Alternatives to ONSemi NCP303LSN09?

[kster59]

I looked into this and think the NCP303LSN10 would work by setting the voltage of the core to 1.05v since XMOS l1-64 can run .95 to 1.05V. The max reset on the NCP303LSN10 is 1.02 so that would work with pretty minimal redesign. Just substitute a resistor or two to get the right voltage.

[sjalloq]

If you can guarantee that all your supplies are good to 2% accuracy then it will work.

The reason they use a 0V9 detector for a 1V0 rail is that it allows some leverage should the supplies not be bang on the money. The datasheet specifies a range of 0.98-1.02V for VDET- for the 1V0 detector. So, should your 1.05V supply be out of spec, or have any ripple, you only have 2.58% to play with as opposed to a full 10% when using the 0V9 detector.

You may be lucky though.

[kster59]

The .9V detector puts the device out of spec also on the low side.

The xmos chip is supposed to run .95-1.05 so if you drop to .9v then it is out of spec.

If I run 1.05V and it drops to 1V then it is ok.

I have reconfigured all my designs to run at 1.05V and then I can use either the .9V detector or the in stock 1V detector without any further changes.

The ripple is typically 1-2% on the chip I use so hopefully 1.05V+2% isn't too much out of spec to be damaging...

What does XMOS think?

[Corin]

We would not recommend the 1V0 voltage detector, because as pointed out there is not much margin to work with and you have to run the 1V0 bus at 1V05.

Another alternative is to use a 1V0 supply with a power good output and a delay that will trip the reset.

Regards,
Corin

[kster59]

What's the abs max voltage the XMOS can reliably take?

My only concern is running at 1.05V + any ripple. I can substitute the .9V when it comes back in stock but would like to leave the default voltage 1.05V instead of 1V.

I'm producing large quantities so would like high reliability but at the same time don't want to wait 6months for the .9V part to come back into stock.

[sjalloq]

The .9V detector puts the device out of spec also on the low side.

The xmos chip is supposed to run .95-1.05 so if you drop to .9v then it is out of spec.

That's not the point. You still run the power supply at 1V0 but use a 0V9 detector. The detector is only there to provide a power on reset so once the core voltage ramps through 0V9 on its way to 1V0, the reset will be de-asserted at time t after the voltage hits 0V9. Time t is dependent on the RC time constant set by the external capacitor.

Make sense?

[sjalloq]

I'm producing large quantities so would like high reliability but at the same time don't want to wait 6months for the .9V part to come back into stock.

If you are producing large quantities and I guess are planning on selling them then I would not build something that is out of spec. Worst case scenario is that you build a load of boards that work in your lab but have intermittent problems over temperature once shipped to customers. The possible problems and headaches are just not worth it IMHO.

[leon_heller]

It might be possible to use one of the tiny PIC10F200 MCUs as a replacement for the NCP303LSN09. They have a comparator and the delay function could be implemented in software.

[kster59]

The NCP303LSN09 costs $.146 from mouser in qty over 2000 and requires no programming. I doubt you can get the same thing with a microchip part. I'd replace it with something XMOS recommends that is also cheap and they tested in their design.

I would think XMOS doesn't want to wait around for 6 months selling no chips because someone else is out of stock.

The NCP part is supposed to offer 2 properties:
1) on reset wait a fixed time after reaching .9v
2) reset if it drops below .9v

.9V is out of running spec so if all you wanted was a fixed amount of time on reset and /forget/ the .9v monitoring then you can just add a reset delay or supervisor to the 3.3v power good line which drives the NC7WZ07.

So basically any of the NCP303LSN from 1v to 3.2V could be used on the power good line of the FAN2558S33x to provide the needed reset delay.

This gives a lot more options as supply from several manufacturers can make chips in that wider range.

But I would think the point is also to make use of (2) which resets the device if voltage drops too much.

I am also not sure why reset logic couldn't be built into the XMOS chips. Delays are pretty easily made up in hardware with a few gates and XMOS wouldn't need to rely on several outside manufacturers to provide power support chips.

It is what it is I guess.

[mon2]

The Rohm BU4209G is a pin and footprint compatible replacement part, but unfortunately that is also proving hard to source. There seems to be a general shortage of 0V9 voltage supervisors at the moment.

Regards,
Corin

Hi kster59 / Corin. Thanks for this posting. Invested a few days on this topic and found that the lead time for the Onsemi component is estimated to be Feb / 2011 !! We wish to test our design within a few weeks and found Corin's post on Rohm P/N # BU4209G. In chasing down this part through Rohm, the lead time is much better. There is no allocation issue with this version of the drop in replacement. We expect to order 2 T&Rs on Monday with an estimated lead time of 12-14 weeks (expected to improve after we place the order). Samples are being provided within 4 weeks for our testing of our design. This should align well with our prototype run schedule. We are using Future Electronics for our sourcing here in Canada..with assistance of the local Rohm rep (Sigma Components).
==========================
BU4209G-TR $.115ea 12-14wks mpq 3k reels
I ordered the rohm samples, no stock at factory, delivery is approx 4-5 weeks
==========================
Also review the Austria Microsystems P/N # AS1926 which looks like a winner (but NOT pin-for-pin compatible) with apparently short lead times. We are considering the same for another XMOS based platform since it also offers a manual reset button option (see Figure 9 in the datasheet). From their sample / online system, the AS1926-BSTT-D07 is available (0.81 volt threshold). Could this be used for the XMOS devices ? Or is the AS1926-BSTT-D08 recommended ? (0.855 volt threshold). Another idea was to just apply a timer that is guaranteed to exceed the time required for a power ramp up but not practical for those who already have the NCP303LSN09 footprint in place in their designs. Hope this helps ! :D

Regards
Kumar

PS: kster59 - Have you had a chance to review the schematic ? We are making some updates with a few corrections. Thanks.