JLCPCB - I give up

[Les Bird]

I've built multiple CPU boards from a known good design and they just don't work. From the old reliable 2.x GIDE design (JLC boards) and from my new Z80 design. For my new designs I ordered a new batch and it's hit or miss. Some work, some don't. I've swapped parts from a known good working board to the bad board in case there was a bad part and still didn't work. I put the chips from the bad board on the good board and the good board worked perfectly. Super frustrating. It's at the point now that when I get a new design and it doesn't work I can't tell if it's in the design or just a defective board.

Doing some research I found this from the owner of OSH Park when someone asked why JLC was so cheap. After reading this I just can't trust JLC anymore so I think I'll be shopping around for a new PCB fabricator.

I will say their prices are spot on for hobbyist but I can't trust their quality.

OSH Park on the other hand is ridiculously expensive. Based on their cheapest price of $5/square inch a 12x6 board would cost $360 for 3 prototype boards.

"Their boards use low quality Tg130 or Tg140 substrate. It begins to burn and break down at normal soldering temperatures, so you have to be quick and careful or you'll get lifted pads. We use Tg180, which has a MUCH higher breakdown temperature. You can hold the iron on it as long as you want without damaging the board.

Their boards use leaded HASL finish. That works fine enough for most uses, but if you're doing surface mount parts, it's nice to have the flatness of ENIG. HASL is more prone to causing shorts in my experience.

They use crappy low resolution silkscreen. We use high resolution LPI silkscreen, so you can use smaller text and have it be clearer. They mess with your silkscreen to add order numbers to it. We don't.

We have better registration tolerances. The Asian fabs tend to widen the window around your pads, so they can get away with lower registration tolerances. This can remove soldermask between pads and open you up to shorts during manufacturing.

In our competitive analysis, their tolerances on mask and silk to copper is 5-7 mils. Ours is maximum 2, and 95% of the time it's well under 1.

Their drill position tolerances tend to be +/- 5 mils. Ours is 2. Their drill size tolerances are +/- 5. Ours is 1.5.

They claim e-testing, however our competitive analysis boards routinely come back with 70% bad. The analysis boards are designed to be difficult to manufacture, but we can fab it right 100% of the time."

[Glenn Roberts]

In might be worth having Norberto look at the gerber files, sort of a “peer review”. Double check that it passes all the automated design checks. I don’t think we’ve seen this kind of failure rate on our other JLCPCB boards?

Sent from my iPad

On Jul 13, 2023, at 1:28 PM, Les Bird <lesb...@gmail.com> wrote:

I've built multiple CPU boards from a known good design and they just don't work. From the old reliable 2.x GIDE design (JLC boards) and from my new Z80 design. For my new designs I ordered a new batch and it's hit or miss. Some work, some don't. I've swapped parts from a known good working board to the bad board in case there was a bad part and still didn't work. I put the chips from the bad board on the good board and the good board worked perfectly. Super frustrating. It's at the point now that when I get a new design and it doesn't work I can't tell if it's in the design or just a defective board.

--
You received this message because you are subscribed to the Google Groups "SEBHC" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sebhc+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/7757479b-8cf6-4f6d-8d84-92d13762ce6cn%40googlegroups.com.

[Joseph Travis]

For lack of a better phrase, "that sucks!"  I know it's a pain in the posterior to troubleshoot new boards at times.  Glenn has a good suggestion but, it doesn't really help you with the boards you have.  About the only thing you can do is verify all connections point-to-point paying close attention to the vias.  Good luck!

Joe

[smb...@gmail.com]

Of all the H8 boards I've made lately, I only recall one that turned out to be bad -- a surface scratch, and I think it might have been me who scratched it.

My smaller boards, Atari 5200 boards and ET-3400 boards have had a few known-bad rejects, including one that made it to a customer. This caused me to visually inspect all boards before shipping, and I have a small reject pile of these small boards (most of which are probably fine, but would either need closer inspection or have visual blemishes which might cause the customer to think they're bad when they're not.

I don't care about the quality of the silkscreen (so long as it's usable). The drill accuracy, soldermask accuracy, etc., is all acceptable. Maybe if you were using tiny SMD packages right up to the design limits it would matter, but for our through-hole boards the accuracy of production is sufficient.

I'm a fan of root causing things -- if it was me, I'd take your boards and desolder all the chips and find out if it's a board problem. An advantage two layer boards, with components removed, you can see everything. If there's a bridge or a dropout or a scratch, you'll see it.

Also, always inspect boards before building -- it's much easier to spot a problem before you've put the components on it. Most common defect I've run into is a scratch, and scratches will be obvious under close inspection.

I used Osh Park prior to jlcpcb. Good boards from Osh Park -- back when I used them all boards were ENIG which was nice. One annoyance was little tabs left between boards when they were panelized, which I had to cut off. 

Scott

[Royce Taft]

I use Aisler (https://aisler.net/). Although you pay in euros, in the U.S. my small (3x PCBs in one order) PCB projects have delivered within a couple of weeks. Prices were reasonable, but my PCBs were small and only double sided. 

I’m not sure if their service will suit your needs, but I figured I would add them to the discussion.

Royce

On Jul 13, 2023, at 10:46, smb...@gmail.com <smb...@gmail.com> wrote:

Of all the H8 boards I've made lately, I only recall one that turned out to be bad -- a surface scratch, and I think it might have been me who scratched it.

--
You received this message because you are subscribed to the Google Groups "SEBHC" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sebhc+un...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/b1e7e5a7-afc4-4214-8358-cb363192d66bn%40googlegroups.com.

[Lee Hart]

I normally use Advanced Circuits <4pcb.com> for my boards. They cost more, but out of 1000+ PCBs I've only had a few defective ones. I've tried JLCPCB and PCBway; and while they are indeed cheaper, they are also lower quality and I've had a number of bad ones. So I've used them for prototypes and cheap low-complexity projects, where people care more about price than quality.

One thing that helps with any PCB is not to "push" the resolution. I use larger holes (0.030" min), wider traces (0.015" min), and wider spaces (0.015" min) than most.

[Les Bird]

Glenn, it's beyond the Gerber review period at this point. The boards have been known to work so it's a board quality issue at this point in time.

Scott, I've had a batch of H17 boards that came with visible defects (I posted about this earlier). The defect caused a trace to be shorted to GND (from +5). It was at that point when I started inspecting all the new boards under a lit magnifying glass. For the Z80 boards I'm having issues with I can't see any visible defects so now I'm wondering if it's the cheap TG material they are using that is causing pads to lift or short (based on what the OSH Park guy said). Maybe I need to be quicker with the soldering iron. This brings up another question, what temp do you guys usually use for soldering? I have mine set at 710 degrees. Is that too high?

Royce, thank you for your suggestion. I will consider them for some of my future projects although a couple weeks is a long time to wait. One thing I like about JLC is they'll ship within a day or two and I get the boards in my hands in about a week from when submitting my Gerbers.

Lee, good advice. I've been using 0.01" width traces on a grid of 0.025" (so tracks are 0.025" apart). Might be a good idea for me to bump those up a bit. One thing I did in one of my new designs is I started spacing out the vias more when there are two next to each other so that they don't short together. I will take a peek at 4pcb.

Les

[Les Bird]

...also regarding the question above about soldering iron temp, I've been using lead-free solder.

[norberto.collado koyado.com]

No issues so far with the JLCPCB shop. I have built a lot of boards and shipped out others as well. On desoldering parts, I have removed bad IC sockets and resistors without any issues. On my Z180 board I removed twice the header for the DUART module which is 40 pins and for the extra RAM board and still working fine. 

I do believe in doing the pcb board checkout to make sure we have proper clearance between traces and vias. Also, on the silkscreen I bold the letters/numbers so that they look great, otherwise they might fade away when printed at the pcb shop.

When I did the Z80 V2.x board and created the schematics for the first time, to develop the Z80 V3. x board, the pcb board checkout light it up as a Xmas tree. I had to manually re-traced them again. It was time consuming and painful but needed to be done.

Not sure what else to say as I have not seen any functionality issues. I just solder the parts and it works.

I do share with Les his pain as it is sad to solder all the parts onto a board and not to work due to pcb issues. It is unnecessary waste of money and time. 😥

I have been sending early prototypes to Joe Travis as he is great in finding issues to assembled and no issues with the board functionality. 

My latest H8-H47 design went to Scott, he put it together and it worked. 

Recently I asked Todd to add more copper to the new boards to provide additional protection to avoid lifting the pads when desoldering parts.

Les, I can review the board files for your most problematic board if you want. 

Best Regards,

Norberto 

Get Outlook for iOS

---

From: se...@googlegroups.com <se...@googlegroups.com> on behalf of Les Bird <lesb...@gmail.com>
Sent: Thursday, July 13, 2023 11:57:27 AM
To: SEBHC <se...@googlegroups.com>
Subject: Re: [sebhc] Re: JLCPCB - I give up

 

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/2e0f0b4f-66fa-4277-b073-0fc54769ba5cn%40googlegroups.com.

[norberto.collado koyado.com]

Also, you must factor in manufacturing process variations and if tolerances are tight that could lead to shorts on traces or vias.

Get Outlook for iOS

---

From: se...@googlegroups.com <se...@googlegroups.com> on behalf of norberto.collado koyado.com <norberto...@koyado.com>
Sent: Thursday, July 13, 2023 12:52:32 PM
To: se...@googlegroups.com <se...@googlegroups.com>

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/SN6PR01MB38559CADF936830C872CA9C5F737A%40SN6PR01MB3855.prod.exchangelabs.com.

[glenn.f...@gmail.com]

So I admit I’m a near total novice at the science of PC board design/mfg.  but I observe that whatever Norberto has been doing seems to have led to some pretty robust boards. I can imagine that (perhaps) insufficient clearances between traces, or some other design issue could lead to problems that would be sporadic?  He’s volunteered to look at the design and it might be worth taking him up on it.  My sense is there’s a lot of “seat of the pants” experience needed to produce reliable boards. Also, if a board is not working it’s worth running it down to root cause before blaming the PC board itself.  Someone needs to get out a scope and do some detective work.  It’s still possible there’s another cause lurking in there somewhere.

 

I’ll probably get criticism for this but I often set my iron to 800 degrees especially when working on areas that have large grounding planes. I haven’t had any problems, and I can’t remember the last time I left a cold solder joint.  I would definitely caution about being “quick with the soldering iron”. I always wait for that little shimmer you see when you know the whole pad of solder is flowing.

 

If you’d like any of us to try assembling and testing one of your boards give a holler. I’m happy to do it.  You’ve been extremely generous with your time on this effort!

 

Good luck Les, and thanks for all you do for the group (and have done for many years)!!

 

• Glenn

 

 

From: se...@googlegroups.com <se...@googlegroups.com> On Behalf Of Les Bird
Sent: Thursday, July 13, 2023 2:57 PM
To: SEBHC <se...@googlegroups.com>
Subject: Re: [sebhc] Re: JLCPCB - I give up

 

Glenn, it's beyond the Gerber review period at this point. The boards have been known to work so it's a board quality issue at this point in time.

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/7f33c501-5b24-4d25-9cd4-b4b1d8ad9272n%40googlegroups.com.

[norberto.collado koyado.com]

On traces width I used 0.016” and for spaces. I might come down to 0.012” on difficult to route traces.

Get Outlook for iOS

---

From: se...@googlegroups.com <se...@googlegroups.com> on behalf of glenn.f...@gmail.com <glenn.f...@gmail.com>
Sent: Thursday, July 13, 2023 1:14:37 PM
To: se...@googlegroups.com <se...@googlegroups.com>
Subject: RE: [sebhc] Re: JLCPCB - I give up

 

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/031a01d9b5c6%24a5ea0630%24f1be1290%24%40gmail.com.

[norberto.collado koyado.com]

I agree with Glenn's feedback. I can debug any storage board that you might have not working such as the H17. Just send it to me to debug it.

👍

Get Outlook for iOS

---

From: se...@googlegroups.com <se...@googlegroups.com> on behalf of norberto.collado koyado.com <norberto...@koyado.com>
Sent: Thursday, July 13, 2023 1:19:57 PM
To: se...@googlegroups.com <se...@googlegroups.com>

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/SN6PR01MB38559D143CDBDBF287FEEDFDF737A%40SN6PR01MB3855.prod.exchangelabs.com.

[norberto.collado koyado.com]

Also, I can help debug your broken Z80 board.

Get Outlook for iOS

---

From: se...@googlegroups.com <se...@googlegroups.com> on behalf of norberto.collado koyado.com <norberto...@koyado.com>

Sent: Thursday, July 13, 2023 1:26:54 PM

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/SN6PR01MB38554911D8C1A9F8F8E5FAADF737A%40SN6PR01MB3855.prod.exchangelabs.com.

[Todd Goodman]

I've had great success with PCBCart with thousands of S100Computers boards are initial boards I do here.

I've never used them for prototype boards though and not sure they're good for that.

Also, prices have gone up with them as with everything else.  Board cost seems pretty much the same, but shipping and tooling charges have gone up.

I'm also getting dinged by customs on every single order now (used to only be on orders over $1000).

I've been happy with their hard gold on the S-100 board fingers too.

And they actually seem to do something for the tooling costs and have people checking the design.

They do do testing too and will let me know when they have boards fail testing and are delayed due to that.  For panelized boards they just use a sharpie to put a line through the board.

Just my $.02 + $.02 customs duty,

Todd

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/7f33c501-5b24-4d25-9cd4-b4b1d8ad9272n%40googlegroups.com.

[Les Bird]

Glenn, you are right. I shouldn't blame the PCB fabricator until I can prove it's a defect in the board. Here's what i know so far...

1. I had batch #1 which all worked perfectly when I built them.

2. I "re-ordered" (no changes to the design) a 2nd batch, some of them worked, some of them didn't and some of them were just flakey (worked/didn't work/worked depending on time of day).

3. I "re-ordered" a 3rd batch thinking the 2nd batch was bad. I built 2 boards and neither one worked at all. I will build a 3rd and cross my fingers.

4. All re-orders were from the same Gerbers, no changes.

5. As part of my debug process I swapped all interchangable parts between them. The parts from the bad board worked fine in the good board. The parts from the good board did not work in the bad board.

So based on that info, the only conclusion I could come up with was the PCB itself was defective. Also the fact that I had an obvious/visible defect in one of my batch of H17s didn't help my confidence in JLC so right away I just assumed the PCBs were defective. Then reading the comments from the OSH Park owner supported my theory that board quality from JLC is just not good.

A couple things changed with batch #2 and #3.

1. I ran out of VNT-A17 16.384Mhz oscillators and had to start using oscillators from Tayda. The timing for the boards working/not working was almost exactly when I started using these new oscillators. I thought maybe it wasn't supplying a good clock signal. I looked on the scope and it seemed ok. I unsoldered the Tayda oscillator and took a VNT-A17 from a good board, that didn't make a difference. So I ruled out the oscillator as being the problem. I also put a Tayda oscillator in the good board and it worked just fine.

2. I ran out of LM555s and had to re-order new ones (different brand). Again, I swapped it with one from the good board and that didn't make a difference. So I ruled out the LM555 as being the problem.

3. I had to order new LS10s and LS04s and LS540s (from oddball sources). Again, swapping these out with ones from the good board. No issues with these parts - they work in the good board.

3. I ran out of solder so had to order some more from Phoenix Enterprises. Pretty sure this is not the issue though but who knows? Is there such a thing as "bad solder"?

4. I ran out of IC sockets so had to order more from Phoenix. With an ohm meter I can see from a random inspection that the IC pins are reaching their goal so pretty sure it's not the IC sockets, but who knows?

I've gone over all of my solder joints and as far as I can tell from a visible inspection they look good. I touched up joints that I think were questionable.

I've run out of ideas at this point.

I could check each trace but that's an enormous task as I'd have to check not only is it going where it's supposed to be going but also check it against every other trace to see if there's a short somewhere. That seems like an all week task so maybe I'll pursue that at some point.

What if batch #1 had a defect that made them work perfectly and batch #2 and #3 were good PCBs with no defects and there's a flaw in my design? Haha. Hmmm..

Norby, thank you for your generous offer. I may take you up on your offer and send you a good Z80 board and a bad one and see if you can find the issue with the bad one, using the good one as a frame of reference. I can't think of anything left to try. But first I will build out a 3rd board.

Todd, if I decide to try a different fabricator I'll take a look at PCBCart as well as those suggested by other members. Thanks for the heads up.

Les

[norberto.collado koyado.com]

Please send both boards to check them both. Hopefully, we can find the issue on the bad one fast. 👍

Get Outlook for iOS

---

From: se...@googlegroups.com <se...@googlegroups.com> on behalf of Les Bird <lesb...@gmail.com>
Sent: Thursday, July 13, 2023 4:41:11 PM

To: SEBHC <se...@googlegroups.com>
Subject: Re: [sebhc] Re: JLCPCB - I give up

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/2c50596b-0701-47c5-8096-c397d510b9f3n%40googlegroups.com.

[Scott Baker]

On the subject of IC sockets, I have run into a handful of machined sockets lately that will occasionally have a single socket pin that is missing the guts that holds the IC pin. I do buy my sockets from the cheapest China ebay supplier I can find, and you get what you pay for. As such this has turned into another one of those "inspect before you solder them" scenarios. The first time I ran into one of these bad sockets, it confused the heck out of me trying to diagnose the problem.

Scott

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/2c50596b-0701-47c5-8096-c397d510b9f3n%40googlegroups.com.

[Steven Hirsch]

On 7/14/23 03:11, Scott Baker wrote:
> On the subject of IC sockets, I have run into a handful of machined sockets
> lately that will occasionally have a single socket pin that is missing the
> guts that holds the IC pin. I do buy my sockets from the cheapest China
> ebay supplier I can find, and you get what you pay for. As such this has
> turned into another one of those "inspect before you solder them"
> scenarios. The first time I ran into one of these bad sockets, it confused
> the heck out of me trying to diagnose the problem.

I encountered this last winter with a socket from Phoenix Enterprises. Most
of the pins were missing the internal portion. To add insult to injury I
didn't spot it until after the socket was in place, then lifted a trace while
removing it.

[Les Bird]

Ok, crisis averted. The problem was not the board, it was me.

Allow me to explain my stupidity...

When I build my boards for testing I never solder in the filter caps. Usually the boards will run just fine without them. When the boards pass my tests I do the final step of soldering in all the 0.1UF filter caps for each chip. I do this to save time, especially with prototype boards that I know will not work and I expect to discard them after debugging them. This has never been an issue before but for some reason with boards from batch #2 and #3 the filter caps are mandatory for them to work at all. I suspect there's either (1) potentially some quality issues in the boards or (2) some of the chips I ordered from oddball sources like eBay (China) and Amazon are cheap and need the filter caps or (3) the massive solar flare storm we are encountering this week is wreaking havoc on our old systems and they need extra filtering to work properly. Whatever the reason for it, that was the fix.

When preparing the "bad" board to ship to Norberto I went ahead and soldered in the filter caps to finalize the board and when I plugged it in to the H8 it worked. Shocked by this simple fix I decided to build another board this morning. I built everything BUT the filter caps. I plugged it in and it did not work. I then immediately installed all the filter caps, plugged it in and it happily came right up and is running without a hitch now.

Ah well, lesson learned. In the future I will install the filter caps even for the prototypes. Extra soldering and extra time but...

Les

[Joseph Travis]

Congrats Les!!!

--
You received this message because you are subscribed to the Google Groups "SEBHC" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sebhc+un...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/1e67456b-7f1e-4427-9184-c74839515870n%40googlegroups.com.

[Glenn Roberts]

I always thought this “filter” capacitor nonsense was just a conspiracy by the manufacturers to fatten the bottom line. I’ve had to buy hundreds (thousands?) of these little suckers over the years. So these things really work? Who knew?

Glad you got it working!

😀

Sent from my iPad

On Jul 14, 2023, at 2:53 PM, Les Bird <lesb...@gmail.com> wrote:

Ok, crisis averted. The problem was not the board, it was me.

[norberto.collado koyado.com]

The tiny decoupling caps are very important for the operation of the logic inside each IC. 😳

Get Outlook for iOS

---

From: se...@googlegroups.com <se...@googlegroups.com> on behalf of Glenn Roberts <glenn.f...@gmail.com>
Sent: Friday, July 14, 2023 1:46:52 PM
To: se...@googlegroups.com <se...@googlegroups.com>
Subject: Re: [sebhc] Re: JLCPCB - I apologize (formerly I give up)

 

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/A14F3AD2-CCEE-4B14-ACB0-4EF6BA398F71%40gmail.com.

[norberto.collado koyado.com]

Once the logic starts switching it is going to expect the voltage at the IC pins to stay steady as current increases. No decoupling caps, at a given moment the voltage can be less than 4.75 volts causing failures as current will need to come from the 7805 regulator and not from the decoupling cap. 

Get Outlook for iOS

---

From: se...@googlegroups.com <se...@googlegroups.com> on behalf of norberto.collado koyado.com <norberto...@koyado.com>
Sent: Friday, July 14, 2023 2:26:09 PM

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/BN7PR01MB38443AD884EFA6D4CC74D3ABF734A%40BN7PR01MB3844.prod.exchangelabs.com.

[Les Bird]

Thanks Norby for that description. This brings up an interesting point too, I did change 5V regulator brands as I ran out and had to order new ones. This could be the reason why I all of a sudden need the filter caps because maybe my old 5V regulators  could handle the current and these new ones can't. I bet that's the issue I'm experiencing.

Les

[norberto.collado koyado.com]

It all depends on the output voltage. I see on some regulators output of 5V and on others 4.85V, etc. I hope you are using 0.1 uf. At least per specs for the EPROM and the memory should be 0.1uf. For regular logic 0.01uf should be fine. I used on the new boards 0.1uf as default. On the Pololu regulators I always get 5.01V and not less.

Get Outlook for iOS

---

From: se...@googlegroups.com <se...@googlegroups.com> on behalf of Les Bird <lesb...@gmail.com>
Sent: Friday, July 14, 2023 7:18:17 PM
To: SEBHC <se...@googlegroups.com>

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/cf44cc42-c492-46b4-85e2-924ad14762b0n%40googlegroups.com.

[norberto.collado koyado.com]

I buy at Amazon 0.1uf 100-pcs bag.

Get Outlook for iOS

---

From: se...@googlegroups.com <se...@googlegroups.com> on behalf of norberto.collado koyado.com <norberto...@koyado.com>

Sent: Friday, July 14, 2023 9:31:00 PM

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/SN6PR01MB3855C3EA029A14CB24BC7F4AF735A%40SN6PR01MB3855.prod.exchangelabs.com.

[smb...@gmail.com]

I have found the larger h8 boards to be more temperamental on power than many of my smaller projects. I think the length of the traces, together with the additional component count and complexity starts to matter. I’m not surprised the decoupling caps made a difference.

On one of my iterations of the 8085 board, glitches would go away if I added a wire from the 5v regulator directly to one of my bus transceivers. Just better more consistent current flow. Changing between LS and HCT also made a difference on that board.

Dave jones at eevblog did a good video a few years back on progressively removing decoupling caps until it made a difference.

[Douglas Miller]

The MMS 77422 "network controller" used used a sort of power bus bar, partly to save space but it also equalized/strengthen the power distribution and I think added extra capacitance (bypass). I wonder if those are still available.

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/1303287c-3b3b-448c-a460-8ae6acaba42an%40googlegroups.com.

[smb...@gmail.com]

I've seen those on other vintage boards too. I wonder if just upping the thickness of copper on the board would be a reasonable compensation. The other option is a 4-layer board, which surprisingly aren't all that expensive from a place like jlcpcb.

One thing about Norberto's boards is they usually have excellent power layouts, often including big fat traces running around the perimeter. My own boards are generally not nearly as good.

Scott

[norberto.collado koyado.com]

Another procedure, I do is to add 2.2uf caps at each end of the board, the center of the board, and on the middle of the board to store additional energy if needed by the logic.

 

If there is an IC that is power hungry and to avoid a dip on voltage I will add an inductor or a ferrite bead along with a OS-CON low ESR used as a smoothing capacitor (47uf + .1uf) to keep voltage steady.

 

Here are examples:

 

 

 

 

Z180:

 

 

 

Norby

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/25d534e4-3863-420c-bdcc-071d42b12350n%40googlegroups.com.

[Glenn Roberts]

All very interesting. I’ve seen these design elements as I’ve assembled the boards of course, but never knew their purpose.

And, much as I hate soldering components to those large ground planes (giant heat sinks that they are), I do understand why they’re there…

Sent from my iPad

On Jul 15, 2023, at 4:13 AM, norberto.collado koyado.com <norberto...@koyado.com> wrote:



Another procedure, I do is to add 2.2uf caps at each end of the board, the center of the board, and on the middle of the board to store additional energy if needed by the logic.

 

If there is an IC that is power hungry and to avoid a dip on voltage I will add an inductor or a ferrite bead along with a OS-CON low ESR used as a smoothing capacitor (47uf + .1uf) to keep voltage steady.

 

Here are examples:

 

<image001.png>

 

 

<image004.png>

 

Z180:

 

<image002.png>

 

<image003.png>

To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/SN6PR01MB3855016C8A09C4BA5657A2AFF735A%40SN6PR01MB3855.prod.exchangelabs.com.

[Steven Hirsch]

On 7/15/23 00:31, norberto.collado koyado.com wrote:
> It all depends on the output voltage. I see on some regulators output of 5V
> and on others 4.85V, etc. I hope you are using 0.1 uf. At least per specs for
> the EPROM and the memory should be 0.1uf. For regular logic 0.01uf should be
> fine. I used on the new boards 0.1uf as default. On the Pololu regulators I
> always get 5.01V and not less.

DEC PDP-11 boards use 0.47 uf decoupling caps. Never seen anything else with
values that high.

[Douglas Miller]

After spending 3 days chasing a glitch on a PDP-11 bus, I'm  glad they
didn't use SMALLER bypass caps!

[smb...@gmail.com]

I don't think I'll order PCBs during one of the China extended holidays again. 

The good news is that JLCPCB publishes a calendar and they actually met that calendar. I submitted my order on 10/2 and they shipped ("awaiting carrier pickup") on 10/5. According to the calendar, DHL should have picked them up on 10/5 (or maybe 10/6; sometimes JLCPCB logs them as shipped a day before DHL picks them up). But ... nothing. No tracking updates. DHL never received the boards. Fast forward to today, where I get an automated email that the PCBs were scratched during production and need to be remanufactured.

Ah well, better they replaced them rather than shipping them here and then replacing them. Still, I'm surprised they were able to determine the boards were scratched while they were fully packaged up and waiting for the shipper to pick up for 3-4 days. ;)

Scott

[Les Bird]

Yeah I had to wait an extended period for my "shipped" boards to start moving. Same as you, they sat waiting for carrier pickup for several days.

Suprised they caught them before shipping. Wonder how that happened, like you said, after they were packaged up and waiting for pickup. Kind of a bummer that you waited through the holiday only to get that message.

Les

[smb...@gmail.com]

I can think of a couple possibilities:

1) They found other boards that were scratched that happened to be in the same run as mine, and assume that all boards from that run are damaged. Maybe the usual QA or technicians were off due to the holiday week.

2) They weren't scratched at all, but were "lost" in transit to the shipper.

3) They weren't scratched or lost, but just never got put out for the shipper, the original boards are going out tonight, and the only code they can enter into the system for delayed orders is "delayed due to scratched board".

Scott