LM3S9D92 revision a1

Hello Bill,

The errata sheet format has applicable bugs mentioned by revision.

Regards
Amit

Thanks for the response Amit.

The errata sheet I found was dated October 23, 2012/Rev. 2.2.

It mentions A1 in only 2 spots and they both have to do with flash memory.

Is this the only issue with A1 parts?  Is there another document that is better suited to my task of determining is the A1 parts are safe to use in production?

I ask because we were originally using the B92 (256K) part.  The early version of that part had an issue that required an additional voltage regulator to insure that the voltages came up in a specific order.  We did that but subsequently designed it out when we added WiFi to the product.  By then the newer step part was out and we didn't need the extra regulator.  

Is there any concerns like that in the A1 version of the D92?

Hello Bill,

Can you please attach the specific errata document to the forum post, so that both of us are on the same page.

Regards
Amit

Amit,

I don't how to attach a file but here is a link.

datasheet.octopart.com/LM3S9D92-IQC80-A2-Texas-Instruments-datasheet-15359586.pdf

Bill

Hello Bill,

In this case Revision A1 will also have the same issue as described in the errata.

Regards
Amit

Amit,

I'm sorry but your answer is a bit ambiguous. When you say that "revision A1 will also have the same issues as described in the errata" are referring to just the 2 issues mentioned in the errata I referenced? Or do you mean it will have the same A1 issues that the B92 part had? It seems to me that since the D92 came out well after the B92 part that those older issues I mentioned above such as the extra voltage regulator would have been fixed in the initial release of the D92 part. I apologize for being pedantic here but this is too important to get wrong.

Hello Bill

Issues are put by version of the device. So even if a device came later than another but the Revision Major-Minor code is earlier than the published errata, the errata shall be applicable in totality

Regards
Amit

Amit,
So you are saying that the A1 D92 parts have the same flaws as the A1 B92 parts. Is that right?
If so, then I submit that your errata for the D92 part is dangerously misleading since it does not mention all the problems in the A1 revision.
Please confirm this and I promise to leave you alone.
Bill

Hello Bill,

Sorry: I am confused here. You are referring to LM3S9D92, errata is for LM3S9D92 but now you are asking for the correlation with LM3S9B92.

Regards
Amit

Amit,

I guess we are both a little confused.
I believe you told me that that the A1 revision for both the B92 and the D92 parts share the same errata. That didn't sound right to me.

We can't find any D92 parts with revision A2. We have found some D92 parts but with A1 revision. All I want to know is, Is it safe to use them?

The D92 errata sheet doesn't have much information about A1 parts. But the B92 errata sheet does. When the project started, we designed in the work-arounds for the B92 suggested in the B92 errata sheet. However when the firmware grew beyond 256K we went to the D92 and we designed those work-arounds out of the design since we were using A2 chips.

So the question remains, is it safe to use D92 chips with revision A1?

Specifically I am asking about the extra voltage regulator that was required to bring up the VDDC signal. This is covered in the B92 errata sheet but is absent from the D92 errata sheet. If what you said is true, then the D92 rev A1 chips we have, share the same problem that the B92 rev A1 chips had.

Bill

Hello Bill,

So after all the confusion: D92 A1 shall have the same errata as D92 A2 parts. Since I would need to dig details on when A1 and A2 were both in production and replacements made in revision, it will take me time to make sure that there are no additional A1 errata that got addressed in A2.

I would however strongly recommend to switch over to TM4C12x series (even though they may not be pin compatible with LM3S).

Regards
Amit

Hi Amit,

So after all the "Sturm und Drang", I believe we agree that it is ok to use the D92 revision A1 parts.

I strongly agree with you about the need for a redesign. It is in the works but in the mean while production of the old design must continue.

Thanks for all of your help.

Bill

Please pardon this intrusion - my small tech firm (and clients) too have suffered w/the demise/withdrawal of LM3S.

In the particular case presented here (this thread) may I suggest that you, "Plan & Implement" for the worse case?    Specifically - I'd provide for the "harshest errata."   Include that "extra voltage regulator" & trace routes as an, "insurance policy."   Should those parts/routes "not" be required - you may "DNF" (do not fill) those component footprints.   But - if they "are" required - this "worse case accommodation" saves you the time, effort, cost, lost morale required by (yet another) board "spin."

These MCUs are old - and the "deep detail" you seek (and require) appears (from my read) to not be fully nor readily available...

Its been a while since I've addressed this thread. My client was able to find a source of LM3SD92 parts but with A1 stepping. They built 5 boards using this part just to be sure before committing to mass production. I understand that these parts are EOL but we are in a bind.

The parts don't seem to program. Manufacturing is getting errors using the Segger j-Link and I am getting errors using my IAR platform which also uses a J-link.

Any thoughts from TI (or anyone who can help) would be appreciated.

You're silent as to whether you adopted the (past) suggestion to, "design for the harshest errata." As you hint at "mass production" the few extra dollars to design for, "worst case" surely is justified.

My firm uses (paid) IAR and J-Link - our boards program via SWD - always error free. What were those errors - have you seen these before? (surely your pressured & frustrated - yet incomplete tech info is "not" your friend.) Did you employ external pull-up resistors @ each JTAG line - and route those pins directly - w/short (and matched lead length) to the JTAG Header?

The use of "mass production" troubles - it's doubtful that you can source more than a few hundred of such "dead/buried" devices - and if you're especially "unlucky" (other) issues may enter this fray.

I fear for your continuation w/so problematic a device... Vendor here is far larger/smarter/resourceful than you/I - and they "pulled the plug" on your device. That's quite telling - is it not?

Hi CB1,
We are using the SWD interface and the 4 signals have pull ups. Reset does not. Signal lines should be short and nearly the same length as one another. It has worked well for years up to this point. The board was not changed but I don't think this is an issue with the errata since power is stable when the firmware is flashed. The Segger direct download that manufacturing is experiencing say something about blank check finding an unexpected value. My IAR download says "The flash loader program reported an error", When OK is clicked it says " The debugger session could not be started. Either the debugger initialization failed or else the file (filename.out) was corrupt or of an unsupported format." Don't get me started about updating to a more modern and available part. It fails pretty quickly so little or no data gets transferred. Next I'll try scoping the lines and compare them to a working unit.

Bill Brown said:

4 signals have pull ups. Reset does not.

All of our MCU reset pins tie to a pull-up w/bypass cap to ground - both adjacent to that MCU pin.   You don't mean that your "Reset" pin floats - do you?

Now the J-Link may "pull-up" that line - but I don't believe that's especially wise.  (I'd prefer as above)

Cannot you test your SW and J-Link by attaching to a, "Golden Board?"   To my (bit experienced) diagnostic mind - you've got too many variables (now) in play.

Those error messages seem to indicate that your file names or types may have changed - either at the J-Link or at your IDE.   (I do recall IAR going thru substantial changes on/around Ver 4 or 5.   (we're up to 7.4 now - and that may not be the newest)

The Golden Board test wins my vote - and it appears that you do have (some) working units.

While you direct "don't get started" - that (bit aggressive) stance may have caused/contributed to your plight.

FYI:  The RESET signal IS pulled high on a different page near the reset chip.  

So called "golden board" is any older board populated with the A2 chip.

You have an interesting turn of phrase and a tendency towards condescension that I find a little intriguing. 

Thanks for the help anyway even if you don't have the information I am seeking which is; what are the differences between the A1 and A2 step parts?

Unless you've a "doppelganger/namesake" we've collaborated (successfully) w/in the M3 forum - years past.   I've no knowledge (and little interest) in A1 vs. A2 - you may  note that vendor here "rejected both" - that tilts, "Risk-Reward" far against you/your client!

My firm has had first hand experience - clients caught just as are you - and each/every end result proved unpleasant.   

I've suggested that you provide for any & every "known" errata condition - it appears that you resist - and seek to "guard against" a "subset" of that errata - even though such "subset" proves questionable.   (and such is not condescension - it is (instead) simple statement of fact...)

Might you be hypersensitive to the "high-lighting" of your adamant rejection (i.e. "Don't tell me to migrate") of vendor's - and my - suggestion to, "Do the obvious."   Your old "horse" is long dead - even artificial heart (and new shoes) will not propel it to Churchill Downs - first Saturday in May...    New horse - glowing and eager - sits in an adjacent stable - and the clock ticks...

When dealing with grey market sources you need to keep in mind the possibility that what you get and what you ordered may be different. The vendor who sold them to you may not be aware either. The probability of this happening rises the longer from eol and the further off of the standard supply chain you trod.

Using designs with parts no longer in production is a risky proposition. At some point you have to decide either to drop the product or redesign. If you wait you will be forced to drop the product, a commitment to redesign at that point only affects how long you will be out of production, not whether you are out of production. It sounds like you may be at that point.

Robert

Robert Adsett72 said:

Using designs with parts no longer in production is a risky proposition.

Indeed Robert - and even worse - this (past LM3S) condition has been LONG known!   Poster and his client/advisers surely had sufficient time to react - yet it appears they (only now) seek great detail.

Now we (really) don't know if their "rescue" devices stem from the "grey market."   Yet - vendor here made the call that, "enough was wrong" - and clearly advised against such LM3S use - first via "NRND" proclamations - and later by formal EOL!   

Poster speaks of "mass production" - should that prove (at all) the case - would not poster's FAE and/or local Sales Office be far better equipped to investigate & advise?   It seems patently unfair to "dump this issue" upon the blameless, Amit - clearly he was not here during the "death march" of LM3S.

I remain sympathetic to this poster/his client - yet I've not drinken the "mass production" Kool Aid - nor can I agree with poster's (unexplained & hostile) resistance to migration...   (and such is simple logic - not at all, "condescension...")

If the devices are eol for some time that forms strong suspicion of grey market. Taking a long time to find a source makes it that much stronger. It seems unlikely they would be this hard to find if the original distributors still had stock.

Robert

I am sympathetic as well, I've had to face the prospect myself.

The stark choice however unpalatable must be faced. Those that pay for production and design must be aware of the risks. It is not just a technical issue.

It wouldn't be the first time someone in management was told of the risks and decided to ignore them though. Sometimes the dice roll their way, sometimes they don't.

Robert