridgid]

Would it be false to assume that there will never be 2 phases on at any time with a controller using microstepping?

Yes - microstepping or not.

Is there an approximate method for calculating power requirements for stepper motors? I couldn't find any literature on the net about it.
If I'm driving three of those motors listed above how would I calculate it? I was estimating that worst case two phases would be on at a time, so 2.8A x3, but I didn't even do any calculations (aka V=IZ etc) only reading that 1.4A per phase from the data sheet. I've got a basic understanding of simple circuits (including LRC, I've takin the first EE class) and I can spec out power requirements for DC motors, but I'm not sure about steppers. Steppers create a current when back driven right? They have to create an induced voltage and therefore some amount of current, right?!

You wouldn't happen to have a large photo copier to strip?
(They're usually 24V)

I should, but I'll have to look around. I recently grabbed a board from a broken HP 2200 printer. I've attached a photo of it. It turns on and I'm able to find some 24V potentials if I look all around it, but I think you have to do something to turn the current at 24V on/up with the relays. Just a guess, but they flip on and off when I used my DMM to check some things when its on.

2. series 2x 12V atx power supplys
           all the designs seem to use a floating ground which seems pretty dodgy

Dodgy? how?
Problem is, that they regulate on the 5V side (And doesn't have that much current to give at 12V without some rebuilding).

Well I've got two ATX power supplys atm (though I can certainly get more of the old dell ones). One is a pretty nice Antec 500W PSU that I just got from my personal comp after black friday upgrades. I thought the newer ones regulated on the 12v. I use my power resistor 'load' setup on he 12v on that guy. I've also got a shoddy (old) 300W dell psu thats prob over 6 or 7 years old. Should I be putting my power dump resistors on the 5V side instead?
The series power supplies sounds dodgy because if one of them has a floating ground, what happens if you do something stupid. Does it just blow up the one of the floating ground? Does the overcurrent protection save both of them? Isn't it illegal to have an 'ungrounded' supply?

3. modify atx power supply to output 24v
           most of the info i can find on this is poorly documented at best. it seems doable if you can swap out select components (mostly resistors) and upgrade a bunch of caps, though all ive found simply disable the overcurrent protection which im hesitant to do (or upgrade it, which sounds like a nightmare to calculate since you will most certainly not have the original schem)

If you have never worked with SMPSUs, that would not be the best option (to put it mildly).

Yeah, the 'guides' make it sound easy, but it looks like a damnned nightmare. That and even though I use fuses liberally (man if we didn't have fuses in mechatronics lab....)but I wouldn't like removing the overcurrent pro, though I suppose if you're modding the whole deal, you could upgrade the overcurrent pro.

First, a 24V transformer will give you around 32V DC when peak rectified.
You'll need an 18V (AC/RMS) transformer for getting 24V DC
If you want to regulate down to 24V, shoot for 27..28V rectified DC.

You don't need a PCB for a simple rectifier. I usually solder 4 diodes and a capacitor directly to the transformers terminals and then wires from that to whatever further circuitry.

Your HP supply can be measured to find the max current.

How much does your 24V transformer weigh and what's its dimensions?
Those numbers will give you a rough estimate of what power you can expect it to put out.

Ok, the attached photo has a picture of the transformer. I've searched pretty well to try to find it's datasheet, but nothing. So I have no idea what it's rated for. I can run any test you'd recommend. It has 4 leads on the thicker side (primary?) and 8 on the small side (next to the heat sink. Yes I'm aware the leads don't really help in identification). It's got two sets of numbers on it. After looking around quite a bit, I'm pretty sure the logo is Matsushita Panasonic (at least it looks exactly like the second photo for them found here http://www.classiccmp.org/rtellason/logos/semiconductorlogos.html ). So it has that [M] thing with the following string RH3-0291 (next line) 011D28IS. The black part on the top says SMTJ35A18T. I can't find anything on the second serial, the first one only brings up replacement parts for the board, not datasheets.
I can unsolder it if you'd recommend, but if I can use the board in part I'd like to avoid that (so I can't estimate it's weight). It's dimensions (of the light grey EI part) is 1.5" x 1.75"). There are several other smaller transformers, but I figured that bigger guy south of the largest cap is the 24V one.

When you say 4 diodes and a cap, you're talking about building a simple diode bridge and a cap to smooth out the ripple? How do I spec out the cap/diodes? This board definitely has IC diode bridges I can identify and plenty of caps.

Also, I just found http://www.mpja.com/ which has some power supplys that a much cheaper than I was finding before. So I might just buy a purpose built one for a bench supply. But in anycase if I can built/modify something for cheaper, that'd be a better solution imo. Self-support college student on a budget and yadda yadda yadda

[Soeren]

Hi,

Ok, the attached photo has a picture of the transformer. I've searched pretty well to try to find it's datasheet, but nothing. So I have no idea what it's rated for. I can run any test you'd recommend. It has 4 leads on the thicker side (primary?) and 8 on the small side (next to the heat sink. Yes I'm aware the leads don't really help in identification).
[...]
I can unsolder it if you'd recommend, but if I can use the board in part I'd like to avoid that (so I can't estimate it's weight). It's dimensions (of the light grey EI part) is 1.5" x 1.75"). There are several other smaller transformers, but I figured that bigger guy south of the largest cap is the 24V one.

Oh boy... That's an offline switchmode supply and the transformer is NOT a regular mains transformer!!
A regular iron cored mains transformer is larger and heavier (even a 50VA transformer is over 1kg).

Perhaps I should prepare a guide on building a basic PSU when time permits.

When you say 4 diodes and a cap, you're talking about building a simple diode bridge and a cap to smooth out the ripple? How do I spec out the cap/diodes? This board definitely has IC diode bridges I can identify and plenty of caps.

Yes, a common Graetz bridge and a cap.
The diodes must be able to handle the current with a fair margin - at the very least 12A (8A*sqr(2) + a little extra) and the voltage rating should be 40..50V (or higher).
The cap needs to be of a similar voltage rating and for 8A, you'd need at least 15,000µF (for a ripple of 3..4V) - higher if you want less ripple.
Such values/voltages means physically large caps, around the size of fairly large pill bottle.

Also, I just found http://www.mpja.com/ which has some power supplys that a much cheaper than I was finding before. So I might just buy a purpose built one for a bench supply. But in anycase if I can built/modify something for cheaper, that'd be a better solution imo. Self-support college student on a budget and yadda yadda yadda

That sounds like a good idea, as I can see that you're out of your safe-zone on this one - and your budget means nada if you kill yourself

The http://www.mpja.com/24V-10A-Hengfu-Power-Supply/productinfo/18436+PS/ should suit your purpose.

[fridgid]

Oh boy... That's an offline switchmode supply and the transformer is NOT a regular mains transformer!!
A regular iron cored mains transformer is larger and heavier (even a 50VA transformer is over 1kg).

Perhaps I should prepare a guide on building a basic PSU when time permits.

How does one tell the difference? When I took apart my UPS (the battery was dead) it had a very heavy and large transformer. Prob 10 or 12 pounds and was about the size of an apple. Is size and weight pretty much it?

That sounds like a good idea, as I can see that you're out of your safe-zone on this one - and your budget means nada if you kill yourself

The http://www.mpja.com/24V-10A-Hengfu-Power-Supply/productinfo/18436+PS/ should suit your purpose.

Alright final stab at saving budget. As you recommended I was able to grab the main ps board from a prob 4-5 year old large multifunction copier. It looks much more promising than the old board. I've attached a photo of it.
(The blue/red wires go to the original power switch.) I can hear something click when I turn it on, but I can't find any voltages on the output pins (see the second photo, these look very similar to motherboard connectors and each pin is labeled (!!) .)

Now I'm assuming you need to ground a logic pin or two to turn the output voltages on. (Also, looking at the top right, it has the input and output voltages AND amp ratings (see bottom of second photo). The ratings look to be perfect for what I'll be wanting to use it for.) Any ideas on turning the rest of this guy on?

I suppose I should check the fuses first. Fuses are good. Now, there are some curious things written on the board. The bottom left has as large 24v relay, which looks like it might be the key. The coil pins (well two pins, which from the data sheet look like they should activate the relay) go to one of the small connectors (CN113 in the photo, top left) and are labeled TRELAY and T24V.

So in all, there are two connectors which have curious labels and are smaller and use much lower gauge wire (I'm assuming logic/control).

There has to be something simple with the smaller connectors that makes the magic smoke happen.


Random question: There's a plate that covers the top that has a transformer looking like object attached to it (its the yellow/grey metal object resting on top of the center heat sink in the first photo). It only has two wires though and its connector pins are not labeled. What the hell does this thing do? From my meek EE221 knowledge, this could only be an inductor? Transformers need a minimum of 4 wires no?

So, bring me some good news so I can wrap this puppy up in a grounded box and call it a day

P.S. Super thanks in advance. I'm sure you get tired of schooling newbies on how to not kill themselves.

[Soeren]

Hi,

How does one tell the difference? When I took apart my UPS (the battery was dead) it had a very heavy and large transformer. Prob 10 or 12 pounds and was about the size of an apple. Is size and weight pretty much it?

An Apple computer or...?
An apple (the fruit) maed of iron wouldn't be that heavy
But yes, the weight is a fairly safe pointer - that's one of the advantages of switch mode supplies (although an 1.5kW switcher I have is still rather heavy, but then again, if it was made with a conventional transformer, I probably wouldn't be able to lift it).

Alright final stab at saving budget. As you recommended I was able to grab the main ps board from a prob 4-5 year old large multifunction copier. It looks much more promising than the old board. I've attached a photo of it.
(The blue/red wires go to the original power switch.) I can hear something click when I turn it on, but I can't find any voltages on the output pins (see the second photo, these look very similar to motherboard connectors and each pin is labeled (!!) .)

The rectangular black  thing in the lower right corner might be a relay (hard to see) and if it is, it's probably what's clicking.
The supply obviously tries to start and it may be a lack of load that shuts it down immediately, or it may be a defect component, but test it with a load of around 1A on 5V and 24V (one at a time).

Now I'm assuming you need to ground a logic pin or two to turn the output voltages on. (Also, looking at the top right, it has the input and output voltages AND amp ratings (see bottom of second photo). The ratings look to be perfect for what I'll be wanting to use it for.) Any ideas on turning the rest of this guy on?

It may be that some of the connectors are for control logics, but a copier supply will have a large number of sub-systems each needing power.
I guess I could make it run if I had it on my desk for probing, but it's hard to probe a photo

I suppose I should check the fuses first. Fuses are good. Now, there are some curious things written on the board. The bottom left has as large 24v relay, which looks like it might be the key.

Thought so.
Personally, I would try bridging the relay contacts with a wire for a few seconds to see if that kept it out of shut-down, but without knowing which pins are which, it's not safe to do, so you better try getting it to behave with a load.
And beware of the hot side of the PCB when you probe around!
Basically, anything below the white line may be bad touching - including the relay contacts!!

The coil pins (well two pins, which from the data sheet look like they should activate the relay) go to one of the small connectors (CN113 in the photo, top left) and are labeled TRELAY and T24V.

You have a datasheet on the board??
Please post a link to i, I may be able to tell you how to fire it up from that.

You'll probably see a short 24V pulse on the relay when you try to power it up.
Assuming the copier worked when it was trashed, you could try energizing the relay, but you'd need to remove the current drive for it then.

Post the datasheet!

There has to be something simple with the smaller connectors that makes the magic smoke happen.

No, that's made by fairies and lepricons.

Random question: There's a plate that covers the top that has a transformer looking like object attached to it (its the yellow/grey metal object resting on top of the center heat sink in the first photo). It only has two wires though and its connector pins are not labeled. What the hell does this thing do? From my meek EE221 knowledge, this could only be an inductor? Transformers need a minimum of 4 wires no?

Random answer:
Pickled cucumber and bricks

Less random answer:
it's probably a large inductance choke used for noise filtering (unless there are wires not shown going to the PCB.

So, bring me some good news so I can wrap this puppy up in a grounded box and call it a day

Bring me the datasheet and we can try

P.S. Super thanks in advance. I'm sure you get tired of schooling newbies on how to not kill themselves.

You're welcome - my next tut will be on making a hangmans knot on a solid rope 

[fridgid]

Ahh, I guess I didn't word that very well. I don't have the datasheet for the board. (When I said datasheet I meant for the relay itself. I was simply looking up what pins were what because I knew it had mains power running through it and I didn't want to short the wrong thing if I was going to try to activate it.) The logic/control pins on the smaller connectors are labeled individually but that's all I've got. Give me /some/ credit. I'd either have it running or have posted pictures of a fried board if I had the datasheet.


So, the white line that runs around parts of the board means those part have mains voltage running through them? One side (the AC side I guess) says P and the other side says PE (might be backwards but I think that's right, sorry I'm at work now.)

So the connectors have this:

CN105 pins:

TRY (transformer relay?)
HLT1
ZRC
HLT2
DOS
SW
GND

(I've turned it on while grounding the SW pin and something makes a sound on the transformer side, but I can't tell what and nothing has voltages still. I was able to find some 5 and 5.1 voltages around there somewhere but I don't remember what I was doing then. I definitely wasn't not studying for a final or working on a naked psu while sleepy. Nope not at all.)

CN113 pins:

TRELAY
T24V (these two pins go to the coil of the transformer in the bottom left, no shit, k)
GND
GND
GND
24V5
24V5
24V5

Some of the other connectors have pins labeled 5.1Vcc1 5.1Vcc2. Vcc means the collector (+5V or w/e it requires) for a transistor no? Maybe these control the voltages sources?
I've got some 10 Ohm 10W resistors I can use to simulate load (yes they're heatsunk, thermal pasted too ).

So, I guess I need to play around with it.

Also, thoughts on discharging the large cap? I was using a motor, but it arcs which is bad. I've been using several resistors (a 2k one then a 1k, then a 250) to bring it down. Is this horrible for the big cap? Fast discharging = cap death?

[Soeren]

Hi,

I was able to find some 5 and 5.1 voltages around there somewhere but I don't remember what I was doing then.

Maybe the 5V is on at all time when the board is powered - usually it's so, as it's used for standby mode and the 5V logic turns on the rest when the copier is summoned to do its magic.

CN113 pins:

TRELAY
T24V (these two pins go to the coil of the transformer in the bottom left, no shit, k)

And there's no other connections to the relay coil?
If that's the case, you need to energize the relay (or short its mains voltage contacts) - if it's a 24V coil, maybe there's a low current 24V line always on - try measuring all connector pins marked 24V (relative to GND).

Some of the other connectors have pins labeled 5.1Vcc1 5.1Vcc2. Vcc means the collector (+5V or w/e it requires) for a transistor no? Maybe these control the voltages sources?

Vcc is a term used for positive analog voltages - named from the widespread use of bipolar transistors in analog (which for the NPN will have the collector at its most positive potential), while in digital circuits, the positive voltage is named Vdd (named from the FETs drain). The negative equivalents are Vee (analog, emitter) and Vss (digital, source) respectively.

So, I guess I need to play around with it.

Yes.

Also, thoughts on discharging the large cap? I was using a motor, but it arcs which is bad. I've been using several resistors (a 2k one then a 1k, then a 250) to bring it down. Is this horrible for the big cap? Fast discharging = cap death?

Use around 100 Ohm per volt - for a 24V charged cap a 2.2kOhm resistor will be fine and for a 120V charged cap, 12..15kOhm will do (but then use it in a way that isolates it from your pinkies and similarly sensitive stuff

If your voltmeter leads are of the type that allows piggybacking another plug, you could make a short wire with plugs and a resistor and then use your voltmeter probes for discharging (set the meter to a suitable range). Then you will both have isolation and be able to monitor the cap charge as it dwindles down to a safe level.

[fridgid]

Update: Got it! Apparently using a screw driver to short the SW pin to GND was a poor method. I guess it needed to be fully on at power up. Shorting this pin makes it all work. I did a bunch of looking around and following traces and took out fuses that provide power where I don't want it. I'm still not sure what the relay is really for. I don't think I need it, but we will see. It's currently removed.

It's a SPDT relay and one of the throws is mains voltage but I can't figure out what the other one is. In either case the supply seems to be working fine (at low current anyway, I'll doing some heavier loading tonight) without it. If that relay takes 24V's to activate, would it really make sense for it to 'activate' some high current side of the board? I boxed the whole deal up so I could have it on my bench and not be worried about all the exposed things and whatnot. I attached a photo for S&G.

It gives me 2x +24V (each at 6.3A assuming the fuses are correct, total of 9.5A), +12V (0.3A) , -12V (0.1A), and  +5.1V (5A). So I think this should work perfectly. The whole box is grounded sheet steel and from my minimum tests last night, everything seems to be working. So thanks for your help Soeren.

Time to spec some motor controllers...

[Soeren]

Hi,

Update: Got it! Apparently using a screw driver to short the SW pin to GND was a poor method. I guess it needed to be fully on at power up. Shorting this pin makes it all work.

Great

I'm still not sure what the relay is really for. I don't think I need it, but we will see. It's currently removed.

Maybe it routed mains to some other part of the copier, but it would make more sense if it powered up something on the PCB.

It's a SPDT relay and one of the throws is mains voltage but I can't figure out what the other one is. In either case the supply seems to be working fine (at low current anyway, I'll doing some heavier loading tonight) without it. If that relay takes 24V's to activate, would it really make sense for it to 'activate' some high current side of the board? I boxed the whole deal up so I could have it on my bench and not be worried about all the exposed things and whatnot. I attached a photo for S&G.

A functional box!  (Isn't that a nice way of putting it )
Well, if the supply will do what you need without the relay, then who cares what it does - just forget it if it loads OK.

It gives me 2x +24V (each at 6.3A assuming the fuses are correct, total of 9.5A), +12V (0.3A) , -12V (0.1A), and  +5.1V (5A). So I think this should work perfectly. The whole box is grounded sheet steel and from my minimum tests last night, everything seems to be working. So thanks for your help Soeren.

Hey, you did all the work

The fuses will always be higher rated than the max. Otherwise they'd keep blowing.
Crossing my fingers, hoping you can load it with the full 9.5A without further changes.