Does this Schematic look right? (Firing Electronic Matches)

[funman1]

Awesome... More to think about

Well there are others driving these right off the Srduino with no issues, so that's why I thought they would be perfect?
They have been building secquencers where the Arduino just fires after a specified time on each channel, and people have been using these wonderfully.

So now I'm back to the drawing board


On the quick software I whipped up I just used Visual Basic Express 2010
Free and works AWESOME for quick programing on Windows.
http://www.microsoft.com/express/Downloads/#2010-Visual-Basic

 

[NickS]

I am not familiar with "Srduino".

You don't have to scrap your plan just know your circuits limitations. I think you should stick with the FET you chose before and just take care to ensure you have some resistance(at least one ohm worth of wire) between the possible short and your FET.

 

[funman1]

Sorry I meant Arduino (NOT Sdruino)

Well I think I'm going to go with the larger ones just because bigger is better right?
Also the traces for outbound power are rated for about 2 amps, so this should keep it in check too?

So here's where I'm at now

Here is an update with the PCB.
It's a 4 copper layered board (Bottom Traces, Ground Plane (Split), Power Plane (Split), and then top traces) Along with 2 solder masks (on each side) and then silkscreen so 7 layers total.

E1-3 = 2342-12-000 Relays
V1-2 = Standard 5Volt Regulators
D1 = Run of the mil 1/2 Watt Diode
Q1-10 = IRFZ44N Firing Power Mosfets
Q11-Q12 = IRF9910TRPBF Power Mosfets
X1-10 = 2N3906 Transistors
Resistors Marked as shown
Arm Relay = 275-248 PC Relay
Connectors as marked.
Arduino as marked.


Components

Botom Layer

Top Layer

Ground Plane

Power Plane

Top & Bottom Layer Combined on view

 

[NickS]

Not the same at all. Having a small resistance accumulate over a long stretch of cable will not cause any significant problems. Impeding the current through a one inch strecth of thin copper on a PCB could vaporize traces on an expensive board.

When I am doing high current on a PCB there are a couple things that take priority over the rest of the board layout.
(1)Thick and Wide Copper. Not a standard 1oz copper. Especially if you don't have room to widen traces try to get board fabbed with thicker material (3 or 4oz)

(2) Short, Direct Routing. Layout can be tricky but you want your high current lines to be as direct as possible. You have done a pretty good job of that, but you need to get rid of the 90 deg bends. They cause dead spots(in the corners) where current does not flow. Thus every 90Deg bend is a higher impedance bottleneck. Try to use 45deg bends whenever possible.

Now I know you are confident with the Arduino driving the FET's because you have seen it. But if your Arduino is working at 3.3V(which I suspect from previous posts) then the datasheet for the IRFZ44N guarantees that you will not even be turning some of the devices on. They call out a Vgs min of 2V(these will turn on) and a Vgs max of 4V! That is the maker of the FET telling you the designer that if you try to drive this FET with less than 4V you can expect failure. For all the work you have put in I would really challenge you not to cut this corner. Not to mention that the gate charge and capacitance both rocket up for the IRFZ44n this is a big deal since you drive direct from your controller. So here is the final breakdown with the new FET that you like, your controller will have to work harder to turn it on it will turn on slower(which means more heat) and due to the spec you may never even turn it on. You say more is better but for the way you are driving it I promise you will not get more out of this FET than you will with the first one you chose.

 

[funman1]

They call out a Vgs min of 2V(these will turn on) and a Vgs max of 4V! That is the maker of the FET telling you the designer that if you try to drive this FET with less than 4V you can expect failure.

1) I have no control over the the copper thickness (This is the special from ExpressPCB) Hence the size limitation. But the load is so fast, just a few ms's I can't help but think it won't matter too much.

2) I will try and convert them to 45 angles if I can. I don't understand the "dead spots" When you bend a wire, you don't change it enough to matter, so why would it matter on a board? I'm very confused by that?

3) I admit I do not understand how to read schematics correctly, but I thought "Max" was the absolute highest allowable before breakdown (Destruction)? So it was my very simple understanding that if you were between the min and the max that's where the unit would operate reliably and safely? I'm very confused by this too....

I'm not questioning you're answers, but trying to understand them, because they seem to go against what little I have learned in the past.

 

[NickS]

I am not near my textbook right now to give you a good quote on the 90deg thing. But if you think about a trace compared to a wire your wire is usually cylindrical so you can make a tight bend but not a square bend. If the wire did square then you have probably broken it on the outside edge. The PCB trace is very thin and flat. It was described to me like water flow. If water was moving down that path most of the water would flow through the bend in a radius fashion but in the corner the water would spin in reverse and be essentially stagnant.

Express PCB gives this reason:
" avoid sharp right angle turns. The problem here is that in the board manufacturing process, the outside corner can be etched a little more narrow. The solution is to use two 45 degree bends with a short leg in between."

I will post up again when I dig up my textbook on layout.

Questioning what your told is nothing to apologize for(its good operating procedure). On the IRFZ44N FET lets look at a couple things.
First the absolute maximum rating for Vgs(page 1). Vgs Gate-to-Source Voltage +/-20V this is the safety rating. If you exceed a voltage of magnitude 20 between these two parts then you will start to break down the insulation between Gate and Source.

Next lets look at electrical characteristics. These are different from the max ratings because this is concerning performance not physical destruction. So page 2 4th line into the table tells you where the device will turn on Vgs(th) threshold. This is not a single value because the parts can range slightly from batch to batch based on many factors like the effective doping and process variables. So they give you a range from 2V to 4V. This means they have produced some of these devices that turn on as low as 2V but they have also tested devices that don't tun on until 4V. As a designer you would interpret this as
(1) If I want to ensure that the device is off I must hold Vgs below 2V (usually 0V)

(2) If I want to ensure that this device is on then I hold Vgs to at very least 4V. Realistically to ensure to are saturating the FET you would look at the figure 1 on page 3 and say to ensure that my device can handle 10A I must drive the gate to at least 4.5V. The fact that you want to operate in a range that is not even shown on the plot is a bad sign. With the gate capacitance just think of that as more current your controller has to source to charge up before the voltage on the gate can rise to the desired level. It becomes a serious issue with power mosfets which is why they are usually buffered with a FET gate driver that can typically source 2A of current

Sorry if I am coming across as overly critical. I want you to have success on the first try.
-Nick-

 

[funman1]

Ahhh ok I'm begining to understand the spec sheets now!
That little tid bit of information makes a huge difference in how I read them and that makes perfect sense!

So using my new found logic How does this mosfet look.
It looks like it's rated for 55 Amps AND max gate voltage is 2.0 meaning 3.3 should be PLENTY to operate this gate correctly?
I think I'm begining to understand the graphs now too! Heheh

http://www.jameco.com/webapp/wcs/st...amecoall&ddkey=http:StoreCatalogDrillDownView

OK I'm working on getting all (as many as I can) 90 deg bends out of the circut design for PCB.

Will update the pics when I get that done...

Let me know what you think of the above little guy.
Thanks again so much...
~Steve~

 

[NickS]

Much better.

 

[funman1]

OK the pictures of the board layout above have been updated.
Does that look better with very few 90 bends?

And, is the 1K resistor enough to protect the gate on the transistors (2N3906) from the 12V?
Gate can only handle 5V max?

Also I just ordered the Mosfets and the transistors
I'm getting excited again hahahah.....

 

[funman1]

Any idea what these connectors are called?
I would like to use this type of connector and pins to connect to the board but can't find what they are called, and thus can't purchase them?
These fit perfecty with the pin headers I want on the board.

Any ideas? I thought there was some type of tool for the pins too, but I can just solder them for the connector if needed (Dont need the tool but would be nice if the price is right)

 

[NickS]

1) LAYOUT ROUTING: Better yes. But if your layout tool has a 45deg bend mode(which all layout tools should) then that is the preferred routing method. Not something you need to redo but for next time definitely try to use it. It is not just about aesthetics, any time you have an acute angle between traces it is high risk area for bad etching. This is hard to avoid if you are going point to point with traces but if all your traces are vert, horiz or 45deg between then it becomes less of an issue. For example your bottom layer on the upper right hand section you have two acute angles between traces. In this case(T junctions) you forgo the bend and bring the branch off at 90deg(hopefully it is a low current trace). The other thing you could do is shown in the attached pic.

2) LAYOUT GROUND PLANE: I am looking at you ground plane layer and wondering with it being a split plane how are you attaching the ground for the FET's to the battery. I understand having a split ground plane but they need to be tied together someplace usually through a FB or choke. Did you do that and I am just not seeing it? Also does your layout tool have some sort of schematic/Layout verification tool to ensure that all nets are properly routed(DRC or LVS)?

3) CONNECTOR: As for that connector it looks like a run of the mill 0.1" pitch rectangular housing. The good news is that there are thousands of mating parts but the bad news is I have never seen a reasonably priced pair of crimper's. If you can verify the pin spacing(pitch) then I will link some options in my next post.

 

[NickS]

4) BIAS RESISTOR: With a 1k in place the max Vbe should be -0.9V and a base current of less than 11mA. If you want to run it lower that is just fine you could use a 10k and bring your Vbe down to -0.8V with Ib ~ 1mA. I like to turn the device on hard so that its internal power dissipation(Vce-on) goes down.

 

[funman1]

1) LAYOUT ROUTING: Nope the software does not have that option What do you expect for free I guess? But glad it looks better (Technically better, not Aesthetically)


2) LAYOUT GROUND PLANE: Yes the software does support linking to schematic, though I did not use that feature. All this was hand done from my head.

The Ground Plane: OK, since someone could get killed if this unit malfunctions, what I did was seperated the gorund plane for the Firing mosfets (IE: they have no complete circut (Ground) UNTIL the ARM RELAY is engaged. So if you look on the top trace layer you see the large trace from the Arm Relay going to the Ground Plane with all the Firing Mosfets. So once the Relay Engages then and only then, could the system ever fire anything. I'm also contemplating putting the relay power through the (Enable/Disable Pod) Switch, SO the switch would have to be set to Enable, the unit would have to have an ARM Command to engage the relay and then it could physicaly fire, This extra point would help if the relay ever failed. Does that make any sense?

3) CONNECTOR: Yep the standard .1 spacing and I was able to find a crimper for $20
I looked for servo connectors and found the 1x3 connectors but that's it and still don't have the "name" for them...

 

[NickS]

1) Bummer

2) I would recommend using it. It has saved my butt on several occasions and its just good practice.

3) ok on the ground plane

4) The search description is connector > rectangular housing. It doesn't hold a special name because there are probably 6 large mfr's that make the same connector. Now if you already found crimpers then you are locked in to using the pins that go with that crimper which also means using the housing that mates with those pins. I usually search the connector first and then find the crimpers. The crimpers and pins are usually proprietary to the MFR which is why you try to stay with one mfr so you don't have to buy everyones crimpers.
Here is a link to get you started just select
1 row
0.1" pitch
3 pos
then check the "in stock" box and click on the "apply filters" button

 

[funman1]

Connectors, yeah I was able to find some 3's like I said, but I need a 1x10 and that search only comes back with 2x5 and no 1x10

 

[NickS]

Then you are over constraining your search. I just used the previous link
selected 10 pos, 1 row, .1" pitch and was given 13 instock options for 1x10 housings that will work.

One other thing to note is that once you decide on a housing and select it you will find a section near the bottom of the page called "mating products". That should link you to acceptable pins for use with that specific housing.

 

[funman1]

Ahh perfect thank you...
That's exactly what I needed.

I just picked up some rainbow ribon cable at a local suppiler too

Also here is a link of what the system will be firing 95% of the time (J-Tek)
http://www.electricmatch.com/product_us.html

 

[funman1]

OK Mosfets and Transistors came in today so it was breadboard time!
However I just discovered a small flaw, I think I now need to add a diode?

So the diagram below (I know my skills suck at drawing them, Never learned how to properly) Anyways...

So if an Ematch is connected; it works perfectly as designed and conceived.....
HOWEVER, if the ematch is removed, the Red Leds lights up EVEN though it's not actually firing..

Which I belive this is because the Green Led is leaking gnd(Neg) and flowing backwards to the gate of the Transistor, and allowing it to open...
When the Ematch is there it's enough Pos to push it down and keep it from opening the gate.

So, add a bigger resistor to the gate on the transistor?,
OR put somekind of diode between the green led and the gate?,
OR put somekind of pull up resistor on the gate?

So exciting it's getting close
Thanks..

 

[NickS]

I was not aware that you were concerned with the LED state when the Ematch is out of circuit. Do you want both LED's off when the e-match is out?

Well that is not such an easy thing to rectify with your configuration. Sure diode blocking the green side would correct that problem but then the green would never turn on. Changing the bias resistor or adding a pull up does not fix the issue either. Let me think about this for a while and I will get back.

 

[funman1]

Yeah my thought was, off for both means no continuity at all (Nothing connected)
Green meant the ematch was connected correctly.
Red meant the system was actively firing the channel.

So there's no way to prevent neg from flowing backwards though? I thought the diode would allow neg coming from the mosfet though, but could prevent it from going the other way back to the mosfet?