Although the freewheel teeth may look perfect, worn sprockets (and particularly older freewheel sprockets with relatively tall teeth) are the most likely culprit.
Even the slightest wear at the base of the teeth allows the chain to run in a slightly advanced position relative to the teeth, moreso under load, so that the rollers can just clip the corners of the driven side of the teeth as each roller is starting to engage between the teeth after leaving the derailer pullies.
If the roller hits the corner of the tooth and hangs there long enough for the next roller to begin to engage, that roller and the next few that follow will invariably all rest atop the next few teeth.
As the lifted section of chain rotates around toward the fully tensed upper run of the chain, slippage will occur at this time until rollers start falling between the teeth once more.
Although slippage is the symptom here, the real problem is failure to engage. This happens more often with increasing chain tension not because the chain "slips under power" but because increased tension advances the position of the chain relative to the sprocket (due to the minute flex between the drive force path from side plate to sprocket teeth, i.e. through the rollers) and the chain starts failing to engage.
A good test if this is the problem would be to note that the problem happens in some gears but not others, since the sprockets almost never wear equally in normal use.
I have tested this theory thoroughly. By grinding a ~1mm bevel at the tip corner of the driven side of each tooth (Takes a minute or two using bench grinder or Dremel) I have eliminated chain skipping from many a used sprocket that wouldn't sustain drive with a newer chain. To date, I have salvaged large numbers of sprockets. I try to keep the bevel at a 30-degree angle to the driven surface of each tooth, which has a maximal effect on allowing too-advanced rollers to clear the corners and fall in between teeth.
Heavily worn teeth which lean foreward from a high-riding, excessively worn chain cannot be restored. The mechanisms for "slippage" in such cases are actual slippage rather than failure-to-engage.
While the bench grinder is the fastest method, holding an individual sprocket in your hand, a Dremel with a 3/8" grinding stone can be used on fully-assembled freewheels and cassettes, even with them still on the bike!
After 15 hours of resto on my Raleigh Pro last Summer, I headed out for it's first major outing one Saturday morning, riding to an out-of-town training ride. The first time I hit the 13t cog under load, slippage commenced and I headed straight home. Without any disassembly, I did the Dremel treatment, leaving the grinding dust still suck to the used freewheel and new chain, and headed back out. I made it to the start in time for coffee, 18 miles away, and have been riding slip-free ever since.
Always think before discarding hard-to-find sprockets!
I often service mountain bikes and hybrids that get ridden for long miles in the top two cogs, and with these the small cogs must be "dressed" with the first chain replacement to prevent slipping under even the lightest loads. Any second-time-around chain replacement usually gets by with the previous grinding exercise, meaning the rider can wear out 3 chains on the original sprockets, saving money. The littlest sprockets can even wear faster than the original chain to the point where the not-too-worn chain gives slippage, but judicious grinding can usually take care of this.
While I found older chains rarely but occasionally had stiff links from the factory, I have found modern chains to be assembled with even better tolerances. When I get a stiff link using modern chain it's either the attachment pin (the one I pressed in) or one of the links got traumatized in use, usually the price of my impulsive off-road meanderings. For a quick diagnosis of this it's usually easiest to get off the bike and watch the backpedaled chain through several passes through the rear derailer cage. The chain bends sharpest there, so a stiff link is usually visable. In use, a tight, kinked link exiting the derailer cage will, you guessed it, cause failure-to-engage.
David Snyder
Auburn, CA
> My practice with new chainsaw chains is to soak them in a bath of light
> oil before installing. I read about it somewhere. If I didn't mind the
> additional time that would be required to wipe down a
> bike chain, I'd consider doing it to that too. A chainsaw chain tight
> link will soon loosen up running at 12000 rpms, but a bike chain might
> require considerable mileage to work smooth. Then there is the additional
> cost of the liquid refreshment along the way. Jerry's idea is probably
> better.
>
> Dennis Young
> Hotaka, Japan