Pseudotectonic
Members
- May 7, 2022
- 476
- 307
Introduction: The lost art of capacitors in airsoft
The mythical powers of "capacitor banks" is not a new idea in airsoft, but the original sources and theories are lost in time, with only faint forum posts remain (all the links are dead in https://www.airsoftsniperforum.com/threads/capacitor-banks-revisited.8688/).
In my pursuit of the ultimate trigger response in AEG, power supply remains a big unresolved issue because batteries (LiPo or NiMH) do not really give enough oomph for lack of a better word. We need something better, I thought, we need supercapacitors. And so I set out to rediscover the elementals of such dark magic from scratch (but based on the same principles). And I believe I have revived this magic which I will share my formula below.
And dare I say, the surface of this dark art has not really been scratched, because the predecessors either did not use enough capacity (some examples were in the order of millifarads which is not nearly enough), and did not really use the correct type of supercaps (stacked coin cells hybrid types are not capable of high amps). And perhaps the right modules were not available until very recently, so this study will also represent a fresh look into this subject.
If you want the bottom line:
Why supercaps at all, isn't LiPo good enough?
In short (excuse the pun), LiPo cannot provide sufficient peak current for the amp spikes during start of motor which is often in the region of 45-60 A. These micro-events of shorts (from the perspective of the battery) is also a main pathology of LiPo degrading and swelling over time.
These spikes happen every shot in semi, and in the first shots in full-auto. This status quo practice of relying on abusing LiPo as sacrifice for performance does not sit well with me, because firstly it is electrically primitive, and secondly it is a safety risk in this toy gun industry that I think is unnecessary and should be done away with, at least as much as possible.
The whole premise of using supercaps is to provide a voltage reservoir for surge power during these initial amp spikes. This is needed because these amp spike often exceed the capabilities of the battery (even LiPo) and the batteries drops voltage in response (for milliseconds) and unable to feed all the current needed for this initial spin-up of the motor.
In theory, it can achieve the following:
My evidence and theorisations are as follows.
What is the actual energy requirement?
To begin to figure out the solution, we need to ask, what is the minimum capacity to be effective?
We can analyse these two graphs from airsoftlab.eu for theorycrafting. http://www.airsoftlab.eu/docs/experiments/motor_current/
To figure out the ballpark of energy levels we are talking about, I have picked two cases representative of a high stress build (with a 16 TPA motor) and a more optimal build (with a 28 TPA).
The first graph is with 16 TPA motor, SP130, 13:1, to illustrate a typical suboptimal setup with a low torque motor paired with a high stress spring and gear ratio.
The second graph is with a more comfortable 28 TPA motor with SP130 and 13:1 to show a more efficient setup.
Just by eyeballing the graphs (and the blog), we can character these spikes (and the energy required to tame them) to be about 45A to 60A, for a duration of about 60ms.
16 TPA:
View attachment 120575
28 TPA:
View attachment 120576
What do the numbers mean? There is a simple answer, and a more complex one.
A simplistic way to translate this to farad (the capacity we need), using the 16 TPA setup as example, 60A for 60ms is 3.6 coulomb, which at 11.1V is about 0.324 F. For the 28 TPA example, this would be 2.7 coulomb, at 11.1 V this is 0.243 F.
Therefore we can say our ballpark figure is about 0.25 F to 0.33 F as a minimum target. If we are aiming to create a unit that can work with worst case scenario, let's go with 0.33 F.
However because supercaps discharge their voltage linearly (meaning their 11.1 V drops down to zero as it discharges over the 60 ms), we can think of it as sharing the workload with the battery in a 1:1 ratio (this is very simplified), meaning it is only doing half of the work while the batteries still need to supply the other half of the current (which is a big improvement already).
Here is a prediction:
View attachment 120577
In this predicted scenario, the amp spike seen by the battery should be reduced to less than half, because the supercaps will react faster to supply the spike, but as the supercap runs out of voltage, the battery will notice the difference in voltage and catch up, eventually supplying the full load in full-auto, but at a much improved stability. And as the trigger is released, the current stops, the supercaps are charged up again to battery voltage as the battery recovers from voltage sag.
We can offload even more work from the batteries if we multiply the supercap capacity (say 3 times, to 1 F) so the they will perform in a 3:1 relationship i.e. 75% of the work are done by supercaps, which will stabilise the current even more. If we go bigger, say 1.66 F, the ratio will be 5:1, or 83% of work being done by supercaps.
(The actual result will probably be better, the ratios e.g. 1:1 at 0.33 F are just conservative notional numbers I made up for ease of explanation. Supercaps are more responsive than batteries, and the overall lowered battery stress should further reduce overall voltage sag, so the battery should see less than 50% of the peak current, but I don't have the equipments to proof this hypothesis.)
So let's say our notional baseline is 0.33 F, now we just need to implement this theory.
Designing a supercaps unit
Here is our goal:
1. Supercaps with total capacity of 0.33 F or more, bigger the better.
2. Voltage should be ok for airsoft usage. (Say a fully charged 11.1 V LiPo is around 12.6 V)
3. Overall size to be small enough to fit in a typical buffer tube, with room for cable management for most cases. Smaller the better.
4. (Bonus feature) built-in safety to drain residue voltage when unplugged.
5. (Bonus feature) LED to indicate presence of voltage.
The obvious (and probably the only viable) strategy is to use 3 no. of 5-6 V supercaps in series to give us a 15-18 V headroom for the maximum 12.6 V we are expecting from a fully charged 11.1.
The first problem is selection of supercaps. The second problem is designing the whole package that can physically fit inside the buffer tube.
Long story short, here is my blueprint: (Just connect the supercaps in series, and then parallel with the batteries, I don't have a drawing)
The Eaton supercaps I am using are the highest capacity that can still fit inside a typical buffer tube and wiring, and with one of the better ESR in its class, and can theoretically suffer 115 A of thermals over 60 ms, and tested to MIL-STD 202G for shock and vibration.
They are wired (I'm using some fancy SPEC 44 16 awg wires) to a connector that goes between the AEG wire and battery wire, so it is completely plug and play, and removable for safety and for storage. It can in fact be stacked up (if you have multiple units) to give extra performance.
For extra safety I have included a bleeder resistor to discharge the residue voltage in maximum 3 hours after it is unplugged. Also for safety (and aesthetics) I added an LED for visual indication of voltage presence.
I am tempted to call it the PASTA 1000 (Pseudotectonic Advanced SupercapaciTor Array 1000 mF) but any suggestions welcome.
View attachment 120578
Here is what it looks like in real life:
View attachment 120579
Testing
AEG: Specna E-19, completely stock (which has an X-ASR mosfet preinstalled.)
Battery: 9.6 V NiMH, fully charged, measuring about 11.32 V when testing.
This is just a simple A/B test to see if the supercaps work at all. I will simply alternate between plugging and unplugging the supercaps several times, then take measurements with groups of 6 to 7 shots, until the data are fairly consistent and/or a pattern can be identified. The groups are measured in Audacity and averaged and rounded to nearest millisecond.
Also note I am not testing with any magazine inserted because A. that is not going to make massive difference either way and B. it is one less variable to worry about and C. if a BB goes off it is going to affect my measurements with sound.
The results:
Stock setup (without supercaps), group #1:
Trigger response: from trigger action = 83 ms, from motor spin-up = [data missing]
ROF: ~17.17 RPS
With supercaps, group #1:
Trigger response: from trigger action = 81 ms, from motor spin-up = 57 ms
ROF: ~17.48 RPS (+1.8%)
Stock #2: (I stopped measuring full-auto because it is getting too loud for the neighbours)
Trigger response: from trigger action = 96 ms, from motor spin-up = 64 ms
With supercaps #2:
Trigger response: from trigger action = 74 ms (-23%), from motor spin-up = 62 ms (-3%)
Stock #3:
Trigger response: from trigger action = 90 ms (+22%), from motor spin-up = 69 ms (+11%)
With supercaps #3:
Trigger reponse from trigger action = 71 ms (-21%), from motor spin-up = 62 ms (-10%)
Stock #4:
Trigger response: from trigger action = 84 ms (+18%), from motor spin-up = 68 ms (+10%)
With supercaps #4:
Trigger response: from trigger action = 76 ms (-10%), from motor spin-up = 62 ms (-9%)
Stock #5:
Trigger response: from trigger action = 80 ms (+5%), from motor spin-up = 69 ms (+11%)
With supercaps #5:
Trigger response: from trigger action = 75 ms (-6%), from motor spin-up = 61 ms (-12%)
Analysis of results: Definitely a noticeable audio difference in trigger response. The only way to describe it is it sounds more "instant" and there is less of the spin-up whine. I am not sure why the groups vary quite a bit (maybe battery and/or gearbox settling) but I think it is fair to say the supercaps are making a difference. If we average the data after group #3, with supercaps, the overall trigger response is about 14% improved, with the cycling time from motor spin-up is remarkably consistent at about 11% improved. The shorter lag time from trigger action to motor spin-up can be explained by the voltage stability provided by the supercaps. Overall I didn't know what to expect but I would say 14% improved trigger response is pretty good. It is definitely not a negligible difference, and definitely noticeable when compared side by side.
But going by feel alone, it definitely feels a bit more snappy.
ROF is probably improved a little bit, but more testing needed.
The installation: This may be the only draw back. To actually fit the unit inside the buffer tube along with the X-ASR is a massive hassle, which I have to actually remove the original long heat shrink around the three wires to get them to flex, and re-crimp two out of three of the spade connectors to the X-ASR because they were damaged by too much bending. And even when the supercaps are in, it is still very stuck and you need to wrestle the wires to get the battery on and close the butt plate. Although I have done it with the stock fully collapsed and if I install it with the stock a few positions out, it will be easy. Once it's on, it works. But it is definitely not ideal if you have any in-line mosfet like mine. However if you use a proper mosfet inside the gearbox and just have wires in the buffer tube, it should fit very easily, potentially upping the supercaps to 5 F ones for even better trigger response.
There is no noticeable sparks or heat or anything when install and in use. It simply lights up when you plug it in and it just works with zero drama.
Here is how it looks like installed, with a fully collapsed stock, and without the stock: (it barely fits)
View attachment 120580
View attachment 120581
To uninstall: If the battery is disconnected with just the supercaps plugged into the gun, the gun will barely able to do one shot and the second shot will be stopped by the mosfet because the voltage will have dropped too low. This is just as expected and fairly consistent with the maths.
The LED: When you unplug everything, the LED stays on but slowly dims down over the course of about an hour (just as designed). This shows the draining resistor and the LED are all working as intended. I could also feel zero heat from the resistor, which is great and again matching expectation. The LED still visibly faintly glows even at as low as 2 V so it works perfectly for its purpose as a voltage indicator. The red LED is pretty to look at, although I might change the colour to something like blue or green, because the red can be mistaken as error from the mosfet. I might also move it to the "top" side of the "plug".
Conclusion of experiment: The prove of concept is a success. Most importantly there is certainly a performance benefit (14% in my test). All the maths check out. The unit is fully functional and fully match the expectations. It literally is plug-and-play. The installation can be a hassle for wire management but that is purely down to physical space and should not be a problem if you are not using in-line mosfet.
Conclusions and speculations
Here is a summary of what this device can do.
Conclusion is, and I am probably biased, supercaps could be the next best thing in airsoft. If the space problem can be resolved.
Please do comment if you spot any issues in the theory or in the blueprint.
Thank you for reading.
The mythical powers of "capacitor banks" is not a new idea in airsoft, but the original sources and theories are lost in time, with only faint forum posts remain (all the links are dead in https://www.airsoftsniperforum.com/threads/capacitor-banks-revisited.8688/).
In my pursuit of the ultimate trigger response in AEG, power supply remains a big unresolved issue because batteries (LiPo or NiMH) do not really give enough oomph for lack of a better word. We need something better, I thought, we need supercapacitors. And so I set out to rediscover the elementals of such dark magic from scratch (but based on the same principles). And I believe I have revived this magic which I will share my formula below.
And dare I say, the surface of this dark art has not really been scratched, because the predecessors either did not use enough capacity (some examples were in the order of millifarads which is not nearly enough), and did not really use the correct type of supercaps (stacked coin cells hybrid types are not capable of high amps). And perhaps the right modules were not available until very recently, so this study will also represent a fresh look into this subject.
If you want the bottom line:
14% improved trigger response! All with a plug-and-play module you simply plug in!
Why supercaps at all, isn't LiPo good enough?
In short (excuse the pun), LiPo cannot provide sufficient peak current for the amp spikes during start of motor which is often in the region of 45-60 A. These micro-events of shorts (from the perspective of the battery) is also a main pathology of LiPo degrading and swelling over time.
These spikes happen every shot in semi, and in the first shots in full-auto. This status quo practice of relying on abusing LiPo as sacrifice for performance does not sit well with me, because firstly it is electrically primitive, and secondly it is a safety risk in this toy gun industry that I think is unnecessary and should be done away with, at least as much as possible.
The whole premise of using supercaps is to provide a voltage reservoir for surge power during these initial amp spikes. This is needed because these amp spike often exceed the capabilities of the battery (even LiPo) and the batteries drops voltage in response (for milliseconds) and unable to feed all the current needed for this initial spin-up of the motor.
In theory, it can achieve the following:
- Surge power for much more responsive motor, resulting in snappier trigger response.
- Stabilised voltage supply for full-auto, resulting in better ROF.
- Protecting batteries from surge currents, prolonging their lifespan and minimise risks of LiPo fire.
My evidence and theorisations are as follows.
What is the actual energy requirement?
To begin to figure out the solution, we need to ask, what is the minimum capacity to be effective?
We can analyse these two graphs from airsoftlab.eu for theorycrafting. http://www.airsoftlab.eu/docs/experiments/motor_current/
To figure out the ballpark of energy levels we are talking about, I have picked two cases representative of a high stress build (with a 16 TPA motor) and a more optimal build (with a 28 TPA).
The first graph is with 16 TPA motor, SP130, 13:1, to illustrate a typical suboptimal setup with a low torque motor paired with a high stress spring and gear ratio.
The second graph is with a more comfortable 28 TPA motor with SP130 and 13:1 to show a more efficient setup.
Just by eyeballing the graphs (and the blog), we can character these spikes (and the energy required to tame them) to be about 45A to 60A, for a duration of about 60ms.
16 TPA:
View attachment 120575
28 TPA:
View attachment 120576
What do the numbers mean? There is a simple answer, and a more complex one.
A simplistic way to translate this to farad (the capacity we need), using the 16 TPA setup as example, 60A for 60ms is 3.6 coulomb, which at 11.1V is about 0.324 F. For the 28 TPA example, this would be 2.7 coulomb, at 11.1 V this is 0.243 F.
Therefore we can say our ballpark figure is about 0.25 F to 0.33 F as a minimum target. If we are aiming to create a unit that can work with worst case scenario, let's go with 0.33 F.
However because supercaps discharge their voltage linearly (meaning their 11.1 V drops down to zero as it discharges over the 60 ms), we can think of it as sharing the workload with the battery in a 1:1 ratio (this is very simplified), meaning it is only doing half of the work while the batteries still need to supply the other half of the current (which is a big improvement already).
Here is a prediction:
View attachment 120577
In this predicted scenario, the amp spike seen by the battery should be reduced to less than half, because the supercaps will react faster to supply the spike, but as the supercap runs out of voltage, the battery will notice the difference in voltage and catch up, eventually supplying the full load in full-auto, but at a much improved stability. And as the trigger is released, the current stops, the supercaps are charged up again to battery voltage as the battery recovers from voltage sag.
We can offload even more work from the batteries if we multiply the supercap capacity (say 3 times, to 1 F) so the they will perform in a 3:1 relationship i.e. 75% of the work are done by supercaps, which will stabilise the current even more. If we go bigger, say 1.66 F, the ratio will be 5:1, or 83% of work being done by supercaps.
(The actual result will probably be better, the ratios e.g. 1:1 at 0.33 F are just conservative notional numbers I made up for ease of explanation. Supercaps are more responsive than batteries, and the overall lowered battery stress should further reduce overall voltage sag, so the battery should see less than 50% of the peak current, but I don't have the equipments to proof this hypothesis.)
So let's say our notional baseline is 0.33 F, now we just need to implement this theory.
Designing a supercaps unit
Here is our goal:
1. Supercaps with total capacity of 0.33 F or more, bigger the better.
2. Voltage should be ok for airsoft usage. (Say a fully charged 11.1 V LiPo is around 12.6 V)
3. Overall size to be small enough to fit in a typical buffer tube, with room for cable management for most cases. Smaller the better.
4. (Bonus feature) built-in safety to drain residue voltage when unplugged.
5. (Bonus feature) LED to indicate presence of voltage.
The obvious (and probably the only viable) strategy is to use 3 no. of 5-6 V supercaps in series to give us a 15-18 V headroom for the maximum 12.6 V we are expecting from a fully charged 11.1.
The first problem is selection of supercaps. The second problem is designing the whole package that can physically fit inside the buffer tube.
Long story short, here is my blueprint: (Just connect the supercaps in series, and then parallel with the batteries, I don't have a drawing)
The Eaton supercaps I am using are the highest capacity that can still fit inside a typical buffer tube and wiring, and with one of the better ESR in its class, and can theoretically suffer 115 A of thermals over 60 ms, and tested to MIL-STD 202G for shock and vibration.
They are wired (I'm using some fancy SPEC 44 16 awg wires) to a connector that goes between the AEG wire and battery wire, so it is completely plug and play, and removable for safety and for storage. It can in fact be stacked up (if you have multiple units) to give extra performance.
For extra safety I have included a bleeder resistor to discharge the residue voltage in maximum 3 hours after it is unplugged. Also for safety (and aesthetics) I added an LED for visual indication of voltage presence.
I am tempted to call it the PASTA 1000 (Pseudotectonic Advanced SupercapaciTor Array 1000 mF) but any suggestions welcome.
View attachment 120578
Here is what it looks like in real life:
View attachment 120579
Testing
AEG: Specna E-19, completely stock (which has an X-ASR mosfet preinstalled.)
Battery: 9.6 V NiMH, fully charged, measuring about 11.32 V when testing.
This is just a simple A/B test to see if the supercaps work at all. I will simply alternate between plugging and unplugging the supercaps several times, then take measurements with groups of 6 to 7 shots, until the data are fairly consistent and/or a pattern can be identified. The groups are measured in Audacity and averaged and rounded to nearest millisecond.
Also note I am not testing with any magazine inserted because A. that is not going to make massive difference either way and B. it is one less variable to worry about and C. if a BB goes off it is going to affect my measurements with sound.
The results:
Stock setup (without supercaps), group #1:
Trigger response: from trigger action = 83 ms, from motor spin-up = [data missing]
ROF: ~17.17 RPS
With supercaps, group #1:
Trigger response: from trigger action = 81 ms, from motor spin-up = 57 ms
ROF: ~17.48 RPS (+1.8%)
Stock #2: (I stopped measuring full-auto because it is getting too loud for the neighbours)
Trigger response: from trigger action = 96 ms, from motor spin-up = 64 ms
With supercaps #2:
Trigger response: from trigger action = 74 ms (-23%), from motor spin-up = 62 ms (-3%)
Stock #3:
Trigger response: from trigger action = 90 ms (+22%), from motor spin-up = 69 ms (+11%)
With supercaps #3:
Trigger reponse from trigger action = 71 ms (-21%), from motor spin-up = 62 ms (-10%)
Stock #4:
Trigger response: from trigger action = 84 ms (+18%), from motor spin-up = 68 ms (+10%)
With supercaps #4:
Trigger response: from trigger action = 76 ms (-10%), from motor spin-up = 62 ms (-9%)
Stock #5:
Trigger response: from trigger action = 80 ms (+5%), from motor spin-up = 69 ms (+11%)
With supercaps #5:
Trigger response: from trigger action = 75 ms (-6%), from motor spin-up = 61 ms (-12%)
Analysis of results: Definitely a noticeable audio difference in trigger response. The only way to describe it is it sounds more "instant" and there is less of the spin-up whine. I am not sure why the groups vary quite a bit (maybe battery and/or gearbox settling) but I think it is fair to say the supercaps are making a difference. If we average the data after group #3, with supercaps, the overall trigger response is about 14% improved, with the cycling time from motor spin-up is remarkably consistent at about 11% improved. The shorter lag time from trigger action to motor spin-up can be explained by the voltage stability provided by the supercaps. Overall I didn't know what to expect but I would say 14% improved trigger response is pretty good. It is definitely not a negligible difference, and definitely noticeable when compared side by side.
But going by feel alone, it definitely feels a bit more snappy.
ROF is probably improved a little bit, but more testing needed.
The installation: This may be the only draw back. To actually fit the unit inside the buffer tube along with the X-ASR is a massive hassle, which I have to actually remove the original long heat shrink around the three wires to get them to flex, and re-crimp two out of three of the spade connectors to the X-ASR because they were damaged by too much bending. And even when the supercaps are in, it is still very stuck and you need to wrestle the wires to get the battery on and close the butt plate. Although I have done it with the stock fully collapsed and if I install it with the stock a few positions out, it will be easy. Once it's on, it works. But it is definitely not ideal if you have any in-line mosfet like mine. However if you use a proper mosfet inside the gearbox and just have wires in the buffer tube, it should fit very easily, potentially upping the supercaps to 5 F ones for even better trigger response.
There is no noticeable sparks or heat or anything when install and in use. It simply lights up when you plug it in and it just works with zero drama.
Here is how it looks like installed, with a fully collapsed stock, and without the stock: (it barely fits)
View attachment 120580
View attachment 120581
To uninstall: If the battery is disconnected with just the supercaps plugged into the gun, the gun will barely able to do one shot and the second shot will be stopped by the mosfet because the voltage will have dropped too low. This is just as expected and fairly consistent with the maths.
The LED: When you unplug everything, the LED stays on but slowly dims down over the course of about an hour (just as designed). This shows the draining resistor and the LED are all working as intended. I could also feel zero heat from the resistor, which is great and again matching expectation. The LED still visibly faintly glows even at as low as 2 V so it works perfectly for its purpose as a voltage indicator. The red LED is pretty to look at, although I might change the colour to something like blue or green, because the red can be mistaken as error from the mosfet. I might also move it to the "top" side of the "plug".
Conclusion of experiment: The prove of concept is a success. Most importantly there is certainly a performance benefit (14% in my test). All the maths check out. The unit is fully functional and fully match the expectations. It literally is plug-and-play. The installation can be a hassle for wire management but that is purely down to physical space and should not be a problem if you are not using in-line mosfet.
Conclusions and speculations
- More testing is needed with other setups, if you would like one for testing I can make you a copy for a fee.
- The tech tree can potentially branch into AKs or other platforms or even external compartments, but I don't have any of these for development.
Here is a summary of what this device can do.
- The biggest feature for me is safety, because when the LiPo (or any battery) is shielded from stress they are much less likely to puff up over time and starting a fire.
- Another key thing is of course the performance. It works very well in my very first little experiment.
- Electrically the overall voltage floor and current ceiling is improved. Adding supercaps is a bit like transitioning from NiMH to LiPo, but on steroids. There is also where the drawbacks are, because a mosfet is probably a good idea for such power, and if you want to collapse the butt stock you will need a more advanced mosfet that sits inside the gearbox rather than the buffer tube, for cable management reasons. (If you copy what I did you will risk breaking some wires)
- In terms of use cases, it will benefit NiMH the most because the performance is suddenly brought closer to that of LiPo because the performance gap is effectively closed, making NiMH a viable option again. In fact it makes NiMH better than LiPo because NiMH is much safer.
- This is also a must if you are chasing the state-of-the-art trigger response or battery efficiency in any build.
- It will also help with cold weather performance.
- And the best thing is, this is essentially a "free upgrade" because it is an entirely new component added to the system, it does not replace or compete with any existing parts, and it doesn't need any complicated installation, it is literally plug-and-play.
- The only downside for now is with wire management inside the buffer tube to make room for it. I am not sure if there is enough space for guns other than an M4 but maybe you can find creative ways to fit it e.g. longer wires.
- This can potentially benefit rental fleets because it is the easiest upgrade possible with zero overhead on tech, plus it makes your NiMH or LiPo inventory much safer and longer lasting. The savings in overhead for battery management could be worth the investment. And when a gun dies you can very easily transplant it to the replacement gun.
- In theory, you can actually stack multiple units for extra performance (the only problem is finding the space to put them).
Conclusion is, and I am probably biased, supercaps could be the next best thing in airsoft. If the space problem can be resolved.
Please do comment if you spot any issues in the theory or in the blueprint.
Thank you for reading.
