If you already noticed by yourself, or read some last testings of new toyquadcopters, all of them now are suffering of a new disease: a poor control range out of the box. For example: Cheerson CX-10C, WLtoys Q282, V686G (2nd revision), JJRC H11D, H20 et many more. What are their common point ? All of them are using the XN297 RFchip, a new chinese and cheap model in theory compatible with popular nRF24L01 model. This move for the XN297 have been done by most of the toy compagnies 6 months ago approximatively. The former chip used was a Beken model, i.e. the BK2423 perfectly 100% compatible with the nRF24L01.
Here is one of the packaging found for the XN297
Here is the main tech caracteristics of this chip
The most important lines are:
Transmit Power: -11~11dBm
RX Sensitivity: -88dBm@1Mbps -85dBm@2Mbps
If you look datasheet of a BK2423
Transmit Power: -10~5dBm
RX Sensitivity: -91dBm@1Mbps,-84dBm@2Mbps
One thing to know, for most of toyquadcopters, the datarate is fixed in general to 1Mbps (except the last syma protocol fixed to 250Kbps)
What does these numbers mean ?
The RX sensitivity is the lowest energy of the signal which allow to decode correctly the received data packet. Below this value is not possible to keep the radio link. Values are expressed in logarithmic scale but expressed in natural scale, the lowest value is 1/10^(-91/10)= 1258925411 times smaller than the 0dBm reference signal (impressive)
The effect of the Transmit Power (from the transmitter side) permits to increase the RX senstivity indirectly. Here is equation bilan for the receiver side
If we assume than max power is transmitted, we have
XN297: -88dBm – (+11dBm) = -99dBm of new RX sensitivity
Let’s see for the BK2423
BK2423: -91dBm – (5dBm) = -96dBm of new RX sensitivity
So it means here that the XN297 should be even more sensitive than the BK2423 (if max power is set of course).
It’s strange, coz I said exactly the opposit few lines earlier ? What is the trick ?. The trick is the usage of a PA+LNA stage in the TX module (PA for Power Amplier, LNA for Low Noise Amplier)
Look the photo of a nRF24L01 transmitter bundle
With this configuration, the emitted power of the nRF24L01/BK2423 is set to 0dBm then send the signal to PA+LNA parts.
The final output power is now 20dBm !!!! (100mW). Have a look again to the equation
BK2423 (PA+LNA): -91dBm – (20dBm) = -111dBm of new RX sensitivity
So as you can read you have 111-99 = 12dBm of difference and you have to know that every 6dBm, in theory the distance control is doubled. It means that the range of BK2423 is at least multiply by a factor 4 !!!!! and if we admit that max power have been set to the XN297.
To conclude, I think the main problem of these new toys is the absence of a PA+LNA stage in the transmitter module. Can this one be added ? I don’t know yet…
EDIT: In fact, the MJX X600 have a daughter RF chip including a LNA+PA stage. It explains why this model offers a very good range
How to improve a bit the radio range. Since the range is divided by a factor 4, a must to do is now to optimize the antenna positionning. Since IMHO, you are more flying in the X-Y plan than the X-Z one, I strongly advice to install antennas vertically. Another radical solution is to fly your quadcopter with a Devo controller if the protocol have been hacked. In this case, the TX emmits 20dBm
Most of these toy’s compagnies hire an external subcontractor to design the RF part… In order to decrease production costs, they offer the XN297 w/o PA+LNA without clearly taking into account the consequences. Who to blame ? For me, more than subcontractors, it’s more the main compagnies who wanted to reduce costs at ANY PRICE…. It’s now time to the market to show them they were wrong to downgrade performances.