bah, I say, to ATX supplies

I've gone through 3 supplies now: original I had laying around, a used gamer-grade 750W, and a new 450W. They all died on me shortly after use. My max (steady) current draw is about 16 Amps for the heat bed and one hot-end, plus the motion stuff. The 750W was rated for 60 amps! It still died though, partly I think because I may have accidently shorted the output, although were no sparks or anything.
I finally decided I had wasted $80 trying to save money and instead found a dedicated OEM power supply, same brand I tend to use for designs at work: $50. I'm sure you can find a lower power supply
Mean Well, NES-350-12. Tight regulation of 12V and 30 amp output, more than enough for future expansion. Run all day at full load. No minimum load. Regulated self-cooling fan. Over-voltage, shorted-output, temperature safety circuits. Hard to bust this guy. I opened up the dead 450W supply and its electronix were really cheap and puny compared to the Mean Well, for only a few dollars less and a whole lot of PC cables and connectors I don't want.

[amzn.com]

feedback?

There has been a lot of discussion about this on the Ormerod forum. The RepRapPro Ormerod 1 kit shipped with a 550W ATX PSU and an adapter board to put a minimum load on 2 of the outputs. Several users reported problems with the PSU and it was barely sufficient to heat the bed to ABS temperatures due to poor regulation. I changed to a cheap Chinese 12V 25A LED power supply. It's fanless, so silent (the 30A supply in the same range does have a fan). Cost was about GBP20. It has a voltage adjustment, so it can be turned up to about 14V for faster heating. It's been working perfectly for nearly a year. I published a design for a printed cover for it at [www.thingiverse.com].

For the Ormerod 2 kit, RepRapPro switched to a similar PSU but rated at 20A.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I just picked up a Mean Well SP-600-24 (24 volts, 600 watts) last week. Got tired of waiting 40 minutes while my heated bed struggled to reach 110C. The fan is louder than a PC-based supply, but the thing isn't breaking a sweat and gets the 300mm diameter Onyx up to 110C in 6 minutes. Wish that I upgraded sooner!!!!!

Edited 1 time(s). Last edit at 01/26/2015 01:47PM by vreihen.

I think the issues people have with ATX is that they get them so cheap. If an ATX supply puts out 60 amps AND has all the connections/voltages/features for a computer, why do they cost far less than a dedicated supply? The expectations people have for power vs. cost are simply unrealistic.

I did the whole cheap $50 ATX thing and hated it as well. A buddy of mine recently gave me the power supply out of his old gaming rig- it's only rated for like 30 amps but it's really nice and cost him $180 new. I get to a full 110 degrees in four minutes without fail.

Quote
shadowphile
I've gone through 3 supplies now: original I had laying around, a used gamer-grade 750W, and a new 450W. They all died on me shortly after use. My max (steady) current draw is about 16 Amps for the heat bed and one hot-end, plus the motion stuff. The 750W was rated for 60 amps! It still died though, partly I think because I may have accidently shorted the output, although were no sparks or anything.
I finally decided I had wasted $80 trying to save money and instead found a dedicated OEM power supply, same brand I tend to use for designs at work: $50. I'm sure you can find a lower power supply
Mean Well, NES-350-12. Tight regulation of 12V and 30 amp output, more than enough for future expansion. Run all day at full load. No minimum load. Regulated self-cooling fan. Over-voltage, shorted-output, temperature safety circuits. Hard to bust this guy. I opened up the dead 450W supply and its electronix were really cheap and puny compared to the Mean Well, for only a few dollars less and a whole lot of PC cables and connectors I don't want.

[amzn.com]

feedback?

Hi,

I am a developer, and I strongly recommend good quality ATX power supplies instead of cheap, noname OEM 12V PSUs. I wrote quite a bit of documentation about it in the wiki, here: [www.reprap.org]

which I suggest you read before posting your anecdotal, and apparently biased information about ATX PSUs.

You don't mention the brand of the 450W ATX PSU which had cheap "electronix" in it but I am quite certain it was not, for example, a Corsair or Antec or XFX or Seasonic PSU. It's quite true there are tons of cheaply built, badly designed ATX PSUs that you can find on eBay which are totally inadequate for either use in a PC or in a RepRap. In general, you have to spend at least $40 to get a new good quality ATX PSU from a reliable brand with a power rating around 350~400W. In the EU I suggest to use be quiet! power supplies, open one and you won't find anything cheap inside. German quality...

As for using an old ATX PSU, that's also a common mistake. Don't. Just don't.

I took one look at the 550W power supply shipped with the original Ormerod kits and I immediately knew they had chosen a bad ATX PSU. Unfortunately kit suppliers, just like individual builders, sometimes make mistakes (and won't admit it). Modern good quality PSUs ("Haswell Ready") do not require balancing resistors.

Well I DID ask for feedback!
But I thought I had selected a decent 450W supply because I know that PC supplies can be vastly under-powered compared to their rating (I paid $45 for it). And the 750W was a Corsair.
I'm just used to buying OEM supplies for work (and I use them for general purpose applications and never thought of them as 'LED' drivers; guess that has been a rising consumer application).
I guess we will see how well my power supply lasts.

AndrewBCN, I agree with you that good branded ATX PSUs are better for running 3D printers than cheap ATX PSUs. For a while I used a Corsair CX430M in place of the Alpine 550W PSU supplied with my Ormerod kit, even though the Alpine PSU worked OK for me. But I don't understand what you have against 12V OEM power supplies. Lots of people are using them. Have you any evidence that they have a high failure rate, or are causing problems for 3D printer owners? You talks a lot about PFC in your wiki page, but that is more relevant to the electricity distribution companies than it is to 3D printer owners. Modern 3D printer electronics doesn't need a very well regulated 12V supply - modern boards (not cheap RAMPS clones) are typically tolerant of anything from 12V to 35V - so even if a top-brand ATX PSU might have better 12V regulation than a 12V OEM power supply (which in turn will probably have better regulation than a cheap ATX power supply), this is of little or no consequence.

Unless I am presented with any evidence that 12V OEM power supplies are problematic (and my experience and that of other Ormerod owners is that they are not), I will continue to recommend them to users who are comfortable with doing mains wiring. And I recommend a good modular ATX power supply such as the Corsair CX430M to those who are not.

One thing you didn't mention in the comparison table on your wiki page is that the 12V OEM power supplies have a voltage adjustment. This is very useful if your heated bed is slightly underpowered.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

ATX supplies usually require both 5V and 3.3V to have some minimal load before GOOD 12V regulation kicks in. The requirement vary widely and the manufacturers would not publish numbers on what is required to run a good, regulated 12V only output (since they don't design the things to be run that way). It's really a crap shoot for the sake of day one savings.

Worth noting that neither an ATX power supply nor a 12V OEM power supply that may be intended primarily for powering LEDs is designed for the load profile of a 3D printer. The load of a 3D printer may vary from 100mA or so when the motors and heaters are all off, to 20A or more when all motors or heaters are on.

The OEM supply might be optimised for a constant load at or close to its maximum rating. At light loads (e.g. when the heaters and motors of a 3D printer are off), its regulation may suffer and the output voltage might reasonably be expected to rise. This should not matter for most 3D printer electronics, which are insensitive to the voltage on the 12V rail as I have said before.

The ATX power supply is designed to work with substantial loads on the 5V and 3.3V rails, which it does not get from a 3D printer, especially if the printer 5V and/or 3.3V rails are derived from the 12V rail, as is often the case. This is IMO a worse mismatch to a 3D printer than the OEM supply.

I took some voltage measurements on the 12V 25A OEM PSU that powers my Ormerod:

- With all motors and heaters off: 12.25V

- With XYZ motors on: 12.21V

- With XYZ motors, bed heater and 1 extruder heater on full: 11.84V

These voltages were measured at the power terminals of the printer electronics, i.e after the voltage drop in the cable from the PSU. From these figures I am satisfied that with this particular PSU, regulation is not an issue I need to worry about.

Edited 2 time(s). Last edit at 01/28/2015 04:58PM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

My main issue with ATX power supplys is that they don't come in 24 volt versions.
The whole system benefits from higher voltages (if designed for it), so I don't consider The ATX route a viable option.

Quote
dc42
AndrewBCN, I agree with you that good branded ATX PSUs are better for running 3D printers than cheap ATX PSUs. For a while I used a Corsair CX430M in place of the Alpine 550W PSU supplied with my Ormerod kit, even though the Alpine PSU worked OK for me. But I don't understand what you have against 12V OEM power supplies.

Please note that I don't have anything against any PSU, brand or person. And I try to keep my opinions (I do have strong opinions about some things - sorry if I sounded harsh, shadowphile) separate from facts. I have presented the facts in favor of and against using LED strip PSUs in RepRap projects in the wiki. From that point on people do as they wish.

Quote
dc42
Lots of people are using them. Have you any evidence that they have a high failure rate, or are causing problems for 3D printer owners? You talks a lot about PFC in your wiki page

Actually I don't, there is a short paragraph about PFC and a link for further reading.

Quote
dc42
, but that is more relevant to the electricity distribution companies than it is to 3D printer owners. Modern 3D printer electronics doesn't need a very well regulated 12V supply - modern boards (not cheap RAMPS clones) are typically tolerant of anything from 12V to 35V - so even if a top-brand ATX PSU might have better 12V regulation than a 12V OEM power supply (which in turn will probably have better regulation than a cheap ATX power supply), this is of little or no consequence.

Unless I am presented with any evidence that 12V OEM power supplies are problematic (and my experience and that of other Ormerod owners is that they are not), I will continue to recommend them to users who are comfortable with doing mains wiring. And I recommend a good modular ATX power supply such as the Corsair CX430M to those who are not.

Yes, the Corsair CX430M is indeed a good choice for RepRap projects. An unnamed, unspecified LED strip PSU, on the other hand, in my opinion is a very, very bad recommendation.

BTW any evidence that you or I could present here in this forum would be purely anecdotal i.e. not really useful. You really need to either:
1. Collect statistical data (realistically thinking this is quite impossible at the moment).
2. Provide a professional analysis and comparison of the circuits in ATX PSUs and LED strip PSUs, with a detailed description of the components used and their failure modes. That is also beyond what anybody here in this forum can possibly do.

So, your request for proper evidence is like asking for a piece of the moon. Neither you nor I can provide any evidence here. However, there are facts, which I have presented in the wiki page.

Quote
dc42
One thing you didn't mention in the comparison table on your wiki page is that the 12V OEM power supplies have a voltage adjustment. This is very useful if your heated bed is slightly underpowered.

Are you aware that the very small regulator on the Arduino is just up to the task of regulating down from 12V to 5V (and actually the RAMPS board has a diode that already brings that down to approx. 11.4V, which is a good thing)? If you raise the voltage of the OEM PSU significantly above 12V you could overload that regulator and get the dreaded smoke coming out of your RepRap electronics. I don't really consider that an advantage or anything that I could recommend that people try. But again, that is my opinion. You can recommend whatever you want and people will do whatever they want.

If a heated bed cannot reach 105C with the 12V from an ATX PSU I would rather recommend that people insulate it (see "heatbed insulation" in the wiki) rather than any other hack involving raising the voltage from the PSU. Simpler, safer, and in the end cheaper too. Again, imho.

Quote
anwe79
My main issue with ATX power supplys is that they don't come in 24 volt versions.
The whole system benefits from higher voltages (if designed for it), so I don't consider The ATX route a viable option.

That is a rather blanket statement with a big "if". The fact is that at the moment, RepRap electronics are designed for 12V. If you want to use 24V you are pretty much in experimental territory and you'd better know what you are doing. Electricity and smoke often go together in my experience.

Quote
shadowphile
Well I DID ask for feedback!
But I thought I had selected a decent 450W supply because I know that PC supplies can be vastly under-powered compared to their rating (I paid $45 for it). And the 750W was a Corsair.
I'm just used to buying OEM supplies for work (and I use them for general purpose applications and never thought of them as 'LED' drivers; guess that has been a rising consumer application).
I guess we will see how well my power supply lasts.

Sorry if I sounded harsh above.

Quote
dclarkm
ATX supplies usually require both 5V and 3.3V to have some minimal load before GOOD 12V regulation kicks in. The requirement vary widely and the manufacturers would not publish numbers on what is required to run a good, regulated 12V only output (since they don't design the things to be run that way). It's really a crap shoot for the sake of day one savings.

As I mentioned above and in the wiki, "Haswell Ready" ATX PSUs (in other words, most PSUs sold nowadays, in January 2015) do not require balancing load resistors, they are designed to provide good 12V regulation with small or negligible loads on the 5V output. Also note that practically all the ATX PSUs sold in the last 5 years or so derive the 3.3V output from the 5V output, so if you need a small load to get good 12V regulation, just providing a load on the 5V output is enough, the 3.3V output can be totally ignored.

Quote
AndrewBCN
An unnamed, unspecified LED strip PSU, on the other hand, in my opinion is a very, very bad recommendation.

I guess we'll have to agree to differ then. If there were lots of horror stories of OEM PSUs catching fire or being dangerous in some other way, then would agree with you - but there aren't.

I agree that neither of us can collect enough evidence to have statistical validity. OTOH I have seen more than a handful of reports of cheap ATX PSUs not being up to the job, and none at all of OEM PSUs not being up to the job. All the posts I have seen from users who replaced a cheap ATX PSU with an OEM PSU have reported good results. This may not be statistically valid data, but it is enough for me to reach the tentative conclusion that the OEM PSU is better suited to the job than the cheap ATX PSU, and that it is probably good enough.

Also worth noting is that RepRapPro changed from an OEM PSU on the Mendel kit, to an ATX PSU on the Ormerod 1 (to make the wiring easier), and back to an OEM PSU on the Ormerod 2 because the ATX PSUs proved problematic. Of course, they are working to a budget in a competitive market, so an expensive top-brand ATX PSU was not an option.

Quote
AndrewBCN
Are you aware that the very small regulator on the Arduino is just up to the task of regulating down from 12V to 5V (and actually the RAMPS board has a diode that already brings that down to approx. 11.4V, which is a good thing)? If you raise the voltage of the OEM PSU significantly above 12V you could overload that regulator and get the dreaded smoke coming out of your RepRap electronics.

Firstly, IMO 3D printer controllers based on 8-bit Arduinos are obsolete. I (and many others) don't use them. Secondly, it is a myth that the 5V regulator on the Arduino Mega can't handle more than 12V input. For a long time now, Arduino Megas have been specified up to 20V. The reason that the Arduino team says the maximum recommended input voltage is 12V is because most Arduino users are incapable of working out the total load on the 5V rail and hence the power dissipation in the regulator. The Arduino team makes it hard to do by anyway failing to specify the current consumption of the Arduino alone. An Arduino Mega driving a RAMPS board and 4 stepper drivers takes around 90mA. At 15V input, the regulator is not too hot for me to keep my finger on. So there is absolutely no problem with running the Mega with 14V input. OTOH if you add a 12864-based LCD display that includes a backlight taking about 100mA from 5V, then even with a 12V input the regulator will much hotter than I would accept - but nobody ever mentions that.

Quote
AndrewBCN
I don't really consider that an advantage or anything that I could recommend that people try. But again, that is my opinion. You can recommend whatever you want and people will do whatever they want. If a heated bed cannot reach 105C with the 12V from an ATX PSU I would rather recommend that people insulate it (see "heatbed insulation" in the wiki) rather than any other hack involving raising the voltage from the PSU. Simpler, safer, and in the end cheaper too. Again, imho.

If you go over to the Ormerod forum and look at some of the posts about ABS printing, you will find several people reporting that even with an insulating blanket on top, it takes too long to get the bed to ABS temperature, and the temperature can't be maintained when the blanket is removed to start printing. The option of increasing voltage has been very important to those people. I did get my heatbed up to 110C using an ATX PSU, but it took 20 minutes - which is too long to wait before each print.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42
If you go over to the Ormerod forum and look at some of the posts about ABS printing, you will find several people reporting that even with an insulating blanket on top, it takes too long to get the bed to ABS temperature, and the temperature can't be maintained when the blanket is removed to start printing. The option of increasing voltage has been very important to those people. I did get my heatbed up to 110C using an ATX PSU, but it took 20 minutes - which is too long to wait before each print.

Then go to any Delta forum and read about the 300mm diameter Onyx heated bed taking 30-40 minutes while it struggles to reach 110C. Mine could not hold 105C wired directly to 12V using 14 gauge wires. Thanks to the magic of Mean Well and a shiny new SP-600-24 (600W, 24V) power supply two weeks ago, my heatbed now hits 110C in 6 minutes (power supply cranked down to 19.2V). I'm using an SSR to control the heatbed, and bought a 12V buck converter to power the logic/heaters/steppers from the 24V supply. Like I said, I wish that I had spent the $79 for this power supply a year ago.....

Quote
AndrewBCN
That is a rather blanket statement with a big "if". The fact is that at the moment, RepRap electronics are designed for 12V. If you want to use 24V you are pretty much in experimental territory and you'd better know what you are doing. Electricity and smoke often go together in my experience.

Yes agreed, you need to check that every part on the printer can manage 24 V before going that way. Posting from phone so tried to keep it short. I wouldn't agree that there is such a thing as standard RepRap electronics though. Even RAMPS handles 24 V pretty well if you take proper precautions. I've run that way since day one on my printer. Fans have turned out to be the biggest problem, 24 V fans are very rare compared to 12 V.

Quote
AndrewBCN

As I mentioned above and in the wiki, "Haswell Ready" ATX PSUs (in other words, most PSUs sold nowadays, in January 2015) do not require balancing load resistors, they are designed to provide good 12V regulation with small or negligible loads on the 5V output. Also note that practically all the ATX PSUs sold in the last 5 years or so derive the 3.3V output from the 5V output, so if you need a small load to get good 12V regulation, just providing a load on the 5V output is enough, the 3.3V output can be totally ignored.

I have an ANTEC 450W power supply, never having done this before, watched a couple of youtube videos, opened the case and started cutting wires off. bundled the yellow 12v and black ground for each rail, to get two 12v 18A circuits, that, in theory could power two 3D printers.....but then I find out I need to have a load on the 5v rail in order to get it to work properly. I have the switch green connected to black ground in order to have the on/off switch work.
I have googled and googled, trying to find out how to properly add resistors to the 5v rail....I can work with household electrical DIY but this is all new to me.
What resistors, how to connect, need wiring diagram, video or something with enough detail to get me going in the right direction.
Thank you
Richard

Quote
RTurnock
I have an ANTEC 450W power supply, never having done this before, watched a couple of youtube videos, opened the case and started cutting wires off. bundled the yellow 12v and black ground for each rail, to get two 12v 18A circuits, that, in theory could power two 3D printers....

Be very careful! I don't know about the Antec in particular, however it may be that the 5Vand 12V negative side connections are internally connected on the power supply PCB. Use a multimeter to check. Also, the cheaper ATX PSUs (again, I don't know about the Antec) have a single voltage regulation feedback loop that feeds a proportion of the 12V rail and a proportion of the 5V rail back to the primary circuit. This will feed back completely the wrong voltage if you separate the 5V and 12V grounds.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42

Quote
RTurnock
I have an ANTEC 450W power supply, never having done this before, watched a couple of youtube videos, opened the case and started cutting wires off. bundled the yellow 12v and black ground for each rail, to get two 12v 18A circuits, that, in theory could power two 3D printers....

Be very careful! I don't know about the Antec in particular, however it may be that the 5Vand 12V negative side connections are internally connected on the power supply PCB. Use a multimeter to check. Also, the cheaper ATX PSUs (again, I don't know about the Antec) have a single voltage regulation feedback loop that feeds a proportion of the 12V rail and a proportion of the 5V rail back to the primary circuit. This will feed back completely the wrong voltage if you separate the 5V and 12V grounds.

Wow! thanks....also, I finally found this discussion from a year ago:
[forums.reprap.org]

Hmmm, separate the 5v and 12v grounds? I will take the cover off and look again.....seemed like all the black/ground wires were coming off the same rail.
Thank you for the warning...I will do some testing.

It seems how to design fail safe device is sacred knowledge in electronics field.

There is well known phrase "You get what you pay for!"
In ideal situation every builder can afford decent components that have proven reliability. But in reality especially hobbyist, tinkerer who do electronics as second activity, money is always an obstacle. Every time I decide to build something, I spend good time to figure out involved logistics - cost and time: buying a brand new product, cheap clone product, used one, building from scratch.

I had spent few days just to modify and troubleshoot two ATX PSUs for RAMPS 1.4 controller.
I learned a lot from it and the bottom line is I wouldn't hook up any modified ATX PSU or 12V power supply to the heatbed that consumes 11A of current.
I'll run the heatbed with 220V heating device that has temperature control feature.
Two 450W ATX PSUs I modified fails to run 110W of actual load that consumes around 8A in maximum.
The thread link is here. (http://forums.reprap.org/read.php?13,463988)

I bought 220VAC 20W Hakko presto few years ago that three or four times expensive than Chinese soldering irons.
It consumes only 90 mA in operation but the power cord is so thick. Now I can understand why the builders designed the iron the way it is.
Any device that consumes over few amps is dangerous. My understanding of electric danger is mostly high current not high voltage.

Five year old, 450W ATX PSU I modified for RAMPS 1.4
It has +12V@20A. I intend to run it for maxium 5A to 6A without heatbed control.




Replaced two polyfuse with 15A, 5A blade fuses and installed fuse holders for easy maintenance.
I think it's okay to run the RAMPS board as is. When it's blown, buy another one because it's so cheap under $20.
Unfortunately I'll have to way two to five weeks to get it.


15A circuit breaker unit I plug test equipment. whenever I run.