Hi, could anyone educate me on static ram speeds. I've got a faulty one I think on a raiden 2 board and it's a CY7C185-35PC, the nearest exact replacement would come from Poland with a large postage fee. The -35 is it's speed ? If so could I for example use a -15 because it works faster or are my thoughts wrong ?
Static ram speed ?
- Thread starter bones
- Start date
You can always use faster RAMs than required, but you have to have a look into the datasheet to know the meaning of the marking. The extension -35 will most likely be the speed, but only the datasheet for this special RAM can tell you what it means. The manufacturers of storage devices used different systems to mark the speed. On EPROMs, its usually 10x the number attached - so e.g. a 2732-45 is 450ns access time. On other chips, e.g. RAMs, -55 can mean 55ns access time. Some RAMs are marked e.g. -2 or -3 - there's a table in the data sheet telling you the meaning.
NivagSwerdna
Active member
Short answer... See above.
Long answer... See the datasheet. You will see the 15 or 35 appear in the min access time ns column. So 15 is faster than 35 so good.
Many of these more modern devices are multiples super fast compared to their original design specs on the board.
Long answer... See the datasheet. You will see the 15 or 35 appear in the min access time ns column. So 15 is faster than 35 so good.
Many of these more modern devices are multiples super fast compared to their original design specs on the board.
Thanks for that,makes sense. I've just ordered 2 of these with -20 on the end(UK seller) and now feel confident that they'll get my raiden 2 up and running again, hopefully.
The problem was pin 19 on this chip being permanently high(4.19) from power up. If I quickly tapped it to ground then missing background,sprite and colour faults would correct themselves. Pin 19 was described as I/O 5 I believe so I'm rather hoping it's outputting and not being driving high by the square surface mount chip I've so far traced it to. I do have in mind that a through hole means that it also goes somewhere else that could be the issue but without schematics I've yet to trace it out,so I've taken a punt fingers crossed lol
The problem was pin 19 on this chip being permanently high(4.19) from power up. If I quickly tapped it to ground then missing background,sprite and colour faults would correct themselves. Pin 19 was described as I/O 5 I believe so I'm rather hoping it's outputting and not being driving high by the square surface mount chip I've so far traced it to. I do have in mind that a through hole means that it also goes somewhere else that could be the issue but without schematics I've yet to trace it out,so I've taken a punt fingers crossed lol
GadgetFreak
Active member
Some boards are timing sensitive to ram speed changes but these are usually the older boards from the early 80's.
Also a -20 isn't always 20ns it could be 200ns, it depends on the manafacturer, though this is most common with 6116 type sram.
2114's are different again altogether as often listed as -1 upto -4, where you have to refer to the correct manafacturers datasheet to get the exact timing.
But as you are talking about 8Kx8 skinny sram you should be fine with what you have gone for.
Also a -20 isn't always 20ns it could be 200ns, it depends on the manafacturer, though this is most common with 6116 type sram.
2114's are different again altogether as often listed as -1 upto -4, where you have to refer to the correct manafacturers datasheet to get the exact timing.
But as you are talking about 8Kx8 skinny sram you should be fine with what you have gone for.
I did read something about that and decided to stick with the same manufacturer (Cypress) and hope their labelling was consistent in regards to the speed. I guess I'll find out hopefully by Saturday at the latest.
I sort of played the game in its current state and found it to be credit hungry but it could my reactions dulling down with old age lol. Tried playing hyper Olympics on joystick but found that I couldn't get the speed I used to. I found something I could use to cheat and got a 8.15 second 100 metre dash having started well after the gun and can get the speed up beyond 1400 for the other events, sneaky chesty me lol.
favouredson
Super Star
- Credits
- 1,260CR
If you check the datasheet you will see that the -35 does refer to the speed (35ns) and that the -15 needs more current.
Also the pinout Cypress used is not the same as, for example, a Motorola MCM6264. The address lines are all on different pins and so you do need to stick with the Cypress IC.
GadgetFreak
Active member
It makes no difference what order the address lines are in because every time the board requires a certain address it will be the same location. It may not be continuous on a chip with swapped address lines but that doesn't make any difference to its operaion at all.
The only thing non standard about the CY7C185 is that it has a second Chip Enable which is active high on pin 26 which if used could cause issues as this is not standard for this size of chip. But if this line is held permanently high it sould not be an issue.
NivagSwerdna
Active member
They look the same to me. As above the numbering on the A's and the DQ's/I/Os is not significant. What goes in comes out the same place. Both are 300mil wide DIP.
GadgetFreak
Active member
It could also be where the signal goes. Depending on how the ram is interfaced it could go straight to a custom or if it is memory mapped to a CPU then it will likely go into a 74LS245 or quivalent which would have direction control.
Well update time, new ram arrived and when installed made absolutely no difference. Serves me right for jumping to conclusions without fully investigating.
I managed to trace the high reading to an output of a ls245, I think someone mentioned that. The input for that was not bouncing about like the others. This I traced back to the output of another ls245 but also appeared to go to a large square chip. The latters trace I could visibly see going to this chip but continuity to its pin was not present. Further inspection with my cheapy digital plug into the phone microscope showed a blob of flux like material on part of the trace. I cleaned this off with alcohol and a cotton bud to reveal a 'puncture' in the trace. I scraped the silk screen off of the trace both sides of the break. I powered up the PCB and then used a bit of wire to temporarily make the connection and everything looks good apart from some lines in places at certain times. I will make the trace good now I'm sure it wasn't deliberately done and will have to check that I haven't created a problem when I desoldered the ram. It was a nightmare to remove, I used a desolder gun, handheld solder sucker and chem wick braid to get this bugger out but nothing would clear 5 of the holes. Never had this before, I ended up having to pop a fresh resistor in each hole and draw the solder onto its legs while rotating it round and round whilst withdrawing it. I very nearly had to get the 0.5mm drill bit out lol. Anyway thanks for all the advice and hopefully I can get it running 100% later on.
I managed to trace the high reading to an output of a ls245, I think someone mentioned that. The input for that was not bouncing about like the others. This I traced back to the output of another ls245 but also appeared to go to a large square chip. The latters trace I could visibly see going to this chip but continuity to its pin was not present. Further inspection with my cheapy digital plug into the phone microscope showed a blob of flux like material on part of the trace. I cleaned this off with alcohol and a cotton bud to reveal a 'puncture' in the trace. I scraped the silk screen off of the trace both sides of the break. I powered up the PCB and then used a bit of wire to temporarily make the connection and everything looks good apart from some lines in places at certain times. I will make the trace good now I'm sure it wasn't deliberately done and will have to check that I haven't created a problem when I desoldered the ram. It was a nightmare to remove, I used a desolder gun, handheld solder sucker and chem wick braid to get this bugger out but nothing would clear 5 of the holes. Never had this before, I ended up having to pop a fresh resistor in each hole and draw the solder onto its legs while rotating it round and round whilst withdrawing it. I very nearly had to get the 0.5mm drill bit out lol. Anyway thanks for all the advice and hopefully I can get it running 100% later on.
NivagSwerdna
Active member
Sometimes getting the solder out of holes can be a real b*****. The problems really start when you have a hole which is part of a big copper pour without thermal relief; just done a 28 pin socket and pins 14 and 28 were a nightmare.
Using a bit of solid core wire and dragging it through is a good tactic.
I recently invested "Stainless Steel Hollow Needles Desoldering Tool Core Desoldering" from EBAY. Not sure if they are really any good and only the smallest size is small enough. It was a good psychological crutch during my recent stuggles.... a piece of wire would probably have been just as good.
Good Luck
Using a bit of solid core wire and dragging it through is a good tactic.
I recently invested "Stainless Steel Hollow Needles Desoldering Tool Core Desoldering" from EBAY. Not sure if they are really any good and only the smallest size is small enough. It was a good psychological crutch during my recent stuggles.... a piece of wire would probably have been just as good.
Good Luck
Warming up the PCB before desoldering may help. Best done with a regulated infrared preheating station, of course. If not available, using a good (regulated) heatgun may do the job as well. But it's important to be careful not to heat up the PCB too much. The temperature of the PCB has to remain below the temperature the solder is melting. I'd recommend approximately 150°C for SnPb40 solder (almost every arcade board of the 80s/90s). Prevent delaminating the PCB or stressing the components by overheating or boiling the PCB for a long period of time. As much as neccessary, as short as possible. The reason this helps is the soldering iron has to lift the temperature by e.g. 30° instead of 160° if not preheated, so much less power is required.
How good or bad desoldering works depends a lot on the PCB. Sometimes it's better to fill the holes again with new solder and remove it again at once afterwards instead of trying to get solder out of a hole where 3/4 of the solder has been removed already. Desolder wick also takes away a lot of heat, so a soldering station providing enough power is essential.
Despite working on PCBs for decades now, I still haven't found a perfect method that works all the time. Watch the PCB while doing and stop if there are beginning signs of delamination.
How good or bad desoldering works depends a lot on the PCB. Sometimes it's better to fill the holes again with new solder and remove it again at once afterwards instead of trying to get solder out of a hole where 3/4 of the solder has been removed already. Desolder wick also takes away a lot of heat, so a soldering station providing enough power is essential.
Despite working on PCBs for decades now, I still haven't found a perfect method that works all the time. Watch the PCB while doing and stop if there are beginning signs of delamination.
cliff_poole
Active member
Well done on finding the problem. So many times I've spent hours looking for faulty components only to find damaged tracks. I guess we have a tendency to overthink sometimes.
I always apply liquid flux when desoldering, more mess to clean up afterwards but it does help. Also as previously suggested, don't hesitate to add fresh solder. I'm not sure if it mixes with the old solder or if it is just the new flux, but it does seem to work a lot of the time. It is all about getting heat transferred to the right place, and the flux is there to aid conduction.
I always apply liquid flux when desoldering, more mess to clean up afterwards but it does help. Also as previously suggested, don't hesitate to add fresh solder. I'm not sure if it mixes with the old solder or if it is just the new flux, but it does seem to work a lot of the time. It is all about getting heat transferred to the right place, and the flux is there to aid conduction.
Well managed to bridge the gap with a single copper core from a wire, my word this is tricky with these tiny traces the square chips require. My eyes are bad now so I had to set up the phone microscope and perform this almost remotely, Daytona USA rams may be beyond me lol.
Bonus is that it looks half fixed, the plane,colours and missing backgrounds look back to normal. The start up screen where the plane 1st rises is still corrupted with lines as to are some areas where the ground bushes are but not the background tiles. I don't think I created a problem as the fault does look the same as before I changed the ram. By removing a 40pin EPROM next to the 2 gal/pal chips and then powering it up I can see that the only affected areas are the ones I have the issues with so at least I have an attack plan. A quick dab with my rubbish logic probe showed these pal/gal chips seem to be nice and active but maybe like the last galaxian board I looked at they may have weak signals or have been corrupted.
Bonus is that it looks half fixed, the plane,colours and missing backgrounds look back to normal. The start up screen where the plane 1st rises is still corrupted with lines as to are some areas where the ground bushes are but not the background tiles. I don't think I created a problem as the fault does look the same as before I changed the ram. By removing a 40pin EPROM next to the 2 gal/pal chips and then powering it up I can see that the only affected areas are the ones I have the issues with so at least I have an attack plan. A quick dab with my rubbish logic probe showed these pal/gal chips seem to be nice and active but maybe like the last galaxian board I looked at they may have weak signals or have been corrupted.