[picture of RadeonLE w/TV-out] [zoomin of RadeonLE RF-section]

I finally removed the last 3 inductors, thereby completely bypassing the output VGA RF-filter. Unfortunately, there is scant visible improvement, but I sleep better at night knowing I've done all I can. :)

In other news, I have found an apparent limitation of the Radeon's overlay controller. At 1600x1200x32bpp 60Hz, all is fine. Video playback, whether DVD or regular AVI/MOV/MPEG, works fine. Moving up to 75Hz, which is the highest I can squeeze out of my CTX 17" PR710, the video-window suffers from some horizontal pixelation and fuzziness. Upon closer inspection, the video-pixels in the playback window, are updated 2 screen pixels apart, horizontally. But if I zoom the playwindow 2.0x horizontally, the effect disappears, and the video-resolution/quality looks normal. What does this mean? At desktop video modes equal to or higher than 1600x1200 75Hz, the RadeonLE overlay-controller enforces a 'minimum horizontal stretch factor' value of 2.0x. For proper full-resolution playback, the playback window must be zoomed (horizontally) at least 2.0x.

I removed the remaining three capacitors from the RadeonLE's RF/filter, to no avail. As you'll recall, the Radeon/LE's RF-out section comprised 6 capacitors and 6 inductors. Thus far, all 6 capacitors are gone. 3 inductors are gone, with the remiaining 3 shorted out (but still mounted on the board.) Looks like I'll have to entirely desolder those last 3 inductors. It's fun to practice soldering skills on an expensive new toy. Please note, at least 2 versions of RadeonLE exist. (1) TV-out (Rage Theater chip), and (2) VGA-only (no TV-out.) The TV-out version has a Rage Theater chip, and is the version I own.

First thing I did was stare at the board for a few seconds. I bought the TV-out version, which has S-video and composite video out jacks, both powered by ATI's RageTheater chip. Moving on to the board's RF-filter, the entire filter section comprised 6 capacitors and 6 inductors. That's 12 more discrete components than I wanted. Nevertheless, the board went straight into my PC. After driver installation and a bit of refresh rate tweaking, the card was roughly as sharp as my RF-modified Savage4. While not bad, it could probably be better. Moving some windows around on the desktop was silky smooth, even at 1600x1200 32bpp. On paper, the DDR memory-interface on the RadeonLE nearly doubles the bandwidth on a Geforce2/MX, but both felt like unrestricted scrollchamps.

The RadeonLE also supported more refresh rate options. I upped the 1024x768 desktop refresh rate from 85Hz (on the Savage3D/4), to 100Hz. At 100Hz, the display lost a bit of definition. Could it monitor-related? Recalling my experience with a modified Cardexpert Geforce2/MX, I knew it was the Radeon/LE.

The following day, I took the board to work, where I could use my workplace's soldering-iron to 'fix' the RF-filter good. I shorted all 6 inductors. Actually, I soldered over 3 inductors to short them out. The other three inductors I removed from the board, then shorted the trace-sites under each inductor's terminals. I told other people not to practice soldering/unsoldering surface-mount components on expensive equipment, but hey, live and learn. I now know that tiny surface-mount components wiggle around if I apply just enough solder to cover the target component. In my previous VGA modification jobs, the components were a bit larger, and I the solder-tip was only large enough to cover one terminal.

After a hard day at work, soldering my RadoenLE, I returned home to test the 'purified' board. Upon booting up, scant imporvement at 1024x768 100Hz. Switching out to 1152x864 100Hz, 1280x1024 85Hz, and 1600x1200 75Hz (to push the RAMDAC higher) showed some sharpness gain, but still behind the modified Cardexpert Geforce2/MX. A quick three twists with $2 Sears pliers, and the RadeonLE was lberated from 3 capacitors.

Reinstalling the board, the improvement was more pronounced. With the RF-filter down to just 3-capacitors, 1024x768 100Hz looked like 1024x768 85Hz before. This is good news, because I'd like to use the higher-refresh rate as my default Windows desktop mode.

Now, moving on to actual Windows apps. I launched the included ATI DVD player, Mutimedia Center version 7.0, and put in Princess Mononoke. Playback was smooth, as expected. Image-quality was good. I'm familiar with ATI's claims of 9-bit motion-compensation framebuffer (versus the usual 8-bits), but without a side-by-side frame-by-frame comparison, the Radeon's VGA DVD-output looks identical to the Rage128 Pro. I saw a few quirks with the ATIDVD software (Cinemaster.) When enabling the subtitles, sometimes the subtitles would appear with no smoothing, so the subtitles looked very blocky. Other times, the subtitles were filtered and blended with the background DVD video (like the Rage128Pro, Savage3D/4, Geforce.)

In Halflife Counterstrike, the Radeon exhibited no display glitches (none that I could see.) Obviously, the Radeon's superior 3D-core my previous Savage4, there's just no comparison. The Stealth III was silky smooth at 640x480x32bpp 160Hz, and smooth at 800x600x32bpp 85Hz.

I still need to install and play "No One Lives Forever." It's been sitting on the shelf ever since I obsessed with DVD-playback on low-end VGA cards.

The RadeonLE might be considered a 'value-edition' of ATI's Radeon 32MB DDR. The RadeonLE has all the features of the Radeon 32MB DDR, except for the 'HyperZ' technology. Other differences include a slower clockrate of 150MHz (for the RadeonLE.) The retail Radeon 32MB boards are clocked at 166MHz, and the 64MB Vivo boards are clocked at 183MHz. For more information, please the following links :

www.rage3d.com - Latest ATI news, tweakfiles, etc.

For ATI's offical RadeonLE FAQ, click here.