So far I've soldered on the SRAM modules and caps. I had no problem using this Hakko bent tip, T18-BR02.

Next up will be adding these flat pins. It's the last of my flat pins.

You have soldered SRAM chips, not modules. The module is where chips got soldered onto 🤣.

It looks fun, I hope it works without problems. SOJ and PLCC are a bit pain to hand solder, especially with minimal clearance...

feipoa wrote on 2025-10-02, 13:07:

So far I've soldered on the SRAM modules and caps. I had no problem using this Hakko bent tip, T18-BR02.

From what I can see on the pictures, the soldering job looks great: No visible damage or burns to the PCB. Obviously, I can't judge the quality of the joints, but I trust you that the joints are good.

I completely failed when I tried it on the v1.0 layout with the shorter pads and no bent tip. The bent tip might be a real game changer to get to the pads with the small clearance between the SOJ chips.

How did you solder them? Using drag soldering or using paste, or in some completely different way?

Also, it is great that you didn't mess up the orientation of the chips. It's quite tricky on that PCB, as the two rows are oriented in the different way. The reason for that is that the data bits are tapped in the center row between the chips, and all chips are oriented with their data pins towards the center.

I noticed the batch number printed on your PCB. I assume you didn't have the PCB manufactured by JLCPCB, because that manufacturer will place the batch number where the silkscreen says "JLCJLCJLCJLC". Do you know what I could have done to hide the patch number with you manufacturer? Maybe I could add that marking below a different chip.

Tiido wrote on 2025-10-02, 14:48:

You have soldered SRAM chips, not modules. The module is where chips got soldered onto lol.

You are right. Sometimes my fingers jump ahead of my thoughts. That should have read, "So far I've soldered the IC's onto the SRAM module."

Joints look good enough. I didn't spend any time on beautification. Some pins have a larger solder blob than necessary.

I didn't drag solder - just went one by one on the pins using some CHIPQUIK NC2SW.015, which is SN60/PB40, 0.38 mm, no-clean. Iron at 330 C. I wanted to use solder paste, but thought it would be more meaningful to see if these modules were suitable for a traditional iron.

I caught the chip orientation early on, but I can see how someone not paying attention might flip the IC's on the other end.

The PCB was from JLCPCB. Maybe they aren't always consistent with their JLCJLCJLCJLC.

I have a very unstable HOT-433 rev4 board (it's stable as long as I turn off L2), so I'm using this thread as a reference to design a board for it.
For now, I'm making it with a design that looks like the UUD board with one bank removed, but I'm a little unsure if this is correct. If anyone has any knowledge about this, I'd appreciate some advice.

feipoa wrote on 2025-10-03, 00:52:

The PCB was from JLCPCB. Maybe they aren't always consistent with their JLCJLCJLCJLC.

I suspect they are manually checking the "Specify Position" option and replacing "JLCJLCJLCJLC" with the order number, so inconsistencies are common.
I often forget to turn on the "Specify Position" option, but sometimes "JLCJLCJLCJLC" is replaced with the order number.
(Plus, I've seen blog posts where a typo like "JLCJLCJLCJCL" was replaced with the order number.)

feipoa wrote on 2025-10-03, 00:52:

Joints look good enough. I didn't spend any time on beautification. Some pins have a larger solder blob than necessary.

Indeed, this looks very nice, even if the blob sizes are not fully consistent. I reckon there is a high chance for the PCB to work on the first attempt.

feipoa wrote on 2025-10-03, 00:52:

The PCB was from JLCPCB. Maybe they aren't always consistent with their JLCJLCJLCJLC.

kagura1050 wrote on 2025-10-03, 03:53:

I often forget to turn on the "Specify Position"

And I didn't mention you should turn on "Specify Position" in the documentation for the project, that might be a sensible addition. By default, JLCPCB does not look for the JLCJLCJLCJLC mark, only if you choose the non-default (but still free) option to specificy the position.

Because I'm using flat pins without a standard header anchor, I decided to bring the PCB down closer to the SRAM sockets. To do this, I needed to swap the 10 uF electrolytic capacitor for a shorter capacitor and to notch the PCB to accompany JP5. One more, equally important, reason I decided to set the PCB down lower was the difficulty of soldering the pins with the top block connector. If I did not bring the PCB down, the distance between the top block connector and the PCB was only about 2 mm. If I were to remove the top block, the pin alignment would be off, even though they are anchored in the SRAM socket. Here's some photos of this work.

And the finished module. Please let me know if I made any glaring mistakes.

Do you know if JLCPCB allows for these notches in the PCB layout?

If I were to reorder PCB's, I might choose black colour. The orange-green contrast is a bit much. What do you guys think?

Does Mouser or Digikey still supply these SRAM IC's new? If not, has anyone found a reliable source for new/original pieces? I might make a few more of these if they workout well.

mkarcher, where did you route your A19 bodge wire?

mkarcher wrote on 2025-10-03, 05:16:

Indeed, this looks very nice, even if the blob sizes are not fully consistent. I reckon there is a high chance for the PCB to wo […]
Show full quote

feipoa wrote on 2025-10-03, 00:52:

Joints look good enough. I didn't spend any time on beautification. Some pins have a larger solder blob than necessary.

Indeed, this looks very nice, even if the blob sizes are not fully consistent. I reckon there is a high chance for the PCB to work on the first attempt.

feipoa wrote on 2025-10-03, 00:52:

The PCB was from JLCPCB. Maybe they aren't always consistent with their JLCJLCJLCJLC.

kagura1050 wrote on 2025-10-03, 03:53:

I often forget to turn on the "Specify Position"

And I didn't mention you should turn on "Specify Position" in the documentation for the project, that might be a sensible addition. By default, JLCPCB does not look for the JLCJLCJLCJLC mark, only if you choose the non-default (but still free) option to specificy the position.

does the batch number really bother you that much? I couldn't care less about it

feipoa wrote on 2025-10-03, 11:46:

And the finished module. Please let me know if I made any glaring mistakes.

Looks fine. I don't see any glaring mistakes.

feipoa wrote on 2025-10-03, 11:46:

Do you know if JLCPCB allows for these notches in the PCB layout?

If I remember correctly, non-rectangular boards are supported, but more expensive. I didn't care about the notch, because I built the board in a way it is located above the jumpers. While the board is installed, I don't need to change the cache configuration, so I don't care some jumpers are covered.

feipoa wrote on 2025-10-03, 11:46:

Does Mouser or Digikey still supply these SRAM IC's new?

I "recently" (about a year ago) ordered some from mouser. Today, I can't find cy7c1009 chips at mouser any more.

feipoa wrote on 2025-10-03, 11:46:

mkarcher, where did you route your A19 bodge wire?

Re: 486 board with UMC 8881E/8886B: The winner is: EDO without L2 (if your only other option is L2 at 3-2-2-2) has a photo. I recommend to fix the bodge wire with nail polish or super glue as strain relief for the solder joint if you use that approach.

In this image, download/file.php?id=165901&mode=view , there is a pin via next to C6 which isn't soldered. What is the function of this?

I understand why you didn't notch your PCB. For those using flat pins, notching the PCB and shortening the capacitor will make soldering substantially easier.

I forgot about you taking the bodge wire directly to the chipset. I'm not sure if I'll do that. Depends on the test results.

feipoa wrote on 2025-10-03, 23:12:

In this image, download/file.php?id=165901&mode=view , there is a pin via next to C6 which isn't soldered. What is the function of this?

That's an error in that PCB revision. That pin was supposed to be a ground pin, but it is in the wrong location, so I had to omit it.

feipoa wrote on 2025-10-03, 23:12:

I forgot about you taking the bodge wire directly to the chipset. I'm not sure if I'll do that. Depends on the test results.

It's not soldered directly to a chipset pin, but I removed solder mask on a trace from the CPU to the chipset, and soldered the bodge wire (enemalled copper, not litz wire) onto the trace.

Haven't been able to test the board yet, but I decided to order some CY7C1009D-10VXI chips from Aliexpress. They were only $1.42 each chip, after shipping. Seller stated they are new/original. They look pretty legit. What do you think? Here's a side-by-side with a digikey sourced CY7C1009D.

feipoa wrote on 2025-10-21, 02:08:

Haven't been able to test the board yet, but I decided to order some CY7C1009D-10VXI chips from Aliexpress. They were only $1.42 each chip, after shipping. Seller stated they are new/original. They look pretty legit. What do you think?

Obviously, these chips were manufactured on different machines. They are also 1.5 years older, so Cypress might have changed on what machine they produce these chips on, or Cypress might have multiple factories. On the other hand, I don't think Cypress sells a big lot of these chips. Sure, a guess there is a steady demand for them, as they are currently in stock at the major retailers, but I wouldn't expect anything than legacy designs or boutique low-volume devices to still require 5V SRAM. So I don't think it makes sense for Cypress to produce at multiple places, unless geopolitical reasons make shipping them between different regions too risky or expensive (read: war, embargo or tariffs). Also, these chips are a very mature design that can be run on machines that are already "ready for decomissioning", so I don't see why they would need to change machines, unless a machine breaks downs and they use a more modern machine as replacement, which of course could happen.

I think I really should grab my scope and do comparative measurements between my Mouser chips and my chips from a random Chinese seller I forgot (DealExtreme? Banggood? AliExpress?), and post pictures. I already did take a first measurement at response times of the Chinese chips when I was disappointed by the timings I obtained from that board, and as I reported, the timings I measured were consistent with what you would expect of 15ns chips, but absolutely not what the Cypress 10ns chips are specified to do. To exclude the possibility that I don't know how to properly measure signals at that speed, or that higher load on the outputs increases response time, I won't put ultimate trust into my measurements until I've cross-validated that I do measure the expected times on the Mouser chips.

So, with your chips, I expect them to be 128k x 8 SRAM chips, but I have no confidence that they are acutally Cypress chips or actually 10ns chips, yet they might be.

I see your AliExpress chips indicate "Philippines" as manufacturing country (in the mold), while the Digikey chips indicate "Taiwan" (TWN printed on the bottom side). You might want to research whether Cypress ever produced in (or outsourced production to) the Philippines, if you want to dig deeper. I have no idea which sources are useful for that research, though.

Your thoughts paralleled those of my own.

I see listed on the Wikipedia page that Cypress has manufacturing facilities in San Jose, Ireland, Philippines, and India, however I could find no reference to production in Taiwan for 2018-2022. Maybe the Digikey supplied IC's are the fakes? Or maybe Cypress outsourced production to a firm in Taiwan.

Your IC's have a 2012 datecode. What country is stamped on the bottom side?

feipoa wrote on 2025-10-21, 07:39:

Your IC's have a 2012 datecode. What country is stamped on the bottom side?

I have a low-quality photo of these suspected fakes at Re: 486 board with UMC 8881E/8886B: The winner is: EDO without L2 (if your only other option is L2 at 3-2-2-2) which primarily shows a serial number on the back side of the chip. The case with the two dips in it is like your "TWN" chip. I will take a look at the suspected genuine products and the fakes the next days and send a better photo.