Wires Left Over After The PSU Mod

I am doing the PSU mod on an ATX unit. Since the wires are AWG 18 gauge only four yellow +12V wires will fit into the 5AMP and four into the 11AMP connectors on the RAMPS board (same number for the black). Still that's quite a chunk of copper. [Which makes me wonder if so much copper is actually needed when the traces inside the RAMPS board are much finer and might be able to carry less current? But I am an electronics ignoramus so I ask until I learn.]

Using up the sixteen wires above still leaves three yellow and more than three black wires. Rather than terminate those short inside the PSU like I plan to do with the other extras I figure to insulate them and coil them for future use for items - like maybe an E3D hotend fan, etc. - that could be controlled with individual switches. So I will save three yellow and three black wires for such.

Then, after soldering in the two resistors (as per the Mendel90 instruction PDF) to the black, green, sense and orange wires I will still have a bunch of wires left over.

Thus my main question. Will I ever need these others or can they be cut? I don't anticipate doing fancy stuff with the printer beyond the above mentioned E3D hotend fan, etc. Thus I only suppose I will need a few +12V wire pairs (the three above). But I don't want to cut the rest without knowing what I could thereby be limiting in the future. Could they be useful?

Am I making any sense? Thanks for any replies!

A 5V line is useful if you ever want to run a Raspberry Pi (or anything USB powered) - I regret cutting mine so short now as I have to dismantle it to get at a spare wire or tap into one of the load resistors. My OctoPi is currently still running off a spare Samsung micro-USB adapter as a result.

At the very least, cut them short and insulate using heat shrink but leave enough that it's easy to get at them without removing the PSU from the frame!

Yes the manual suggest cropping to 50mm and that is long enough to get at and solder a wire to later. I use 5V standby to power the RPI so it might be worth keeping that one long.

With wire gauge and PCB track width there are two different considerations: the current capacity and voltage drop.

For current capacity it is all about the heat generated, how it is dissipated and the temperature rating of the insulation. So, for example wire buried in a wall will have a lower current rating than equipment wire of the same gauge in free air. Higher temperature insulation allows a higher current rating with the same wire gauge. Thicker insulation for high voltage will reduce the current rating, etc. When wires are closely bundled they have to be de-rated.

PCB tracks are wide and flat so can get the heat out much better and therefore don't need as much copper for the same current capability. There are on-line calculators where you put in the current, desired temperature rise and the thickness to give the width. I have no idea if the RAMPS copper is correct, I have never checked it. A lot of Reprap boards are designed by amateurs and don't have enough copper.

For voltage drop the length is important, so because the wires are much longer than the PCB tracks it is more important that they have a lower voltage drop per unit length. I.e. they need to have a lot more copper area to keep the losses low, beyond what is needed for carrying the current.

Edited 6 time(s). Last edit at 09/26/2014 07:21AM by nophead.

---

[www.hydraraptor.blogspot.com]

Thank you both for advice! Wonderful explanation of capacity and voltage; thank you Chris.

So I will keep a +5V wire long. Do I need to keep one black wire for it?

I prefer to take the ground from the Melzi to minimize the noise on the USB ground. I.e. you want the ground of the RPI at the same potential as the Melzi's ground. If you take it from the PSU then any drop in the black wires caused by the high currents will cause a small offset.

I use one of limit switch ground terminals or the expansion connector.

Edited 1 time(s). Last edit at 09/27/2014 04:53AM by nophead.

---

[www.hydraraptor.blogspot.com]

Awesome! Thank you. I'm not planning to add one now but never know what will happen down the road. Now more good advice is here just waiting to be read.

I made a mistake and did not differentiate between the 5V standby and normal 5V (red wire). So I saved a red wire. Will this be a problem? I'm guessing the benefit of using the standby line would be it is live always and therefore the Raspberry Pi can always be communicated with?

Also, If I am using RAMPS, which would be a good place to terminate the ground from the Raspberry pi?

5V will run the RPI OK. The reason I use the standby 5V is that I switch the main PSU on and off with the RPI.

Not sure where a good ground on the RAMPS is. I don't use them because they don't have screw terminals or proper MOSFETs.

---

[www.hydraraptor.blogspot.com]

Thanks Chris. May I ask a few questions about the connectors and MOSFETs?

I am assuming you like screw terminals because they provide a more secure connection than the push fit? IF so, does this apply only to how easy they are to pull off the pins and thus they could come loose during operation?

I searched a bit on MOSFETs but don't really know what is the difference between RAMPS MOSFETs and Melzi MOSFETs. Is there a better MOSFET that could be used to replace the stock RAMPS FETs? Do all RAMPS makers use the same FETs?

It seems not too difficult to secure the connectors to keep them firmly in place, and if better FETs were available they should be easy to replace as well.

I'm just wondering ... Thank you.

Hi Garry,

Push-fit connectors such as the Molex KK series can be problematic because the contact area is just one side of the square pin. The best types of push-fit connectors provide a gas-tight contact on all four sides of a square pin.

Regarding MOSFETs, there are two parameters that are important.

The first is the on-state resistance (R_DS) which needs to be as low as possible to prevent excessive heating of the device. Devices with ultra-low R_DS (a few milli-ohms) cost significantly more than those exhibiting tens of milli-ohms. The power dissipation needs to be considered if the electronics is operated at an elevated temperature e.g. in an enclosure.

The second parameter is the gate-source voltage (V_GS). R_DS is specified for a particular V_GS (because it is related to it). Quite often this can be 10V and below the specified value R_DS increases. A MOSFET must be properly specified for use with logic-level drive which will be 5V or less.

Some MOSFETs are quoted as being logic-compatible. What this often means is that the device exhibits a low-ish R_DS at typical logic levels. However, unless the device also includes gate drive circuitry, it is still an analogue device and, in order to achieve minimal power dissipation, it must be driven with an adequate V_GS and sometimes typical logic levels do not achieve this.

I do not have a very high opinion of the design philosophy behind many RepRap electronics. The designers seem to go all-out for lowest cost and neglect reliability. This is evident in the measures that Chris has had to adopt to make his Melzis more reliable and that should not be necessary.

Regards,
Neil Darlow

Edited 3 time(s). Last edit at 11/18/2014 05:52AM by neildarlow.

---

I try to write with consideration for all nationalities. Please let me know if something is unclear.
Printing with Mendel90 from fedora 25 using Cura, FreeCAD, MeshLab, OpenSCAD, Skeinforge and Slic3r tools.

Hi Neil,

Thank you for that excellent information about MOSFETs. Again, lacking any background in electronics, it has taken me a while to digest the technical considerations you mention. While not fully understanding everything you write about the switching threshold value I think I have grasped MOSFET basics.

1) Cheaper MOSFETs will have higher resistance and therefore get hotter. This might/would necessitate cooling but might/could be overcome by replacing the poor MOSFET with a higher quality MOSFET. I am understanding? If so, which MOSFET would be best to substitute into the RAMPS board for highest reliability? This seems an easy fix and well worth doing to achieve more solid performance if indeed it was needed.

2) I am assuming the MOSFET for the heated bed would be subject to the most stress and would be the one that should be higher performance? Personally if I were going through the effort of sourcing and buying a better MOSFET I would replace all three. While replacing the MOSFET should not be difficult I really have no idea how to choose the right one and would have to rely on someone giving me a recommendation.

If I'm reading you correctly it comes down to the quality of the components used. And I am assuming that Chris has actually had his Melzis made with better components than some vendors? I would rather pay more for a better board if I knew the components were very high quality. Stability, reliability and longevity are high on my list of priorities.

P.S. Doesn't the RAMPS board use Dupont connectors? And don't the female connectors make contact with the four sides of the square male pin?

Thanks again!

>And I am assuming that Chris has actually had his Melzis made with better components than some vendors?

The MOSFET on the Melzis we supply are these FDD8870

30V, 160A, rDS(ON)= 4.4mΩ, VGS= 4.5V, ID= 35A

A 10A bed doesn't trouble them!

Edited 1 time(s). Last edit at 11/20/2014 03:38PM by nophead.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
A 10A bed doesn't trouble them!

And that is why I love my Mendel90; it surely doesn't give trouble. The heated bed just cranks up hot; and I never even knew it could cause problems for the electronics until I learned some people have issues with other machines! That's the way it should be.

For RAMPS which needs three pins would this be a similar MOSFET? FDP8870 I assume the 156A is the correct one of the two listed.

Yes, interesting the smaller surface mount version can handle a little more current than the through hole T0220.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
Yes, interesting the smaller surface mount version can handle a little more current than the through hole T0220.

Thanks! Then I will get a few of these and upgrade the MOSFETs for a fun soldering exercise and a bit more peace-of-mind.

The surface mount MOSFET of the Melzi is smaller but has no heatsink so I am assuming the larger through hole MOSFET could live without a heatsink as well? But I'll use the heatsink anyway.

The surface mount parts use the copper on the PCB for a heatsink.

The junction to ambient thermal resistance for the TO-220 package is 62°C/W, so the worst case temperature rise at 10A is

I^2 . rDS(ON) . RθJA = 10 * 10 * 0.0046 * 62 = 29°C.

So less than 50C in a warm room, so I wouldn't bother with a heatsink. The datasheet quotes a max of 19A without a heatsink although that is with 10V gate drive.

Edited 2 time(s). Last edit at 11/21/2014 05:38AM by nophead.

---

[www.hydraraptor.blogspot.com]

Sweet! Thanks Chris. Now as long as I use good thick cabling to the bed I'm thinking the heated bed loop should be fine.

If it is a moving bed then thick wires need a large minimum bend radius. I think it is probably a least a square law on wire gauge. That is why I use ribbon cable with lots of thin wires.

---

[www.hydraraptor.blogspot.com]

Yes, I am going to use your ribbon method from the heated bed to the frame base, and then wires to the controller. Sorry, I should have mentioned that. Your method is working perfectly on my current machine!

Curiously, for the wire section, is it OK if I use a thicker wire for the positive than the negative? Or must they be the same? That's just because I have wires in different sizes now. And also, I know thick wires are good but aside from being overkill, and a waste, could too thick wires cause a problem?