Folks, What happens to the output of the ADUM1201CR if it receives an input pulse less than the minimum input pulse width (PW 40 ns) stated in the data sheet on page 9? Regards, Marco
The ADuM1200 contains a glitch filter on each input that rejects pulses narrower than the specified limit. You'll note that the typical minimum pulse width for the C grade is 20 ns, allowing for some margin at the maximum throughput rate of 25 Mbps.
I don't find specifications about this filter in datasheet, is it the refresh timer ?
It's separate from the refresh. In the simplified block diagram below you can see the refresh timer and glitch filter in the input path. The glitch filter rejects any signals narrower than the specified pulse width - they simply are not transmitted across the barrier. The refresh timer counts after each input edge and retransmits that state if another edge isn't received after approximately 1 ms.
The watchdog counts edges on the output side and will put the outputs into a default state if no signals are received
after approximately 3 ms.
I've tryed to replicate the issue and I've found a problem that point out my question, here below the trace :
TRACE C1: Input Signal, TRACE C2: output signal, Z1-Z2 Zoom of trace C1-C2
What is the problem ? The pulse duration is 23ns and glith filter must eliminate this signal
There's not a minimum pulse width specification for this product, so it's performing as expected. The typical minimum pulse width is 20 ns, which is approximately what you are observing. Some of our products do have guaranteed minimum pulse widths.
ok but in this case the device operate like a pulse amplifier.... is not the same of an opto coupler...
It seems me that the low level doesn't propagate correctly over the device because the signal from the pin is not qualified.
I need a device that filter input glitch, can you suggest me one ? Propagation delay is a must, typ I need 25nS.
The ADuM128x family has a guaranteed minimum pulse width of 10 ns and typical propagation delay of 18 ns (24 ns max).
Hi Dave, I've just tested same sample of ADuM1281 and seem works like ADUM1201..... I need a device that refresh output every 100-200ns otherwise I will transmit wrong information or alternatively level sensitive input....
Our refresh interval is approximately 1 us on all devices, nothing faster than that currently.
I don't think that refresh every 100-200ns can eliminate the 23ns glitch just like you shown above.
You need a filter can eliminate the 23ns pluse, It is impossible when you want to transmit 25M Hz signal.
Because the period is 40ns, so that 23ns pluse could be recognised with useful signal. (half period)
I don't think the glitch filter of icoupler can distinguish 23ns and 20ns.
I guess, in your application, the useful signal is less than 25M Hz.
You need to place a filter ahead the iCoupler for filtering the 23ns pluse when you used ADUM1201CR.
I think you can choose a lower speed one for this case if you signal frequency is lower. Such as ADUM1201BR.
If 10M Hz is too low. I think you must be use a filter in this design.
BTW, where is from the 23ns pluse? power supply interference? layout EM coupling? or other device power on?
Above is my personal opinion. FYI
Retrieving data ...