Yarrb - redesigned

[hoglet]

There was no notice like "Don't use this part because it is discontinued".

As far as I'm aware, the XC95288XL-10TQ144C has not been discontinued by Xilinx.

The last product discontinuation notice (PDN) affecting the XC9500XL CPLD family was in 2015:
https://www.xilinx.com/support/document ... n15006.pdf

And this was for the 44 pin PLCC package.

So this is purely a JLCPCB desision that they don't want to stock this part.

Dave

[roland]

I got an answer from JLC:

Actually that depends on the demand for this part XC95144XL is high or low .
If it's high demand,then I believe we will purchase this part soon.
If low demand,I think there won't be purchasing plan recently.

So actually, you can never be sure if any non-standard component is available which IMHO makes this company quite a risk if you're developing small amounts of boards for professional purposes....

Maybe next week they have another batch of 288's, maybe next year, and so for the 144....

[hoglet]

Hand soldering 0.5mm pitch QFPs is possible, with a little practice and lots of flux

I've done this exactly once though!

I'll post a picture later.

Dave

[roland]

I was playing with the calculator at PCBGOGO.com. Assembling 5 - 15 of these boards costs only $50. Adding 5 more raises the costs to $145
It looks like it's much cheaper to order 2 x 10 boards than 1 x 20 boards. Parts not included.

The calculator at allpcb.com looks very similar to the one at pcbgogo.com, however this site is much cheaper. Assembling only at $30 and $5 shipping to the Netherlands. In the past I ordered Atom2k15 boards at allpcb.com so I decided to get a quote from them.

Edit: I don't think I want to destroy my board and components for only $3, besides the time that it takes to solder 41 SMD components. I will leave that for Chinese children machines

[Ramtop]

Hand soldering 0.5mm pitch QFPs is possible, with a little practice and lots of flux

I've done this exactly once though!

It's actually surprisingly easy with some practice. I've soldered a couple of hundred 0.5mm QFP and TSOP parts in this past year without ever terminally messing one up, despite bad eyesight and shaky hands. If I can do it anyone can!

It takes liberal use of good quality flux, some form of magnification (I use cheap 3.5x goggles from Amazon) and a knife tip on the soldering iron. Most people try drag soldering, but I find that works poorly with very fine pitch parts. My method is to put a little solder on the tip and tap it quickly against the pins. But to my mind he real secret to consistent, repeatable good results is increasing pad length on the PCB. Long pads will soak up more solder before bridges form, which makes it much less important to get exactly the right amount of solder on the iron.

[bprosman]

Hi Roland,
Another supplier like "Allpcb" is not an option ?

[roland]

Another supplier like "Allpcb" is not an option ?

Yes, see my post at Tue Oct 29, 2019 10:31 pm

It's actually surprisingly easy with some practice. I've soldered a couple of hundred 0.5mm QFP and TSOP parts in this past year without ever terminally messing one up, despite bad eyesight and shaky hands. If I can do it anyone can!

I could start practising this, but there are also people who really don't want this. I already have a small board to practise ($1 from Ali including P&P) and I could try to remove and remount a large chip from a broken PC mainboard or any other faulty device. I really must start with that

[bprosman]

Its not that difficult. I did a 100 Pin version last year.

[roland]

Today I got my quotation for the Yarrb2 board and assembling the SMD components. Without the LED's (those are out of stock), the price is about €420,00 including VAT and import taxes for ten boards. And then you still have not all components (e.g. the EEPROM, voltage regulator and headers) so the complete Yarrb2 is quite expensive.

I tend to do some hand soldering when I hear these prices. Assembling is about $145.00 for 10 boards and the components are $160.00 however I think that is quite reasonable as the CPLD is not very cheap.

Yesterday I practised some SMD soldering and I still need some practising, but I think that even for me it is doable to solder a CPLD to a board, especially when it is a 100 pin device. Like somebody said, make long pads and the chance of soldering pins together will be smaller.

For those who have shown interest in this board: what is your opinion?

[Ramtop]

I say give hand soldering a try. Given sufficient practice it becomes very possible to solder fine pitch parts with accurate, repeatable results; I soldered ten CPLDs in 0.5mm QFP packages this evening and it only took me about 40 minutes to do, because I work with those regularly.

To start I'd suggest getting some cheap 2-layer practice PCBs made up. Just plop down a CPLD footprint (with long pads). JLCPCB will do ten of those for about $10 delivered.

Then just keep practising. If the PCB gets a bit crusty, hot-air the CPLD off and use a fresh board. It doesn't really matter if the CPLD gets cooked, the aim is to be able to solder it in place with no bridges, time after time. Practice is the key.

This is the type of board I can do fairly easily with just a soldering iron and magnifying goggles; 144-pin FPGA, plus fine pitch level converter chips. The only challenging component on the whole board was the little 4-pin clock oscillator, because the pads are underneath the package.

[KenLowe]

Not sure I understand what options you've priced. Sounds like one option is pre-assembled units @ €42 + extra components per board. I would be happy to pay that price.

Regarding the hand soldered option, I'm not clear on the costs, Are you suggesting $14.5 + $16 per board? Does this include the bare pcb? I assume that's extra? If bare PCB cost has to be added, then given the likely marginal difference between hand soldering and pre-assembled, I would probably recommend the pre-assembled option - especially if you're not overly confident about soldering the SMD components. The cost of a couple of failed soldering attempts could start to mount up!

[gob33]

See PCBWay in Shenzen.

[marcusjambler]

Still interested Roland

What will the top speed be?

[roland]

I asked for a quotation at PCBway. Naturally they have other specifications for the BOM and CPL files So I might reconstruct them again. Luckily there are only 41 SMD components on the board

The speed of the board will be limited by the 6502, 6522 and 8255. So probably it will run at 1 or 2 MHz, but you can use higher speeds for example 4 MHz when the Atom clock circuit is used. By using the on board oscillator you can even get higher speeds. It seems useless when the 8255 cannot go faster but you can switch to high speed for calculations and when I/O is needed you can switch back to lower speed.

[BillG]

Hi,
I'd be interested in the new Yarrb board too in either format. Getting more use to SMDs now.

[roland]

Good news: I have ordered the first ten boards at PCBway, now let's hope they work

After all, they charge only $30 for assembling those 10 boards, which is $3 per board. I cannot do that myself for that price.
The price will be lower than Allpcb, however I have to wait what the taxes will be.

[KenLowe]

Nice one. Let's hope they work ok!

[caspian]

Looks like you may not need to solder it yourself now, but I saw an easy-looking soldering technique used for fine pitch parts on the basicengine.

https://basicengine.org/hardware.html#_assembly
https://youtu.be/DTTEZJDEUIA

quick description: you drag solder over the PCB pads with a soldering iron, maybe remove excess solder, then put the component on, then re-melt the solder with hot air from a heat gun to connect it to the part.

I had to look up drag soldering to check it wasn't the same.

[Ramtop]

The technique demonstrated in that video is a very poor way of soldering surface mount ICs. Putting solder on the pads causes them to be lumpy, making it more difficult to align the chip. Its pins will tend to slide into the gaps between the pads. And using a hot air gun is slow and subjects the chip to unnecessary thermal stress.

I really should try to find a suitable camera to video my way of soldering QFP chips, just to show how quick and simple it can be. People tend to drastically over-think SMD soldering.

[hoglet]

The technique demonstrated in that video is a very poor way of soldering surface mount ICs. Putting solder on the pads causes them to be lumpy, making it more difficult to align the chip. Its pins will tend to slide into the gaps between the pads. And using a hot air gun is slow and subjects the chip to unnecessary thermal stress.

Seconded!

I really should try to find a suitable camera to video my way of soldering QFP chips, just to show how quick and simple it can be. People tend to drastically over-think SMD soldering.

That would be really great.

I'm still pretty new to surface mount assembly, and I'm still pretty amazed at the results you can get with drag soldering and lots of flux.

The most challening thing I've done to date is to replace a faulty 0.5mm pitch FPGA on one of Revaldhino's GOPs. These are nice little DIP compatible FPGA models, that are sadly now discontinued. The FPGA blew because the switching regulator blew (or may have always been faulty). So it was running the core of 3.3V rather than 1.8V.

The hardest part was desoldering the old part (using hot air) without dislodging any of the surrounding 0603 components. Masking everything possible in Kapton tape did the trick though:

After carefully cleaning all the pads with solder braid, it almost looked like new!

Soldering the replacement part litereally only took a few minutes. Accurate alignment is the key. I tend to use a couple of blobs of blu tack, then tack down a couple of opposite corner pins. The blu tack stops the part moving sligtly when tacking down, leaving both hands free to solder.

And the final result, after cleaning the board with IPA.

The riskiest part of the procedure was buying the replacement part from China. It only cost ~£7 though, and is now working perfectly.

Dave

[roland]

I do agree that you can practise smd soldering IF you can see what you are doing and you have the right equipment. On the other side I can also understand that there are members here that even have trouble with soldering normal through-hole components but still want to buy a new board. I sometimes get the question to deliver a board completely build.

I already tried to drag solder a memory chip to my 1MHz FPGA board and I managed to connect all 16 pins on one side of the memory chip together. I needed the help of somebody else to remove the chip and resolder it the right way. Soldering 1 pin at a time with a magnifier went better, however it requires some practise and good equipment.

I am glad that PCBWay had a good offer for producing these boards but the project triggered me to practise more on smd soldering. I encourage others to do the same: buy a few cheap practise circuits on Ali (about $1 including P&P) and start practising

Perhaps it's a nice idea to work out for an ABUG meeting: a workshop in smd soldering.

[anightin]

Perhaps it's a nice idea to work out for an ABUG meeting: a workshop in smd soldering.

I’d definitely sign up for that! SMD soldering is on my todo list for sure.

[eXne]

Folks,

fine pitch smd components soldering is done either with hot air but applying solder on the pads is a tricky one. Normally a stencil is used as applying the solding by hand is difficult.
The other way is ensuring all solderpads are clean and even. Next align the smd component and fix it by soldering a few pins at the corners.
Next apply flux on all pins and pads and solder the pins by using a hollow soldertip filled with solder, simply striking the pins in one flow.
The solder will be extracted from the hollow solderpoint. It will take some practice.
@Roland: removing your smd component requires a special removal tool. You need to heat all pins at all together and then removed the smd component otherwise you will damaged the pads and/or pins.
If on the component is not important you can cut the pins and removed them. But this is not the applicable for you now.

[daveejhitchins]

I use one of these with a block of aluminium on to enable a stable temperature. Set to about 150C and using a fine nozzle on the hot-air-gun it takes seconds just going around the pins and then lifting with a fine point.

Dave H

[balford]

For soldering fine pitch devices, I tack the IC in place by soldering opposite corners in. Once the IC is in the right place, I add loads of flux across all pins and drag solder the other connections. The extra flux means you don't get shorts between pins as long as you're not using too much solder.

Dave from EEVBlog has some quite good videos on the subject, from which I learnt this technique.

Oh, and I'm still interested in a board

[roland]

A few weeks I ordered my pcb's at pcbway.com. They have a good price for the boards, components and mounting the smd components. But unfortunately, my footprints went wrong again:

I had the wrong footprint and I did not notify that. Luckily pcbway first checks if everything will fit so they did not spoil any components yet. I had to redesign the board so the RAM chip would fit. Then I submitted the modified board and something strange happened: suddenly the board was too large for the special offer: although I only replaced the RAM footprint, pcbway mentioned that the board was 0,4 cm too large for the special offer. As I had enough room to trim the board, I did so. Due to the new footprint they also needed to make a new stencil.

To avoid paying import duties (the price went up because of the second run) I asked them to ship the boards with China Post. Bad luck, they just shipped them with DHL. And DHL always makes sure that I have to pay my import duties

But the boards look nice and I am going to build one in the next couple of days...

My next step will be ordering the EEPROMS and the voltage regulators so I can program those EEPROMS and the CPLD.

To be continued....

[roland]

Today I programmed the CPLD and the EEProm for these boards and they almost worked the first time. I got the ACORN ATOM banner and a prompt. Entering commands also worked. But switching the ROM mode to AtoMMC did not work, neither could I switch on any of the LEDs. I checked my Phi2 signal with my scope and that was clean.

As every thing else seemed to work, I re-read my VHDL code and I saw a mistake in decoding #BFFC/D/E/F: I had b4 and b3 equal to 0 but that should be 1. So those registers were at the wrong place in the memory map. A quick check on poking to #BFF2 literally turned a light on. After fixing that, I could switch between the memory profiles. Except for the AtoMMC, that did also not work. It turned out that the jumper for the WD65C02 was in the wrong position. It still must be in position 2-3 for a normal 65C02 but the position of the three jumpers is turned 180 degrees compared to the Yarrb-1 boards. So it looks like another position but it isn't.

After setting that jumper to the right position also the AtoMMC works. My Atom works stable with this board, it passes the "Chuckie Egg Test" so I can go figure out what it costs and inform the people who did a pre-order for a board by PM.

Yarrb2 with some Leds turned on.....

[hoglet]

Very nice indeed Roland!

[1024MAK]

Looking very nice

I’ll answer your PM and send over a dumper truck full of cash tomorrow, it’s been another long day, and I’m falling asleep here...

Mark

[KenLowe]

Ooooh. Waiting in anticipation. Maybe not in time for the xmas stocking, but will be a nice new year pressie to myself .