46 replies to this topic

[Mr.RC-Cam]

Inside the popular wireless A/V receivers is a SAW type bandpass filter. The filter is in the Rx's I.F. (intermediate frequency) stage and it ultimately determines the video bandwidth. The filter's bandwidth performance has to be sufficiently wide to avoid compromising the video quality, but narrow enough to provide a good signal-to-noise ratio. Generally speaking, wider bandwidth increases RF noise, which reduces receiver sensitivity. So making it too wide is a disadvantage.

As it turns out, the typical 2.4GHz A/V receiver uses a 17MHz wide filter. But the 900MHz-1.3GHz receivers have a 27MHz wide filter on their 480MHz I.F. stage. The wider bandwidth is not needed and theoretically would reduce receiver sensitivity by a couple dB.

Given the typical 4-5MHz video bandwidth, and the related audio sub-carrier that is located 5-6MHz from the main carrier, the filter only needs to be about 14MHz or so. As it so happens, the Ham ATV folks have been swapping out 27MHz filters with a easily obtained 17MHz equivalent, and this grants them a bit more receiver performance. I've seen other references to the filter mod in the past, but this ham club newsletter published last year is nicely done: http://comtech.hampu...ring_2008-2.pdf

The reason for the 27MHz filter is due to the RF tuner module used in the A/V receiver. Unlike the 2.4GHz and 5.8GHz A/V systems that were purposely designed for A/V applications, the 900MHz - 1.3GHz models are older designs that use a baseband RF tuner developed for the satellite receiver industry. The most popular supplier for the tuner module is Comtech, but there are an endless number of other offshore sources that seem to have copied each other. Not all offer good RF performance, so don't let anyone tell you that all wireless A/V receivers are the same. That's why all those eBay Hong Kong A/V receiver bargains are not always bargains after all.

So here's is where it gets confusing. Some A/V'ers have modified their 900MHz - 1.3GHz receivers by swapping out the filter and have seen noticeable improvement. Others have made the swap and reported that their video system worked worse. The theory suggests we should see about 2dB of sensitivity improvement and the more narrow bandwidth should help reduce interference from neighboring RF sources.

Or so the theory says. But what really happens? We'll, since no one else has bothered to measure and report some real numbers, I decided to do that. So, here is sort of a blog of my efforts, with measurements and a few photos.

[Mr.RC-Cam]

First thing to do is to measure a stock 900MHz receiver. So, I grabbed a brand new one (fresh from the box) and prepared to measure it. I needed an accurate 900MHz fixed RF source that had about -80dBm signal levels that could be adjusted.

My solution was to use a 900MHz / 500mW (+27dBm) A/V transmitter and install a fixed 30dBm attenuator on it. That gave me a nice -3dBm signal source, which I verified on my Spectrum Analyzer. Next I added a 110dB step attenuator. The net effect is that I now had a verified signal source that could go from -3dBm to -113dBm. That's plenty of range.

[Mr.RC-Cam]

I decided to measure a few things besides signal strength, just in case the filter swap causes some side affects.

I first tried to determine receiver sensitivity by attenuating the RF signal until I saw noise in the image. For the tests I used CH1 (910MHz) and the Tx based RF Source used a common color bar test pattern from a NTSC video generator. At about -83dBm the image started to show a tiny amount of snow. That was nice to see since a good quality RF tuner would have about -85dBm sensitivity and what I saw reflects that.

But, it became obvious to me that just trying to make measurements based on the amount of snow I saw on the video monitor would not allow for accurate data. So, I worked out a scheme that involved measuring the RSSI voltage when the AGC was at its weak signal limit. That would give me a hard number to compare.

Here is what I measured:
-78dBm: Max AGC, RSSI = ~3.3V
-75dBm: RSSI = 2.56V
-73dBm: RSSI = 2.37V

I also measured video levels (for reference) and the video bandwidth. I found that the video bandwidth was 4.9MHz. I've seen some wireless A/V systems that had as little as 3.8MHz of video bandwidth, which causes a loss in camera resolution. So, the 4.9MHz bandwidth is great to see here.

So, now I had something to compare to.

[Mr.RC-Cam]

Now it was time to replace the SAW filter in the 900MHz receiver. I've done a couple before, so this is not something I look forward to. Honestly, it is a tedious job.

Here is the original SAW filter. It's center frequency is about 480MHz and it has 27MHz bandwidth. It needs to come out.

[Mr.RC-Cam]

Removing the SAW was done with a 47W Weller solder station, my trusty hand solder pump tool, some desolder wick, and a lot of care. I've been in a few different tuner modules and this part's installation can vary, so some are easier to remove than others.

One cumbersome issue on this particular tuner is that there is more holding the SAW than the three leads soldered to the PCB. There is a solder bridge from the SAW case to the metal chassis (an important RF ground), which is not hard to remove. But what is not obvious is that the bottom of the SAW filter's case is also soldered to the PCB! That's right, under the part, were you cannot see, there are hidden PCB pads that are soldered too (these are RF grounds). Overall, this is not a task for someone that does not have a lot of PCB rework experience.

The old SAW filter is finally out, as seen here.

[Mr.RC-Cam]

Getting the new one in is easy. Just install it like the original. The SAW filter component I used is part number ECS-D480A. It is offered by several sources, including www.dpcav.com. That's where I work, so for me it was a quick walk to the stockroom to get one .

Even if the original SAW did not have the solder bridge from the SAW's case to the metal chassis frame, I highly recommend that you include it on every tuner that is upgraded (see arrow in photo). The filter is in a critical high frequency RF signal path and this extra RF ground is quite helpful to the SAW filter.

Before installing the new SAW filter, be sure to use proper PCB cleaning chemicals and remove the flux residue from the PCB's surfaces (both sides). RF performance can be affected by such things, so clean is best.

Keep in mind that the SAW is in a impedance controlled 480MHZ RF signal path. You can't connect it using hook-up wires or sockets. Install it directly on the PCB just like the original.

Also, if you drop the SAW filter on a hard surface then it will be damaged. Internally, this fragile part involves crystal construction and hard impacts are not a good thing to them. So, if yours has fallen on the tile floor, just sweep it into the trash can and grab another.

[Mr.RC-Cam]

With fingers crossed I measured the modified receiver. The news is good.

Video bandwidth was still 4.9MHz, so video performance (resolution, color, etc.) on this receiver will remain exactly the same. So, that shows that the reduced SAW filter bandwidth is still appropriate.

I found the video level had dropped by 10%, which was a puzzle. Then I noticed that the video level pot on the Rx's main PCB was very sensitive to movement and during the rework it must have moved from all the handling. Keep that in mind -- it's something to watch out for. It's a simple pot adjustment, so video amplitude was simply tweaked back to the original level.

But the best part is that the Rx had +3dB more gain. Before the modification the RSSI showed the AGC bottomed out at -78dBm. With the new SAW filter the same setup was measured at -81dBm. I have no idea if the improvement was due to the reduced bandwidth or by the higher quality SAW filter (I suspect it is a bit of both).

What does this mean? Well in this particular conversion, the upgraded receiver offers the same performance as would be achieved by doubling the transmitter's RF power! Not bad for one hours work and $4.

I have no idea why some modified receivers have reportedly worked worse than when they were stock. That certainly does not seem to jive with what I saw during this hack effort. I guess I should proclaim my good fortune with the customary "winner-winner-chicken-dinner" yell.

BTW, in case anyone should wonder: Sorry but I don't have the time to do this mod for others. Heck, if I had some spare time I would use it to go fly!

[Terry]

Thanks Thomas, a very interesting read. You got me thinking about my 1.3Ghz system, if only I didnt have 10 other projects to do
But I just had to have a peek at the filter anyway, it says F480-2.
I looked it up and its a 21Mhz filter, is it worth changing ?
Would a more narrow filter be worth fitting if some quality loss is acceptable ?

Terry

Edited by Terry, 09 July 2009 - 11:42 PM.

[Mr.RC-Cam]

I looked it up and its a 21Mhz filter, is it worth changing ?

In a perfect world, the improvement would be minor, perhaps +1dB. But, if the existing filter is a low quality component (good chance for that) then a swap would help more than usual.

Would a more narrow filter be worth fitting if some quality loss is acceptable ?

That would make an interesting experiment if you didn't make it too narrow. The main problem is finding other suitable filters. Although they are offered in many different bandwidth configurations, many suppliers require purchasing 1K pieces minimum.

EDIT: I just looked at the SAW filter I pulled out and it is also marked as a F480-2. According to the Comtech module's data sheet, these are 27MHz filters. So, there is good reason to believe you will achieve a sensitivity improvement that is similar to what I measured on my 900MHz Rx.

But you do bring up a good point. Maybe the reason some folks are not successful with the SAW hack is due to the specific filter that was originally used in the Rx. Depending on the supplier, their specs (insertion loss, I/O impedance, pinout order, etc.) may vary. Maybe we should use this thread to report about the modified Rx results and the SAW filter part numbers. That may help identify which Rx's are the best candidates for the ECS-D480 component.

Edited by Mr.RC-Cam, 10 July 2009 - 01:58 PM.
F480-2 is 27MHz?

[Terry]

Thanks Thomas that is interesting, I'm even more keen to see what happens now.
Is there any chance of finding a more norrow filter in another module I could pinch to use or are they all the same?

Terry

[Mr.RC-Cam]

These particular filters were developed for the satellite set top box Rx industry. So, finding a suitable part in the junk box is not likely. Besides, 27MHz seems to be the dominant choice.

I found the data sheet to a F480-2 (21MHz), which is probably the same one you saw. The F480-2 docs I read listed it as a dual channel IF filter for satellite receivers, but it has four signal pins. The "F480-2" in our A/V receivers is a single channel version with only three pins. So, they share the same part number, but are not the same.

[kist2001]

I have no extensive rework experience. Yet with some difficulty I was able to remove the stock filter.

A four step process is best:

Step one is to try and cajole a friend with extensive rework experience to do it for you.

Step two is to use solder braid and a solder sucker to remove all the solder from the three legs and the solder bridge above.

Step three is to use pincers to destroy the metal can of the filter. You need access to heat up the metal base which is attached to the PC board with solder.

Step four is to heat the metal base from above with a high wattage soldering iron while prying the base away from the PCB with a screwdriver. I used a large Weller designed for stained glass. I have heard of others using a hobby torch. You need more heat than you think to get it off.

I wish you the best of luck and few burnt fingers. -Ken

PS: I have not tried the modified receiver in the air yet.

[Mr.RC-Cam]

Glad to hear you got it out and swapped. To help in the data collection, can you post the Rx freq and model number, tuner module part number (usually stamped on the RF tuner's tin lid), and the number on the filter you removed?

I used a large Weller designed for stained glass. I have heard of others using a hobby torch. You need more heat than you think to get it off.

A hobby torch might not be the best pick for this rework. In one case I used a hot air wand (surface mount rework iron) to assist in the part removal, but these are temperature controlled tools that are designed for this kind of activity. Basically, what is needed is enough heat to remove the part, but not so much that the nearby SMD parts are accidentally reflowed or PCB damage occurs.

To those considering this mod, just be prepared for a worse case result (fancy looking paper weight). But as they say: no pain, no gain. Good luck!

[recep]

Hi! will that mod be good with double out put receiver. And does working with 60 w ruin the components.?
http://www.dpcav.com...o...=270&page=1

[Mr.RC-Cam]

will that mod be good with double out put receiver

The 900MHz dual video output Rx has a typical RF tuner module in it, so it is possible to upgrade the SAW filter component.

And does working with 60 w ruin the components.?

In good hands, it should be fine. Like the saying about guns, soldering irons don't kill components, people do.

The ideal soldering irons are those with a temperature controlled tip. I'm using 800°F.

[galaxiex]

The 900MHz dual video output Rx has a typical RF tuner module in it, so it is possible to upgrade the SAW filter component.


In good hands, it should be fine. Like the saying about guns, soldering irons don't kill components, people do.

The ideal soldering irons are those with a temperature controlled tip. I'm using 800°F.

Speaking of guns...

Would it be ok to use a Weller 120W soldering gun (not an iron) you know the one, big heavy "transformer in a pistol" looking thing.
To remove this filter?

Any danger of the strong AC field causing damage to other parts?

(I used to use these things to demagnetize tape heads, back in the day, the AC field at the tip was great for that...as long as you didn't let the tip touch the head....)

Thanks!

[Daniel Wee]

Thanks Thomas, for the test. I believe the gain in performance comes directly from the reduced bandwidth. I am however, exploring some filters with far lower IL (typically 4-5dB) which would represent a 16dB improvement. Unfortunately these filters need to be custom made but it would be really interesting to see the results of such a filter.

Daniel

[Terry]

Wow Daniel, that sounds fantastic. Keep us posted please

Terry

[Mr.RC-Cam]

I am however, exploring some filters with far lower IL (typically 4-5dB) which would represent a 16dB improvement.

Yes, please report how they work out. When do you think you will have the custom filters?

Would it be ok to use a Weller 120W soldering gun (not an iron) you know the one, big heavy "transformer in a pistol" looking thing. To remove this filter?

The magnetic field should not cause any issues. However, the Weller solder gun is not usually a suitable tool for use on electronic projects. It is probably better for use in repairing rain gutters. When it comes to PCB rework irons, bigger and/or hotter is not the ideal solution. The best thing is a decent temperature controlled iron and a bit of expertise behind it. So please seek hands-on help if you have any doubts on how to replace the SAW filter.

[galaxiex]

........


The magnetic field should not cause any issues. However, the Weller solder gun is not usually a suitable tool for use on electronic projects. It is probably better for use in repairing rain gutters. When it comes to PCB rework irons, bigger and/or hotter is not the ideal solution. The best thing is a decent temperature controlled iron and a bit of expertise behind it. So please seek hands-on help if you have any doubts on how to replace the SAW filter.

Thanks!
Happy to know the magnetic field should not cause any issues.
I have some experience with soldering electronics, mostly old school thru hole stuff.

I do realize the Weller gun is a very crude tool and not really suitable for this kind of work,
but I tried my 45watt iron with the largest tip I have, and it would not even begin to melt the solder on the can.

I was able to use the 45W iron to desolder the pins, and in the end I used the big gun to get the filter can loose.

I was very careful while using the 120W gun and as soon as solder started to melt into the wick, I lifted the tip away.
Carefully applying heat in this manner I was able to change the filter.

Got it all back together and the Rx still works!

I'm sure you think this is very "bush league amateur" way of doing things but hard to argue with success.

I have the Oracle and it now seems to prefer the signal from the modified Rx, so I will now change my other Rx filter.

Thanks again!