Warning: Switching LED Power Supplies

So I bought one of those cheap 12V 20A switching LED power supplies (drivers) off of ebay for 20 bucks. The ones with the metal case with all the holes (because they don't have a fan) and the screw down terminals on one end. Unless I just got a bad one, these are junk. As soon as my heat bed kicks on the thing drops 1.5V and my fans slow down. This makes makes the current low on the stepper so they don't move an accurate distance by tiny tiny amounts so your layers are all wavy.

Has anyone else tried one of these out?

That sucks. I have not used these, I'm still using modified PC power supplies, but thanks for the warning.

I wonder up to what amperage it will provide full voltage?

---

Yvan

Singularity Machine

A 30A PS would be a better solution.

---

_______________________________________
Waitaki 3D Printer

Don't most of the printers use 11+ 5 amps? Regardless, any supply that states a current capability should be able to hold its stated voltage at that current.

Cheap chinese stuff is often rated very generously. You want to take those numbers with a large grain of salt. That number might be a max instantaneous current or some other such dodge, or it might only do 20A if there is an external fan on it, etc. So like waitaki said, a 30A supply would likely get the job done. Or you might just have a bad one.

Okay I just recieved my 12v 20A LED power supply, with all the holes and no fan just like yours, appdev(from A2Aprinter.com). We will see how it holds up. I'm thinking I'll test it with the heated bed only. If it can't handle that....

---

Yvan

Singularity Machine

I will be eagerly waiting for you results Yvan.

I've been using a 20A 12V supply I got off amazon without any real problems. It was unhappy once when I had a short on the heated bed and quickly reset itsself. But with everything set properly it's been chugging away no problem on the .8 ohm load of my current setup. It even worked just fine cranked up to the highest voltage at 14v when I was trying to get my old Helios up to 115* (I think it's like 1 or 1.1 ohm load maybe, but I've got it mounted to a 300mm heat spreader).

The adjuster is fairly sketchy, but it has yet to fail on me.

I think you must have just got a bad one, because I have used well over 40 of them without any problems at all. I must say though that I use mains powered heated beds, so that saves quiet a bit of the current that you may be using. I have run dual hotends without any problem as well as the fans and motors, all on 20 amp supplies.
Could be from a different source though, hard to tell when buying from China, so many small companies that sell from a major manufacturer.

I have a 30a version of this power supply that I bought through eBay. It has been working flawlessly for over a year - well enough for me to purchase the same unit for my second printer. No experience yet with the second unit.

Out of the 100 that we bought 11 were no good right out of the box. Several more crapped out on our customers and we had to replace. That being said, the first one that we wired up has been in service pretty much every day for over a year now. The longest print went 22 hrs. and it managed to heat the bed up to 140 degrees (trying get polycarbonate to not warp ). So I guess it's a crap shoot. We complained to the manufacturer about the problems we had encountered and they suggested that the heavier fan cooled model was better built and better suited to our application, so we switched.

Rule of thumb: If the voltage drops more than a tick when everything is turned on, it ain't happy, send it back. If it gets hot enough to burn your finger after running the machine for an hour, send it back.

---

Blog

appdev,

Can you return your defective unit? Some ebay sellers are good with that.

Finally getting the chance to test my 20A power supply. Printing upgraded X axis ends designed by Herr Jonas Kuehling.

We will see how it goes...

---

Yvan

Singularity Machine

Oh they have offered me return, but didn't mention who is paying shipping. I only paid $18 for it, so it's not even worth it. I hooked it up to heatbead only and using my existing PC PSU to run the rest. Things have been going OK so far.

extent - You got a link to same one you bought. I'm willing to give a try from another supplier for my next printer.

regpye - Yes, it wanted to run the electronics OK, it was when the bed kicked on it crapped out. My oscope says the power wasn't perfectly clean on it little humps left at the top, so that's why I chose the bed for it.

Bruce, I think you probably have the truth of it.

[www.amazon.com]

But who knows if you would get the same part. It's all no-name, no-brand.

Well at least you are getting some kind of use out of it. Mine seemes to work just fine, very little voltage drop while under load.

Something happened through 3 hours into the 4 hour print though. Had to restart everything to get things working again. Don't exactly know what caused that. I'm guessing it probably isn't the power supply that did that...

I'm very glad this topic came up, I like knowing what to avoid. Don't like unreliable stuff at all!

---

Yvan

Singularity Machine

Yvan, that problem 3 or four hours in could have been the power supply. That's what I was getting. After two hours of print X,Y or both would skip. Now that I have went to two psu I've printed for fourteen hours without fail. 2 or 3 hours at a time. I believe it was too much current for the psu when eveverthing lined up just right. All five motors running, hot end kicked on, heat bed kicked on, extruder fan kicked on, and its too much.

I have seen circuits that had to have just a particular set of circumstances to fail.

Yes, it takes a while before I truly trust or discount something. The problem was odd. No skips, everything just froze. Hot end and bed temp were maintained, but motors didn't move, and MacPronterface was not able to control the motors in manual mode.

Just now my modified PC power supply has just fried the connector at the RAMPS 1.4 board. I'm lucky, it's just the removable side of the connector that is trashed, not the soldered on half.

I'm restarting things with the 20A power supply, we will see how it goes -this- time. Same part, attempt #3!

---

Yvan

Singularity Machine

Yvan, I think you have a short somewhere. If I were you, I would unhook everything from ramps and do an ohm test on it. If I remember right each of the motors too coils should have like 2 ohms. End stops should be 0 or infinite depending on your type. If you have the three connector end stops then one poll should be 0 and the other should be infinite. Thermistors and heater RR resisters should be what ever values you purchased. PCB heated bed is what something like 0.8?

If no shorts are found there, pull up ramps schematic and start probing it. Myself, I would go on until I did all this.

appdev, okay, pulled everything apart minutes after you posted! Better safe than sorry, and I might just learn something.

I have another backup RAMPS board installed on the machine now, so I should be able to resume printing while I test out the RAMPS that just had a near death experience.

Thanks for the warning, and we will see what we find...

---

Yvan

Singularity Machine

Now if the short is in a motor, thermistor, heat resistor, end stop or heatbed you are putting your spare ramps at risk. I wouldn't hook up a new one until you have checked everything that hooks up to the ramps.

Actually now that you mention it I have seen that freezing behaviour a few times since switching from a 18A medical power supply to the noname 20A. I narrowed it down to something that happens whenever the dishwasher is running at the same time on the same circuit. That's the only time it's ever happened to me. I thought it might be because of the much heavier load that it's under with my current printer vs the old 6" bed, but it's more likely just the crappy power supply.

Your supply could even have more bumps on the oscope than mine. DC is suppose to be a straight line. Then their is the dish washer. It could be putting a lot of noise on the line and these cheap power supplies aren't filtering it out. there's some filtering on the ramps but it's probably not much.

you're heat bed won't care about the noise so you could try hooking the cheap PSU up to the 11 amp hook up to power the heat bed Sandy Hook your good PC power supply which has a lot of filtering up to the five and pick up to power the motor s and extruder. this is what I did and my printer has never worked better so far.

appdev007 Wrote:
-------------------------------------------------------
> Now if the short is in a motor, thermistor, heat
> resistor, end stop or heatbed you are putting your
> spare ramps at risk. I wouldn't hook up a new one
> until you have checked everything that hooks up to
> the ramps.

I had started wondering about that very same issue. Good thing I didn't turn anything on yet. I am learning things from all of this!

I suspect the issue is we are expecting to much of these LED power supplies. LEDs just don't need a lot of filtering, and it's not built into these units.

---

Yvan

Singularity Machine

@ appdev007

could you post some pics of your wavy prints.. I suspect i am having a similar problem with my 30A chinese psu but not so sevear...


thanks

Sorry, but I trashed it. I was ticked. The layer wavered like 0.05 mm back and forth along both axis. Only one like that since I got my axises smoothed out. The pololus use fine current control to do partial stepping, which is how we get micro sgepping. The voltage changing causes the current level to be off so the step isn't as accurate.

appdev007 Wrote:
-------------------------------------------------------
> The layer
> wavered like 0.05 mm back and forth along both
> axis. Only one like that since I got my axises
> smoothed out. The pololus use fine current control
> to do partial stepping, which is how we get micro
> sgepping. The voltage changing causes the current
> level to be off so the step isn't as accurate.

Interesting. I've made two attempts at printing the same part, first with the LED power, the second with the modified PC power supply. The second part looks smoother than the first.

I've now got LED power feeding the bed, and PC power for the rest.

Took a close look at the fried connector and RAMPS from ealier, and only the bed heat V+ is damaged. Heated bed checks out at 1.4 ohms. I don't know....? Aluminium dust? There is a bit of aluminium dust around the work area.... I'm stumped.

I have an ancient oscilloscope I should dig up and learn to use. At the very least, it would look very nerdy.

---

Yvan

Singularity Machine

@ yvan

pics please .. i think i have the same problem...need to verify before i buy a psu for my second printer

You are all attributing a lot to these power supplies, and I realize that its very possible to get a bad one, but my experience tells me that the situation is not that dire.

I used to have issues with the print stopping in mid-print - sonetimes simultaneously with something in the house turning on or off (usually fluorescent lights). I eventually isolated it to a weak USB connection. I got better and shorter cables, added some ferrite chokes, and haven't had a single problem in months.

I had waviness in my prints, particularly after the print head came around a corner. The problem was solved by reducing maximum acceleration.

Throughout all of this, my "cheap" Chinese LED power supply has been chugging along with no issues. It had nothing whatsoever to do with the issues I was having which sound a lot like the issues some of you are having. Bottom line is - it could be your power supply, but don't use the power supply as a scapegoat without considering other possibilities.

The pololus are carefully tuned circuits. Noise is going to affect them.

jbernardis Wrote:
-------------------------------------------------------
> I used to have issues with the print stopping in
> mid-print - sonetimes simultaneously with
> something in the house turning on or off (usually
> fluorescent lights). I eventually isolated it to
> a weak USB connection. I got better and shorter
> cables, added some ferrite chokes, and haven't had
> a single problem in months.

Better PSUs like the ones for PC have built in ferrite chokes as well as other circuitry for cleaning line power. Filtering the source power is part of a good PSUs job as well as not introducing new noise. From what my oscope is telling me these cheap LED (or I've even seen them labeled security camera) power supplies don't do either well.
I'm going to say whatever was putting interference on your USB cable was also present in the supply. I'm not sure if you added those ferrite chokes to the USB cable or the output of the supply. Either way you eliminated some noise and enough to allow the circuit to work. Had the PSU not been allowing or making so much noise you probably would have been OK

> I had waviness in my prints, particularly after
> the print head came around a corner. The problem
> was solved by reducing maximum acceleration.
These were the only prints I had ever had this issue on and it was an object I had printed several times before.

> Throughout all of this, my "cheap" Chinese LED
> power supply has been chugging along with no
> issues. It had nothing whatsoever to do with the
> issues I was having which sound a lot like the
> issues some of you are having. Bottom line is -
> it could be your power supply, but don't use the
> power supply as a scapegoat without considering
> other possibilities.
Again, noise could have been introduced thought both the supply and the usb cable. If you eliminated only the USB noise, then that could have brought the noise down to a tolerable level even though the PSU is still making noise. You maybe could have eliminated the PSU noise and left the USB cable alone, achiving the same result.

This is just my experiance. What I have gathered is that my single existing PC PSU was over loaded causing several issues. The LED PS wouldn't handle the load despite being rated 20A. With the PC PSU handing the heat bead over to the LED PS my machine now runs smoother and without occasional axis skips.

ekaggrat, technical difficulties are preventing me from taking a photo right now! Silly. In any case, I spent three hours fighting bugs to print two new test pieces, one with only the LED power, and the other with the PC power feeding everything but the heated bed.

You can just barely see more roughness on the surface printed with LED power, but it is not substantial. Few would notice. It could just be a fluke, more testing would be needed.

On the Techzone Mendel, with the monotronics board, we had problems we eventually traced to a Samsung cell phone chager. The truth is, silly things like that should not be interfering with a 3D printer! It's a matter of building reliablity into these things from the ground up.

---

Yvan

Singularity Machine