Minor corrections marked by *** change ***... Not particularly big changes
On 1/24/23 19:49, Jonathan Brandenburg via pacsat-dev wrote:
> Thank you for responding with your insight, Bob!
>
> On 1/24/23 19:15, Bob Stricklin via pacsat-dev wrote:
>>
>> First the Excel spread sheet I sent is a early look at currents
>> needed. Since I put that together some of the parts have changed and
>> some have been added.
>> I am sure we have a power issue but taking the position of just
>> trying to get everything we want done then we can back down on
>> capability and reduce power later.
>> There is not a limit or budget on power at this time.
> Thank you. I meant to mention I considered the spreadsheet a first
> order approximation but I may have missed that in my revisions.
>>
>> Each time you add one of these current monitors to the design you
>> introduce another part that can fail due to latch-up and other reasons.
>>
>> The action taken for each monitor added may be different. Latch-ups
>> are possible from radiation exposure. These can be single event or
>> they can result in a hard failure of a part. When there is an event
>> and high current the plan may be to power down and wait for a period
>> of time and then try to restart. If it is the processor with an issue
>> then you are restarting everything if it is a sub circuit then you
>> may be able to do a quick recycle. There are different types of
>> current monitors to help you with your action plan. It may also be
>> necessary to build a subcircuit to get the results needed.
> We're not necessarily dealing with hard failure of a part with this
> current switch. We are specifically dealing with single-event upsets
> leading to latchup from a radiation effect that further results in
> unregulated power consumption. This result is considered transient and
> is resolved with a power cycle, hence the use of this ***type
> of***part in Fox and now Golf. From our recent experience, hard
> failure of a part seems relatively rare and we haven't had a recent
> satellite with batteries that lasted long enough to deal with total
> ionizing dose, for example. (I don't know for sure which AMSAT
> satellites used non-hardened integrated circuits and thus would be
> ***suseptible*** to that affect.)
>> <snip>
>>
>> I worked on optical ICs and since these were exposed to light we had
>> to be careful not create an issue with latch-up. When a new design
>> comes out of wafer fab it is one of the early test you do to see if
>> you have issues. If you find a problem you have try and fix it by
>> changing the die layout, adding more metal or modify the circuit.
>> When a device is “radiation harden” this should also be done and
>> hopefully the TMS570 had this done. Still could fail with radiation
>> though.
>
> One thing to point out... I don't believe the TMS570 is radiation
> hardened. I understand it's used in safety critical equipment and has
> special circuitry to detect failure modes. But I wouldn't expect it to
> be immune to single-event upsets. In the case of bit flips that impact
> processing, the TMS570 could detect that as a failure when comparing
> the results of the two cores and assert a failure. In the case of the
> RT-IHU this would result in failover to the mirror processor. In the
> case of the PACSAT payload, which I believe is running a single
> TMS570, the failure line could be tied to the power circuit to reset.
> If the power circuity of the TMS570 suffers a single-event upset that
> latches up a power rail I'd expect we'll depend on the current switch
> to detect and recycle power to recover. (On a related topic, it's
> pretty fascinating to examine the Fox telemetry and observe the impact
> of the SAA. I don't know if Fox reset every time it traversed the SAA
> but it was quite impactful.)
>
> As long as we're talking about radiation affects, nothing we're doing
> will mitigate total radiation affects that will ultimately degrade and
> cause failure of our chips.
>
> Jonathan
>
--
Jonathan Brandenburg
Radio Amateur Satellite Corporation
1-214-213-1066