Hi Jonathan, I rebuilt your circuit to drive my heatbed. I have a GT2560 board. This board as a LED for the heatbed output.

Now, when the bed is switched off and I connect the bed power source, the board LED will switch on and the bed starts heating up (-> the board cannot switch it off), as if there is some voltage at X1........

Any guess what goes wront? I am extremely sure that I wired it correctly. However, I used a 30N06L instead of the one you proposed...

if you allow me some days I can post mine. It is not much more

complicated than the original design. It just needs an additional cheap

optocoupler.

Update 2 (where did the edit button go?)

Actually I found that it isn't that difficult to make a good version using an optocoupler. The new design has 2 components more than before, but is much better due to the complete isolation of the two power supplies.

Soldering is a little bit trickier, but really just a bit. If you can do V2, you can also do V3...

BOM consists of 3 "building blocks" depending on your voltages, but that should be manageable.

It'll take me some time to make the 3D files for V3, so stay tuned ;)

Update:

I put some thought into that upgrade. I came to the conclusion that I won't do it because the design would become too cluttered and complex for what this is meant to be - a simple and easy to build hack for the UMO.

You could either power the RAMPS1.4 using a 19-24V power supply (in that case the hack should work - note that I don't have a RAMPS so I can't test it. But since the schematics say the boards are identical in that regard...) or buy a "proper" heatbed relay board instead.

Hi Ariel,

The RAMPS 1.4 actually has the same heater outputs as the Ultimaker 1.5.7 board (for which my hack is made). So, connection is the same as described in the documentation ("19.5V" -> "+" and "hot bed signal" -> "-").

However, my hack won't work with 12V, only with 19-24V power supplies (on the RAMPS side) because I intentionally used a simple way to set the switching mosfet's gate input voltage using the two resistors R2 and R3. Of course there are better solutions, but this is the easiest one to reproduce while producing good results at the same time.

Generally my solution works so well because it's tailored to the specific voltages on the UMO (19V UMO supply, 12-24V hotbed supply). Many other solutions such as SSRs offer much more flexibility, but sacrifice performance in the process because they are more complex.

Now, for a 12V RAMPS, this won't work as it is now. One could mess with R2 and R3 to produce the correct gate voltage for 12V input, but since these resistors are also needed to ensure proper switching of the mosfet I won't recommend anything without trying first.

The reason why I'm calling this a "hack" and not a proper solution is - while it certainly works very well - it is not a "clean" design because it doesn't separate the two power supplies properly from each other.

I could make an upgrade with a proper input and GND separation, but that requires a few more components. It seems like people would still want it, so I can look into making that upgrade when I have time..

Hi I was following Louis Cypher comment about adapting your idea to work with Ramps 1.4.

Since my knowledge in electronics is limit It will be great if you can do this adaptations for the use Ramps users.

Thanks in advance,

Ariel

They use a 25A MosFET. Which is... well... I used a 260A MosFET which may be a bit overkill - but the point is to get nearly 100% efficiency (no losses in the MosFET, no heat generated).

They could've taken a better mosfet and left the heatsink away instead. But - as usual - they just take the cheapest components available, or maybe just didn't think far enough to make a good product...

http://www.aliexpress.com/store/product/3D-Printer-Parts-General-Add-on-Heated-Bed-Power-Expansion-Module-High-Power-Module-for-3D/2173157_32658370869.html

found the original. Its for something else.

I found a page maybe it is useful for somebody

https://developer.mbed.org/users/4180_1/notebook/relays1/

@Marius

There are no schematics so there is no telling what they put together there...

The fact that they need a heatsink means they either don't know what they're doing, or they chose a bad MosFET. Because a good MosFET does not get hot and does not need a heatsink.

Anyways, it looks like they're using an optocoupler to drive the FET which means it's probably isolated (probably! There is no telling without knowing the schematics...) which is better than my solution. Efficiency is probably worse - why else would they need a heatsink... And I don't see any ESD diodes (which is why that chinese stuff usually doesn't work reliably for a long time).

I ended up just using a DC-DC 3-32V Control 5V-110V 24-36V SSR. $15AU on ebay including Heatsink and thermal paste.

http://r.ebay.com/fbAXTN

I put a fan on which i think was over kill as after a 2 hour print it was cool to touch. though cool things are reliable things.

Turned the 24v power supply to max of 28.5V and the SSR takes .3 so still getting 28.2 to bed. The bed gets to 170deg and 100deg in near nothing.

That is unless you have a Ultimaker where this rocks. I should have read more before making one. If you in Melbourne Aus I have one if you want it.

Is this something like what you have designed?

I'm lazy.. but if you tell me that yours is better i will do yours.

Thank you

http://www.aliexpress.com/store/product/Free-shipping-New-3D-printer-hot-bed-Power-expansion-board-Heatbed-power-module-MOS-tube-high/2191056_32668453447.html?spm=2114.12010108.0.131.DNjUmf

Solid state relays suck. They are expensive and have a very high on-state resistance compared to my simple and cheap mosfet. They do have the advantage of being well protected and universally usable (if they're good quality).

As for the switching question:

The answer is simple - it's cheaper and easier to switch the low-side. An N-channel mosfet (which is the simple and cheap part with high performance) can only be used as a low-side switch. If you wanted to switch the high-side, you would need a P-channel mosfet. These are usually more expensive, have higher on-resistances and it takes a bit more thinking to be able to use them correctly (usually you would need an N-channel mosfet to switch that P-channel mosfet).

Nowadays there are lots of good and relatively cheap P-channel mosfets. But that hasn't always been the case. If you look at some reprap designs or other stuff, you can see that people are sometimes still using bipolar tranistors which is stone-age technology. Sometimes, new technologies take a long time to actually be used by the average engineer / hobbyist...

Thank you for the reply.

How come you didn't just use a solid state relay?

Also on a side note. Why would you switch the ground instead of the active side. I have seen this a lot lately in things such a my daughters electric car. Everything is ground switched and a time is perminatly connected.

Update: I corrected the schematics as per my comment below.

Hi Troy,

If you're using anything else than the UMO electronics, you have to make sure that your board is built the same way as the UMO.

The UMO hack is really just that - a hack for the UMO. It's not a good design, it's not flexible, not universal. It only works for output stages that switch the negative line, not the positive line (It's incorrectly named in my schematics, "Bed Signal" should be "V+ UMO" and "GND" should be "Bed Signal").

If you have a lower control signal voltage, then you have to adjust the voltage divider R2 and R3 to be ~10V again. Also, the LED's dropping resistor (R1) should be adjusted so that it's not too dim or bright.

For 28.8V, you would also need to use higher-voltage ESD protection diodes (D1 and D2), since the current ones are designed for 24V (or lower) operation.

Does it matter what the input and output voltages are for this unit?

As my current output from the GT2560 is 12v and want to run the heat bed at 28.8v. It's a 24v power supply and bed but at power supply max of 28.8 is gets to 160deg and heats to 100 in a minute.

Thanks.

You can't use this for anything else than what the description says. You can't use the same PSU because the mosfet works by switching the GND side of the wire, thus needs to be able to sever the GND connection from one PSU to the other.

You could just remove the connector and solder the wires to the board directly. That "should" get rid of the excessive heating.

Is it really necessary to use different power supplies? Using a Ramps board and my PSU can provide enough power for the bed. I'd like to make this because the connector gets too hot.

Just as a comparison:

A 0.4mm diameter wire (corresponds more or less to the thinnest THT "legs") has a copper cross section of 0.5 mm2.

That corresponds to a 14mm wide copper trace on a standard 35um PCB, and we're talking about the thinnest wires on THT parts. The power components in my board use 2x 0.7mm (or maybe they're even thicker) wires as traces. This is much more than could be achieved even with thicker copper cladding on standard PCBs.

Interesting technique. For some people it might be easier to use the diode laser instead of a traditional UV light & acid bath. Can't help you with the result, I've never made PCBs this way. I suppose you either put it into the acid too long, or your foil suffered a bit from the laser cutting. With UV & bath, it looks a bit better. But you can't get very good results this way.

The achievable resolution will never be good enough for fine-pitch SMD boards. But then again, these need to be multilayer in almost every case anyways, so don't bother...

What I like about my 3D-printed circuit board is the fact that you can put in really heavy duty "traces". One of those 0.7mm diameter diode legs has a huge copper cross section compared to a 35um PCB trace. So, for those crude, but heavy duty designs, this method is really cool. (pun intended)

Hi Markus,

Using an optocoupler is exactly what you need to do in your case. My design was always meant to be a fix for Ultimakers - not a universal solution for connecting heated beds.

The key of my design is to keep things simple, which is why I didn't isolate the power domains.

You're welcome to make a fork and extend the design.

A hint on optocouplers and mosfets: Optocouplers are current amplifiers, they work best when there is some minimal amount of current flowing. A mosfet gate however draws almost no current at all. You might need to add a small load (maybe 1 mA? - see optocoupler datasheet) to the optocoupler's output in order to make everything work reliably.

Best regards,

Jonny

Thanks for the great idea. I built this setup and found out that I should have read your instruction more thorough. At least with a RAMPS board the heatbed is on all the time. The reason is that the the MOSFET on the RAMPS switches the lowside. If input and output use the same ground connection the voltage between the highside on the RAMPS connection opens the gate even in case the MOSFET on RAMPS switches off.

My suggestion (tested) is to use a cheap optocoupler such as VISHAY 4N35 to separate grounds. You can use the highside (connected to PIN1) of the input as input for PIN 5 (input of the gate) and connect PIN 6 to the gate (of course needs a resistor to limit current 1k did the job for me). PIN2 gets connected to ground via a 4k7 resistor or the like. Maybe you want to update your design. In case not I can make a branch.

Best,

Markus

Hello Jonny. Very nice Work. I was planning on using a couple of these in an UM2 clone Im Building. I have a rather beefy 12v PSU at 45amp and was planning to use 12v to Ramps, use up-amp to get 24v and use 3 of these converters to control Heated be and 2x nozzles.

... untill I read your final warning about having to use a seperate PSU for this.

Does it mean I can use the same PSU for all the parts? Asking to make sure, as a friend of mine firmly believed I could, but I understand it otherwise, so I'd rather ask to be safe and try and be sorry.

Hope you are still active here :)

Regards,

Morten - Denmark

I believe that there will be a printer capable of printing and laying down conductive material quite soon ;)