The problem of burning of 5V voltage regulator while connection to battery with XT30 is well known. In particular it affects new 2S Mobula 7 and Eachine Trashcan whoops. HappyModel did some research and proposed protection with external capacitor. (If the regulator is already damaged it can be bypassed with external one (search e.g. Albert’s Kim youtube video “Fix your CrazyBee”)).
In summary: When 2S whoop is powered by two 1S batteries connected in series with two PH2.0 everything is fine. But if cable for 2 batteries is replaced with a wire with single XT-30 connector sometimes voltage regulator (step down buck converter) gets damaged at the moment of battery connection. The voltage is the same, burning out of voltage regulator is sporadic, it may or may not happen.
Fig.1 2x1s battery PH2.0 cable and single battery 1x2s XT30 (picture from the web).
Here I analyze why it may happen and how to prevent this damage.
The reason of the damaged is more-or-less clear, it happens because of inductance of the wires or battery. At the first moment of connection LC circuit is formed and a resonance leads to appearance of strong oscillations (sometimes called transients), the peaks of this oscillations can exceed threshold of damage of voltage regulator located on the FC board (around 12V for this type).
Here is a simulation model made with Matlab simulink.
Fig.2 Matlab simulink model. Rbat=20 mOhm, Rwire+Rcontact is a resistance of the wire and connectors (for the values see discussion below). Lwire is inductance of the wire from battery to connector (includes also inductance of the battery itself) and from connector to the battery (50 nH each). Cin is an input capacitance of step down voltage regulator (10uF), this capacitor also includes any capacitors connected to power line, ESC, etc.). Battery connection is simulated with the perfect switch operated by voltage step delayed by 10 us from the moment of simulation.
We use realistic values: 20mOhm for internal battery resistance, 50+50=100nH for the inductance of 10 cm length wire (assuming both parts to- and from- connector equal to 5 cm).
A. Two 1s batteries connected in series with two PH 2.0 connectors.
In this case we may assume that total resistance of the two connectors and wires would be around 60 mOhm (see PH 2.0 specs). Here is what we get:
Fig.3 Simulation result for two 1S batteries connected in series via 2xPH2.0 connectors. Total resistance of the connectors and wire is assumed to be 60 mOhm. X scale is in microseconds, Y scales are voltages at the battery (top) and input of the FC (bottom)
We see some spike, but the value do not exceed 10 V. This is may be safe for the regulator.
B. One 2s battery connected with XT30 connector.
XT30 is much more powerful and gold plated. We can assume that total resistance of the wire and connector would be around 10 mOhm.
Fig.4 Simulation result for one 2S battery connected via XT30 connector. Total resistance of the connector and wire is assumed to be 10 mOhm.
The result is drastically different. Oscillations are much stronger and may exceed safe level for the regulator’s chip. This happens because of oscillation are less dumped with the total resistance Rbat+Rwire+Rconnector.
So we see that in both cases there are oscillations, the amplitude depends on total resistance of the battery, connectors and wires. We cannot get exact values of the spikes voltages with our rough model but nonetheless we can see that it is dangerous in either case. Better quality (less resistance) of the battery/connector/wires produce stronger oscillations. Length of wires, type of used battery affect strongly amplitude of the spikes, even small change of the setup may cause dramatic difference. That is why some people experience the damaged and others are not.
If Cin is already charged, the oscillations are much lower and disappear if Cin is pre-charged to the battery voltage.
Fig.5 Simulation result for single 2S battery connected via XT30 connector. Cin of buck converter is pre-charged with 0/5/7 V
C. Fix with additional resistor
From the above, we can fix the problem if some external resistor will be inserted just before voltage regulator (e.g in the cut of trace between battery input and Cin of voltage regulator; this way we will not affect high-current part of the circuit). We have estimation of about ~50 mOhm-100 mOhm resistance to dump the oscillation. If voltage regulator current is 1A, power dissipation will be i^2*r=2*2*(0.05-0.1)~200-400 mW. This is quite small resistor (~3×2 mm and very lightweight), we lose power, but this power is much smaller compared to the power of the hovering whoop ~(8V*4A)=32W, so we will have less than 1.5% of additional loses.
D. Fix with external capacitor connected at the XT30 (proposed and sold by HappyModel).
The idea is to dump spikes with external capacitor.
Fig.6 External capacitor connected at the XT30. Rwire+Rcontact=10mOhm, Cfix=100 uF, ESR=0.2 Ohm (HappyModel parameters)
Fig.7 XT30, 2S battery, external capacitor 100 uF, ESR=0.2 Ohm
The oscillations get lower, the effect of dumping depends strongly on ESR value of the capacitor, that is why HappyModel sets ESR<0.2 Ohm. If we will use perfect capacitor with ESR~ 0.02 the oscillations are almost gone:
Fig.8 XT30, 2S battery, external capacitor 100 uF, ESR=0.02 Ohm
Unfortunately it is hard to find large capacitors with good ESR, that is why in high-speed electronics 2 types of capacitors connected in parallel are used: relatively slow large capacitor and ceramic fast small capacitor (big capacitor filters low frequency noises and fast capacitor filters high frequency noises). Let us see if this trick is working in our case:
Fig.9 Simulation model: XT30, 2S battery, external capacitor 100 uF, ESR=0.2 Ohm, Cfix fast is 100nF, ESR=0.001
Here is the result, not any helpful, oscillations are exactly the same to the ones without fast capacitor (Fig. 7). This is expected because our Cfix fast is still has much low capacitance compared to fast Cin.
Fig.10 XT30, 2S battery, external capacitor 100 uF, ESR=0.2 Ohm, Cfix fast is 10nF, ESR=0.001
In conclusion, external capacitor connected at XT30 reduces oscillations, but must be a very good quality (low ESR). If regular 100uF electrolytic capacitor is used it may not prevent the regulator from damage!
E. Fix with external capacitor connected at the FC.
Let us consider situation when external capacitor is connected right at the regulator.
Fig.11 External capacitor connected at the FC. Rwire+Rcontact=10mOhm, Cfix=100 uF, ESR discussed below
If capacitor is the same to one used by HappyModel: 100 uF and ESR=0.2 Ohm (Fig 6, Fig. 7) we get
Fig. 12 Cin=100 uF, ESR=0.2 Ohm (compare to Fig. 7).
Much better! Even relatively slow capacitor dumps oscillations effectively!
In conclusion, if external capacitor is used better connect it near the FC board input. In this case regular electrolytic capacitors should work well.
F. Fix with a diode.
Let us see if we can use a diode to get rid of the oscillations.
Fig.13 Simulation model: XT30, 2S battery, a diode (0.6V forward voltage drop, internal resistance 0.3 Ohm)
Fig.14 Simulation results: XT30, 2S battery, a diode (0.6V forward voltage drop)
It works perfect, no any spikes, in this case internal resistance of the diode does most of the work. The drawback is 0.6V drop of the input voltage, it’s is OK for 2S (in any case 2S-0.6V is higher than 5V output of the regulator), may be a problem for 1S, but who cares if we are going to use XT30 to have 2S whoop only. Another disadvantage is a heat dissipated at the diode. If input current is 2A, the power will be 0.6*2=1.2W. Not very good, the diode should be big enough, but may be better ( smaller/lightweight ) than external capacitor used in method D. Also it must be inserted right before voltage regulator (with PCB trace cut) to keep the pathway to high-current part (ESC+motors) unaffected. I believe the resistor method C is better (both diode and resistor dump oscillations because of resistance), the resistor is smaller and can be adjusted precisely. The diode can be used as a temporary replacement.
G. Fix with a ferrite ring/bead.
Ferrite ring or bead is a common fixer of spikes in the wires, it can be seen in many places like PC monitors / USB cables etc. Actually it acts as an induction coil connected in series to the wire.
Fig.11 Ferrite ring over power line. Equivalent is an external coil connected at the FC. Lext=1-10uH.
The idea is to make oscillation period significantly longer and have a half-period less than start-up time of buck converter. After the buck converter started it should suppress oscillations because of load. May work or may not work. Depends on how big should be the ferrite ring (1 turn of 2mm ring will produce about 1 uH of inductance) and how the coil interact with the buck-converter. At the same time, the inductance should be small enough to do not interfere with the high pulsed current of PWM at the motors (~20us). This balance is delicate. The solution is simple but needs experimental investigation.
Fig 12. External inductor of 2 uH (ferrite ring/bead equivalent). It increases period of oscillations, we hope that back converter start will happen before amplitude maximum and this may dump further osculations. This is just my guess.
H. Fix with Zener diode.
Another common solution to prevent damage caused by extra voltages at the input is so called Zener diodes. If voltage at the diode exceeds some level (Vz) the diode gets conductive with some differential resistance Rdiff (dV/dI), if voltage is below, the resistance is very high.
Fig.13 External inductor coil. Vz=10V, Rdiff see below
For perfect Zener diode (Rdiff is very small) it works. Here is what happens if Vz=10V and Rdiff=0.01Ohm:
Fig. 14. Zener diode Vz=10V, Rdiff=0.01Ohm
We can see limitation of spike amplitude at ~10V as expected. Unfortunately real Zener diodes have much higher differential resistance in the range 1-10Ohm (best case). Even if resistance is only 1Ohm Zener diode almost has no effect:
Fig. 15. Zener diode Vz=10V, Rdiff=1 Ohm
The solution that works very well for low current circuits is not applicable in our case.
I. Fix with soft-start circuit.
Imaginary circuit that may solve the problem. When battery is inserted MOSFET opens gradually eliminating any oscillations. When voltage regulator is started we have just fully opened MOSFET with low resistance.
Conclusions:
please note, all solutions below must not affect performance of the quad. Battery is directly connected to the ESC. Here we are talking about how to prevent damage of 5V voltage regulator used to power MCU and the camera. If something is inserted in series between the battery and the FC (diode, resistor) the path to the ESC/motors must remain untouched!
1. Damage is caused by extra voltages at the voltage regulator. The extra voltages appear because of oscillations at the LC resonance circuit formed at the moment of connection. This LC circuit has high Q factor because of low resistance of elements typical for powerfull quads and requirements to have low ESR capacitor for proper operation of buck converter (voltage regulator).
2. To reduce the chance of damage use wires as short as possible to reduce the induction.
3. Better connector, better wire and better (or more mAh) battery enhance the effect. By chance parameters of 2x PH2.0 are below the damage level, but XT30 is above.
4. The chance of damage can be significantly reduced if some low current charge of the input capacitor is done just before connection. For that the positive terminal of the battery wire may be touched with the finger (just before connection) with the hope the capacitor will be charged by EMS from the body. Or better, connect positive terminal of the battery to the connector with some resistor first, then plug the battery and disconnect the resistor. As a resistor a finger-to-finger resistance can be used. This may be a temporal solution (or not). Not tested.
5. Another option is to insert resistor of 50-100 mOhm just before the input capacitor of the voltage regulator. Low enough to do not disturb proper operation of buck converter but high enough to dump oscillations. Power dissipation will be 0.2-0.5W. Such resistor is about 3×2 mm in size and very lightweight.
6. Protection can be done with the external capacitor of about 100uF attached at the XT30 connector or at the FC. In the first case the capacitor should have extremely good ESR value (<0.1 Ohm), in the last case capacitor with ESR 0.2-1 Ohm can be used. I would not trust very much HappyModel wire with capacitor, to me it may not work well (because of connection at XT30, not at the FC).
7. Diode connected in series will prevent damage for sure. Not very clear if buck converter will work with such supply, in any case rectification diode of >1A should be used. The diode must be inserted just before voltage regulator, to have other electronics unaffected.
8. Ferrite coil or bead over the power line. May be OK, but too many uncertain parameters are involved. Need experimental tests.
9. Common method of spike removal with Zener diode will not work.
10. I believe the problem will be solved in the next releases of FC, for that voltage regulator with higher damage threshold can be used, or with larger capacitor on the board, or with extra MOSFET to pre-charge input capacitor of the buck converter (soft-start circuit).