I need info on what is a good replacement chip for the sram, ie sharp lh5216ad-10l.
I couldn't even find datasheets for it...
kyuusaku wrote:
Any 0.3" 6116.
Could you be more specific, i'm memory stupid...
6116 is a generic code for 2k x 8 asynchronous SRAM with a JEDEC pinout. Most 6116 come in a DIP package with a 0.3" width, which I assume is what you want, but some are 0.6" wide like other SRAM with more pins.
Here's one for sale:
http://www.jameco.com/webapp/wcs/stores ... _242676_-1
kyuusaku wrote:
6116 is a generic code for 2k x 8 asynchronous SRAM with a JEDEC pinout.
Thank you kyuusaku, this was enough info to figure out that Mouser and Digikey don't have anything exactly like this in stock, but since they are my main sources, would anything else take its place? Maybe something cheaper.
kyuusaku wrote:
I bought all they had left at the time, 31, I hope they get more in, but I can confirm they do work in the top loader just fine.
Still would like to know if there are other chips that can be used in place of the originals and where to get them.
Of course there are more chips, practically every memory manufacturer has made 16Kbit RAM. Why do you need so many? Surely not every broken NES needs their memory replaced... Monopolizing the world's remaining supply?
I just want options, mostly I want a cheaper option.
Also, as far as I know mouser and digikey have none and I bought the rest jameco had and ebay stock is all at least 4X higher than jameco was.
But, like I said earlier, I have little knowledge of these chips and what would work in their place etc.
No I hope not every broken top loader I have needs these or I won't have enough and none for future repairs.
How is $1 not cheap enough?
It's just SRAM nothing special. If you took the time to read the data sheet and or snoop around on the net you could learn how it works. Once you did that you could configure nearly any SRAM that was 2KB or greater to work. The only thing you might have difficulty with is the package and pinout. But here again the datasheet saves the day.
infiniteneslives wrote:
How is $1 not cheap enough?
I guess when i'm so used to finding logic chips for mere pennies, yes over $1 per seems high, but like I said I have no experience with these ICs, maybe it is a great price, compared to ebay sellers it is/was.
infiniteneslives wrote:
It's just SRAM nothing special. If you took the time to read the data sheet and or snoop around on the net you could learn how it works. Once you did that you could configure nearly any SRAM that was 2KB or greater to work. The only thing you might have difficulty with is the package and pinout. But here again the datasheet saves the day.
I'm not a total newb, I did snoop around and it is the package and pinout that is the problem, at least for the top loader.
The chips are tucked in between the cart connector and other chips so using .6" dip chips is not an option.
MB8416A-15-SK
HM6116ASP-15
TMM2115BP-15
LH5116D-12
CXK5816SPS-15L
All of them .3". Need more specific?
80sFREAK wrote:
MB8416A-15-SK
HM6116ASP-15
TMM2115BP-15
LH5116D-12
CXK5816SPS-15L
All of them .3". Need more specific?
Very cool 80sFREAK!
How did you come up with the numbers?
I checked findchips.com and most are not available, some MBs are around, but higher prices than the ones kyuusaku found on jameco.
I believe 6116s haven't been manufactured for many years now, and so anything that you're able to buy is going to be new-old-stock or salvaged. Fortunately with RAM all the address lines are completely interchangeable, so rewiring to fit an oversize one is a fairly straightforward task.
Quote:
I guess when i'm so used to finding logic chips for mere pennies, yes over $1 per seems high, but like I said I have no experience with these ICs, maybe it is a great price, compared to ebay sellers it is/was.
Well these aren't LOGIC chips like the others you're comparing them to. They arw 16,192 bit SRAM. So you should expect them to be reasonably more expensive than a NAND gate. Just to put it in perspective for you the SRAM you're looking for has somewhere around 70,000+ transistors inside. A '00 NAND gate should have less than 100 transistors inside. Not that transistor count directly relates to price, but it has a factor in difficulty to manufacture and test which does have an effect on price.
lidnariq wrote:
Fortunately with RAM all the address lines are completely interchangeable
Data lines are interchangeable as well.
This whole post still boggles me. I can't figure out why someone would need these so badly? Are you just stocking up on them to be able to repair top loaders for eternity? Because if you are I would think that having 30+ of them in your stock would be enough to repair 1000's of toploaders because the chances of this part failing are low. Even still if you NEEDED 100s of them just spend some time learning and you'll be able to replace it with any 0.3" SRAM >=2KB. I'm just saying all this because you keep claiming to be a newb. I'm just letting you know that once you understand these parts you will NEVER have a hard time coming up with a replacement regardless of how many you need.
akaviolence wrote:
80sFREAK wrote:
MB8416A-15-SK
HM6116ASP-15
TMM2115BP-15
LH5116D-12
CXK5816SPS-15L
All of them .3". Need more specific?
Very cool 80sFREAK!
How did you come up with the numbers?
I checked findchips.com and most are not available, some MBs are around, but higher prices than the ones kyuusaku found on jameco.
Ummm it's a long story and i think i missed couple more
So how many chips you want? Be more specific, so you might find answer more easy.
infiniteneslives wrote:
Just to put it in perspective for you the SRAM you're looking for has somewhere around 70,000+ transistors inside. A '00 NAND gate should have less than 100 transistors inside. Not that transistor count directly relates to price, but it has a factor in difficulty to manufacture and test which does have an effect on price.
Ah, makes sense!
infiniteneslives wrote:
This whole post still boggles me.
The only reason I started this thread was to get some HELP finding some replacement chips and I did and I appreciate that. I keep asking questions so that I can learn. Shouldn't I be allowed to ask questions to ppl who will actually know the answers and provide assistance?
infiniteneslives wrote:
I would think that having 30+ of them in your stock would be enough to repair 1000's of toploaders because the chances of this part failing are low.
I've only started repairing my stash of broken consoles and the first one had bad sram x2, cpu, ppu, and reset cap (weird). Yes I want to be ready to have to replace a bunch of them, and if none of the rest of the consoles need them then i'll have plenty of extra and if anyone wants them i'll gladly part with them.
infiniteneslives wrote:
They arw 16,192 bit SRAM. So you should expect them to be reasonably more expensive than a NAND gate. Just to put it in perspective for you the SRAM you're looking for has somewhere around 70,000+ transistors inside.
1024 * 16 = 16384
SRAM cells are typically 6 transistors, so 16384 * 6 == 98304 T, in a 128 * 128 bit array. That means there's also a 7 bit row address decoder == 128 * (7 + 1) = >1024, and a 4:16 decoder for the column + a bit of glue. It's more like 100k+
If we're going by transistor count then the SRAM should cost 1.5x what a NMOS 68000 costs.
akaviolence wrote:
I've only started repairing my stash of broken consoles and the first one had bad sram x2, cpu, ppu, and reset cap (weird). Yes I want to be ready to have to replace a bunch of them, and if none of the rest of the consoles need them then i'll have plenty of extra and if anyone wants them i'll gladly part with them.
You forgot to add blown 7805 on the list. Most likely spike or either wrong power supply. If there could be more electrolytic capacitors, they usually popped as fireworks at double voltage.
Quote:
NMOS 68000 costs
in purple ceramic? WANTED
80sFREAK wrote:
akaviolence wrote:
I've only started repairing my stash of broken consoles and the first one had bad sram x2, cpu, ppu, and reset cap (weird). Yes I want to be ready to have to replace a bunch of them, and if none of the rest of the consoles need them then i'll have plenty of extra and if anyone wants them i'll gladly part with them.
You forgot to add blown 7805 on the list. Most likely spike or either wrong power supply. If there could be more electrolytic capacitors, they usually popped as fireworks at double voltage.
Oddly enough, nope, 7805 and the other electros were fine...
So the Logic brand sram chips I got from Jameco have been causing major issues, so i'm still looking for more help.
Could access times that are too different cause video glitches?
Also, can I use sram with larger organization, like 8kx8?
Here is a video of the video glitches I was getting on the top loader.
What's the actual difference in access time?
Are you 100% certain that they are a pin for pin replacement? Not something like two /CE's on them or something.
There was some discussion of WRAM (work) in the video. That shouldn't be able to cause graphical glitches like that. If the WRAM was bad the game would crash, not graphic glitch.
Make sure the chips have really good connections to the cartridge. /CE and A10 for those VRAM chips are driven by the cartridge. So if those signals were weak you could easily have problems like this. Maybe even more suspect based on the fact they are toploaders. Things like coke can get spilled inside and corrode tracks.
Assuming those chips you got aren't super slow, I'd blame connections with the cartridge CIRAM /CE and CIRAM A10 specifically, CHR /A13, A10, & A11 as well. No other connections could cause problems like this. It's highly doubtful that you've got a mess of bad chips.
Seeing as how I swapped the new and old srams in and out so many times, swapped in better sockets, and swapped in/out and cleaned the carts and connectors on two different top loaders, its not likely the connections, but i'm not ruling it out yet.
I was also seeing different little glitches with different sram chips, but eventually I got two that showed zero glitches, already have 5 chips that are either totally fried or still show some sort of glitch.
I had several instances where a game would not boot, most likely wram had a bad sram chip.
As far as access times and pinout, I don't know, as stated in the first post I can't find a datasheet for the original srams.
akaviolence wrote:
As far as access times and pinout, I don't know, as stated in the first post I can't find a datasheet for the original srams.
What do the actual chips say on them. If they are too slow you'd have this issue. Hard to know if the datasheet isn't available and the part number doesn't give away the speed.
As stated in the first post of this thread, the old ones are sharp lh5216ad-10l.
The ones I got from jameco are logic l6116pc10, according to the
datasheet they are 10ns.
So maybe they are both 10ns chips?
I wasn't sure if those old posts were accurate or not. Yeah they are both 10ns which is plenty fast.
IDK, maybe there is a reason they stopped making them. They are all crap! Either way you should be able to get other 0.3" SRAM to work even if they are larger. It might be easier to bend all the pins up and slap em on the bottom side of the mobo. Remember you don't have to worry about matching the address lines up perfectly. A10 on the mobo could connect to the datasheet's A14. An address line is and address line and same goes for data. A data line is a data line, you CAN'T trade an address line for a data line though or vice versa if that wasn't obvious. So that might make wiring up 8-32KB SRAMs a little easier.
I would bet extremely good odds that the original NES RAMs are 100ns, not 10ns. (My NES-001 uses 0.6" LH5116-15s, which are150ns)
Also, I think the CIRAM's problems in the video look a little like crosstalk.
lidnariq wrote:
I would bet extremely good odds that the original NES RAMs are 100ns, not 10ns.
Yeah you're right on that, I forgot about the dang old x10 factor. The data sheet does say 10ns for those new ones though. I don't think being faster should cause issues...
Being faster might cause issues if the enable is glitchy. PRG RAM is traditionally decoded using NAND(PRG /CE, A14, A13, M2), such as the 74HC20-based circuit in Family BASIC. The trouble is that PRG /CE (used to distinguish $6000 from $E000) changes a few dozen nanoseconds later than the other signals, meaning writes to mapper ports at $E000-$FFFF might momentarily write to $6000-$7FFF if the RAM is too fast. Old versions of the PowerPak's MMC3 implementation had this problem, clearly visible as map corruption in M.C. Kids and Crystalis.
Could this also be exacerbating my powerpak issues?
Very cool to know I can try 8kx8 though, mouser actually has some for $1, and its 70ns, does it have to be _kx8 or can it be _kx16 etc?
The *kx16 is for platforms with a 16-bit data bus, like Sega Genesis. You can theoretically use it with an NES if you're OK with wasting half the chip.
Although what Tepples is saying about the delays with PRG /CE are true it wouldn't contribute to the issues you're having with VRAM. Now there is some similar logic going on with CHR enable signal A13 and /A13. /A13 is merely an inverter's output from the CHR A13 signal. So /A13 comes one gate delay behind A13. BUT, you still have CHR /RD and /WR controlling /WE and /OE and CHR A13 and /A13 are ALWAYS valid when either /RD or /WR are active. So the interlock there with /OE and /WE prevent any sort of timing issue from arising on the CHR bus from too fast of memories.
Also it would be a huge pain to use 16 bit memories. Technically you should leave all the unused data lines pulled high or low with their own individual pull up/down resistor. So not only is your memory excessively large because of the extra data lines but you should also have 8 extra resistors on there. It may very well work with out the resistors, but you could get weird intermittent issues from leaving those inputs floating. So unless you want the headache stick with 8bit memories.
infiniteneslives wrote:
So unless you want the headache stick with 8bit memories.
It's really coming down to a price versus availability issue.
What options are you looking at?
These look like the a good option to me at $1.50-1.80 ea. I didn't find any 16bit that would even fit the bill for 0.3" DIP. Granted I didn't look anywhere besides digikey. You could save ~50 cents if you went soic not worth the hassle IMO though.
http://search.digikey.com/us/en/products/CY7C199CN-15PXC/428-2158-5-ND/1206029
Update on powerpak problem, its not sram related, the original 373 chip labeled as 74HC373 had failed or was not really an HC version, I replaced it with a new TI 74HC373 and it works perfect.
I also happened to find a front loader board laying around and noticed it had .3" sram chips, I had only really looked at one other board before and just assumed they all had .6" srams in them, not so apparently, good news for me.
MB8416-15-SK. Findchips showed some, but not cheap, most say mfg is Fujitsu.