thermionic
Well-known member
- Joined
- Jun 3, 2004
- Messages
- 1,671
Hi,
A while back I seem to remember someone here (Jakob maybe?) say that the ubiquitous SSM2142 line driver didn't like being connected to discernable DC offsets and was prone to fail if so treated. I've recently had to replace a couple of non-working SSM2142s. I have no idea what prompted them to fail, but one was completely dead and the other was (slightly bizarrely) outputting 1dB below what it should have (can't remember if THD was raised, but I double-checked and the output jumped by a dB when I subbed a new IC).
So...this has made me wonder about whether specifying the 2142 in a new design is a good move? Would a more rugged solution be (albeit less convenient) to lash up a balanced buffer from a couple of, say, OPA2134? The 2142's spec is decent, but I doubt a buffer comprised from the 2134 would measure any worse - in fact, maybe better.
Have you seen many 2142 failures? In some designs where the other ICs aren't socketed, the 2142s have been socketed...an admission they're prone to failure, or just a pragmatic decision, bearing in mind they are the interface with the outside world and may get connected to all kinds of unsavoury equipment?
Thanks in advance.
edit - Oops! Dropping balls today... Meant to post this in the DB... Please move if it's convenient.
A while back I seem to remember someone here (Jakob maybe?) say that the ubiquitous SSM2142 line driver didn't like being connected to discernable DC offsets and was prone to fail if so treated. I've recently had to replace a couple of non-working SSM2142s. I have no idea what prompted them to fail, but one was completely dead and the other was (slightly bizarrely) outputting 1dB below what it should have (can't remember if THD was raised, but I double-checked and the output jumped by a dB when I subbed a new IC).
So...this has made me wonder about whether specifying the 2142 in a new design is a good move? Would a more rugged solution be (albeit less convenient) to lash up a balanced buffer from a couple of, say, OPA2134? The 2142's spec is decent, but I doubt a buffer comprised from the 2134 would measure any worse - in fact, maybe better.
Have you seen many 2142 failures? In some designs where the other ICs aren't socketed, the 2142s have been socketed...an admission they're prone to failure, or just a pragmatic decision, bearing in mind they are the interface with the outside world and may get connected to all kinds of unsavoury equipment?
Thanks in advance.
edit - Oops! Dropping balls today... Meant to post this in the DB... Please move if it's convenient.