stan2004 posted:Norton posted:gunrunnerjohn posted:Pete, I see that Adrian beat me to it, 8V P-P is too low, that indicates that probably two of the four gates in the 244 have already failed. You should be getting around 14-15V P-P with a properly functioning TIU, see my previous post before and after replacing the 244 for the failing channel.
Ok, I rechecked mine using a transformer on Fixed 1. I get 12v P-P on Fixed 1 and 14v P-P on Fixed two which I never use. I expect the two variable channels are fine as week as they never get used. I will order parts to have on hand but hold tight until something breaks.
I have to say it seems odd that these devices fail gradually instead of with a bang.
Pete
Whoa! This is a new twist. So in your initial measurement (8V P-P) the 12V DC wall-wart was into Fixed1?
Anecdotally, DCS operates better on DC than AC. But in this sample of 1, it appears going from DC to AC increases the DCS signal level from 8V P-P (on all channels) to 12V P-P (Fixed1) and 14 P-P (Fixed2). Hmm.
As described in earlier posts, the ACT244 is the originator of the DCS-signal. That's a 5V IC chip. It does not matter what the TIU power source is (AC or DC). AC and DC supplies are or course not created equal, so it could be that your particular AC and DC supplies load down the DCS signal differently than two random other supplies. I don't think we want to go down this rabbit hole (yet). So to keep the conversation focused I think it should be specified that we are talking about an AC transformer power source so that everyone is measuring the same thing.
For comparison you probably all want to be measuring the unloaded condition (the Thevenin voltage). So power (lets keep it AC for the disucssion) flowing into the INPUT side, and only the scope/filter on the OUTPUT side. That's the condition I measure under.
Also I make sure not to measure the overshoots, otherwise everything becomes complicated. Instead I measure the flat-tops of the line code signal. The overshoots are a higher voltage but they don't exist across the entire symbol time so there's no guarantee they are there at the sampling instant in the decoder. Also they are a function of the frequency response of the layout, test cable or w/e.
Here's a figure showing what I mean:
Theoretically the best you can do with the ACT244 driver (5V Vdd) operating fully differential is 10V. Then driving the 1:1.414 (sqrt[2]) turn ratio output transformer gets you to 14V. If you are measuring more than 14V you are probably considering the overshoot portion of the waveform.
