20 replies to this topic

[Old Man Mike]

I've been doing a lot of testing of various Chinese RX and TX 1280 Mhz systems for FPV. As part of that testing I built a test fixture which shows the FPV video quality as the signal levels is reduced from -80dBm to below -100dBm. Here is a video showing what it looks like with one of the better receivers in which I had replaced the SAW filter:



For comparisons I use the lowest level where I can read all the OSD level as the defined sensitivity for the receiver. For this receiver I marked it as -95 dBm. Notice how the 1Khz reference audio signal drops out about the same time as the video.

By the way, it is not an easy thing to make accurate measurements at these low signal levels. I wanted to be sure that I could do this with an absolute accuracy of less than 1 dB. Here's a quick summary of the process:

1) A 60mw Video TX was used as the source. It was mounted inside an RF tight box with high quality chassis mounted bypass filters for the video, sound and power inputs. Copper Tape was used to reduce ground/RF leakage between the 60mw TX and the inside of the box. In-line attenuators were used inside the box to bring the signal down to -23 dBm at the SMA output of the box.

2) The -23 dBm level was verified with a well calibrated HP8594 Sectrum Analyzer. This was measured without modulation.

3) A PIC controller was used with a programmable step attentuator to provide the final -80 dBm to -100 dBm signal at the receiver input. Double shield cables were used for all RF connections. The PIC display was set to show the calibrated signal level at the receiver and setup so that the FPV camera could see it as the signal was reduced.

Notice that the video sync is maintained all the way down to -97 dBm where the signal where the image is just barely visible. To achieve this you must have the TX and RX video modulation and demodulations set to the correct levels. For that you need a good scope which can accurately show you the video levels.

One big discovery is that unlike the Lawmate receivers, all the Chinese transmitters I tested had a high input resistance rather than the standard 75 ohms for the video input. This really destroys weak signal reception because the TX will often be overdriven by the video camera expecting to see a 75 ohm load. That will cause the sync to drop out at higher signal levels well before major video degradation. For those of you that may have experience this, try adding a 75 ohm resistor in parallel with the video input of the transmitter.

OMM

Edited by Old Man Mike, 14 April 2010 - 10:17 AM.

[Terry]

Interesting test, just the sort of thing we like to see here, thanks

Terry

[Mr.RC-Cam]

-95dBm is fantastic. Is the tuner module inside the Rx a generic OEM no-name, or something with a recognizable name on it?

It's also nice that the audio drops out at the same time as the video. The usual behavior I've seen is that audio is normally lost before the video. Having them stay together is a treat.

The one thing I learned from testing Chinese A/V receivers is that out of the box, no two seemed to be the same, even "identical" models. So, if you have a duplicate Rx, test it too to see if its performance is the same.

[saabguyspg]

Hi Mike, thanks for this!

I have a few dumb questions if you don't mind.

First are you saying to put a 75ohm resistor in parallel on the video wire from the camera to the video TX? Would the video signal not just take the path of least resistance and not go through the resistor?

second I have discovered that my 5V video transmitter and camera does not work well at 5.0v. It seems like it is overpowering the system as you describe here.

I have hooked up an anyvolt micro and reduced the voltage down to 4.1v and the video is better. It sounds like I am doing the same thing that the resistor is doing but in a different way. I would like to install the resistor so that I can run my TX on 5.0v again.

I have attached a sketch, am I doing this correctly? Should the resistor not be inline like sketch #2?

Finally is there something we can do at the receiver end to clean up the video so I don't have to add resistors to all my video TX's?

I apologise for the green questions ;-) and thank you in advance!

-95dBm is fantastic. Is the tuner module inside the Rx a generic OEM no-name, or something with a recognizable name on it?

It's also nice that the audio drops out at the same time as the video. The usual behavior I've seen is that audio is normally lost before the video. Having them stay together is a treat.

The one thing I learned from testing Chinese A/V receivers is that out of the box, no two seemed to be the same, even "identical" models. So, if you have a duplicate Rx, test it too to see if its performance is the same.

Attached Files

•  resistor location.PDF   17.44KB   350 downloads

[Kilrah]

Great testing

I have attached a sketch, am I doing this correctly? Should the resistor not be inline like sketch #2?

In parallel with video input = between video and ground. So none of your sketches fit

Finally is there something we can do at the receiver end to clean up the video so I don't have to add resistors to all my video TX's?

No. As the OP explained, having optimal performance requires the whole chain to be correct. And it would take a lot more work to open each of your receivers and adjust them, for subpar results, than to quickly add a resistor on your TX

Edited by Kilrah, 01 April 2010 - 09:41 AM.

[saabguyspg]

In parallel with video input = between video and ground. So none of your sketches fit


ah well that makes sense now!! hahah... thanks guys. I will give this a shot.

[Old Man Mike]

-95dBm is fantastic. Is the tuner module inside the Rx a generic OEM no-name, or something with a recognizable name on it?

It's also nice that the audio drops out at the same time as the video. The usual behavior I've seen is that audio is normally lost before the video. Having them stay together is a treat.

The one thing I learned from testing Chinese A/V receivers is that out of the box, no two seemed to be the same, even "identical" models. So, if you have a duplicate Rx, test it too to see if its performance is the same.

I've bought quite a few Chinese RX/TX units in addition to several LawMate units to get an idea of the variability of these systems. Here is a sample group:



The older RX units had adjustments for both Audio and Video gain:



The newer RX units have only the audio adjustment.

As for the tuner, all Chinese and the LawMate receivers appear to use the same unit:



The SAW filter is marked F480-3 which is 27 Mhz wide.

The sensitivity level measured as shown in my first post and before replacing the SAW filter ranged from -86 dBm to -96 dBm. The -96 dBm unit is quite interesting since it has a better sensitivity then the two receivers with replaced SAW filters. They both came in at -95 dBm which was about 3 to 4 dB better than where they started. I've been a bit reluctant to take the -96 dBm unit apart since I'm using it as a reference. But I need to see if perhaps the Chinese have started producing some units with the more narrow SAW filters. It is hard to believe that a 27 Mhz wide receiver could have -95 dBm sensitivity. I've even tested the NF to make sure that was not the reason for improved sensitivity so it almost has to have a narrow filter.

OMM

[Mr.RC-Cam]

I've been a bit reluctant to take the -96 dBm unit apart since I'm using it as a reference.

For the sake of science, pop the tuner's lid (painless work) and check the SAW's part number.

Your sensitivity numbers are averaging more than 6dB higher than what I have measured on several different 900MHz Rx's (1.2GHz Rx designs are the same, just different firmware). So, either you have some dynamite tuner modules from China (which would be a welcome surprise) or there are differences in our test setups. But the fact that you are seeing substantially higher gain in the special Rx is very cool.

On some Rx's (and/or their companion Tx) I've seen some interesting things happen when the frequency is switched. Does the gain or video level change when you change the channel? Nearly every system I've tested has at least some variation in video levels as channels change, but some of the Chinese models I validated were simply horrible at this. Especially those that provided 8 or more different channels.

Have you field tested the hot (-96dBm) Rx to see if it provides 3X more range than the common -85dBm module equipped Rx? If it can do this then gold plate that puppy and put it on the trophy shelf.

[Old Man Mike]

For the sake of science, pop the tuner's lid (painless work) and check the SAW's part number.

I'll eventually do it after I've completed other testing (lots of balls in the air right now). My worry is that if it shows the wide SAW filter then I will just have to replace it to see if it will do better than -96 dBm.

Your sensitivity numbers are averaging more than 6dB higher than what I have measured on several different 900MHz Rx's (1.2GHz Rx designs are the same, just different firmware). So, either you have some dynamite tuner modules from China (which would be a welcome surprise) or there are differences in our test setups. But the fact that you are seeing substantially higher gain in the special Rx is very cool.

Yours are 6dB worse than -86 dBm or are you saying 6dB worse than -96 dBm? Over half of my samples were -90 dBm or worse.

On some Rx's (and/or their companion Tx) I've seen some interesting things happen when the frequency is switched. Does the gain or video level change when you change the channel? Nearly every system I've tested has at least some variation in video levels as channels change, but some of the Chinese models I validated were simply horrible at this. Especially those that provided 8 or more different channels.

Sorry but I've only been working with the 1280 Mhz channel so I have no data on video levels for other channels.

Have you field tested the hot (-96dBm) Rx to see if it provides 3X more range than the common -85dBm module equipped Rx? If it can do this then gold plate that puppy and put it on the trophy shelf.

No. But what I did today was to setup my -23 dBm reference source with a 2 dBi antenna 200 feet from the receiver. At that distance the signal was snowy but quite visible (a little better than the picture in my first post) using the -96 dBm receiver. I switched out the receiver with a -90 dBm receiver and nothing was visible. By the way, using freespace Pathloss calculation and including the antenna & feedline losses, the receive level calculated out at -94 dBm which is nice secondary confirmation of the sensitivity measurement. Actually I was a bit surprised that it came out that close.

My efforts have been centered on developing a multipath resistive system. The LOS range of my current system is many miles beyond what I would ever fly. My personal challenge is to be able to fly in a non LOS area where the signal has to travel thru 500 feet of trees. The 30 dB of attenuation can be overcome but the real killer is the multipath. I've finally finished some custom circular polarized antennas with a high axial ratio (VERY important) which look promising.

OMM

[Terry]

Fantastic thread, I look forward to hearing what comes out of this.
Also the circular polarized aerial is something I have never managed to get, even my helical has a prefered 'up' but I have no way of telling just how oval it is.

Terry

[Mr.RC-Cam]

Yours are 6dB worse than -86 dBm or are you saying 6dB worse than -96 dBm? Over half of my samples were -90 dBm or worse.

I tested a couple dozen Chinese Rx's looking for something worthy. Some Rx's were complete junk, but those that weren't obvious dogs ranged from about -83dBm to -86dBm (stock, no mods). I would have been tickled pink to see the numbers you are measuring.

Some of the China models I evaluated were messy hand assembled, perhaps in the worker's kitchen. Poor construction is one of the reasons the performance often varies with the cheap imported devices (even when they are from the same vendor). With two identical purchases, one guy gets a good one, the other guy gets a dog. I seem to always be the other guy.

[Old Man Mike]

Fantastic thread, I look forward to hearing what comes out of this.
Also the circular polarized aerial is something I have never managed to get, even my helical has a prefered 'up' but I have no way of telling just how oval it is.

Terry

Hi Terry,

Circular polarization is one of the most misunderstood items I've seen in the various RC forums. Many people think that it refers to a "circular" Az pattern (it does not). Some think that it is just useful because it responds to both horizontal or vertical polarization (true but with a 3dB loss). The real power of circular polarization for FPV video is it's unique ability to reject multipath when BOTH aircraft and base stations are running circular polarization.

To understand this first consider that multipath occurs when the reflected wave cancles the primary wave. Many objects including the ground, buildings and trees will reflect RF waves. When a linear polarized signal like the typical vertical polarization from a whip antenna is reflected or "bounced" it remains vertical. When the direct wave and reflected wave arrive at the receiver it is likely out of phase so the signal is reduced. This effect often occurs when you are in a room or flying close to the receiver. The resulting in and out picture condition for what should be a strong signal area can be quite hard to understand for a person seeing it for the first time.

Circular polarization is actually a spiral type wavefront. Like a screw, it can be Right Hand Circular Polarization (RHCP) or Left Hand Circular Polarization (LHCP). A key point to understand is that a pure RHCP signal cannot be received by a pure LHCP antenna. Now consider what happens when a LHCP signal bounces off the ground or a tree: it reverses direction or becomes a RHCP signal. If you are using a LHCP transmit antenna and a LHCP receive antenna, the reflected signals will be RHCP and therefore will not be seen by the LHCP antenna so there is no multipath interference!

In practice nothing is perfect so multipath interference is still possible. A key factor is how much rejection a circular polarized antenna will have for the opposite circular polarization. The ratio of LHCP to RHCP is called the Axial Ratio. As you can imagine, this is a very important factor in the antenna design if you want to use it for multipath rejection. There can still be a double bounce condition which will end up reversing the waveform back to the original polarization but this happens a lot less often than the more typical single bounce.

It is also worth noting that the very popular diversity systems do not perform that well aginst multipath. The reason is that the video TX signal is wideband so all areas of the waveform (video, color burst, sync, etc) are not effected equally by the multipath interference. This often results in a very distorted video signal that still has sync so the diversity box does not know to switch to the other channel. It is interesting to consider that a specialized form of processing might be able to recognized the unique multipath distortion and make the decision to switch channels. For now the use of circular polarization to reduce the amount of multipath interference is probably the best approach. The biggest problem has been designing an omni circular polarized antenna with a good axial ratio for the airborne platform. After months of modeling, building and testing I finally have a design that is working very well in actual flight testing.

Sorry for getting a little pedantic with this post but I thought that with the generally higher technical level of this group it might be acceptable. More details on the circular polarized antenna designs and my other quadcopter efforts can be seen in this thread:

http://www.rcgroups....d.php?t=1147430

OMM

[Terry]

Hi Mike, I did know about circular polarized aerials since I made my first one about 10 years ago which was posted on the RC-CAM forum before this one. I had great success with my first 16 turn helical which transformed my then pethetic 2.4Ghz system into a useable system and became even better when I added a low noise pre-amp.

My point is that I still had to mark the helical with an up arrow to get the best from it, may be down to dodgy building I dont know.

Yes an omni for the plane is very hard to do but a 4 turn helical works well on the plane if pointing down for high alt flight.

Good to have your input here


Terry

Edited by Terry, 03 April 2010 - 01:46 AM.

[Old Man Mike]

Hi Mike, I did know about circular polarized aerials since I made my first one about 10 years ago which was posted on the RC-CAM forum before this one. I had great success with my first 16 turn helical which transformed my then pethetic 2.4Ghz system into a useable system and became even better when I added a low noise pre-amp.

My point is that I still had to mark the helical with an up arrow to get the best from it, may be down to dodgy building I dont know.

Yes an omni for the plane is very hard to do but a 4 turn helical works well on the plane if pointing down for high alt flight.

Good to have you input here


Terry

OK Terry. Good to meet someone with some real antenna experience. I designed my first helix for detection of the Mars Observer back in 1996. It was a 23 turn 10 foot helix for 435 Mhz. The receiving system used with the helix antennas was able to detect signals down to -179 dBm! I have a confirmation card from NASA confirming that I was able to detect the spacecraft 1.3 watt beacon when it was 5.6 million Km from earth. Here is the info:

http://www.af9y.com/radio40.htm

A 4 turn helical would have a little over 9 dBic. The problem using it on the plane is that the gain would fall below 0 dBic for an angle of 45 degrees or less. The flying rule for that system would be simple: Ground Distance should be no greater than Altitude.

OMM

[Terry]

Thats very impressive Mike, well beyond my attempts.

Yes the 4t helical was always flown with much more Altitude than ground distance. Ground distance was always under 1/4 mile but Altitude was hmmm, lets say quite high

I have always wanted to try horrizontal polarized omni aerials but DIY is the only way to go and I dont have the time to do the job right so its still a dream.


Terry

[Old Man Mike]

Here's a very interesting demonstration showing how a low noise preamp can provide a greater S/N improvement than expected. In the video below, the left side shows a receive signal that is only barely visible and certainly not useable. This signal was received from a distant location by my CPOD antenna with a 4 dB loss coax run to the receiver:



A 13 dB, 1.5 dB NF preamp was used at the receiver in the left video and then at the antenna for the right video.
(that secondary OSD bargraph is jumping around because it is injected at the receive side and sensitive to sync noise)

Now consider that all we did when moving the preamp from the receiver to the antenna was to eliminate the 4 dB of coax loss which is the same as reducing the effective NF of 5.5 dB (Coax loss + Preamp NF) down to 1.5 (Only Preamp NF). Looking at the video in my first post of this thread you can see that the apparent improvement is much more than 4 dB. It is at least 7 dB improvement when you consider that the left video is certainly no better than the -98 dBm level and the right is certainly no worse than the -91 dBm level. So how can this be?

The answer is that at these very high frequencies, a well designed antenna can see less noise than the 50 ohm attenuator used to measure the minimum receive signal. In this case, there is an extra 3 dB of improvement beyond the normally expected 4 dB. This effectively makes the minimum receive signal 3 dB lower. Bottom line: the measured -95 dBm minimum signal for useable video becomes -98 dBm when a 1.5dB NF preamp is used at the antenna. If I switch from the 1.5 dB NF preamp to a 0.5 dB NF preamp I should have a system capable of useable video down to -99 dBm!

OMM

[Terry]

Yep as I said it realy worked for me together with the helical. I had about 100ft range before the mods and about 1/2 mile after. My first 2.4Ghz RX was a real dog!

This is the preamp kit I used http://www.g3wdg.fre....uk/preamps.htm


Terry

[Mr.RC-Cam]

Thanks for posting the test information.

[liteglow]

oh..

I got the second system on your picture (The 800mW system that FAIL, and only give 60mW) !
Seems like that can be my problem... I only got 2-300 meeters with a >800mW< system.
That cant be correct...

Any suggestion what to buy when I now throw this junk of sunsky in the trash ? :-)

[Old Man Mike]

oh..

I got the second system on your picture (The 800mW system that FAIL, and only give 60mW) !
Seems like that can be my problem... I only got 2-300 meeters with a >800mW< system.
That cant be correct...

Any suggestion what to buy when I now throw this junk of sunsky in the trash ? :-)

Sorry to hear that. It certainly is a challenge to buy from China since you never know for sure what you will get. I got lucky with the 500 mw transmitters. I have now gone to all Lawmate for the TX (1 watt) and RX.