PIC18F4620: new silicon revision B4

T.K · July 30, 2006, 4:50pm

INFO: It turned out that the new chip revision still contains the EUSART bug, therefore this article is obsolete.

The original article was:

So, PIC18F4620: according to a recently published errata sheet (http://ww1.microchip.com/downloads/en/DeviceDoc/80282a.pdf), Microchip has finally released a new B4 silicon, where the EUSART bug is fixed. These are really good news for all future MIDIbox projects, because this means that in some months PICs will be available where the MBHP_IIC_MIDI module won’t be required as a workaround solution for a reliable MIDI Out port anymore!

My proposal is not to buy PIC18F4620 devices yet until some of us have verified that the right revision is available in the appr. shops. In general I think it makes sense to wait until november to ensure, that older revisions are out of stock, thereafter we could try the new one (the revision number can be checked with a program). Maybe companies like Mouser or Farnell could also be asked via email or phone before…

For those who cannot wait, they can also use in future the combination of CORE and IIC_MIDI module - it works perfectly, but increases the complexity (-> lower success rate for electronic newbies)

A new revision of PIC18F4550 (USB device) has not been announced yet.

Best Regards, Thorsten.

stryd_one · July 30, 2006, 5:19pm

That’s great news TK

It seems that after all, they had heard your comlaints, but just decided not to say anything : I wonder if they will have introduced another new bug for you to work around? heheheheh

T.K · July 30, 2006, 5:22pm

We will see, I’ve enough MIOS code for a validation under realworld conditions

Best Regards, Thorsten.

Wilba · July 31, 2006, 4:21am

TK, is there some way we can test for this EUART bug with a Core using MIOS?

i.e. some MIOS app that echos MIDI In to MIDI Out, and send a lot of messages to it from a PC and watch for a zero byte sent back?

T.K · July 31, 2006, 7:07am

I’ve developed an algorithm which can reproduce the bug on a deterministic way (see also this article: http://forum.microchip.com/tm.aspx?m=85120), but it requires a second core (with another program which is not described in this article), and a scope to catch the triggers.

However, there is an easier, but not so deterministic way: enable the MIDI merger, and send & receive a large SysEx dump with MIDI-Ox. When you compare the dump (function provided by MIDI-Ox) you will sometimes (!) see additional zero bytes

Best Regards, Thorsten.

docbrown · July 31, 2006, 10:12pm

Yay ;D ;D ;D Finally!! its fix!!

Thanks for the info TK.

I wonder if you order a sample from MicroChip, if they will give you a older version just to get rid of there defective PIC inventory.. just a thought.. So beware for the newbies..

T.K · August 1, 2006, 5:34pm

I wonder if you order a sample from MicroChip, if they will give you a older version just to get rid of there defective PIC inventory..

thats the reason why I will wait at least for one month before ordering samples

Best Regards, Thorsten.

docbrown · August 2, 2006, 4:55am

Just FYI:

I ordered the 18F4620 I/P about 2 1/2 weeks ago from MicroChip and I received a PIC with a Packaging Information number " 0605 4R3". Pay attention on the bolded numbers.

According to this PDF under Packaging Information: http://ww1.microchip.com/downloads/en/DeviceDoc/39626b.pdf

This PIC was (I guess) made on the week of January 29, 2006

MTE · August 2, 2006, 7:42am

quote from the pdf (page 365)

Legend: XX…X Customer specific information*

Y Year code (last digit of calendar year)

YY Year code (last 2 digits of calendar year)

WW Week code (week of January 1 is week â€˜01â€™)

NNN Alphanumeric traceability code

Note: In the event the full Microchip part number cannot be marked on one line, it will

be carried over to the next line thus limiting the number of available characters

for customer specific information.

i have 3 pic18F4620-I/P with the number 06054R1…other 3 with number 06205WW…

when came out the model without the bug ???

Greetz

MTE

stryd_one · August 2, 2006, 8:55am

It came out like last week, so you’ll want at least 0630 or so (+/- a few weeks because I’m too lazy to count them)

MTE · August 2, 2006, 9:00am

I’m too lazy to count them

;D

prot a noblem

thnx for the info

Regards

MTE

docbrown · August 2, 2006, 2:33pm

I guess if we wait for the November production, the chip will have a +/- number of 0645 XXX.

OrganGrinder · August 2, 2006, 10:10pm

hi

my sample 4620 where shipped 11 june with the date number 0620

counting the weeks, my chips where sent about 1 month after manufacture

this should give an idea when to order to get the new revision

OrganGrinder

dcer10 · August 5, 2006, 10:09am

I also have got a couple with the same numbers 06054R1, did you figure out if they were the ones with the bug?

Could these chips be loaded with the bootstrap loader from the JDM programmer? If so does anyone have a guide on how to load it up ang get going with the new pics or is it the same as with the old ones?

Also if all of that works whats the chances of TK allowing people with suitable new chips access to the applications to try out? Im keen to try the new SID and sequencer apps. I will keep the old pics on hand to swap if the new apps are still buggy.

OrganGrinder · August 5, 2006, 11:27am

Could these chips be loaded with the bootstrap loader from the JDM programmer? If so does anyone have a guide on how to load it up ang get going with the new pics or is it the same as with the old ones?

i don’t know about the JDM, but my 4620s burned ok through my PIC Burner.

OrganGrinder

T.K · August 5, 2006, 1:07pm

Could these chips be loaded with the bootstrap loader from the JDM programmer? If so does anyone have a guide on how to load it up ang get going with the new pics or is it the same as with the old ones?

It’s basically the same procedure like with the old ones. You only need to select the right binaries in the mios_v1_9_update package. The applications are binary compatible and don’t need to be recompiled

Just only take care that you are using the latest MIOS Studio release, which contains a bugfix for PIC18F4620 uploads

Also if all of that works whats the chances of TK allowing people with suitable new chips access to the applications to try out? Im keen to try the new SID and sequencer apps. I will keep the old pics on hand to swap if the new apps are still buggy.

I haven’t started with MBSID V2 yet, but MBSEQ V3 is nearly finished. It’s available via email request.

I also have got a couple with the same numbers 06054R1, did you figure out if they were the ones with the bug?

I’ve now released a small application which allows you to determine the revision ID, it’s available under http://www.ucapps.de/mios_download.html

From the README.txt

This program displays the Revision ID of a PIC18F and sends it
also via MIDI as CC# event

Known numbers (mostly documented in the errata sheets of the PIC devices)

  - 0427: PIC18F452  C1 Silicon (latest and greatest)
  - 0C03: PIC18F4620 A3 Silicon (contains EUSART bug)
  - 0C04: PIC18F4620 A4 Silicon (contains EUSART bug)
  - 0C06: PIC18F4620 B4 Silicon (the good one w/o EUSART bug)
[/code]









Best Regards, Thorsten.

Rio · August 7, 2006, 8:24am

0C06: PIC18F4620 B4 Silicon (the good one w/o EUSART bug)

does this mean that this doesn’t needs a IIC MIDI Module?

stryd_one · August 7, 2006, 12:20pm

That’s right

Well… That’s the theory anyway hehehe

c0nsumer · August 7, 2006, 7:24pm

Just to confirm, IIC_MIDI support will remain, though?

-Steve

T.K · August 7, 2006, 7:35pm

of course, I don’t plan to remove MIOS features

Best Regards, Thorsten.