Updating ematch ignition circuit for altimeter

[profmason]

Here is my current thought on 1S altimeter ignition using 3.3V TTL logic. VBat is a 1S Lipo which ranges from 4.2V-3.2V.

When the low impedance FIRE1 pin is pulled high, the NPN (100V 5A) transistor allows current to flow between VBat and GND. The 1K resistor on the base of Q1 provides ~3.3ma which given the gain of the Darlington will allow ~3.3A of current to flow through the ematch at ~(VBat - .3V) so should provide plenty of kick for the ematch (The ones I use work fine at 2V, 2.5A)

When there is continuity across Ematch1, the opto is powered through the 1K resistor. The opto drops about 1.3V, so we have ~2mA through the ematch all the time. When the opto turns on, current can flow from VBat through the 1K, opto and schottky diode to the CONT1 digital input which has high impedance(~1MOhm) The 1K is not really necessary, just there in case some bounce on startup sets the digital CONT pin to low resistance. The idea is that the NPN in the opto and the schottky give about .4V CE drop + .5V Schottky diode drop, which drops the voltage of the Battery by 0.9V. Fully charged the battery is 4.2V - .9 = 3.3V, discharged the battery is 3.2V - .9 = 2.3V both of which are within the range of 3.3V TTL high.

Finally, when the charge fires, the opto shuts off, but the capacitor C10 holds the CONT1 input high for about 5s +- based on the time constant.
Comments: I had thought about doing this with a P-Channel mosfet, but was concerned about firing on power up as the pins float down at power up for some 10s of milliseconds.

A couple of things I have noted:
1. The use of the opto seems silly here, since I have tied the same voltage rail to both ends! I had first tied the NPN of the opto to the 3.3V rail which made sense but then started to worry that the voltage drops across the diode and the NPN would pull the voltage below the TTL high threshold.
2. It seems like I could eliminate the opto and do this:

3. I thought about doing a voltage divider, but then the diode isn't there to block the cap discharge which holds the continuity input high.
4. I think the TIP122 equivalent is overkill and would like to go do to a smaller SOT223 NPN part which will do 2A.

Any thought on favorite ways to do this?

 

[Voyager1]

Personally, I would use an N-MOSFET switch here. However, I’m confused as to why you need to hold the continuity high after the ignition. You definitely don’t require the optocoupler. I would just use a suitable voltage divider for the continuity sense.

Perhaps I’m missing something with your intentions here.

 

[jderimig]

If you have intermittent continuity what is going to discharge your capacitor to detect that?

 

[profmason]

1. N mosfet switch. I am using a voltage rail of a 1S battery. (3.2-4.2V) and the logic level is nominally 3.3V but due to the LDO, actually 3-3.3V. The mosfet switches on based on the Vgs threshold being reached and the source voltage in this arrangement changes between off (0V) and on (Vin). This means you can’t switch the gate to Vin, you need a different voltage rail that is higher than Vin by at least the mosfets Vgs threshold (At least a few volts!). I don't see how I can use a N channel without some shenanigans. I could use a p-channel, but it would be conducting until the micro booted and pulled it high. I have used n-channels for switching with a 2S supply. If I am not thinking about this correctly, please let me know!

2. "If you have intermittent continuity what is going to discharge your capacitor to detect that? " DOH. that makes good sense. I have filtered out intermittent continuity problems. Eliminate the cap and use software to only detect continuity until fire is initiated.

3. Suitable voltage divider. If I am eliminating the charge hold cap, then I don't need the diode and could use a voltage divider. It is cheaper and more compact.
I updated the NPN to a SOT223 part so that I can shrink the whole thing down. Thanks for the help. Any more input is appreciated.

 

[Voyager1]

If you have intermittent continuity what is going to discharge your capacitor to detect that?

Very good point.

 

[Voyager1]

1. N mosfet switch. I am using a voltage rail of a 1S battery. (3.2-4.2V) and the logic level is nominally 3.3V but due to the LDO, actually 3-3.3V. The mosfet switches on based on the Vgs threshold being reached and the source voltage in this arrangement changes between off (0V) and on (Vin). This means you can’t switch the gate to Vin, you need a different voltage rail that is higher than Vin by at least the mosfets Vgs threshold (At least a few volts!). I don't see how I can use a N channel without some shenanigans. I could use a p-channel, but it would be conducting until the micro booted and pulled it high. I have used n-channels for switching with a 2S supply. If I am not thinking about this correctly, please let me know!

2. "If you have intermittent continuity what is going to discharge your capacitor to detect that? " DOH. that makes good sense. I have filtered out intermittent continuity problems. Eliminate the cap and use software to only detect continuity until fire is initiated.

3. Suitable voltage divider. If I am eliminating the charge hold cap, then I don't need the diode and could use a voltage divider. It is cheaper and more compact. View attachment 452103
I updated the NPN to a SOT223 part so that I can shrink the whole thing down. Thanks for the help. Any more input is appreciated.

Use a MOSFET with a lower Vgs threshold, e.g., https://dlnmh9ip6v2uc.cloudfront.net/datasheets/Components/General/FQP30N06L.pdf
or
https://www.vishay.com/docs/68986/si7232dn.pdf

 

[Brian H.]

N mosfet switch. I am using a voltage rail of a 1S battery. (3.2-4.2V) and the logic level is nominally 3.3V but due to the LDO, actually 3-3.3V. The mosfet switches on based on the Vgs threshold being reached and the source voltage in this arrangement changes between off (0V) and on (Vin). This means you can’t switch the gate to Vin, you need a different voltage rail that is higher than Vin by at least the mosfets Vgs threshold (At least a few volts!). I don't see how I can use a N channel without some shenanigans. I could use a p-channel, but it would be conducting until the micro booted and pulled it high. I have used n-channels for switching with a 2S supply. If I am not thinking about this correctly, please let me know!

Nexperia PSMN5R6-60ylx
Surface mount, 60V, 100A, logic level MosFET.
1.7v gate/source threshold
Overkill for this application, yet affordable at $0.84 each from Mouser
Easily full on switching with 3.3V logic
Current limiting can be controlled with proper circuit design.

 

[cerving]

Use a MOSFET with a lower Vgs threshold, e.g., https://dlnmh9ip6v2uc.cloudfront.net/datasheets/Components/General/FQP30N06L.pdf
or
https://www.vishay.com/docs/68986/si7232dn.pdf

That's an often-overlooked parameter when selecting a MOSFET... your uP has to be able to turn it on! One with a minimum Vgs(th) of about 2V or less is best. Don't use the "typical" specs... always use the minimum/maximum and figure on the worst case. The AUIRL3114Z that we use in the Proton has a minimum threshold of 1.0V and a maximum of 2.5V... well within the "on" output of the processor.

 

[AllDigital]

Any thought on favorite ways to do this?

Mason, what are you using to drive the circuit? Are you still using the ESP8266?

On my board, I keep it simple. I use a 4A opto relay directly connected to a gpio pin to fire the pyro. I prefer the optos for the extra isolation and I like to switch the positive side, instead of the negative side (makes me feel safer). I keep all the pyros grounded and then I have another opto inline as a arm/safety to switch the positive (has to go through two relays). For continuity, I just use another gpio input with a diode attached to the relay side of the igniter to sense continuity. It pulls less than 1ma through the igniter to detect continuity. Below is a busy diagram (for four pyros), but you get the idea.

 

[OverTheTop]

As others have said I would have thought a low-Vth MOSFET would be a much better choice than a darlington pair.

 

[jderimig]

If you end up switching to a Fet, add a pull down to the FET gate. You uP probably inits to a high impedance state on power up.

 

[cerving]

You might also want to look at automotive drivers... they have both driver and sensing in one package, and often have other features such as thermal cutout and reverse-polarity protection. And, they'll save you board space and wiring.

 

[OverTheTop]

You might also want to look at automotive drivers... they have both driver and sensing in one package, and often have other features such as thermal cutout and reverse-polarity protection. And, they'll save you board space and wiring.

I was thinking of them too but do they not have higher voltage requirements? The OP is using a 1s LiPo so voltage is limited, especially as the battery discharges.

 

[cerving]

Yes they do, however what's a few extra grams in a 10 kg rocket?

 

[OverTheTop]

Agreed. Those protected drivers are very robust. If the voltage limit is lifted the electronics would be more robust if those drivers could be used.

 

[profmason]

Mike, Yes still using the ESP8266 the drive. I drooled over the omron SSR, and in particular those in the tiny VSON package, but at $15 each, I looked for other options. I know the dip ones you are using are only $5 each. Staring at your schematic, I thought "no current limiting resistors on the input of the opto?" The opto drops 1.2V, but unless there is some kind of built in current limit, I think a 220 Ohm resistor(3.3V - 1.2V = 2.1V / 220 = 9mA) on the opto inputs means you are not having your the teensy hits is source limit(10 mA) on those pins. The teensy has GPIO resistance of 23Ohm, so it seems like you are putting ~200mW of heat into the teensy whenever you trigger a relay. I mean it will just work they way you have it but .....

I have avoided mosfets in this application since when I went to school they needed 10V to turn on and when I last played with them 10 years ago a "logic level mosfet" was still in the linear region at 5V

Here is a re-think with a P -channel mosfet.




The mosfet in question is a vishay SI2301. It has an Rds on of 0.11 to 0.14 Ohm at Vgs = 2.5V and Id = 2.7A. Vgs(th) is 0.4 to 1V per data sheet but the graph below is more useful.


While I think this will work, pros and cons:
Pros of mosfet:
a. Mosfet is smaller then Darlington (SOT23 vs SOT-223)
b. Mosfet handles more current then Darlington (~12A peak vs ~4A peak)
c. Much cheaper part (0.09 cents vs 0.29 cents in quantity 10)

Cons:
a. 1 additional resistor
b. Have to route to both voltage rails (Fat traces etc)
c. The darlington is dead nuts simple while this mosfet is new fangled mystery magic

Even:
I think the voltage drop across the mosfet vs darlington is a wash... I could be wrong about this, but the Rds on for this part is ~5% of the resistance of the igniter.

So looking at the pros and cons the mosfet makes logical sense, except for my irrational feeling that making depletion layers thin enough for Mosfets to turn on at that low of a threshold can't possibly work, and if it does any bit of static will blow them up.

Am I missing other arguments here? Ways to improve my mosfet ejection circuit? I know I could use an n-channel to switch on the low side but I am getting a killer deal on these vishay parts.

 

[Brian H.]

That circuit will fire immediately when powered on.
MosFETs are a completely different beast that are voltage controlled vs. current controlled bipolar transistors.
Eliminate the 3.3V and 10k resistor to the gate/FIRE1.
Move the 470 ohm resistor so it connects gate to ground.
Connect Fire1 directly to gate.

Edit:
You will want the resistor connecting gate to ground to be the lowest resistance value possible. This will be based on max current output of FIRE1.

 

[profmason]

Thanks Brian for looking at this. I am confused. This is a p-channel mosfet. I pulled the gate to 3.2V on start up and the source is pulled up to ~Vbat(or 4V). This makes Vgs 3.2 -4 or -0.8V which is below Vth so the fet should be off. Now I know that I am pushing it close to the limit here... If I tie the Gate to ground then Vgs is -4 and everything turns on. It is possible that the Difference between Vbat and 3.3V is enough to put the fet into linear mode... I hadn't thought about this. Were you thinking about an N channel or do I have something backward on my diagram?

As far connecting Fire1 directly to the gate, I don't know what the capacitance of the fet, but having some resistance reduces the inrush current. It may be that this capacitance is so small that it doesn't matter.

 

[Brian H.]

I misunderstood.
I thought it was an N-Channel MosFET by glancing at the diagram with the gate arrow pointing away from the gate.
P-Channel MosFETs would require the pull-up resistor which complicates the design.
It would be ideal to go with the N-channel MosFET sinking the igniter to ground as shown in your schematic with the changes I recommended.

Edit:
With an N-channel MosFET, the 470ohm(or otherwise properly calculate resistance) to ground resistor keeps the transistor off at all times until the output of FIRE1 goes HIGH(.5V+depending on MosFET)

 

[Voyager1]

You might want to have the ematch and continuity sense in the drain circuit. The source needs to be tied to Vbat in a P-MOSFET switch.

 

[jderimig]

Why are you using a p fet on the low side? Use an n fet with a pull down r on the gate

 

[profmason]

Ok. Per voyagers suggestion I moved to a high side p channel switch.
I am using a pfet for false economy! It saves me $2 on the protoboard. I will look into availability of low Vth nfets. It will save me the problem of the 3.3V vs VBat issue at the gate of the Pfet for pull up.

 

[cerving]

The SI2302 is the NFET complement to the SI2301... same price. Eggtimer Rocketry uses them for driving buzzers... a relatively low current application, well suited for a FET in a SOT23 package. Our very first remote switch prototype used one for switching the connected altimeter on/off, after blowing a few of them up in testing we switched to a much more robust FET (NTD4965, with a Vgs(th) of 2.5V max); anything in a SOT23 package is probably not suitable for driving a resistive load from a LiPo without some kind of current limiting in place.

 

[profmason]

Thanks cris! Nothing beats your first hand experience saving me from making the same mistakes. I will look for something in a SOT223 package.

Here is the output characteristics of the STN4VF03L



I wondered about needing a transistor to drive the buzzer. I have used 40 Ohm buzzers directly from 5V outputs but they are all through hole. The smd ones seem to be 12 or 16ohm. I was wondering if keeping the 50% square wave duty cycle low would let you run a 16 ohm buzzer within the 10mA output pin current limit, but in retrospect using a small transistor to drive the buzzer makes sense.

I think I will just make up all 4 versions (DPak and SOT223 versions of Darlington, plus the PFet and NFet drive and test them all with ematches. In the PFet version, I will replace the SI2301 with a SOT223 ZXMP4A16G which has the characteristics below.)

I will draft up the PCB tomorrow and get them sent out. Any additional input is appreciated.

Thanks for all the help folks!

 

[profmason]

PCBs in to the fabricator and through DFM analysis. Will have assembled boards back in 2-3 weeks and then will update.


Comparing the BOMs:
DPak Darlington $0.48
SOT223 Darlington $0.69
NFet $0.81
PFet $0.68

In quantity 5, the setup and shipping are high etc so the BOM makes only a small difference.

 

[AllDigital]

DPak Darlington $0.48
SOT223 Darlington $0.69
NFet $0.81
PFet $0.68

That's awesome. You could sell DIY deployment circuits at the .99 cent store and still make margin.
The circuits looks solid, but there is something about running positive voltage through the match and switching ground that always makes me nervous. I think it is all those years working on cars, where everything is surrounded by ground. In this case it really doesn't change the risk profile.

 

[roboticsmick]

PCBs in to the fabricator and through DFM analysis. Will have assembled boards back in 2-3 weeks and then will update.
View attachment 453128

Comparing the BOMs:
DPak Darlington $0.48
SOT223 Darlington $0.69
NFet $0.81
PFet $0.68

In quantity 5, the setup and shipping are high etc so the BOM makes only a small difference.

This is an awesome thread! What was your preference in the end after testing?

 

[Former Squirrel]

Last time I ran the numbers, 2mA through a 40 ga nichrome raises the temp a few degrees per sec. I like a relay interlock on the plus side and close it only for the voltage divider continuity check. And even then, it's under automatic timing when MY fingers are elsewhere. Safety folks!

 

[AllDigital]

I switched all my pyros over to an automotive high-side driver. These are great. They are only about $2 and they operate at 4.5-36vdc, provide 5.5A over current protection, and have a built in continuity check. They interface directly with a 3.3v MCU and require very few additional components. I am using the Rohm BV1HD090FJ-CE2. Datasheet here:

https://fscdn.rohm.com/en/products/databook/datasheet/ic/power/ipd/bv1hd090fj-c-e.pdf

 

[cerving]

That's similar to the ST VN5E160S driver that we use in most of our altimeters now. Same parts availability issue that's endemic to all automotive silicon, though... fortunately we were able to round up some recently.