Exactly, the RAM is being used not only for storage but continually for various background tasks. No calculations on my part but I would expect the RAM to be accessed hundreds of times a second. Yes, most would probably be reads but plenty of it could be writes too. In writing my operating system for the RetroPinball KOD, I know I have a number of "divide by" counters running through the 1 ms and 5 ms background interrupt routines. Fortunately, for my design, using a 8051 core processor, most of that is happening within the built in RAM within the CPU chip, not on the outside data bus. But I still have some tables out there I used for lamp and solenoid on and off times. So those would see hundreds of read and writes even during attract mode.

Unfortunately the 6800 and 6809 bus design does not work like the 8051 Microcontroller cores. There is what is called a "Stack Pointer". This is a register that tells the CPU where in RAM to store values when they are pushed and pulled as well as where to store and retrieve the program counter every time a subroutine is called. A normal machine code program would be jumping all over the memory map many times a second. Even worse with code that is compiled using a higher level language where everything is done using subroutine libraries.

All of this tends to tell me that the better product is to use the Simtek parts and not the Ramtron (personal opinion here), as they perform as regular CMOS SRAM while power is applied and then do a batch transfer of data to the NVRAM portion of the chip on power down.

That I need to get me one of them thar Super Duper Soldering Pots so I can re-socket a Bally AS-2518-35 board in 10 seconds flat!!!

And they still have 9996 lots of 20 chips available. Better hurry!

If soldered in the correct direction, it would illuminate when power (+5v) is applied to the host board.

What I noticed was a number of the cheaper sellers had 38 to 45 days shipping to the US. Normally, from the "full price" vendors, its 10 to 15 days and I often get it sooner.

So, your theory about filling a container verses air post, make sense.

I cannot comment directly on the part you link to but I can share my experience with you about UTSOURCE. I know it has been posted that they are known for selling counterfeit parts but I have used them for a number of different semiconductors over the last couple of years and have always received "working" parts. Can't tell you if they are re-branded or other, just that they worked as required. Also, here are details of a transaction I had with them:

On December 10th of 2014, I purchased some FM16W08 SOT package ram. They had them at an unbelievable price of only $1.20 for a quantity of 16. So, I placed a purchase.

The next day I received an email indicating that the supplier they were using was "sold out" and the new supplier required $4.00 per unit.

I referred them to their current EBay listing where they were still offering to sell 10 pieces for $29.33.

They replied saying they could sell them to me at $2.933 each. Since this was still a good deal to me, I agreed and sent them the balance they requested.

I received my 16 pieces promptly and they were all good. Shipping for 16 pieces was $2.60 and they arrived in under 2 weeks.

Looking at their current Ebay listing price, they are now at $32.50 for 10 with a whopping $2.00 for shipping. Their website currently lists them for $4.33 each or $3.22 if you buy 10. So, always check their Ebay listings before placing your orders. But it appears they seem to match up pretty well now. I think they sell under 2 names, one of which is HKUTSOURCE.

Compared to their website pricing of $5.83 for 10, you can get them for $48.67 shipped off ebay which is basically their price for 11 on their website:

ebay.com link: 10PCS FM1608 120 P Encapsulation DIP 28 64Kb Bytewide FRAM Memory

I wouldn't hesitate to order 10 from them but would not risk buying 100 of them until you got and used the smaller batch first.

No, actually, like we discussed before, in some games the same RAM chip used to hold bookkeeping and high scores is also used as "scratch pad" ram by the main CPU and can be read and/or written to hundreds of times a second.

Yup. For that price, they wrap them in static inducing shrink wrap.

I believe most do what the manufacturer recommends. Some do offer a choice as to what algarythm to use.

What you describe I think was called something like prototype mode. Works but wasn't guaranteed to last the full amount of time as a full burn.

It is my assumption, that only about 4 of you here are looking at this from the business side of things (making and selling). The problem, if that's what you want to call it, is that many here are simply looking at the cheapest possible way to get NVRAM into their collection. Some are fine with the high end devices with customer support, while others are also fine dealing with bottom basement pricing from China for something that gets the job done. And there is apparently, a market in between.

After all, many of us deal in our own repairs at all levels and soldering a few cheap components together isn't much of a stretch to add to the repair skill set. And there are plenty of others that happily pay big bucks for others to do all the work. Two completely separate markets. It's the same for repairing original boards verses paying for brand new replacements.

To the sellers, if onesie twosie orders aren't worth the profit, simply offer singles at a high price and deeper discounts for 5 and 10 Qty sales.

Looks fine. But I have to Laugh. Those are the ones many of us are building ourselves for around $5.00USD now with parts from China.

There are also sellers on Ali that will ship faster than the 31 days shown. for the most part, the longer shipping times (30 to 45 days) are on the sellers with rock bottom prices. The higher priced products (even these NVRAMs) come with faster shipping times.

You really have to pay close attention to the ship time shown in the listing before clicking on BUY. The thing that is a bit weird is it seems the low priced sellers take about 2 weeks to even tell you your order has shipped. Where the shorter ship times usually alert you in only a couple of days. At least, that has been my experience so far.

Since it used in a lot of places, you can add just a tiny bit of PC to your design and run a Ground trace along the outer edge to provide all those grounds to unused pins. It would certainly clean up your design a bit. On the layout program I use, it allows me to hand route some traces prior to running the auto route routine. I don't know if you are auto routing or doing it 100% by eye and hand?

Also, remember unless you plan to read out the stored data in the future, you don't need to route A1 -> A1 and D0 -> D0 etc.. You can use any address pin that is closest to an adjoining address pin and the same for any data pin to pin choices. It doesn't matter how the cells in the device are selected internally or where a 1 or 0 is stored on a data pin. Whatever gets written in, will get read out correctly regardless of your choice of address and data pins. This can also drastically clean up a lot of traces and vias if you just look at what you see now and adjust for it and try another routing.

For example: 5101 Pin 1 is A3. 6264 Pin 2 is A12. In your design, they are right next to each other. If you change your schematic to be 5101 A3 -> 6264 A12, you get a short trace going pin to pin instead of scrolling all around and passing through a via to get to pin 7 of the 6264.

Same holds true for your choice of gates within the 7400. And you might want to look into at least grounding the unused inputs of the 7400. Floating pins can cause problems. You can leave the unused outputs open

I have found sometimes, it helps to slide the chip back and forth with a little tension on the zip socket on my programmer ( not all the way latched) to get a good contact on some older chips.

I see little benefit to using a NVRAM chip while leaving batteries still connected since the RAM will always be powered and simply act like normal SRAM.

Without looking at the data sheet, it is possible that the Ramtron chip may use more power than a low power CMOS static RAM would. And therefore drain the batteries faster.

There are some active links for aliexpress which will get you some of the lower priced NVRAM being discussed beginning at post #34 in this thread.

And yes, this is for 6264 style dip packages that are plug and play is SOME applications.

There is certainly a lot of value in pre-tested chips being shipped promptly from a US seller! Do you really want to wait 49 days to receive something questionable from China?

Not sure what type of Zip/Zif socket your tester has but many Zif sockets contact the IC pin on the edges that are not normally touching a normal single or double wipe IC socket. So, even if the larger flat surfaces look clean, it may be the thin edges that are corroded from poor storage or used pulls.

Yup, cut traces when you pried the old chip off.

Here is a bit if advice to beginners:. Never pry up on an IC or socket. Use you fingers to slide it left and right and up and down. If it won't lift right out gently, you haven't freed all the pins yet. As you move it, you should be able to see each pin move within the hole. And never dig down with a flat head screwdriver to pry out a chip.

Well, at least he got a photo before installing the socket so we know the source of the failure.

Unfortunately, the socket will need to come out again unless you want a bunch of "blue wire" on the back of the board.

Harbor freight does sell close cutters (red handle $4.99). But they are not narrow tips. They are very sturdy and cut extremely flush. They can do the pins at the pcb but not along the sides of the IC.

If you have an Eprom programmer that either reports using either the Ramtron part, or the Dallas equivalent, you can always run a test of them by programming and verifying 00, FF, CC, and 55. Write, then verify, then write the new pattern and verify again. These 4 patterns (all zeroes, all ones, and then alternate 1's and 0's) normally will catch most bad parts. It may not take into account the shortest access time but that won't really matter since the pinball CPUs are running so slow.

Always do 00 last and you will have the chip ready to go fully wiped of any data.

EDIT:. Sorry for the somewhat duplicate info. I guess I didn't open the last page of posts.

No advantage. Just disadvantage since you would need to replace it in 10 to 15 years from date of install.

The clock requires power to the WPC ASIC chip. So, no external battery, no clock.

You folks realize that the black claw is designed to pull PLCC packages (like the WPC ASIC chip), not normal dip ICs.