CMOS 4000 series obsolescence 3

[Philrock]

I am designing a small digital control circuit using CMOS electronics. The circuit will be used in a consumer product and will be manufactured in quantity of about 10,000 per year. I had originally selected the CD4000 series, then an associate told me that this series is headed for obsolescence in the not-too-distant future, and the 74HC series would be a better choice. I found a couple of items on the Internet that support this statement, but when I check availability from distributors, the CD4000 series is more widely available and considerably less expensive.

The devices I'm considering are:

CD4017, 74HC4017 Johnson decade counters
CD4029, 74HC161 4-bit binary counters
CD4093, 74HC132 quad 2-input NAND gates with Schmitt trigger inputs
CD4585, 74HC85 4-bit magnitude comparators

All distributors I've checked have ALL the 4000 series devices - mostly at dirt cheap prices. I haven't found any one distributor that stocks all the 74HC series devices, and the 74HC series devices I have found cost two or three times as much as the corresponding 4000 series devices.

Manufacturers have not been helpful - they won't give generalizations about an entire series. They will only look up status of individual devices. Texas Instruments tech support says most of the 4000 series devices I'm considering are in full production and recommended for new designs, and some of the 74HC devices I'm considering are "made to order" only.

Can anyone provide a bit of an overview on this issue?

 

[IRstuff]

Perhaps you haven't addressed the question of how long your product is expected to be in production. Are you truly anticipating 10K parts/yr for 20 years? or only 3? The lifetime on commercial products is less than 3 yrs, at which time you probably will have an opportunity to design in newer components.

You have several choices, among them:

> make a lifetime buy now - but have the attendant sunk cost up front

> use the most common and available parts and do nothing until the suppliers ask for lifetime buys

> consider alternate designs that can eliminate these components altogether. 4000 series seems hardly compatible with surface mount that usually required to keep commercial costs down. Are there no other parts such as processors or FPGA's that could incorporate these functions?

TTFN

 

[cbarn24050]

Hi, nearly all the lsi chips are heading that way, its much better to use programmable logic, much cheaper for volume products.

 

[designchain]

IRstuff is right on all points he makes. How long you expect this product to be in production and supportable is a key one. For anything over a year or two I'd skip the 40xx family. See my other post on lifecycle regarding this family and some of the concerns about it...but consumer products tend not to have the reliability concerns or longevity concerns of, say, servers so you may not care. And as cbarn24050 rightly points out, you can probably combine all that logic in to a simple PLD like a 20V8 or 22V10. You should do the math on the programming plus assembly cost of a single programmable part vs. the assembly and added acquisition costs for multiple discrete logic parts.

--
Mike Kirschner
Design Chain Associates, LLC

http://www.designchainassociates.com

 

[Chuckles159]

I have designed with the 40000 series since the 70's and have yet been unable to source a device that I used then. It seems to be one of the most available families of parts ever made. I still design with it and have no fear of being unable to get it. By the time you come up to speed on gate arrays and programmable logic you can have had your product using 4000 series logic on the market for months already. As well, it is the programmable logic devices that go obsolete much faster.
I have one product I made in the 70's that used a 1K EPROM. This is all the memory I need. However, today, that is the one part I can no longer buy. I can still buy all the 40000 series logic, I can still buy the op-amps, the 555 and all the passives. I just can't buy the EPROM. Same for all the programmable logic from back then.
4000 series - If it will do the job, go for it.

 

[Comcokid]

I first heard about the death of the 4000 series in the late 80's. Of course, that has now been more than 15 years ago. In the meantime, many other logic families have come and gone. In the past years, more parts with a '4000' family number designation have been introduced into the market (mostly in the form of single-gate logic in a SOT-type package). Certaintly, this part in dip packages may prove a problem, and some of the less used 4000 parts.

I recently opened-up my new laptop and found a few 4000 parts - aparently within the battery monitoring circuit.

I would predict the factor that will most likely determine a problem in getting 4000 series parts would be a fast recovery from a recession, and a resulting capacity crunch on semiconductor lines which would displace lower-margin/volume parts such as 4000 devices. In such cases, sourcing the parts may require going to Korean or other far-east semi vendors.

 

[designchain]

I've been hearing about the impending demise of 4000 series and 74F series for a decade too from suppliers; but some big OEMs are rabid about telling suppliers that they can't obsolete these things no matter how much the suppliers want to. There's really no availability problem using these parts IF your product has a short lifecycle and it won't need support for the next 20 years (that's the topic of another thread). Just consider the options and impact of using these vs. using other parts and having fewer components, a simpler PCB, lower procurement overhead, etc.

Comcokid makes a great point; once demand exceeds supply these low-margin parts are always the first to go on allocation because wafer, package assembly, and test capacity go to product that has higher margin. When the economy turns expect demand to exceed supply faster than it ever has because this recession has resulted in the dismantling of lots of capacity.

Look at other, more recent logic families if you insist on designing with individual gate-level logic devices and stick to the basics.

--
Mike Kirschner
Design Chain Associates, LLC

http://www.designchainassociates.com

 

[IRstuff]

One other point to consider is cost. While 4000 series parts are indeed available, they are on the subset side of the supply curve and the remaining users cannot command much in the way of price breaks.

When I worked in the semiconductor industry, we made some metal-gate PMOS devices when the industry was already going to 74FC parts. We subcontracted the wafer fab, test, packaging, etc. The only thing that we did was to put our logo on the parts. We still had a 75% net margin. When we needed more operating margin, we increased the prices or forced last-time buys.

The moral is that you can design with 4000 series parts, but be prepared to pay premium prices relative to the identical function in another technology and be prepared to redesign.

TTFN

 

[Chuckles159]

The thing to be aware of is that many logic families have indeed come and gone, specially in the 74 series. The 4000 series however seems to have been immune to major changes. Regardless, in the 74 series, while families have disapeared, the functions available have only grown and most any of the old families that have disappeared have been replaced with functionally equivalent devices.

As for the 4000 series, these are made by so many different cmpanies, they seem to be the bedrock of semiconductor logic. There have been no changes to speak of in the devices for 2o years and no forseeable changes in the future. They are the B-52 of logic. Built right in the first place and like the B-52 where the grand children of the early crews are flying them now, it is likely our grand children will be designing with the 4000 series.

 

[47gandalf]

I'd like to add one comment to what has already been said. Although various "standard" logic families have come and gone away over the years (e.g. 74XX, 74HXX, ECL, etc.) there is a reason that the HEF4000 family has survived nearly intact.

It is the only standard logic family that can work with power supply voltages up to 18V! All of the others use supply voltages at or below 5V.

So my experience is that if you need the higher supply (and increased noise margins) then the '4000 family can be used, but if you have a system at or below 5V supply, you can protect against possible obsolescence by using a different logic family.

As an aside, I have always found the HEF4xxx parts (originally from Philips and RCA) have more "bullet-proof" front ends than the CD4xxx parts. That is, they are better protected against stray noise spikes. This was very important when I designed equipment for industrial control, as was the increased noise margin when running +15V instead of +5V supply.

 

[designchain]

That's a great point, 47gandalf. But the fact is that worldwide wafer fab capacity for MOS devices >=0.7 microns is down about 19% in the last three years. My recollection is that the 4000 series is a 3 micron metal gate (not self-aligned) technology...so as the long-lived, and consistent volume products that rely on this technology are eventually replaced or updated, this too shall pass but guessing when requires a crystal ball.

--
Mike Kirschner
Design Chain Associates, LLC

http://www.designchainassociates.com

 

[lcsjk]

47gandalf is probably correct. The 4000 series is the only one that is commpatible with the 12 volts used in many linear circuits that work at 12-15 volts. For that reason it will stay around. However, some devices in the family have ceased production, just like all functions that get very little design-in usage. Stay away from multi-function packages.

If you can use 5 volts, go to the 74HC series. If you must use the 4000 series, check with the manufacturers of the part to see if it is on phase-out status. Just because it is on the shelf at Digikey and Allied or other distributors does not mean it will be available by the time you get well into production.

 

[nbucska]

It would help us to recommend possible alternative if we knew the operating frequency range and the target price ?
Are you going to have other components, too, on the PC board ? Analog ?


<nbucska@pcperipherals.com>

 

[jtsamuels]

I agree with most of the messages posted here, both pro and con. In 1978 I helped design an x-ray processor controller board using 4000 cmos as it rabn at 12V for the noise immunity. Just this year, we are replacing it with a one chip micro for cost reduction reasons, not sourcing problems! In fact, a Motorola 68HC11EOFN based microcontroller board that I designed in 1994 has had 4 revisions due to obsoleted parts, all around the PLD's such as the 22V10's, early FLASH memory, etc. So the warning is that the newer, slighlt more exotic parts (if you can call PLD's exotic) have a much greater probability of going obsolete than the jelly bean discrete logic gate parts, especially the 4000 series. One observation is that there are now 74HC4000 series that are 5V only versions of the original 4000 series but do not cover the same extensive gate styles. I also agree with the gentleman that warned about staying away from any chip that contains multifunctions (or-Nand-Inverter) within the package. I had one of those go on me in 1995.

Also, one of the other attractive aspects of 4000 cmos is its essentially ZERO power dissipation when it is not switching. Whether the output level is high or low, as long as you are not attempting to switch the devices at too high a speed (>50 kHz) the power dissipation is miniscule. In fact, you have to be careful to disconnect any powered inputs to a package when the supply is shut off as the powered input can feed a voltage into the gate, up through the input protection diode and out the Vcc supply pin to the rest of your design. The first time this happened to me (1970's) I thought I had invented perpetual energy as I couldn't figure out why my circuit was still working with the power supply turned off!! LOL

In summary, I have been hearing the call of death for 4000 cmos since the 1980's by our own Material Engineering staff. I am sad to say that in these decades of downsizing that the support engineers that warned me against designing with 4000 cmos have been made obsolete by our company while we continue to buy, build and sell equipment into the medical marketplace with the devices. The average lifetime of our products are 10-15 years. And yes, we will continue to soure the very cost effective, readily available 4000 cmos gates.

 

[designchain]

And to your point, a couple weeks ago just inside the front cover of EBN, the procurement-focused trade magazine, Texas Instruments had an ad touting the breadth of their logic family product line and there it was listed with all the others: the CD4K series. I still wouldn't design it in to a short-lived consumer product (what started this thread) but would have to think twice about an industrial control or other long-lived product.

Mike

--
Mike Kirschner
Design Chain Associates, LLC

http://www.designchainassociates.com