Reply by rickman February 25, 20042004-02-25
Jim wrote:
> > "Jim Granville" <no.spam@designtools.co.nz> wrote in message > news:EJu_b.28084$ws.3169183@news02.tsnz.net... > > > > You can always make the order conditional. > > Quote the device you have qualified, and also your timing (& any > > other, such as Icc?) targets, and give an order, and also ask for > > samples to fully qualify. > > That way, the wheels turn in parallel, and there is more clear > > incentive for the disti to get samples - much more than 'might order if > > they are OK'. > > > > -jg > > > > Thanks Jim - that sounds like good advice. You can probably tell I am pretty > new to this so advice from people like yourself is very much appreciated.
Just be careful with this approach. If you make the order conditional, they may hold placing the order with the manufacturer until you "approve" the parts. Make it very clear what you expect to happen and make sure the "product manager" at the disti is ok with it. That is the guy who will say yes or no to placing the order before you have committed. -- Rick "rickman" Collins rick.collins@XYarius.com Ignore the reply address. To email me use the above address with the XY removed. Arius - A Signal Processing Solutions Company Specializing in DSP and FPGA design URL http://www.arius.com 4 King Ave 301-682-7772 Voice Frederick, MD 21701-3110 301-682-7666 FAX
Reply by Jim February 25, 20042004-02-25
"Jim Granville" <no.spam@designtools.co.nz> wrote in message
news:EJu_b.28084$ws.3169183@news02.tsnz.net...
> > You can always make the order conditional. > Quote the device you have qualified, and also your timing (& any > other, such as Icc?) targets, and give an order, and also ask for > samples to fully qualify. > That way, the wheels turn in parallel, and there is more clear > incentive for the disti to get samples - much more than 'might order if > they are OK'. > > -jg >
Thanks Jim - that sounds like good advice. You can probably tell I am pretty new to this so advice from people like yourself is very much appreciated. Jim
Reply by Jim Granville February 23, 20042004-02-23

Jim wrote:
<snip> I've got a few samples of various brands that should work on 
paper but won't
> be able to try the out until our next prototype is ready in 3 weeks or so. > I'm itching to get the order in before then in case the situation gets > worse. As well as that, being in the UK, the dollar is so cheap for us at > the moment I want to spend some money now! ;) > > I know I should probably hold off but don't want to hold up production when > the time comes.
You can always make the order conditional. Quote the device you have qualified, and also your timing (& any other, such as Icc?) targets, and give an order, and also ask for samples to fully qualify. That way, the wheels turn in parallel, and there is more clear incentive for the disti to get samples - much more than 'might order if they are OK'. -jg
Reply by Jim February 23, 20042004-02-23
"rickman" <spamgoeshere4@yahoo.com> wrote in message
news:403A2559.DC24FC7F@yahoo.com...
> You should get in touch with the FAE at your distributors. They can > work with you to get parts to test and will be able to help you identify > any possible troble areas. SRAMs are actually very, very compatible > parts. So it is unlikely that you will see any problems switching to a > different brand. Like Rob said, you need to qualify multiple brands of > parts for every socket on your board if possible. > > The distis like for you to lock in their brand, but they understand that > most manufacturers won't do that. So work with them and get parts > qualified. The shortage will only get worse, not better... at least for > the next few months. > > -- > > Rick "rickman" Collins > > rick.collins@XYarius.com > Ignore the reply address. To email me use the above address with the XY > removed. > > Arius - A Signal Processing Solutions Company > Specializing in DSP and FPGA design URL http://www.arius.com > 4 King Ave 301-682-7772 Voice > Frederick, MD 21701-3110 301-682-7666 FAX
Thanks for the insight Rick. I definitely don't want to limit the brands if at all possible. This is the first time I've used a SRAM this big (to me it's big anyway) so it's good to know, in general, they are pretty interchangeable. We're not using them beyond their specs so there is no reason why they shouldn't work ok - just a bit new to all this ;) Yes, I suspected the shortage would get worse, so am going to try to get on top of this before it bites us! Jim
Reply by rickman February 23, 20042004-02-23
Jim wrote:
> > Thanks Rob for your good advice. I hadn't realised at the design stage about > 6 months ago that SRAM would be so hard to get hold of come production. Our > product is quite low cost so I picked the cheapest I could find at the time > (from STM). > > I've got a few samples of various brands that should work on paper but won't > be able to try the out until our next prototype is ready in 3 weeks or so. > I'm itching to get the order in before then in case the situation gets > worse. As well as that, being in the UK, the dollar is so cheap for us at > the moment I want to spend some money now! ;) > > I know I should probably hold off but don't want to hold up production when > the time comes.
You should get in touch with the FAE at your distributors. They can work with you to get parts to test and will be able to help you identify any possible troble areas. SRAMs are actually very, very compatible parts. So it is unlikely that you will see any problems switching to a different brand. Like Rob said, you need to qualify multiple brands of parts for every socket on your board if possible. The distis like for you to lock in their brand, but they understand that most manufacturers won't do that. So work with them and get parts qualified. The shortage will only get worse, not better... at least for the next few months. -- Rick "rickman" Collins rick.collins@XYarius.com Ignore the reply address. To email me use the above address with the XY removed. Arius - A Signal Processing Solutions Company Specializing in DSP and FPGA design URL http://www.arius.com 4 King Ave 301-682-7772 Voice Frederick, MD 21701-3110 301-682-7666 FAX
Reply by Jim February 23, 20042004-02-23
"Rob Young" <rwyoung@ieee.org> wrote in message
news:69e5135f.0402230725.fac94d1@posting.google.com...
<snip>
> I won't say it is a bad idea, but if it was me, I would burn up the > phone lines trying to get my hands on some samples of the part to test > and also somebody that has some (any) stock on the part. If the > samples check out then release the PO for the stock to ship. You > might loose a week or two of your manufacturing time. > > But even before I do that I'd take a very close look at the data > sheet(s) and compare timing diagrams and specs. In general SRAM data > sheets of similar parts will have specs and timing diagrams done the > same way but sometimes you find somebody giving the same information > but in a slightly different way. Be careful. If I do get a warm > fuzzy feeling from the data sheet(s) then I would order the parts. > Still want the sample parts ASAP for testing and if I do find a > problem then I may have enough time to halt production or find a > work-around that can be integrated into that production run. > > And finally I'd try to identify two or three (or more in some cases) > substitutes so I don't get pinched down the road. Not always possible > but for SRAM I'd definately try. > > By the way, one more thing. If your company doesn't mind buying old > inventory you could try one of the hundreds of companies that deal in > surplus and seconded inventory. One such broker I've had luck with is > M & R Communications 512-502-0783 and ask for Willy. There are others > like America II, Harry Krantz, etc. (Disclaimer: I don't work for M&R > or otherwise bennifit from mentioning their name, just gotten good > service). > > None of these suggestions are pretty or will help you sleep at night > but it sounds like you are getting pinched.
Thanks Rob for your good advice. I hadn't realised at the design stage about 6 months ago that SRAM would be so hard to get hold of come production. Our product is quite low cost so I picked the cheapest I could find at the time (from STM). I've got a few samples of various brands that should work on paper but won't be able to try the out until our next prototype is ready in 3 weeks or so. I'm itching to get the order in before then in case the situation gets worse. As well as that, being in the UK, the dollar is so cheap for us at the moment I want to spend some money now! ;) I know I should probably hold off but don't want to hold up production when the time comes. Jim
Reply by Rob Young February 23, 20042004-02-23
"Jim" <jim@nospam.com> wrote in message news:<Wzj_b.34$dt1.29@newsfep3-gui.server.ntli.net>...
> For our prototypes we used STMicroelectronics SRAM 4Mbit, 512Kx8, 55ns > (M68AW511AL). > > Now it is getting near production I find that they have stopped making that > memory. Lots of other companies make equivalent SRAM, but can I assume it is > going to work ok? Our application requires an access time of 60ns to the > SRAM, so within the SRAM spec. > > In an ideal world I'd get one of the new types and prototype with it before > ordering, but lead times are so long at the moment. So I want to place an > order (1000 units) for our production run asap. > > Is that a bad idea? > > Thanks, > > Jim
I won't say it is a bad idea, but if it was me, I would burn up the phone lines trying to get my hands on some samples of the part to test and also somebody that has some (any) stock on the part. If the samples check out then release the PO for the stock to ship. You might loose a week or two of your manufacturing time. But even before I do that I'd take a very close look at the data sheet(s) and compare timing diagrams and specs. In general SRAM data sheets of similar parts will have specs and timing diagrams done the same way but sometimes you find somebody giving the same information but in a slightly different way. Be careful. If I do get a warm fuzzy feeling from the data sheet(s) then I would order the parts. Still want the sample parts ASAP for testing and if I do find a problem then I may have enough time to halt production or find a work-around that can be integrated into that production run. And finally I'd try to identify two or three (or more in some cases) substitutes so I don't get pinched down the road. Not always possible but for SRAM I'd definately try. By the way, one more thing. If your company doesn't mind buying old inventory you could try one of the hundreds of companies that deal in surplus and seconded inventory. One such broker I've had luck with is M & R Communications 512-502-0783 and ask for Willy. There are others like America II, Harry Krantz, etc. (Disclaimer: I don't work for M&R or otherwise bennifit from mentioning their name, just gotten good service). None of these suggestions are pretty or will help you sleep at night but it sounds like you are getting pinched.
Reply by Jim February 23, 20042004-02-23
"rickman" <spamgoeshere4@yahoo.com> wrote in message
news:403A166C.86EFCE40@yahoo.com...
> Jim wrote: > > > > For our prototypes we used STMicroelectronics SRAM 4Mbit, 512Kx8, 55ns > > (M68AW511AL).
<snip>
> > In an ideal world I'd get one of the new types and prototype with it
before
> > ordering, but lead times are so long at the moment. So I want to place
an
> > order (1000 units) for our production run asap. > > > > Is that a bad idea? > > It is a bad idea to prototype something to "see if it works". You need > to sit down with a paper and pencil (or spread sheet) and do the math on > all the timing requirements to see if the part will work in your > design. SRAMs are very simple to analyze compared to DRAMs. Just > hooking it up won't tell you unless there is a gross error. The figures > in the data sheets are max over voltage, temp and process. Any given > chip will run faster at room temp than worse case so you aren't really > testing it. > > If you don't know how to analyze the timing, you need to ask for help on > that. If it passes timing analysis, then I seriously doubt that you > need to test a prototype to make sure it works. But with a 1000 piece > order, wouldn't you feel safer? You can always place the order and then > cancel it if the parts fail tests. I don't think a vendor will give you > a hard time for that. > > -- > > Rick "rickman" Collins > > rick.collins@XYarius.com > Ignore the reply address. To email me use the above address with the XY > removed. > > Arius - A Signal Processing Solutions Company > Specializing in DSP and FPGA design URL http://www.arius.com > 4 King Ave 301-682-7772 Voice > Frederick, MD 21701-3110 301-682-7666 FAX
Thanks Rick for your advice - it sounds extremely sensible all that you say. Yes, on paper we've calculated it should work fine. This is my first design to go into production, so I'm pretty tentative about taking it forward without testing the actual part. In my naivity I didn't even realise we could cancel a parts order once it had gone through without a lot of hassle/penalties. That would be worst case scenario and hopefully it wouldn't come to that. Jim
Reply by rickman February 23, 20042004-02-23
Jim wrote:
> > For our prototypes we used STMicroelectronics SRAM 4Mbit, 512Kx8, 55ns > (M68AW511AL). > > Now it is getting near production I find that they have stopped making that > memory. Lots of other companies make equivalent SRAM, but can I assume it is > going to work ok? Our application requires an access time of 60ns to the > SRAM, so within the SRAM spec. > > In an ideal world I'd get one of the new types and prototype with it before > ordering, but lead times are so long at the moment. So I want to place an > order (1000 units) for our production run asap. > > Is that a bad idea?
It is a bad idea to prototype something to "see if it works". You need to sit down with a paper and pencil (or spread sheet) and do the math on all the timing requirements to see if the part will work in your design. SRAMs are very simple to analyze compared to DRAMs. Just hooking it up won't tell you unless there is a gross error. The figures in the data sheets are max over voltage, temp and process. Any given chip will run faster at room temp than worse case so you aren't really testing it. If you don't know how to analyze the timing, you need to ask for help on that. If it passes timing analysis, then I seriously doubt that you need to test a prototype to make sure it works. But with a 1000 piece order, wouldn't you feel safer? You can always place the order and then cancel it if the parts fail tests. I don't think a vendor will give you a hard time for that. -- Rick "rickman" Collins rick.collins@XYarius.com Ignore the reply address. To email me use the above address with the XY removed. Arius - A Signal Processing Solutions Company Specializing in DSP and FPGA design URL http://www.arius.com 4 King Ave 301-682-7772 Voice Frederick, MD 21701-3110 301-682-7666 FAX
Reply by Jim February 23, 20042004-02-23
For our prototypes we used STMicroelectronics SRAM 4Mbit, 512Kx8, 55ns
(M68AW511AL).

Now it is getting near production I find that they have stopped making that
memory. Lots of other companies make equivalent SRAM, but can I assume it is
going to work ok? Our application requires an access time of 60ns to the
SRAM, so within the SRAM spec.

In an ideal world I'd get one of the new types and prototype with it before
ordering, but lead times are so long at the moment. So I want to place an
order (1000 units) for our production run asap.

Is that a bad idea?

Thanks,

Jim