There have been a lot of rumors and discussion about the $99 Eachine EV100 FPV goggles recently. I already did some teardown, component analysis, component identification, and analyzed the “diversity” feature of the receiver in my previous posts.
With the goggles being more and more used, people experience a new problem with these goggles: Depending on the depicted scene a gray bar overlays the video feed and some strange distortion becomes visible:
Many people were complaining about this and Eachine/banggood was promising a “firmware upgrade” soon. Well… They released a “fix” for the problem. But that’s not really a firmware upgrade. More a hardware “upgrade” or let’s say bugfix. The banggood employee (?) BGJim suggests to replace a small capacitor on the mainboard. Interesting… Let’s investigate what this changes. If you are here to see how to fix it in order to get flying again please scroll down, there is a detailed tutorial with pictures on how to swap the capacitor. Just in case you are as curious as me, you might not want to skip the following investigation of the cause and how a single capacitor fixes the issue that follows next.
Before BGJim posted the fix I already had an idea what was causing this. I have been digging into the “wonderful” world of analog PAL/NTSC video signals the past months while designing my custom tinyOSD project and this looked like a bad line sync to me. This gray/dark bar always happens together with some strange image distortion you can see in the previous picture.
An analog video signal consists of different sync signals that basically tell the “television” when to start a new line or frame. Remember, this is all analog, so different voltage levels stand for different parts of the signal. Let’s have a look at the following scope picture (ignore the yellow signal for now):
I am going to keep it simple, in reality things are a bit more complicated. The line sync signal can be seen at the left, followed by a color burst signal, and finally one analog representation of a single video line. Now comes the hard part: which voltage level represents a line sync, which is data?
This is where the PAL and NTSC standard define voltage levels for this type of signals. The sync level is specified at -0.3V, a black “pixel” is represented by 0V and a white “pixel” is around 0.7V. So this is where things get complicated. Various things can cause the signal to shift up or down. This is why AC coupling is used to get rid of this artifacts. You basically add a capacitor in line with the signal to block any DC (= very low frequency with 0 Hz) components of the signal and let only high frequency parts of the signal pass. Maxim has written a really great (but technical) article on this, you can find it here. An excerpt reads:
Using too small a capacitor causes the displayed image to darken from left to right, top to bottom, and can distort the image spatially (depending on the capacitor’s size). In video, this is called line droop and field tilt.
Wait, this sounds familar, right? But more on this later…
BGJim suggests to replace a specific capacitor with 2x 100 uF tantalum capacitors which sum up to 200uF capacity. If you read my blog regularly, you will know that random try’n’error part replacements are not my thing. I want to know exactly what is going on. So let’s have a look at the capacitor and how it is connected:
The video signal originates from the right (cyan), passes a 3157 video mux (which is an electronically controllable switch) and can take either the green or yellow path, depending on the mux status. The yellow path is passing our mystical “firmware” fix capacitor. In addition, both signal paths have a lot of unpopulated filtering and signal reconditioning pads. The designer of this PCB knew what he was doing, he was prepared for a lot of filtering that might become necessary during testing.
The populated part of the filtering circuit looks like what is used on a reference design for a similar chip (MST702 SZDEMO). The CVBS1 and CVBS2 pins of the video decoder (MST706) are just analog video inputs. I did not investigate further when the signal passes the green or yellow path but I would expect analog video input will be routed the green path as external devices usually contain the AC decoupling capacitor.
So what does this AC coupling capacitor do to your signal? How does this look like? Let’s hook up a scope to the capacitor in order to see the voltage levels before and after passing the capacitor.
The yellow plot shows the signal before passing the capacitor and the blue one is the signal right behind the capacitor. You can see that the yellow signal is all positive and has an offset of around 500mV to the blue one. The blue one looks better, most of the offset has been removed by the AC coupling.
You might ask why this is necessary? Well, a PAL or NTSC signal is defined to have certain voltage levels and the input stages have some voltage limits they can deal with. A signal shifted to high might damage the input stage or might not get “decoded” properly. This is what is happening here, depending on the video signal, the signal behind the AC decoupling capacitor (blue) can still contain a DC offset when the capacitor is to small. This causes the “decoder” to misinterpret or miss the start of line voltage level and thus creating the artifacts we see.
In case you are curious what capacity Eachine used: I desoldered the capacitor and measured it: My (crappy) multimeter reads 23 uF, this looks quite small. BGJim recommends to replace it with 200uF. So let’s do it!
This modification is not that hard to do. It’s a pity that this was delivered to thousands of customers and was neither found during initial testing or by the beta reviewers. If you know some tricks this modification could be done by many people of the quadcopter building scene, but be warned, I do not know how Eachine will handle any warranty claims when you soldered stuff to the PCB on your own. Most likely this will void warranty. Anyway, here is a do-it-on-your-own risk tutorial:
 Things you need
Of course you will need a soldering Iron. In addition you will need a capacitor that is large enough. The exact type will probably not matter, I have seen tantalum, ceramic and electrolytic capacitors for AC decoupling in Chinese AV gear before. In my opinion for most of the users it would be the easiest way to solder a wired capacitor to the pads. I used a 220uF 16V electrolytic capacitor for this (e.g. one out of this set should do). You can also add the capacitor in parallel to the existing cap. There is plenty of room for the capacitor.
 Open the case
Unsnap the face plate and use a cross screwdriver to remove the screws on the bottom:
 Remove the PCB
Unscrew the PCB, open the flex cable connectors, and remove the flex cables.
Remove the fan on/off switch connector and remove the PCB.
 Desolder the capacitor
This step is optional, you could also leave the original one connected and solder the new one on top of the old one. Both capacitances will add. There are reports on rcgroups that 47-220uF in parallel fixed the issue.
Removing the old capacitor looks complicated at first, but with a neat trick it becomes be quite easy: Take a medium or big solder tip and add some tin. Now hold it on top of the original capacitor. Make sure to have contact with both sides. The capacitor will now get desoldered and most likely will stick to your soldering tip. If not, use a tweezer to remove it.
(Sorry for the bad picture, I noticed it to late that this is defocussed a “bit”…)
 Add the new capacitor
Now take your capacitor. BGJim recommends to have the minus side of the capacitor towards the incoming signal (remember, tantalum caps have marks on the positive side!). Most video AC coupling circuits I have seen (e.g. this one by maxim) connect the positive side towards the signal, see the plus sign in this picture. I am quite sure this does not really matter in this application as the voltage difference is only 0.5V. Electrolytic capacitors behave bipolar for voltages of less that 1.0-1.5 V.
 Check for shorts & Solder blobs
You are almost done. Do a short visual inspection for any shorts or soldering blobs that might got distributed during your soldering.
Re-assemble the goggles and enjoy your next FPV session without those annoying gray bars. That’s all for now.
I am still investigating the inner workings of this FPV goggle. However, right now I concentrate on finishing my new 16x16mm ESC design. You can expect an all new micro brushless quadcopter design similar to my 45g 2S pepperFIISH design soon. Stay tuned, you will not regret it! If you liked this blog post and want to support future work you can use this link to visit banggood the next time you go shopping. This is an affiliate link, this does not cost you a cent but I will get a small provision from banggood whenever you make a purchase after you clicked that link. This helps me to acquire new stuff and write tutorials like this. Thanks a lot for your support!
A list of previous and newer posts regarding the EV100 can be found in this list:
- Eachine EV100 component analysis: Display module supplier and technical data — UPDATE (12/3/2017)
- Eachine EV100: no audio problem — cause, fix, and technical background — UPDATE (10/24/2017)
- Eachine EV100: no audio problem — cause, fix, and technical background (10/20/2017)
- Eachine EV100 component analysis: Video receiver module identified — Sinopine SP338RX (10/12/2017)
- Eachine EV100: gray bar & distorted image problem — cause, fix, and technical background (10/11/2017)
- Eachine EV100 component analysis: Video decoder chipset identified — MST706 (9/27/2017)
- Eachine EV100 diversity: to be, or not to be, that is the question! (9/22/2017)
- Eachine EV100 component analysis: Display module supplier and technical data (9/20/2017)
- Eachine EV100 $99 FPV goggle — Disassembly, Components & PCB pictures (9/18/2017)