P8X32A-Q44 Shortage Expected: place your orders
Ken Gracey
Posts: 7,392
Hello,
The foundry has suddenly increased our lead time for Propeller 1 wafers. Normally, the materials requisition process (MRP) ensures orders are placed so that this doesn't happen, accounting for lead time, safety stock, weekends, and a typical high month's worth of demand. Normally we'd have several months of supply when more chips arrive, but this time we're going to run out. This will likely occur pretty soon - either after I post this or in the next week.
https://www.parallax.com/product/p8x32a-q44 is the specific product.
The reason for the foundry lead time increase is because new equipment has been brought on-line. At the same time, demand increased on them.
I'm alerting the forum members so they can buy what they need. Chances are that you'll still find some chips at Digi-Key or Mouser while we await for new shipments.
Thanks,
Ken Gracey
The foundry has suddenly increased our lead time for Propeller 1 wafers. Normally, the materials requisition process (MRP) ensures orders are placed so that this doesn't happen, accounting for lead time, safety stock, weekends, and a typical high month's worth of demand. Normally we'd have several months of supply when more chips arrive, but this time we're going to run out. This will likely occur pretty soon - either after I post this or in the next week.
https://www.parallax.com/product/p8x32a-q44 is the specific product.
The reason for the foundry lead time increase is because new equipment has been brought on-line. At the same time, demand increased on them.
I'm alerting the forum members so they can buy what they need. Chances are that you'll still find some chips at Digi-Key or Mouser while we await for new shipments.
Thanks,
Ken Gracey
Comments
Hi Ken, thanks for the heads up, what kind of lead-time are you looking at anyway? I notice too that in your pricing that there is a break at 500+ but it seems tray lots are more advantageous from the supplier's point of view so why aren't the breaks at 480 for instance? I see the combined total of Q44s from Parallax+Digikey+Mouser at around 3,300 but I'm not sure how fast they go through them. Anyway it wouldn't hurt to grab a few trays in the meantime which I hope is not a long time
Do you expect the unmarked to get depleted also?
http://www.parallax.com/product/p8x32a-q44-u
There are >7,500 in stock.
Thanks Jim - yes, I do. As soon as the marked Propellers are gone the unmarked ones will quickly vaporize.
Ken Gracey
There's something else to consider and I'll use Parallax as an example. Our purchasing habits almost never land squarely on any reel or tray quantity for a few reasons:
- With parts being used in multiple products, they're never purchased for a single build of x or y product.
- Build quantities are usually not determined by the Propeller alone, since other parts might have higher cost and dictate a build quantity.
I don't know if it's worthwhile to make the price breaks at tray quantities or not. If it is worthwhile to do this, then it hasn't been done for the simple reason of not having time to coordinate changes internally to make this possible, that's all.
Peter, are your purchasing habits designed for 480-unit build quantities?
Ken Gracey
I'm assuming it's a tray quantity, but I'm not sure. I'd have to take a look to know for certain.
Ken Gracey
The QFN44, being smaller, comes in trays of 260, 10 by 26.
Procuring quantities of different parts, to add up without too much surplus, is a maddening process.
Do you only have a single foundry ?
The process must be trailing edge now, and I have seen chips EOL'd simply because the Foundry made a floor-space decision.
Might pay to check what volumes of other devices are going down the same line ?
Digikey used to repackage the QFPs into Tape as that is how they used to supply them in cut tape. Not sure if they still do this as its a while since I purchased from Digi.
When purchasing qties from parallax they come in trays.
Ken, I wonder what the cost to you would be to get the QFPs in tape or a mix of tape and trays? Would be easier for you to sell broken qties as the shipping is safer in cut tape.
Maybe its all those Defcon boards, and hopefully follow-on orders, that helped cause the shortage
Are more being made?
When these QFNs are gone, they're gone. I'd be happy to be in a shortage situation with QFNs because they move very slow. We will unfortunately EOL this part. The MOQ is too high for the inventory turns we have with them.
If somebody wants to order 12-25K at a time we'll build them; otherwise we can't afford to stock these.
QFNs are not preferred by contract manufacturers due to a lower yield during manufacturing.
Ken Gracey
Parallax is a little company and our volume is interesting enough to entertain one foundry. There's no way we could have two foundries on tap, alternating between them, and ensuring they both give us the best pricing based on 50% volume.
While we can always get quotes from a second source, there is no guarantee that the same process is compatible. We can't simply hand them files like we can do with a printer or PCB company. Our latest transition for P2 entailed review of 60 pages of design rules that must be accepted, waived (at the buyer's risk), or re-designed for in a "compatible" process. At least this is the way it looks to me and I've now communicated with three foundries during my experience at Parallax.
The foundry we use (Austria Microsystems) has no plans to retire the process we are using at this time. At this stage it's a profitable, refined process for them that's not likely to be retired soon. When it is, we'll get several years worth of notice that we can share with everybody.
And in case anybody is wondering if we'd leave them stranded: the answer is NO. I will use the SX as an example. The SX went "obsolete" by the IP owner Ubicom perhaps ten years ago. We still have SX chips in stock if you need them, not to mention we personally satisfied every single consumer of those parts.
I only go overboard in my reply to provide clarity in a situation where customers start to have unfounded concerns about using our parts in their design.
Ken Gracey
My suggestion: buy those two boards immediately. I'm not seeing any more on order to build internally and I'm not sure why - I'll have to figure that out tomorrow morning by talking with a few people.
Ken Gracey
Digi-Key is who you are referring to.
You bet, probably within a month they will be restocked.
Ken Gracey
This is certainly a compounding factor, as you guessed. However, we placed orders in the first of June, the very minute we realized we'd be using near 15K chips on DefCon22 badges. The normal 12-week lead time is actually behind us so we should have chips by now, but the foundry tells us they need a few more weeks.
In regards to your question about tape and reel, the issue is that we test these chips in Rocklin at Parallax. They are not tested with wafer probing (on the semiconductor fabrication line) or at the packaging company. That means we receive 100% yield until we're in testing process in Rocklin. It would not be practical to test them from T&R only to put them back in T&R since we don't have the equipment. Normally, chips are tested in two prior steps: die probing and at packaging.
So, they're only available in trays.
Ken Gracey
Guess I created a bit of a run on Propellers today.
Ken Gracey
I imagine our packaging supplier supports TQFP-44s in T&R, as you suggested. However, since we test these in Rocklin we can't remove them from T&R, test them, and return them to T&R. See my post above.
With very high volume we could justify a high investment in NRE, test fixtures and setup at the packaging company. Propeller 2 chips will be tested at the packaging facility in Taiwan.
Ken Gracey
All of the Propellers are tested in Rocklin, post-packaging process.
Ken Gracey
Ken,
Thanks for the heads up. When you can determine an estimated restock date from "few weeks", please give us a heads up. It'll help planning here.
Jeff
And I was just to about to up my soldering skills from SMD resistors and LEDs to SMD chips. Not to mention that from my local distributor ELFA, the QFN is 40% cheaper than the DIP.
Now, I thought there was an assurance from Parallax that Propellers would be available well into the future. Does this not also include the packaging?
I did guess that the magic 480 number was derived from some tray sizes. It's just that I don't recall price break quantities ever being such non round decimal numbers.
Edit: More realistically my soldering skills will max out at QFP.
Bummer, my M44D40+ module uses them. But as slow as the QFNs move, I think I have plenty of time. I am about due to order another batch of 20 QFNs though. Put a microscopic dent in the inventory............
Sorry Ken, I can' let this one go. With a properly established process, there should be a 0.00% difference in yield between QFNs and QFPs. I make this statement based upon data from our lines' daily performance. With focus on proper reflow profiles, stencil optimization for apertures, automated 3D solder paste inspection, and validated placement accuracy, I have no concerns with using QFNs over QFPs or vice versa. In other words, I would have no problems redesigning any Propeller boards currently using QFPs over to use QFNs. I would bet that board layout complexity is the key factor in the QFP vs QFN decision. You lose a lot of routing real estate under the QFN due to the center pad, so it is much easier to use a QFP. It is easier to rework a QFP than a QFN, but the QFN just needs the right rework equipment and a properly designed pad layout.
Hmm, upfront honesty begets clever marketing? That's funny actually. Anyhow, that would be fine for me, since I can use the undesired QFNs to make the equivalent of a DIP40 by putting just the QFN on my M44D40+ PCB. Funny thing is that my M44D40+ design originated from the Prop DIP shortage back in 2009. I still sell about 5 a month without any advertising, mostly paired with a PowerTwig.
Ken, sorry for the confusion. I meant for you to package them into tape and reel after testing. So, receive in trays, test, then have a third party tape and reel them. It's actually not that expensive to do nowadays. Even so, I would probably only do a small portion of the QFPs for orders that Parallax ships to end customers. For your distributors, let them handle the extra repacking for individual shipments.
It's always interesting to hear from people with real world experience.
What I'm wondering now is how many thousands of a thing do you have to make in a run before you have all those parameters tweaked just nicely to get that 0.00% difference in yield between QFNs and QFPs.
I'm kind of guessing that for small runs those optimizations are never quite made and yield drops.
This is all idle curiosity, but you never know I may want to make a run of a 1000 of something one day.
When do you anticipate resupply into your stock?
I have not wanted to get into QFNs if possible. There is certainly resistance to using them but I have no problems with QFP, even when I was hand soldering them.
Now I understand why you use tray packaging.
Think the new Juki P&P line helps a bit, too? Between your equipment, experience and PCB layout done in-house you have control over the whole process. I'd say you're doing a fantastic job.
Our customers work with a variety of contract manufacturers all around the country and in Asia. Some of the P&P lines in Asia look like they use modified electromechanical pinball machines for P&P, but others have the best equipment from Europe. Collectively, our customer feedback around QFNs shows a much lower yield. Not to mention the packaging cost is at least a $.75 more than a QFP.
As for redesigning your module to switch to QFP, don't bother. We have plenty in stock for several years.
Ken Gracey