FET Bed Heater Board

So I noticed that the MakerFarm Pegasus 10" ships with a Bed Heater relay board and after doing some reading/thinking found it would not be very helpful in the long run and should be replaced with something better.

The design as I have it below has more then just one use and I can see it being used for many things. It is setup to allow any input voltage (change R3 as needed) 10mA device to control a much larger power device. The best Fets to use would have very low RDSon. The Surface Mounted Fet may not have enough cooling for +20A loads and the TO220 would need a heatsink of some kind.

To facilitate this I have started work on a replacement board design that will allow some customization of the parts.
Right now its set up to use 2 different Opto types and 2 different Fet packages. Slow Transistor type Opto's, Fast gate driver Opto's. Dpak(TO263) and TO220 Fets. I chose the packages for 2 reasons.
1. There is a large number of same pinout devices on the market.
2. They are off the shelf parts that are not hard to get.

Attachment 8550Attachment 8551

My Question for you guys is as follows.
1. Should it be the same size as the existing Relay Board and if so what are the dimensions since I dont have my Pegasus yet.
2. Should it be smaller?
3. Should I make it more specialized(like remove the opto and expect whoever uses it to have there own way of driving it)?
4. From my standpoint the standard slow optos should be more then enough, 1-100kHz switching frequency is more then enough to take care of our needs and it reduces losses due to switching(as long as the slow rise/fall does not increase on/off losses). Should I get ride of the Dip8 Gate Driver and just use the Dip4 Opto Transistors. Or should I do a limited run from Oshpark and pull out my trusty O-Scope and compare the Fast vs Slow opto's?
5. Did I make some huge mistake in this design. (considering I have used both circuits at home and at work they should be fine combined, as long as you dont mix resistors/misplace the dip4)
6. Should I replace the 2 large terminal blocks with screw terminals like these http://www.digikey.com/product-detai...798-ND/2171014

Not sure if this should be in one of the larger forums, but for me its first and foremost for the Pegasus and later on it may be used for other things.

Edit
Just noticed I put C1 wrong, should go to pin8 on U2 and GND.
I tried replacing the 2 terminal blocks with those screw terminals. They are kinda big lol. Might make good heatsinks.


Update......

Attachment 8573Attachment 8574Attachment 8575

Got ride of the Pos Supply Post and just added a small jumper point for a wire.
Shank the board down to 2.77in square from 3.17in square. I can probably make it smaller but would like some feedback.....
Attachment 8555
I kinda like the screw posts, they are a little cheaper and they make good large secondary heat sinks.

I don't see why it is so complex. If I needed to do this I would use a P channel and a TL431 to drive the gate of the MOSFET. Making sure -ve was common.

Can't see a need for anything more complex than that other than a couple of resistors to limit currents.

Trying to isolate whatever your driving the load with from the higher voltages the load may have.

Why? There is no need to do that at all.

Fewer things go wrong that way. I would rather desolder a few parts from a sub board then replace a micro controller...

Good design is one that fulfils the requirements but does not overdo it. Opto isolators will be the fail point in your circuit, they always are. They do fail safe usually but without it the circuit will not fail at all for many years. You are driving a mostly fixed resisitive load so the strain on any components is well within the design spec.

Belts and braces do not make trousers stay up any better.

The only time I have seen Opto's fail is if to much current is drawn from them or to much voltage is pushed into them. We use FOD3182's at work to drive 3phase 3HP motors and when the IGBT's go they take everything between them and the FOD3182's out due to the 700V bus voltage.

This is more a generalized design then a specific one case setup, I want to give the end user some leeway in what they put on the board. I am considering removing the dip8 Gate driver opto's since I see very few cases in which you would need to switch at the very high speeds they can produce. The on/off losses with the slower 4Dip opto's is fairly small compared to the overall switching losses themselves.

I agree that the TL431 can be a simpler way in doing it (you could just drive the N Mosfet directly to) but I have preferences towards Opto's due to the protection they provide.

edit..
Added 10K pulldown on the fet gate to actually allow it to switch on and off....
Attachment 8556

For sure if you want an opto then the 4 pin would surely suffice. 70p for the FOD3182 versus 3p for a generic 4 pin represents a huge saving.

I am also not surprised they don't fail at that price. :)

In SMTPs I would guess that the opto is the third most common failure point after the capacitors and the chopper transistor. I keep maybe 20 or so different types in my component drawers, I wouldn't do that if they didn't fail.

We use 4pin opto's for things that don't need +2A to run. I considered the 8pins due to the high speeds they can achieve but I don't really see a need for them, also many have Under Voltage Lock Out built in so they may cause more headaches then help. In the 3 years I have been testing/fixing inverters I have never once replaced a 4pin style Opto. From experience I can say that generally things don't fail unless you use them outside there safe operating areas. I have to keep reminding the engineers where I work to make sure the electrics stay below ratings at all expected temperatures. Ive had to request changes to designs due to oversights like using a linear regulator inside an enclosed inverter that can have internal ambient over 60C. It was melting 200C rated plastics and destroying the 2 caps next to it.

I love having a job where I get paid to find ways to break things :)

Removing the 8pin gate driver style opto. (Removed 8Pin Gate Driver Opto, 1 Capacitor and 1 Resistor.)
Attachment 8557

edit.
I can probably shrink the board down more, not really sure yet. May play with the screw terminals, maybe make it possible to use the plastic blocks or the all metal screw terminals.

Shrank it again and added a zener incase your VCC is higher then your VGS Max. I also added a small solder jumper so you can just jumper the Z1 if you dont need it.

I think if I make it much smaller I may compromise its ability to cool. I may add holes on each side of the Dpak Fet so that a heatsink can be screwed down, or lay the TO220 out so that it can be screwed down with a heatsink onto the board.

Attachment 8559

Moved Q1 around to allow it to be bent down and secured to the PCB.

Attachment 8572

Added changes to first page.

Feels like im talking to myself sometimes...

Quote:

---

Originally Posted by Donziboy2

Feels like im talking to myself sometimes...

---

Well, what you're basically doing is designing and building your own solid state relay. Most people would likely just purchase a commercially available SSR like discussed in other threads and go that route.

I don't have a link to provide as evidence, but this also isn't the first time someone has designed a MOSFET replacement for the mechanical relay board.

EDIT: Here's a post that introduces one attempt at a MOSFET replacement board: http://3dprintboard.com/showthread.p...ll=1#post38740

Quote:

---

Originally Posted by printbus

Well, what you're basically doing is designing and building your own solid state relay. Most people would likely just purchase a commercially available SSR like discussed in other threads and go that route.

I don't have a link to provide as evidence, but this also isn't the first time someone has designed a MOSFET replacement for the mechanical relay board.

EDIT: Here's a post that introduces one attempt at a MOSFET replacement board: http://3dprintboard.com/showthread.p...ll=1#post38740

---

I have noticed that TechMasterJoe vanished a while back. I was looking at his post on ATX boards but he has not been on the forums in a while and all the images where not visible.

On the relay board, I recall maybe a couple of other threads where, in addition to providing a MOSFET switch, the person was adding other useful stuff onto their custom board. For example, adding an ABL servo or a Raspberry Pi Octoprint controller can leave you starved for 5V, so people were adding 5V regulators on their boards. I remember suggesting that providing 12V distribution and a few lightweight MOSFETs for things like fans or lights would be a good idea.

Quote:

---

Originally Posted by printbus

On the relay board, I recall maybe a couple of other threads where, in addition to providing a MOSFET switch, the person was adding other useful stuff onto their custom board. For example, adding an ABL servo or a Raspberry Pi Octoprint controller can leave you starved for 5V, so people were adding 5V regulators on their boards. I remember suggesting that providing 12V distribution and a few lightweight MOSFETs for things like fans or lights would be a good idea.

---

Im actually thinking of following in TechMasterJoe footsteps and using an ATX power supply. For me atm its hard to guess what will and will not be useful, I just got the check today that will be going toward my 10" Pegasus, I have spent the last week or so just looking at things I plan to change from the start. The Mosfet Driver seems to be the easiest to tackle and I can see it being used for other things, I just ordered 3 boards and parts today. p.s. I can see some Chinese guy making and selling them for $3 in a few weeks time lol.

I will say the ATX power supply's nowadays can give more then 1A in standby on the 5V line, this can power all the small stuff until the controller decides it needs more power and turns the supply on. I also looked at reworking the Ramps 1.4, but thats a long battle for later.

Got 2 boards assembled, boy its hard to tell just how small those resistor pads are until your trying to solder them. The worst ones are those attached to large planes, good thing I have a good soldering iron. Used the iron at work to do the large stuff, industrial irons are awesome lol.

http://i.imgur.com/tXG0qR9.jpg?1

Can only comment on the size of the mechanical relay. When I got a 12" piece of glass cut I had to get them to trim a notch in it to go around the relay. It sticks into the plane a full sized built plate needs by about an 1/16 to an 1/8th of an inch. I've since removed the relay and just jumpered the connection and will be using the SSR recommended in other threads for now.

Did some testing at work, the D2Pak case gets to around 85C running at 0.7ohm load at 95% Duty Cycle (500Hz, 12V, 17Apk, 16.5A RMS).
I took the TO220 board and added a large heatsink , it had no problems at the same levels as the D2Pak and seemed fairly stable at around 70C case temp.

I need to increase the size of the PCB to add space for a larger HS/larger copper pad for the D2PAK.
I found reducing the switching frequency also lowered temperatures fairly nicely, about a 10-12C drop going from 500 to 250Hz.

edit..
Got another 10C drop going from 250 to 100Hz :)
At lower frequency's the two packages have close to the same case temperature, both are sitting happy around 50-55C.

Slowly getting things setup. V2 of the board will be larger but I want to see how V1 holds up under load first.
V2 will be the same size as the stock relay on the right.
http://i.imgur.com/4Gj4YNH.jpg

Got my printing doing some test prints today. On my third print of the day using PLA(60C bed) im about 2 hours in and the SMD board is staying below 90F. I have not changed from the stock relay sw configuration yet but it seems I over engineered my torture tests at work.

Got PID control setup on the Heated Bed, the switching frequency is much lower then I had expected (only 7.633Hz). Other then some hot wires(need to increase # of wires from ATX supply to bed) everything is nice and cool.

http://i.imgur.com/SNbmNxN.jpg