Powering 30A 12v Heatbed with a mosfet

Hello,

So for a custom printer I'm building, I'm using RAMPS 1.4 electronics, and a 300x300mm 12v 30A heatbed.
Since I can't use the RAMPS directly for such high current, I bought a couple of mosfets & heatsinks to switch the bed(https://www.fairchildsemi.com/datasheets/RF/RFP70N06.pdf), but I'm not sure how to properly wire it up.

I've unsoldered the heat bed mosfet on the RAMPs board, and what I'm planning to do is to connect the gate of a 70N06 mosfet to the gate trace on the RAMPS board, and directly connect the drain and source of the mosfet between the board's negative terminal and the PSU's ground. I'm buying a 40a 12v PSU for the heated bed, and I have an old ATX PSU for the RAMPs electronics. For connecting the 40a PSU to the heat bed, I have some 1.5 sqmm & 2.5 sqmm copper house wiring cable, which I think should be sufficient to carry the current.

My questions are-

1. Would 1.5mm multistrand copper wiring be sufficient to carry 30A 12v?
2. How do I connect such thick cable to a MOSFET's thin leads?
3. Is this a electrically sound setup?

I have my electronics mounted in an acrylic electrical box, with a fan mounted on top for cooling. I'm not extremely fluent in electronics, although I know my basics, so any help is appreciated.

Thank you,

I have the exact same question here:

[forums.reprap.org]

I am going to put 3 mosfets in parallel. I plan on making the circuit this weekend so keep an eye on that thread for my experience.


I think your chosen MOSFET will need to be run at 10V or higher to get the required performance. If you look at page 3 figure nr. 7 you see that you are almost in the flat region at 30A with 5V gate voltage. This means you get a really high voltage drop at 30A which in turn means your mosfet heats. You need a V(gs) of 8V.. Mosfets are not on/off switches that works at the gate thresshold voltage. If you want to transfer serious currents you need to be well above that value.



Disclaimer: I am not electronic expert. I only "really" started learning about Mosfets the other day when I made that post. The above is my best understanding.

Quote
LarsK
I have the exact same question here:

[forums.reprap.org]

I am going to put 3 mosfets in parallel. I plan on making the circuit this weekend so keep an eye on that thread for my experience.


I think your chosen MOSFET will need to be run at 10V or higher to get the required performance. If you look at page 3 figure nr. 7 you see that you are almost in the flat region at 30A with 5V gate voltage. This means you get a really high voltage drop at 30A which in turn means your mosfet heats. You need a V(gs) of 8V.. Mosfets are not on/off switches that works at the gate thresshold voltage. If you want to transfer serious currents you need to be well above that value.



Disclaimer: I am not electronic expert. I only "really" started learning about Mosfets the other day when I made that post. The above is my best understanding.

Hmm, that's a very important point you've made, I didn't really study the charts properly. Looking at this, I think I need to replace my mosfets, but thankfully you replied before I went on a scheduled trip to electronics town.
In terms of specific parts, how do you find which particular electronic component is suitable for your purposes? Do they have lists of them and their basic specifications for comparison & reference?
Also, what gauge wire are you using? and how do you connect it to the mosfet? I've got access to decent electronic supply stores, but getting specialized equipment or PCBs is unlikely.
Is there any advantage to using multiple parallel mosfets rather than a single, more powerful one besides thinner wires, and better heat dissipation?

Thanks so much,
A

Hey,

I just picked the mosfets that I am using now based on some other people using them here in the forum. I had a shortlist of a few which were all mentioned here in the forum and of those the IRLB8743 seemed the best.

Again, I am no electronic expert, but it seems to me, that you can actually not get a Mosfet of the normal sized type that can do 30A without serious active cooling and so. I am putting mine in parallel because I hooked up one alone, and it got really hot and I just didn't feel comfortable trying to manage that much heat in a setup that I leave on unattended overnight.



I THINK, that if you just want your printer up and running and don't care so much about learning about Mosfets and PCBs and so, then the way to go is a SSR (Solid State Relay). With those you can just pay a few bucks extra and you get it all, except high switching rates, but at the end of the day we don't really need that kind of tools on the heatbed... So if you have not purchased anything yet, think about just picking up a high-end SSR.


A quick google search turns up: [www.phidgets.com]

Well getting the high-end relay is surprising expensive;

aliexpress.com/item/KYOTTO-SSR-KG1040D-New-and-original-in-stock-ready-to-ship-DC-Solid-State-Relay-40A/1328574617.html

Notice that DC42 (a very famous user in here) warned against using the Fotek. If you google them you get a lot of bad stories so seems right.

Hmm, the only cheap ones locally available on amazon are fotek, I guess I'll check in my local electronic stores for any other brands, but I think I'd likely just go for mosfets now.

I was planning to power the RAMPs board from an old ATX power supply I had, and buy a 12v 40a PSU for the heatbed, but do you think it would be a better idea to just buy a 12v 50a PSU and use it to power both RAMPS+electronics and the 30a heatbed? This is taking into account using stronger supply so that it isn't running at the edge of its capacity. Using a 50a power supply for the entire printer also solves the issue with the mosfets requiring a singular ground.

Hey, I power mine with a 50A. I think that was a mistake actually. Should have picked something bigger. It is not good practice to run that close to the capacity of the PSU. Maybe I will pickup a 80A.

Heatbed: 30A
Heater 1: 4A
Heater 2: 4A

Nema 17: 5x1.20A = 6A

Total: 44 A.

Also I think it is wrong to power it with two independent PSUs from a design point of view. Meaning, if you can get one big unit that does it all, why split it on 2 PSUs? Only advantage I see is that you can use a PC type PSU which has the shutdown signal so your printer turns off when print is done. I plan to make something like that myself but with a PC - PSU it would be one less circuit.

From an engineering point of view it is fine with the two PSUs as long as they are on common ground and that the + side is not. You already do this on the RAMPS with the 24V option on the heat-bed.


About MOSFETs vs SSR, - I think at the end of the day you have a cost versus time investment here. Unless you put a real heat-sink on the MOSFET, then you need to put them in parallel like I did.

When I say real heat-sink I mean something like these:






So maybe you could do something like the 2nd picture on an old PC CPU cooler and connect the leads directly to the legs of the MOSFET. Then you don't have to make a PCB.

I have no idea if it would work. If you do it, I highly recommend you install it in a metal box where a meltdown can't set fire on anything.

hey,

If you go with MOSFET and haven't purchased any yet, then you should go with this one I think:

[www.ti.com]
CSD18536KCS


It has half the on-resistance of the one I am using (IRBL8743). It is a lot more rare also.

It seems to be a little rare (the TI one) and you will pay 3.5 USD a piece, but for me it looks to be the best in the market. Otherwise have a look at the Digikey alternatives:

[www.digikey.com]

Thanks, I'll try to see if I can procure the CSD18536KCS or IRLB8743PbF, it really depends on whether my local stores stock it or not. Although the CSD's gate threshold is just at the 5v level, I hope it won't cause any issues.

Here is my current setup, all stuck into an acrylic electrical box-


The heatsink is just temporarily placed on the right, I was thinking of mounting it to the back wall, as I'm installing a fan to direct airflow through the box to cool all the electronics down. (Link to my blog where I've posted pictures of the cooling setup) I'm a bit hesitant about putting anything into an metal box because it might short something due to the uninsulated back of the mosfet, although I'm also concerned about figuring out how thick my heat bed wire gauge should be to avoid combustion.
I've got more heatsinks, so if I find that the mosfet setup heats significantly, I'll migrate it to a separate enclosure and have multiple mosfets with heatsinks & a fan.

I calculated my total power draw, and I estimate it should be around 460w or 38a. Leaving 10a for a safety margin brings it up to 50a, but even if I get a 60a SMPS, it won't draw any more current than the 50a SMPS providing the current draw doesn't exceed expectations, right? In that case, its probably just a better investment to spend a little more and get the 60a supply.

Edited 1 time(s). Last edit at 06/26/2016 01:42PM by A_Designs.

Hey, sounds like a solid plan. Test it and see how it works out.

The TI CSD MOSFET has its Drain-to-Source On-Resistance given at 4.5V as 2.2 mOhm max (typ 1.7 mOhm). That is at 100A power drain (!).

When it is written like that in the sheet it is a guaranteed figure. The curves in the later pages are all "typical" but when it is given in the "specification" part it is trustworthy.

So then you go I * R = V and find you will have a 0.066V drop. Then P = V * I and find that you will have 1.98W. Or just the equation: I^2 * R = P (EDIT: What I calculated on this line is the power dissipation inside the chip. This is the heat you need to get rid of if the MOSFET is permanent ON)

Then go back to the datasheet, Look at :

"Junction-to-Ambient Thermal Resistance " this says 62C/W. This means it gets to 62C hotter per Watt if just let in ambient (25C). So at 2W this means if you just let this Mosfet sit without anything whatsoever it will get to 62 * 2 + 25 = 149C.

Now the datasheet also gives you Junction-to-Case Thermal Resistance which assumes a perfect heat sink is attached. Then it is 0.4C/W. In that case your MOSFET will be 0.4*2 + 25 = 25.8.

Obviously there is no such thing as a perfect heatsink and you then need to design for something in between these two extremes.

Now you might ask; How hot can a MOSFET be allowed to run? Well... The datasheet gives typical charts at 25C and 125C. The max junction temp is 175.

I think it is proper to dimension for 75C as max on the junction temperature.

I am starting to think you might not even need a dedicated fan cooling. Just cool the entire cabinet (like I think it is already your plan) and put a big passive on.


Now I should stress that the above math is HIGHLY dependent on the chip. Double the Rds ON and suddenly you need an entire different setup, and you don't find any other chips that does 100A at Vgs 4.5 with a guaranteed max 2.2 mOhm. That TI chip we are talking about here is really something! I am thinking of picking up a few just to have it on storage for future projects. But at 3.5 USD a piece I imagine it never made a big hit in the industry.

Edited 1 time(s). Last edit at 06/26/2016 04:11PM by LarsK.

A few comments:

1. For bed heaters larger than 200 x 200mm, it is much easier to use 24V power and halve the bed current than to stay with 12V and have to wire for very large currents. But a standard RAMPS doesn't give you that option. Preferably, upgrade to 24V-capable electronics. If you don't want to do that, using two separate PSUS (24V for the bed heater and 12V for everything else) with your RAMPS is a reasonable option.

2. For really large beds, mains voltage bed heaters are the way to go, switched by an inexpensive zero crossing DC-AC SSR. The Fotek ones are OK for this. But of course you need to take extra safety precautions.

3. That TI mosfet does appear to have very good specifications. But in practice, it's the PCB trace widths and other wiring that are the limiting factor, not usually the mosfet, unless you choose a particularly poor mosfet such as you find on most but not all RAMPS boards.

4. There are a couple of relatively inexpensive SSRs that don't suffer from the problem (high voltage drop) of the so-called Fotek SSR-40DD (which you won't find listed on Fotek's datasheet). One is the Power Expander from reprap.me. The other is a conventional 4-terminal SSR from a company in the USA. I haven't been able to find it recently, but I think the manufacturer begins with the letter A.

5. When calculating psu current, just because you have e.g. 5 stepper motors taking 1A each doesn't mean that they will take 5A from at the power supply. If they have 3V drop at 1A then to a first approximation the drivers will draw 3/12 * 1A each using 12V power, or 3/24 *1A each using 24V power. In practice it is a little higher than this because of losses in the drivers and extra power needed during acceleration and when pushing filament, but still much lower than 1A per motor.

---

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
A few comments:

1. For bed heaters larger than 200 x 200mm, it is much easier to use 24V power and halve the bed current than to stay with 12V and have to wire for very large currents. But a standard RAMPS doesn't give you that option. Preferably, upgrade to 24V-capable electronics. If you don't want to do that, using two separate PSUS (24V for the bed heater and 12V for everything else) with your RAMPS is a reasonable option.

2. For really large beds, mains voltage bed heaters are the way to go, switched by an inexpensive zero crossing DC-AC SSR. The Fotek ones are OK for this. But of course you need to take extra safety precautions.

3. That TI mosfet does appear to have very good specifications. But in practice, it's the PCB trace widths and other wiring that are the limiting factor, not usually the mosfet, unless you choose a particularly poor mosfet such as you find on most but not all RAMPS boards.

I looked at trying to get a 24v heatbed, but couldn't find one of the right size and specifications, and I didn't want to wait for one to ship from china.
Speaking of wiring, how can I calculate how thick a wire should be to be able to carry the required amps? Also, most mosfet leads don't look like they'd be able to handle the rated amps, let alone connect to a cable which can. What would be the correct procedure to connect a thicker cable to the leads?

Thanks,

just an off note, but for the heatbed's LED indicator, I should use a 1k resistor, right? If my understanding of ohms law is correct, then a 1k resistor should prevent the LED from fusing, although is there a possibility the LED can damaged by the heat from the bed?

... typical LED's should suffer 60degC for long enough - so check, how hot it goes

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Ok, I finally got a 50a 12v PSU, and started testing the heated bed, WITH the super thin wires it came with, although I redid the solder joints. However, surprise surprise, no spontaneous combustion. I'm guessing that the reason for that is both because the wires must be silicone insulated or the like, the insulation didn't feel quite like PVC, and because the heated bed was not drawing 30a as advertised. Measuring the resistance, it should be drawing 15a, though my cheap multimeter could be at fault. Even after leaving it running in the noontime sun, the PSU only turned on its fan for about 10s every minute, the ground wire of the bed was warm, the positive wire was cool, and the heated bed was hot.
I don't know how I can measure the temperature of the heated bed, I have yet to setup the thermistors, so maybe I should invest in a probe cooking thermometer or an oven thermometer? It wasn't hot enough to boil water, so not even close to 100 at max.

Quote
dc42
4. There are a couple of relatively inexpensive SSRs that don't suffer from the problem (high voltage drop) of the so-called Fotek SSR-40DD (which you won't find listed on Fotek's datasheet). One is the Power Expander from reprap.me. The other is a conventional 4-terminal SSR from a company in the USA. I haven't been able to find it recently, but I think the manufacturer begins with the letter A.

Auber Instruments
[www.auberins.com]

Right, so I finally hooked up my system and started testing it. 3 IRLB8743 in parallel, mounted onto a dotted PCB, using 2.5mm multistranded copper wire, gate traces hooked up to the heatbed mosfet's gate on the RAMPS board, and a 10k resistor from gate to source, mosfets placed between the heatedbed and ground.
I haven't started printing yet, nozzle problems, but so far, I ran the heated bed for half an hour, and there was barely any heating in the wires and mosfets. I think my heated bed doesn't draw as much current as its rated for, probably around 15a instead of 30a. Either way, even if the mosfets were drawing more current, the system is rated for much more.

Hi A_Designs,

Can you post a schematic diagram or a photo of the setup which you mentioned.

Regards,
Prashant.

On request, here ye go.
Super neat and totally not hand-scrawled schematic

bottom view of the mosfet PCB

mosfet PCB in its box.

Notes-
The box is metal, and the mosfet is hotglued onto a wooden plate, which is hotglued into the box. You should probably use another glue, but my mosfets run cold, as I think only 15a is going through them, althought TECHNICALLY they should be able to handle max 300a or so. The wires I used were multistrand 2.5mm, Mosfets irlb8743, ignore the scribbles about a totem pole or fuse, they weren't needed, although a fuse might be nice. The 10k keeps gate at ground voltage, and without it gate will stay on whenever you turn it on and current flows. I've only drawn 2 mosfets in paralell, but I've used 3, and for the connection at the heated bed, I sanded the silkscreen off the long contacts on the PCB, and soldered about 2cm of the wire to the bed for a solid contact. I also blew up the LED, I did use a resistor, so I'm not sure why, but that's immaterial. I unsoldered the ramps bed mosfet, and connected my mosfets to the gate hole on the ramps.
Any other questions?