MiSTer FPGA Forum

LeeW wrote: ↑Thu Aug 31, 2023 4:50 pm

I've completed all boards now.

Bottom:

Top, in case. The camera really picks up the flux residue huh, not this noticeable to the naked eye.

Front cover on:

The fan is very loud and appears to be a 30mm fan so the smallest Noctua fan won't fit... Also if the screws are tightened the fan seems to be in contact with the enclosure while running. Will probably be running this fan-less.

Awesome work!

Are you sure you dont have the fan backwards?

how much did you pay for all?

Only his second born child

LeeW wrote: ↑Thu Aug 31, 2023 4:50 pm

Top, in case. The camera really picks up the flux residue huh, not this noticeable to the naked eye.

If you haven't tried to clean off the flux residue, you should try to clean off the flux residue. "No clean" flux is kind of an oxymoron these days.

https://www.electronicdesign.com/market ... clean-flux

LeeW wrote: ↑Thu Aug 31, 2023 4:50 pm

The fan is very loud and appears to be a 30mm fan so the smallest Noctua fan won't fit... Also if the screws are tightened the fan seems to be in contact with the enclosure while running. Will probably be running this fan-less.

From the picture, I think you have the fan backwards. Label should face the case and blow the air out. See this post: viewtopic.php?p=51020#p51020. According to that post, that will make it run much quieter.

I used a lot of 99.9% IPA on the boards to clean them off, along with cloths and a toothbrush. I've removed as much as I could, but I may look to see if there's something else that could help remove the last remnants on the board. Any suggestions?

Thanks for the link on the fan orientation. I specifically tested the air flow direction and chose to place it in this orientation. Because in general, positive air pressure (more intakes than exhausts) is best for cooling. But maybe in this specific case it is better for temperature as an exhaust fan? Or is it just that this orientation is specifically to counter the noise of the fan, as the fan would be pushing against the case rather than pulling away from it (and rattling)?

Final cost comes to £111.60 per board (not accounting for labour and equipment costs). FedEx just hit me with the shipping and VAT bill from JLCPCB yesterday, since it took so long to get here I thought JLCPCB had taken care of the VAT. But no, they sent it by post and it took a couple of weeks to arrive.

LeeW wrote: ↑Fri Sep 01, 2023 10:30 pm

Thanks for the link on the fan orientation. I specifically tested the air flow direction and chose to place it in this orientation. Because in general, positive air pressure (more intakes than exhausts) is best for cooling. But maybe in this specific case it is better for temperature as an exhaust fan? Or is it just that this orientation is specifically to counter the noise of the fan, as the fan would be pushing against the case rather than pulling away from it (and rattling)?

I think @multisystem said somewhere that the fan is barely necessary in the first place so I think it is the latter. Give it a spin and see what happens.

rezendes wrote: ↑Wed Aug 30, 2023 7:19 pm

Nice! Waiting for my official case to come from the UK to the me in the states.

It might take a couple of weeks. I seem to remember it taking a couple of weeks when I ordered mine.

LeeW wrote: ↑Thu Aug 31, 2023 4:50 pm

The fan is very loud and appears to be a 30mm fan so the smallest Noctua fan won't fit... Also if the screws are tightened the fan seems to be in contact with the enclosure while running. Will probably be running this fan-less.

I assume you mean the fan is sort of rattling because it's in contact? I ended up swapping out my 16mm screws with 20mm so that I could add a washer to the nut. You can use 20s because actually the lower screws fit under the board. But screwed in with long screws and nuts mine doesn't seem to make anything other than 'fan' noise. Maybe if you put washers between the fan and the case you won't get any 'contact' noise.

I set mine on 3.3V and it seems quiet.

LeeW wrote: ↑Fri Sep 01, 2023 10:30 pm

I used a lot of 99.9% IPA on the boards to clean them off, along with cloths and a toothbrush. I've removed as much as I could, but I may look to see if there's something else that could help remove the last remnants on the board. Any suggestions?

I found this to be quite good: https://www.amazon.co.uk/gp/product/B07XDCRLR6/
I also used a bit of IPA from a dropper bottle and a brush just to clean up small areas where I was doing individual bits of de-soldering / re-soldering.

LeeW wrote: ↑Fri Sep 01, 2023 10:30 pm

Final cost comes to £111.60 per board (not accounting for labour and equipment costs). FedEx just hit me with the shipping and VAT bill from JLCPCB yesterday, since it took so long to get here I thought JLCPCB had taken care of the VAT. But no, they sent it by post and it took a couple of weeks to arrive.

If the total cost of your order with JLCPCB came to less than £135 (which I assume it must have done?) you shouldn't owe VAT or import duties. VAT should have been included at the point of sale. https://www.gov.uk/goods-sent-from-abroad/tax-and-duty

emiliom wrote: ↑Sat Sep 02, 2023 4:44 pm

LeeW wrote: ↑Fri Sep 01, 2023 10:30 pm

I used a lot of 99.9% IPA on the boards to clean them off, along with cloths and a toothbrush. I've removed as much as I could, but I may look to see if there's something else that could help remove the last remnants on the board. Any suggestions?

I found this to be quite good: https://www.amazon.co.uk/gp/product/B07XDCRLR6/
I also used a bit of IPA from a dropper bottle and a brush just to clean up small areas where I was doing individual bits of de-soldering / re-soldering.

I was looking at that flux remover but the reviews suggest that it still leaves a sticky residue and it's also fairly expensive. What's your experience with it been like?

emiliom wrote: ↑Sat Sep 02, 2023 4:44 pm

If the total cost of your order with JLCPCB came to less than £135 (which I assume it must have done?)

Actually it was quite a bit more than that haha. Most of the components and assembly was done by JLCPCB.

Here's a full breakdown of my* costs incurred: https://docs.google.com/spreadsheets/d/ ... Kj/pubhtml
*Obviously it goes without saying that your costs could be slightly different even if you were to make identical boards to me. But this is what I paid. Also I'm pretty sure you can pick cheaper parts than I did in some cases. And as mentioned in the sheet I wasn't paying attention to the "Extended component" selection which is a bit of a hidden fee. JLCPCB does tell you about this, but I guess either I misunderstood how it applies or just forgot over an extended period of time due to JLCPCB fucking up my BOM so many times (...due to the originally provided BOM being terrible it kept resetting some of my component picks when it reloaded the BOM).

And the final JLCPCB BOM for my boards: https://docs.google.com/spreadsheets/d/ ... Kj/pubhtml
Note that quite a few components are missing from the BOM as there are no good picks on JLCPCB. Also, this is trimmed down a little bit from my BOM as I had picked a couple of parts wrong, for example I accidentally selected the wrong part for U4, so this is not included on this BOM. But it would seem like an alternative part for U4 that can be chosen from the JLCPCB part picker is TPS2552DBVR (C46506), I accidentally selected TPS2553 instead of TPS2552. The 2 is for the active low version which is important.

At some point I'll probably sell 3 of the boards at £125 each + postage.

LeeW wrote: ↑Sat Sep 02, 2023 9:27 pm

emiliom wrote: ↑Sat Sep 02, 2023 4:44 pm

I found this to be quite good: https://www.amazon.co.uk/gp/product/B07XDCRLR6/
I also used a bit of IPA from a dropper bottle and a brush just to clean up small areas where I was doing individual bits of de-soldering / re-soldering.

I was looking at that flux remover but the reviews suggest that it still leaves a sticky residue and it's also fairly expensive. What's your experience with it been like?

I found it similar to IPA in that respect. It seems to boil down to how hard you're prepared to go at it with the toothbrush . What I did find it better for though was allowing the board to soak for a while. After I soldered the SMD components I let it soak in the MG cleaner for half an hour and then gave it a good scrub with my hog hair brush. It seemed to come up better than using IPA.

LeeW wrote: ↑Sat Sep 02, 2023 9:27 pm

emiliom wrote: ↑Sat Sep 02, 2023 4:44 pm

If the total cost of your order with JLCPCB came to less than £135 (which I assume it must have done?)

Actually it was quite a bit more than that haha. Most of the components and assembly was done by JLCPCB.

Ah, I thought you'd soldered everything yourself. No wonder you got 5 boards done so fast!
I got components from different places so I need to sort out my spreadsheet to work out the averages correctly, but the bulk of your components from JLCPCB seem dirt cheap. My resistors alone were a lot more expensive so I think my component cost was a lot higher.

LeeW wrote: ↑Fri Sep 01, 2023 10:30 pm

At some point I'll probably sell 3 of the boards at £125 each + postage.

Do you have any timeframe on when you might sell them?

I just got my official Multisystem 3D printed case in the mail and sadly they forgot to include the bolts+nuts for the internal fan Here’s my results after a quick trip to the hardware store for 4xM3 20mm flathead machine screws and 4xM3 Nuts…

rezendes wrote: ↑Wed Sep 06, 2023 12:11 am

I just got my official Multisystem 3D printed case in the mail and sadly they forgot to include the bolts+nuts for the internal fan Here’s my results after a quick trip to the hardware store for 4xM3 20mm flathead machine screws and 4xM3 Nuts…

Did you not just get 4 short screws included in the kit to fit the fan? That's what I got. However I replaced them immediately and also put in longer screws with nuts.

So I noticed something a little odd. I can't see LED1 (green) unless I look directly over the top of the case. Initially I thought I'd messed up as the LED I'd selected only turned out to be 15mcd, so I replaced it with a 100mcd green LED, but I still can't see it unless I look directly over the top of the case. Looking down the hole it doesn't seem to be obscured by anything. The red LEDs I selected are 80mcd and I can see those just fine. Anyone else finding this with the green LED?

Same for me regarding the green LED. As far as the screws, no I just didn’t get those 4 20mm screws and 4 Nuts also missing. I assume it happened with confusion of when they were sometimes pre-installing the fans and at one point stopped doing that…

emiliom wrote: ↑Wed Sep 06, 2023 3:12 pm

So I noticed something a little odd. I can't see LED1 (green) unless I look directly over the top of the case. Initially I thought I'd messed up as the LED I'd selected only turned out to be 15mcd, so I replaced it with a 100mcd green LED, but I still can't see it unless I look directly over the top of the case. Looking down the hole it doesn't seem to be obscured by anything. The red LEDs I selected are 80mcd and I can see those just fine. Anyone else finding this with the green LED?

Re: the green LED. Check the current through the LED by measuring the voltage drop across the associated current-limiting resistor (e.g. pins 3 & 6 on N2) and calculate the current through the LED from VoltageDrop/56Ohms. Then see how that current compares to the spec value from the datasheet. And note that green may take more current than red to achieve the same brightness.

I may have a problem, I have been having trouble running the PSX core and found out that when I run the memtest utility from the mister menu it shows a #2 which apparently indicates I have only 64MB of ram installed. I double checked all the solder joints and even redid them all. Both 64MB chips are installed which should equal 128MB. Can someone else run the memtest utility and tell me if it shows a #3 which indicates 128MB? Maybe there’s some problem with one of the U1/U2 components I believe that are associated with the SDRAM circuit, I’m tired so I’m gonna pass out.

rezendes wrote: ↑Thu Sep 07, 2023 10:02 am

I may have a problem, I have been having trouble running the PSX core and found out that when I run the memtest utility from the mister menu it shows a #2 which apparently indicates I have only 64MB of ram installed. I double checked all the solder joints and even redid them all. Both 64MB chips are installed which should equal 128MB. Can someone else run the memtest utility and tell me if it shows a #3 which indicates 128MB? Maybe there’s some problem with one of the U1/U2 components I believe that are associated with the SDRAM circuit, I’m tired so I’m gonna pass out.

Double check U1 on the integrated 128MB memory module, this is a chip select, it toggles high/low to select each bank of 64MB of memory, if the MiSTer menu is only showing 64MB on the Multisystem it's highly likely that U1 is missing, not soldered or has a short to stop CS1/CS2 being enabled.

DeKay wrote: ↑Thu Sep 07, 2023 4:02 am

Re: the green LED. Check the current through the LED by measuring the voltage drop across the associated current-limiting resistor (e.g. pins 3 & 6 on N2) and calculate the current through the LED from VoltageDrop/56Ohms. Then see how that current compares to the spec value from the datasheet. And note that green may take more current than red to achieve the same brightness.

When I designed the Multisystem enclosure, the holes for the three LED's are actually designed to use short sections of 1.75mm 3D printing filament. If you use a transmissive / translusive material or even white PLA you can get a nice low cost light pipe effect. You can also flame finish the ends of the filament to make a slight dome, it looks nice.

We did this in the original prototypes and quite a few testers had the feedback that the LED illumination was too high, so we removed the 'light pipes' from the design and so the LED's are less likely to light up the room, but are more visible directly viewed from above.

As mentioned already you may need to match up brightness levels with SM LED's that have a narrow viewing angle (30/60 Degrees) for the best effect.

multisystem wrote: ↑Thu Sep 07, 2023 10:13 am

rezendes wrote: ↑Thu Sep 07, 2023 10:02 am

I may have a problem, I have been having trouble running the PSX core and found out that when I run the memtest utility from the mister menu it shows a #2 which apparently indicates I have only 64MB of ram installed. I double checked all the solder joints and even redid them all. Both 64MB chips are installed which should equal 128MB. Can someone else run the memtest utility and tell me if it shows a #3 which indicates 128MB? Maybe there’s some problem with one of the U1/U2 components I believe that are associated with the SDRAM circuit, I’m tired so I’m gonna pass out.

Double check U1 on the integrated 128MB memory module, this is a chip select, it toggles high/low to select each bank of 64MB of memory, if the MiSTer menu is only showing 64MB on the Multisystem it's highly likely that U1 is missing, not soldered or has a short to stop CS1/CS2 being enabled.

MMS_Memory_selects.png

This is the part I used for U1 https://www.mouser.com/ProductDetail/Te ... 1NgQ%3D%3D

Would this part be fine in that spot, not sure it’s exactly the same one in the files you released. I double checked and went over all of the soldering in that area as well.

For me the green is less bright but I wouldn't say it's problematic:

I aimed for a target of about 40mcd for my LEDs as I hate having blinding LEDs on things. I just now re-checked the specs on my LEDs on digikey and it looks like my red is 39.75mcd, yellow is 42.75mcd, but the green is only 20mcd. Hmm... I could have sworn the datasheet of the LED I picked showed the mcd around 47 on JLCPCB... Well I think that pretty much accounts for the difference in brightness I'm seeing. I'm pretty sure that a brighter green would have been equally as visible as the red.

Could it be that there's some 3D print stringing or something that is partially blocking the hole? Also, as mentioned, adding a small translucent thing in the top to diffuse the light will make it more visible.

Can confirm that the memtest shows a 3 for me:

LeeW wrote: ↑Thu Sep 07, 2023 5:58 pm

For me the green is less bright but I wouldn't say it's problematic:

I aimed for a target of about 40mcd for my LEDs as I hate having blinding LEDs on things. I just now re-checked the specs on my LEDs on digikey and it looks like my red is 39.75mcd, yellow is 42.75mcd, but the green is only 20mcd. Hmm... I could have sworn the datasheet of the LED I picked showed the mcd around 47 on JLCPCB... Well I think that pretty much accounts for the difference in brightness I'm seeing. I'm pretty sure that a brighter green would have been equally as visible as the red.

Could it be that there's some 3D print stringing or something that is partially blocking the hole? Also, as mentioned, adding a small translucent thing in the top to diffuse the light will make it more visible.

Can confirm that the memtest shows a 3 for me:

Can you confirm the exact part you used for U1? Also where it was purchased if you don’t mind? The original BOM has one part mentioned 04 and another on the comment marked 07 and they appear to be functionally different parts.

This is what I picked: https://www.digikey.co.uk/en/products/d ... BVR/385716

The 04 would be the one to select, my understanding is that the 07 is a "library reference" which is a reference to some arbitrary parts library and not an accurate description. DBV is the SOT-23 footprint. And I think the R at the end is for RoHS compliance.

EDIT: Ah, just double checked and it looks like the actual part that was used on my board was by UMW(Youtai Semiconductor Co., Ltd.) and not Texas Instruments. Looks like there are duplicates with the same name and it probably doesn't matter too much on the specific brand.

Thank you LeeW. @emiliom Emiliom you may need to adjust your BOM to reflect this, I think this is my problem, the 04 is the correct part and the 07 does not work in it's place. Also the pin header with the super long pins should ideally have a different option.

LeeW wrote: ↑Thu Sep 07, 2023 6:24 pm

my understanding is that the 07 is a "library reference" which is a reference to some arbitrary parts library and not an accurate description.

The LibRef field for the the vast majority of the parts seems to be a description of the part, apart from BC856AL which itself matches a part number but not the comment so generic series name? In a few cases, including U1,U2,U3,U4 and the SNAC connector it is the exact manufacturers part no, except for in U1 04 is inverting and 07 is non inverting, so I think it's just wrong. I must not have noticed that the last digit in the Comment field and LibRef field were different so I copied the LibRef field by mistake.

LeeW wrote: ↑Thu Sep 07, 2023 6:24 pm

DBV is the SOT-23 footprint. And I think the R at the end is for RoHS compliance.

DBV does seem to be the footprint, but I don't think R is for RoHS. There are 7 matching parts on Mouser for SN74LVC1G04 and they are all RoHS compliant but they don't all have an R. Given that there is SN74LVC1G04DBVT and SN74LVC1G04DBVR I thought it might be Tape and Reel, but all 7 are available as Tape, Mouse Reel, and Full Reel. The suffixes are T, R, TE4, RE4, TG4, REP, and RQ1. Anything with a T is available in full reels of 250 and R in full reels of 3000 so I think that's what those mean. That leaves E4, G4, EP and Q1.They all appear to have the same basic specs, apart from EP and Q1 which show the voltages to be 1.8V, 2.5V, 3.3V, 5V rather than 1.65V to 5.5V, and appear to be military grade and automotive grade respectively. That leaves E4 and G4. I don't know what those mean. All 7 are described as single inverters in the datasheets. TBH the main difference seems to be the price and SN74LVC1G04DBVR is the cheapest. I don't see a reason to go for one of the others, and if the Youtai one you got is functionally the same then it should be ok.

I don't understand why someone would go for the T rather than the R though given it's twice the price and appears to be the same other than reel size.

LeeW wrote: ↑Thu Sep 07, 2023 5:58 pm

I just now re-checked the specs on my LEDs on digikey and it looks like my red is 39.75mcd, yellow is 42.75mcd, but the green is only 20mcd.

What's the viewing angle on your green LED?

LeeW wrote: ↑Thu Sep 07, 2023 5:58 pm

Could it be that there's some 3D print stringing or something that is partially blocking the hole?

No, I even cleaned it out with a pin to make sure.

multisystem wrote: ↑Thu Sep 07, 2023 10:19 am

When I designed the Multisystem enclosure, the holes for the three LED's are actually designed to use short sections of 1.75mm 3D printing filament. If you use a transmissive / translusive material or even white PLA you can get a nice low cost light pipe effect. You can also flame finish the ends of the filament to make a slight dome, it looks nice.

I literally known nothing about 3D printing other than having witnessed the end result. Are you saying just buy a roll of 1.75mm translucent filament, cut a short length and stick it in the LED holes which will 'pipe' the light up to the top?
Also, how would you 'flame finish' the ends? I'm imagining melting the end of a bit of filament with a match but I suspect there's slightly more to it?

rezendes wrote: ↑Thu Sep 07, 2023 6:44 pm

Thank you LeeW. @emiliom Emiliom you may need to adjust your BOM to reflect this, I think this is my problem, the 04 is the correct part and the 07 does not work in it's place. Also the pin header with the super long pins should ideally have a different option.

I've uploaded a new BOM with the change to U1 and also separated out LED1 and added a note about the issue of light reaching the top. I've added a note about the long pins as well but I'm not changing this part unless someone has an alternative? Lee had one but it was 3 times the price so I kept the original.

emiliom wrote: ↑Fri Sep 08, 2023 2:38 am

rezendes wrote: ↑Thu Sep 07, 2023 6:44 pm

Thank you LeeW. @emiliom Emiliom you may need to adjust your BOM to reflect this, I think this is my problem, the 04 is the correct part and the 07 does not work in it's place. Also the pin header with the super long pins should ideally have a different option.

I've uploaded a new BOM with the change to U1 and also separated out LED1 and added a note about the issue of light reaching the top. I've added a note about the long pins as well but I'm not changing this part unless someone has an alternative? Lee had one but it was 3 times the price so I kept the original.

Where did you upload the BOM to? I had to go back and find your old post on like page 2 or 3 to see the attachment...