Does this Schematic look right? (Firing Electronic Matches)

[NickS]

That is pretty close. Did the site you linked mention why? With LA batteries when you discharge you are basically coating the lead plates with sulfur. If you deplete the battery too far then you can be causing irreversible damage to the plates. I try to set my absolute lower limit at 11.9V (for a 12V batt) but I try to recharge before I reach 12V. A fully charged LA battery sits well above that when fully charged. If it was disposable then sure you drain all you can out of it. But since you want it to keep working for you you must baby it a little for good lifespan.

Is your resistor divider feeding into an ADC pin?

 

[funman1]

Yes the resistor junction feeds into an analog pin around ~4 volts when the system is at 12.2 volts

 

[funman1]

Hahah ok just before I was ready to send this to manf.
I had one of my crew bring up a good point. With the current design, if a Mosfet were to fail, I would not know this without testing each channel independently.

The Red fire led is tied to the controller, so there could be a time with the mosfet failed closed allowing power to flow though yet the LED would not be red.

Is there any way to redo the Opamps to read directly from the gound output of the mosfet, so if a mosfet is spewing power the light would be red.

I'm so close and dying to see it in action, yet I want to only do this once, and would much rather wait and redo the board over and over to make it as SAFE as possible.

I think if we do this I will need a resistor for each opamp again too.

Let me know how I can make this better
Thanks
~Steve~

 

[NickS]

Sure. You could go back to the way it was before you added the op amps. This always signals the red LED when current is flowing through the FET.

I will spin my brain on another way and get back to you later.

 

[funman1]

But it didn't work the way it was before because of the back feed issue..
Which was why I changed to op-amps.
If there was nothing connected the Red fire LED was on with the old Schematic

 

[NickS]

Yes I remember that, I was just pointing out that it did work in the other respect... Hold that thought....
ok I am back
If you look back you will find that I submitted a different solution which you bypassed in your excitement to use op amps. I just re-simulated for the case you have mentioned and it works for all states we have discussed.

However it is growing increasingly difficult to keep them all straight now that there are three variables so I made a truth table that describes the behavior of my circuit. The situation we did not discuss is when the ematch is out and the uC is firing. For that case the red led will light when S1 is pushed even if the ematch is out of circuit. Personally I see no reason to change this since the lack of green OK prior to firing should have indicated that the channel was not ready to go.

Enjoy

 

[funman1]

There's just too many Resistors, and 3 transistors/mosfets
I just dont have the room on the board for that....
This would be multiplied by 10 channels so just out of space...

Not quite understanding the switches. on the panel thats already made there are no switches...

The truth table/description I would like is:

Green led = ONLY lit when there is continuity (IE ematch is connected) Green can even be on when firing, this is fine and will make a dark orange so thats ok.
Red led = ONLY lit when mosfet is allowing gnd to pass though it (either firing or failed)

So yes I would like the led to turn red on even if there is no continuity when the mosfet is firing.
This way when the unit is turned on and armed I can see if there are any faling channels without connecting ematch AND I can test fire the system with no ematch and its working if the led turns red durring a fire command...

I hope that makes sense.. Not trying to say the idea wont work, I just lack the room for all the parts with that schematic.

So the lights are really easy
Green = continuity (yes / no) On / Off
Red = firing / leaking mosfet (yes no) On / Off

~Steve~

 

[NickS]

If you read the descriptors on the left of the truth table you will see the switches are just the simple way that I simulate your circuit. ie where I have S1 you have the uC but for my sim it is easier to use a switch. Where I have S2 is just to simulate the ematch being in circuit or out of circuit and S3 simulates a MOSFET failure by shorting out the drain to source. Under those assumptions and what you said above I think the truth table I presented holds up.

So that leaves us with only one issue remaining...space. Now for all through hole parts I agree with you, but you can easily fit 4 surface mount(0603 size) resistors in the space of one through hole resistor. And most gen purpose BJT's are available in a tiny SOT-23 package. The other advantage to surface mount is that you have twice the real estate because you can use both sides of the board without interference. Some SM parts are hard to put down but the ones you would need are very simple.

So if you are willing I would recommend you try to arrange the parts for one channel worth using the SM footprints then drag it to your board and see if you still think it can't fit.

 

[NickS]

here is a rough demonstration of the amount of space afore mentioned parts would take up

 

[funman1]

Hmm I have a space of 28 x 22 mm for this.
Let me see what I can do.. Though I don't think my solder skills are upto this.
This is extremely close together, Perhaps if I had space to solder but a tiny blob would ruin it. and I would have to do this 10x
Not looking optimisitc

 

[NickS]

My part placement was just a rough cluster to show size. I didn't actually place them how they would be for wiring.

Don't get discouraged. These SM parts actually go down quite easy. For the resistors you pre-tin one pad then using tweezers you place the resistor down while keeping that pre-tinned pad hot. Then you just tin the other side and your practically done. Same thing with the sot-23. After all parts are down you apply flux to all pads and then touch them up with a short tap of the iron tip. You will find it a bit peculiar at first but after you have put a few down it will be no big deal.

Also do you mind breadboarding the design to verify it. I ran it through spice but that can sometimes be mis-leading.

 

[funman1]

I would have to order these new parts in order to breadboard.
And I havn't found a good way to bread board with SMT yet.

 

[NickS]

Sorry I meant leaded breadboarding(the SM parts are functionally identical). I thought you might already have all the leaded parts needed.

 

[funman1]

I have the 3906 but not the 3904's

 

[funman1]

OK getting ready to order all the SMT stuff.
But a few questions plague me

1) Do I really need R1 AND R8, It would seem to me you could just use 1 Resistor here and don't really need two?

2) I dont understand Q3 at all, it looks like when engaged it's just a short from + to Gnd with some resistors in between?

3) Why is R7 1M? the Led's are tied together because it's a single package....

Just trying to understand the reasons on how this works...
I will try and post my list of parts I plan to order tomorrow.

 

[(*steve*)]

1) Yes, you need both of them. They ensure that the base currents are shared and that both transistors turn on.

2) Sorry, I don't understand that either (I haven't been following this thread in any detail)

3) That's 1 milliohm, not 1 megohm. (I don't really understand all the very small value resistors, but see my answer to Q2)

Probably not much help, sorry.

 

[NickS]

Ok lets see here Steve already gave the answer for number 1.

2) You are exactly right on Q3 it does short out Q1 we had to do this because otherwise we were unable to turn off the red LED in the event of no ematch in place. This basically draws the current that would go to the red LED thus starving it and keeping it off for the first state.

3a) Sorry about the 1m ohm resistors I usually clean those out before posting a schematic. The reason I use them is because multisim has a very hard time converging with active parts tied hard to ground or other active parts. So I litter my sims with 1m ohm R's which add no significant change to the functionality of the circuit and keep multisim happy. You can simply omit R7, R6, R5, R9. R3 is not needed for your solution either.

3b) Swap LED2 and R2 places on the schematic and that will support your common cathode LED while serving the same purpose. Remember that I have LED1 as red and LED2 as green.

See my cleaned up schematic.

btw I breadboarded it and it works as specified.

 

[funman1]

ok wow, very cleaned up thanks!
2 more questions..

1) Having the greens shorted out across a bank of 10 when no ematches are put in, how will this effect battery life? Seems it would be really bad on it?

2) R2 should be on the cathode for the LED's right? not on the Anode for LED2?

Also I have begun to respin the board and I think I can make it all fit!
Here is what I plan to order for this.. The 2N3904 was for breadboard testing.

Let me know what you think..

 

[(*steve*)]

1) you're looking at about 30mA per device. That 1K resistors (R2, R4, R8) limit the current in various paths. (10mA if no ematch in circuit)

2) it doesn't matter whether you place the resistor in series with the anode or cathode. It just has to be in series. You could swap it over if you wanted to.

3) the 1K resistors could dissipate up to 0.144 Watts. You should pick a larger SMD resistor that is rated for higher power. R2, R4, and R8 are affected. One easy option is to get twice as many 500 ohm resistors and place them in series since each will dissipate half as much heat.

(I wonder if R1 and R8 couldn't be a higher value? 10K anyone?)

 

[funman1]

1) per channel (300ma), or per full blown unit of 10 (30ma)?

2) I think I just figured out why it's there. Instead of having it on the Cathode shared, we are using two resistors for each red and green on Anode. Got it......

3) I will see what I can find for R4 and R8,
R2, will be a 1/4W right on the green Led Anode .