SDRAM board

[TWINKYTWONKA]

I think this is correct and supports my own findings. Ive found that on jlpcb boards if you solder at 310 and don't drag solder you will get 100% success rate and even the 7ctn chips will test at 150mhz ( which they shouldn't i know). However, if you dwell too much and especially at higher temps then you damage the boards. I got someone at work to check these boards out under their fancy microscope and confirm this. Sadly i didn't keep the images

So in summary. If you don't want to spend an evening soldering,checking, soldering,checking, soldering,checking etc then solder at 310C and don't dwell.

hope this helps someone

I cannot confirm this. I use 400C for manual soldering and have no problems. Time after time i even desolder the chips by hot air and solder back. Still no problems.
I think result may depends on flux used and how good it was cleaned.

and i was jumped on , on classic gaming discord for saying i soldered @ 400c haha , funny how folks are quick to jump to there own conclusions, ive been soldering for years at these temps and never had any problems, also used chipquick to re-use smd parts

ha ha. I have also been wearing my lucky moose socks that i got when on holiday in new hampshire, usa a few years back. Is this the key to the error free 128mb boards ?

[digitron]

The silkscreen on the jlcpcb boards is poor quality..

I think this is correct and supports my own findings. Ive found that on jlpcb boards if you solder at 310 and don't drag solder you will get 100% success rate and even the 7ctn chips will test at 150mhz ( which they shouldn't i know). However, if you dwell too much and especially at higher temps then you damage the boards. I got someone at work to check these boards out under their fancy microscope and confirm this. Sadly i didn't keep the images

So in summary. If you don't want to spend an evening soldering,checking, soldering,checking, soldering,checking etc then solder at 310C and don't dwell.

hope this helps someone

Excellent advice, did a couple boards without drag soldering and went down to 310C, 150mhz tests were good. Thanks!

[antoniovillena]

I have adapted XXS to official IO board thanks for the help of Sorgelig. The new model will be called XXSD. I have published open source here:

https://github.com/MiSTer-devel/Hardware_alternative

Ordered today to JLCPCB. I will try in about 3 weeks. It's a small change from XXS but by caution don't order until I have tested.

[Morpheus1979]

Did anyone ordered some 2.4/2.5 boards from OSH park? Are they good (considering they don't provide any advanced ordering options)?

I tried to order some... but OSH park demands the files for the inner layers to be named .G2L and *.G3L:

https://docs.oshpark.com/troubleshootin ... g-pattern/

So do i have to rename *.G1 -> *.G2L and *.G2 -> *.G3L? Is this the necessary layer order(?):

Top


Layer 1


Layer 2


Bottom

[Sorgelig]

So do i have to rename *.G1 -> *.G2L and *.G2 -> *.G3L? Is this the necessary layer order(?):

that's correct.

[Nat]

Did anyone ordered some 2.4/2.5 boards from OSH park? Are they good (considering they don't provide any advanced ordering options)?

I tried to order some... but OSH park demands the files for the inner layers to be named .G2L and *.G3L:

https://docs.oshpark.com/troubleshootin ... g-pattern/

So do i have to rename *.G1 -> *.G2L and *.G2 -> *.G3L? Is this the necessary layer order(?):

I know terminator2k2 ordered some v2.4 boards and they were incorrectly produced. I am sure he can give more details, but I know they refunded his purchase.

[Morpheus1979]

Damn... that's too bad. I was hoping for some well produced pcbs, since i screwed up an order from JLCPCB i did a couple of weeks ago. I ordered some 2.5 boards from them, using Gerber files exported by myself from Altium (because i had failed to notice the official Gerber file packages ).

The boards arrived last week, i soldered all parts onto the first pcb under the microscope, cleaned the board, did some testing... and... couldn't get it to work past 80mhz without getting massive errors. I soldered two more boards, using different flux. I always used Alliance AS4C32M16SB-6TIN memory with Sorgeligs recommended caps setup (Top -> 10µF/0,1µF/10µF/0,1µF; Bottom -> 10µF/1µF/1µF/10µF/10µF/1µF/1µF/10µF), cleaned all the boards thoroughly and checked them under the microscope:



But always the same: above 80mhz i get lots of errors... with every single one of them.

In the end i assumed it may be an issue with the files i exported. That's why i wanted to order a second batch of pcbs - this time using the official Gerber file package. And since JLCPCB sometimes seems to produce some faulty SDRAM boards, i wanted to try another manufacturer to make sure: the boards aren't the source of any issues.

[Sorgelig]

Original gerber v2.5 from me has no problem. Just sometimes JLCPCB complain about random things. It seems depends on specific worker who audit the files. So sometimes need to insist to accept the files. In v2.5 holes of connector are very close to the edge. But as practice shows it's not a problem for JLCPCB to manufacture it.

[terminator2k2]

yes i ordered some from osh park and so did some one else who ive never met and we got the same results back





they put it down to the camera taking images of the render so i canceled and got a refund just incase there was something else a miss

[Sorgelig]

yes i ordered some from osh park and so did some one else who ive never met and we got the same results back





they put it down to the camera taking images of the render so i canceled and got a refund just incase there was something else a miss

Wow! that's very odd. Wrong traces, almost no rings around the vias. And edge milling as always is crap in OSH. They stuck somewhere 10 years in the past..

[Morpheus1979]

Wow indeed - i'm glad i was able to cancel my OSHPark order before it was sent into production.

[THaase]

The only really good thing for OSH Park is: Castellated Holes are damn cheap compared to others

[retronerd]

Hello there folks! I just got my Boards from pcbway! but what kind of caps shall i use? 1UF on all 12 places on the board? thanks!

[Morpheus1979]

Take a look at page 10-13 of this very thread:

Thanks for caps testing!
i use:
top: 10uf, 0.1uf, 10uf, 0.1uf
bottom: 10uf, 1uf, 1uf, 10uf, 10uf, 1uf, 1uf, 10uf

Got the parts and soldered a 128MB SDRAM card. I used Sorgeligs capacitor configuration (shown below) and it seems to work fine (stable @150MHz).

top: 10uf, 0.1uf, 10uf, 0.1uf
bottom: 10uf, 1uf, 1uf, 10uf, 10uf, 1uf, 1uf, 10uf

[danytyler]

Some pass at 110MHz

That must be a bad soldering.
All 128MB modules properly soldered are passing at least 140MHz in memtest.

The soldering is fine. Just switching DE-10 Nano boards shows different results. I have one DE-10 Nano board that won't pass any 128MB module (no matter which type, the MiSTer version or my own design) at more than 110MHz. Pull the module out and put it in another DE-10 board and it passes 140MHz+. < ads are not allowed and has been removed >

Have you found any problem in the usual use of that DE-10 Nano that does not pass the test to more than 110 MHz ?.
Or does it work normally like any other DE-10?

[bazza12]

This is probably a silly question, I don't know anything about making daughter boards, or soldering.. so feel free to laugh at what i'm about to say.. or just disregard it..

I've currently got a 32meg and 64meg SDcards.. Is there anyway to connect them both to have 96meg? like a double header that can be fitted on to the existing GPIO header?.. or is this just nonsense that would only work in my head..?

[witchmaster]

This is probably a silly question, I don't know anything about making daughter boards, or soldering.. so feel free to laugh at what i'm about to say.. or just disregard it..

I've currently got a 32meg and 64meg SDcards.. Is there anyway to connect them both to have 96meg? like a double header that can be fitted on to the existing GPIO header?.. or is this just nonsense that would only work in my head..?

It's all nonsense, just get a 128MB card!

[bazza12]

It's all nonsense, just get a 128MB card!

fair enough lol

[retronerd]

I just assembled my 128MB mister sdram. When i plug it in it doesnt detect 128mb, only half of the chips. Do i need to change something in the ini file?

[terminator2k2]

I just assembled my 128MB mister sdram. When i plug it in it doesnt detect 128mb, only half of the chips. Do i need to change something in the ini file?

no it should auto detect......if it only shows 64MB and works ok with 64MB then check U3 and U2

[TWINKYTWONKA]

Damn... that's too bad. I was hoping for some well produced pcbs, since i screwed up an order from JLCPCB i did a couple of weeks ago. I ordered some 2.5 boards from them, using Gerber files exported by myself from Altium (because i had failed to notice the official Gerber file packages ).

The boards arrived last week, i soldered all parts onto the first pcb under the microscope, cleaned the board, did some testing... and... couldn't get it to work past 80mhz without getting massive errors. I soldered two more boards, using different flux. I always used Alliance AS4C32M16SB-6TIN memory with Sorgeligs recommended caps setup (Top -> 10µF/0,1µF/10µF/0,1µF; Bottom -> 10µF/1µF/1µF/10µF/10µF/1µF/1µF/10µF), cleaned all the boards thoroughly and checked them under the microscope:



But always the same: above 80mhz i get lots of errors... with every single one of them.

In the end i assumed it may be an issue with the files i exported. That's why i wanted to order a second batch of pcbs - this time using the official Gerber file package. And since JLCPCB sometimes seems to produce some faulty SDRAM boards, i wanted to try another manufacturer to make sure: the boards aren't the source of any issues.

Did you buy any chance buy the chips from mouser? I purchased some 6 tin chips last week and soldered them to the 2.4 boards that that i have used previously for both 6tin and 7tcn chips. Like your post, none of these newer chips would test at more than 80mhz and yet 6tin ( and 7tcn) chips from previous batches all run at 140/150 mhz. I spent a day soldering and desoldering ram chips and despite the mouser 6tin chips having the have the same serial number as previous batches they are unable to test higher than 90 mhz.

I have contacted mouser who have in turn contacted alliance. They have suggested the following


For the power supply filter, we recommend The front stage may uses 4.7uf~10 uf to remove low frequency noise. This capacitor should be near the header.
For this circuit, we need one or two front capacitors should be ok, the rest are the bypass capacitors.
The second stage bypass capacitors use 0.1uf to remove high frequency noise. Generally, the front capacitor the bigger the better, the different values of two capacitor is about 100 times.

I have just emailed them again to clarify capacitor value and positioning and once i receive any updates i will post it here

[Sorgelig]

I have just emailed them again to clarify capacitor value and positioning and once i receive any updates i will post it here

Waiting for their reply.

As for chips - it's weird. Looks like broken or even fake batch. May be re-marked batch of older chips.

[TWINKYTWONKA]

I have just emailed them again to clarify capacitor value and positioning and once i receive any updates i will post it here

Waiting for their reply.

As for chips - it's weird. Looks like broken or even fake batch. May be re-marked batch of older chips.

There is definately some funny business going on as the numbers on my dodgy 6tin chips also match those of the chip shown in morpheus119 post above.

[Sorgelig]

There is definately some funny business going on as the numbers on my dodgy 6tin chips also match those of the chip shown in morpheus119 post above.

can you post hires photo of this chip?

[TWINKYTWONKA]

There is definately some funny business going on as the numbers on my dodgy 6tin chips also match those of the chip shown in morpheus119 post above.

can you post hires photo of this chip?

Here you go:

https://photos.app.goo.gl/dsH9mwZ7X7VAAQpe9