Transmit 32kHz TXO Clock signal over AM26C31/32 pair

Hi Neet,

Thanks for the updated information - I really appreciate it!

So with the propagation delay and rise/fall requirements of the system - I don't see any issue using this part - as you should still have more than enough margin on both propagation delay and transition times that it would be very unlikely you would see an issue. It always depends with clocked data - sometimes its very flexible like this and other times they need less very tight tens to single nanosecond response times. 

The driver has a propagation time of worst case 12ns (typically 7ns) and the receiver has a worst case propagation time of 27ns (typically 17ns) - and at 2-3meter the propagation speed across the a cat5 cable should be ~5.3ns/m - so around 10.6ns to 15.9ns across the cable. So a total propagation delay from source to intended load should be around like 60ns (I overestimated to compensate for a worst case cable) - which is well below the requirement that the MEMs needs in terms of propagation delay.

In terms of rise/fall time you only really care about the 32 device (receiver) as that is where rise/fall time is actually going to be measured when looking at the MEMs device. which has a worst case of 9ns (for commercial and industrial variants - auto and military versions have a worst case of 10ns) but all versions are looking at a typical of 4ns - so no real worry on the rise times.

The one main concern I see is that the MEMs circuit seems to have a VDD of 1.8V and a VIO of 1.8V - the AM device will be outputting 5V - so if the MEMs device can't handle 5V inputs some type of level translator or buffer that has 5V tolerant inputs to bring it down to the 1.8V level. This can be done discretely or using ICs. 

In terms of jitter - I don't really see a hard requirement on the datasheets you have provided, generally longer buses add jitter and pulse skew will add jitter to the system. The 31 device has a skew of 500ps typically and worst case 4ns - we don't spec it on the receiver, but it would be strange if it wasn't similar - essentially very low typically but could have some skew. However with the large range and of acceptable values I don't think this is actually that much of a risk factor for your system.  You avoid jitter by not pushing the data rate too fast for the bus length - but at 2-3 meters that really isn't going to be any concern to the system. Generally a Schmidt trigger buffer wouldn't help too much because if the periodicity is off (which is what the pulse skew/jitter really is) the buffer won't generally correct that  as it will just follow the transitions and generally will add its own skew. Generally an equalization network would be needed to correct for additive jitter - however I just don't see jitter being that large of an issue in this system - I say this because we do have one device with an integrated equalization feature and it really isn't doing much until you are pushing the bus hundreds of meters - not 3.

For the logic families you have shown - I don't really see too much benefit - the biggest issue I foresee is interfacing the 5V signal from the AM device to the 1.8V MEMs device - its needs to be down translated (unless MEMs inputs are 5V tolerance - but I am skeptical that is the case).  I understand that you are using an unused channel from your pre-existing AM device - but I do want to let you know about our other option that may be better suited as it will level translate down to 1.8V without any additional circuitry - the THVD2450V - when operating in 50Mbps mode (this doesn't mean you have to run at 50Mbps - but that it can because we aren't slew rate limiting the output will have similar to better transition times and propagation delays and slightly better skew times (and it is spec'd on receiver and driver sides). Essentially the THVD2450V would work similarly but has the benefit of having a separate I/O supply that will work at 1.8V and therefore any Single Ended signals will be between 0V and 1.8V - while VCC can be 3.3V or 5V (but best performance is VCC = 5V).

So ultimately I think the choice comes down to:

1. Keep AM device - the risk is low. Major concern is that you must down translate to ensure that the MEMs device isn't exposed to 5V signals (assuming it isn't 5V tolerant - if it is you can disregard this concern) 

2. Use THVD2450V which will allow 1.8V I/O supply - removing the need of a level translator - the down side is that it seems you already have an unused channel on your current AM devices - so this would be adding two new devices to the BOM of your system. The DRC package is pretty small - but I can understand why it may not be desirable to add more parts to BOM. 

Please let me know if you have any questions - but I think the AM devices should work in conjunction with a down-level translator but I do think the THVD2450V is a better device that wouldn't require the additional level translator - but I think ultimately they both could be used - just some tradeoffs discussed above that needs to be considered.

Best,

Parker Dodson