A DIY pcb production method suitable for 8-mil spacing?

[swbluto]

Hello. I'm trying to find a home-made PCB method that's compatible with fine spacing requirements. I'm planning on using a home-method due to cost and time requirements and I'm trying to find something that's feasible for TQFP-32 type where the footprint has pads that are spaced somewhere around 8-9 mil apart. Is there such a method available? I've tried the toner transfer method and seem to have occasional problems with anything below 20 mil and I've heard about UV's precision, but is it really suitable for something like 8-mil spacing? I'm trying to prototype this TQFP-32 packaging that seems to have 8 mils spacing between the pads and I definitely don't think it's possible with the Toner transfer method. Or is it? Is Pulsar paper or any other special paper (Even with a special laminator of some sort?) suitable for these kind of requirements?

 

[nike6]

get an adapter PCB.

alternatively, only solder each 2nd pin, and wire the others manually.
for this purpose, bend up every second pin.

I've mounted TQFP44 on ordinary prototype PCB, just milled every pad into two halves.

 

[DirtyLude]

TQFP is 0.5mm spacing. I do that all the time with regular toner transfer and have done smaller. I don't use Pulsar paper, but if your not experienced, I'm sure it would help get the results you are looking for. Photo will definitely do it.

 

[srikanthsamaga]

TQFP is 0.5mm spacing. I do that all the time with regular toner transfer and have done smaller. I don't use Pulsar paper, but if your not experienced, I'm sure it would help get the results you are looking for. Photo will definitely do it.

Which paper you are using.....?
Whats your opinion about transparencies and magazine papers....?

 

[3v0]

The pulsar paper makes the process easier. It also makes fine lines easier because the toner is never attached to the paper, just the starch covering it. I have been doing SOICs for a while and and just made my first PCB with a 44 pin TQFP.

Here is a picture from their site

**broken link removed**

 

[swbluto]

TQFP is 0.5mm spacing. I do that all the time with regular toner transfer and have done smaller. I don't use Pulsar paper, but if your not experienced, I'm sure it would help get the results you are looking for. Photo will definitely do it.

I did notice that the pin's middle-to-middle point was nominally .85 mm and the pin width was .3 mm, so the spacing seems like it should be ~.5mm, but the eagle footprint for the atmel 168 (TQFP 32 packaging) seems to envelope the pin a little bit which seems to reduce the pad to pad spacing down to 8-9 mil just by eye-balling it in the Eagle program. It doesn't seem like I get anything under 20-mil reliably using the toner transfer, but if it sounds like you're not having a problem, then it seems that photo shouldn't be a problem.

Does it really matter how "strong" the UV light is? It seems there's a preference for 1-5 minute development times online and people buy like 20-160 watt set-ups, but I'm just fine with longer exposure times if it means lower setup costs(and, bonus, harder to over-expose, right?). As long as it's the same day, it's sounding faster than professional services.

But I really sound like I have a lot to learn. People mention a "ballast" in their tutorials and explanation, but I have no idea what that is for the photo method.

 

[swbluto]

The pulsar paper makes the process easier. It also makes fine lines easier because the toner is never attached to the paper, just the starch covering it. I have been doing SOICs for a while and and just made my first PCB with a 44 pin TQFP.

Here is a picture from their site

**broken link removed**

Are those kind of results reliably possible using a regular iron? I was getting the impression that "ideal results"(i.e., the kind that one really wants) were only achieved in conjunction with their laminator and 1/2 oz. copper board stock, but that was just an impression.

 

[nike6]

depends.

It is possible to solder one side each time completely, adding tin until you get a blob, and then move the PCB with some velocity.

often just one or two shorts will remain, which you can remove with some skill.

I'd recommend a soldering station, at least 50 watts, so the heat is just applied for a few seconds.

 

[3v0]

The only good reason for making you own PCB's is that you can make them 'right now'. If you only make a few it can be cheaper to use a service like BatchPCB.

Are those kind of results reliably possible using a regular iron? I was getting the impression that "ideal results"(i.e., the kind that one really wants) were only achieved in conjunction with their laminator and 1/2 oz. copper board stock, but that was just an impression.

You are correct but neither the laminator or 1/2 oz copper are big ticket items. In the Yahoo group Homebrew_PCB people are buying $25 laminators on ebay and fitting them with DIY temperature controls. The one from pulsar is about $70.

These days most PCBs are on the small side so the cost of PCB material is not a big deal. For SS work I use surplus sissor cut PCB from Electronic Goldmine. For DS I buy new stock.

It is easy to spend more money on this regardless of the method used to make the PCB. I have a shear for cutting the pcb, liquid tin, conformal coating, temperature controled soldering station, and a hot air rework station. Then there is the ever growing inventory of parts. I only mention this to point out that the paper, PCB, and laminator are a small part of the investment.

As you know you need a drill press or some sort. CNC is nice if you have or can use one.

 

[Chippie]

3v0,

Where did you obtain your shear from?

 

[swbluto]

Thanks. I'm just trying to find anything that satisfies the criteria at minimal cost, and getting the full details helps in the evaluation.

It seems the initial investment for the pulsar+laminator+pcb_stock seems to be roughly comparable to the photo method considering time+money costs. Now the next item to determine are the running costs or, probably more appropriately, what would be more costly for a given area of board? It seems the pulsar paper and the extra cost of specialized pcb is the biggest continuous cost, whereas the extra expense of photosensitive boards and possibly the developer(Though, that lasts nearly forever, right?) seems to be the biggest continuous costs. How would those compare?

And, oh yeah, how does the difficulty in doing double sided boards compare? I've heard the photo method is much easier, but I don't know if the laminator-method significantly eases alignment compared to traditional iron-on transfers. Also, any real life experiences using pulsar's+laminator's and how repeatable the result for fine spacing is? It is an impressive picture shown, but I've noticed that the iron-on method can occasionally get lucky.

 

[swbluto]

3v0,

Where did you obtain your shear from?

Very curious about that too! It seems industrial guillotines are recommended, but that doesn't seem as simple as it could be for pcbs. I've been scoring and snapping the pcb thus far, but that seems to be mainly effective for the entire span of the pcb and not a segment less than the pcb's span.

 

[AlainB]

Hi,

I use this model of shear. It is an expensive tool, often used in engraving shops for cutting plastic and thin soft metal. I did not buy it for the purpose of PCB cutting. I just give the information in case somebody happend to find one at a real cheap price.

It comes with the upper blade for metal cutting, maybe at 85 degree angle or for plastic cutting, maybe 20 degrees angle, like a big commercial paper cutter.

Only the plastic blade work relatively well. If equiped with the metal blade, this blade could be reground for the needed plastic angle.

Alain

 

[DirtyLude]

Which paper you are using.....?
Whats your opinion about transparencies and magazine papers....?

I use Staples Photo paper. They've changed it to Gloss paper or something; I think it's in a purple envelop now. I'm not certain. I have a stack of the old stuff.

I think if you are starting out, the Pulsar paper is a good deal. If you want to move to cheaper alternatives later you can.

Double siding and other details are all in older posts.

You can always change the pad sizes in the Eagle footprint libraries if it doesn't match your requirements for etching. I had to adjust a few of them for a boardhouse with 8mil trace/spacing limits.

Really, I don't have many chips that aren't QFP, QFN, or TSSOP, all with 0.5mm spacing. If I couldn't etch for them, I'd be screwed.

EDIT: Generally I can do smaller spacing than I can do traces, so for 0.5mm leads, I do 10mil traces and let the spacing suffer. I also find with 8mil traces, the copper can seperate from the board much easier, so you can get pads that break loose if you are a little hard on them.

 

[Mr RB]

I've recently switched to pressnpeel blue, I thought it was crud the first 2 times I used (with an iron) but now I got a $25 laminator from the newsagency it is giving really nice results.

I had to modify the laminator which was as easy as opening it up and seeing the electronic temp control pcb inside had a trimpot. Being a cheap one with thin (ie flexible) little rollers it feeds standard PCB perfectly. I also cut away a little of the plastic and opened the guides a bit to make feeding easier without snagging the film.

The other trick is to run the temp as low as possible, the pressnpeel will hold very fine detail if it not overheated to where it shrinks/smudges.

Another possible advantage of pressnpeel is it leaves a black plastic coating on your copper, the last couple of jobs I just left the coating on the tracks and scratched it off only on the SMD pads etc before I soldered, so it acts as anti corrosive. Doesn't look quite as flash as shiny copper (it looks a bit "dirty") but it's functional.

 

[3v0]

Warning: long

I am sort of surprised to see the amount of activity on this thread. I have this harbor freight shear/brake and like it.

**broken link removed**

I have nothing against photo transfer PCB making. I use the pulsar system. Which is better for you depends on a few things.

First you wanted to see a real PCB. This work in progress is the 44 pin TQFP board. I did a shitty job on the 805 parts (do not work when angry). The processor chip is spot on. The unpopulated pads are for a serial EEPROM. Note: If I cared about what people think I would not post this image. This my first shot at this board. The next time I will do a solder mask and everything will be beautiful. I will even clean off the flux

**broken link removed**


Some of the header pins are soldered top and bottom. The bread board holds the loose pins in alignment during soldering. For vias I used .025 in holes stuffed with header pins that are held in by friction. Again ugly but faster than messing with wire. The plastic on the 2x5 header is above the PCB because the pins are soldered on both side.

The following is about the two methods. It has more toner transfer detail because that is what I have been using for the past few years. I have tried to be fair to both.

It used to be that you could get more resolution from the photo method. I do not think that is true anymore. Pulsar Toner transfer puts the toner on the copper. Photo relies on contact print. I am not including P&P here.

Resolution is limited by the printer and the ability of the PCB to hold a narrow trace. Now all PCBs are created equal.

Both methods have steps you have to get right. With photo you need to get the exposure and developing correct. With photo transfer you have to get the toner to stick to the PCB and enough of it.

If you do this on the cheap I would have to say you are more apt to have success with the photo method. Toner transfer using an iron is an art, and a lot of work. Will I get flack for saying that?

One problem with photo is that the photo resist chemicals/coatings age. If you make a board or two a year, or a board every two years this could be an issue.

With photo you need lights to expose the resist. With toner you need the laminator. You can go cheap on either with mixed results. There are people who can and do do it on the cheap, but do not expect it to be easy to master.

Next is the cost. The pulsar paper and foil cost under 3 cents a square inch. How much you pay for the PCB stock depends on where you go.

I have not used the photo method recently enough to speak about the cost.

Alignment for 2 sides PCBs is always an issue. There are several solutions. It can always be done.

I drill my holes first then apply toner to one side and mask the other side. Then I mask the etch sided and apply the toner to the other. For masking I use common/cheap (not scotch or UPS) box tape. Some people apply toner to both sides and etch once.

If you do not have predrilled holes you can align the 2 toner images to form a pouch. I have not tried it.

Regarding fine lines with the Pulsar system. As I said earlier the limitation is not the paper but the printer and the PCB. I like to stay at .010 and larger. It is possible to put finer traces on the PCB but you can easily lift them when soldering. There is a greater possibility that that printer will not print the line well enough to etch.

Note that the starch coating on Pulsar paper does not absorb liquid toner. If an insufficient amount of toner is present it will tend to bead up rather than make a line. With a good toner cartridge this is not a problem.

When doing toner transfer use a brand name toner cartridge. There has been a report (here) of a refill brand that works but I do not recall the name. It is best to use a fresh cartridge for printing PCBs and then move it on to regular use when it gets somewhat used up. I do not do that but it would not hurt. It is all about delivering maximum toner.

3v0

 

[DirtyLude]

My 0805's tend to look like crap whether I'm angry or not.

I have a similar board, but I've let the chip hang off the edge of the breadboard to squish the pins together so I have room to plug in wires on the sides. I also have a SAM7 board that I did but I made it cross over 2 breadboards so I could keep some room for plug in wires.

Why do you care about tinning and masking? Usually, if I know I'm keeping it for a while, I just spray on the acrylic. Otherwise I just let them tarnish, but even my oldest boards aren't that bad.

 

[3v0]

DirtyLude,
My breadboards are setup 3 wide with a power bus between each. This processor module spans a power bus with pins in two BB's.

I spray all the non SMD boards with a very thin acrylic coat prior to soldering. I tried MG Chemicals Liquid Tin for the first time on this PCB. I seems to be OK. Too early to tell but lot easier to use then Tin It.

I have been using spray acrylic on most everything once it is finished. For stuff that goes in the garage and maybe outside I have some actual conformal coating.

My 0805's tend to look like crap whether I'm angry or not.

I have a similar board, but I've let the chip hang off the edge of the breadboard to squish the pins together so I have room to plug in wires on the sides. I also have a SAM7 board that I did but I made it cross over 2 breadboards so I could keep some room for plug in wires.

Why do you care about tinning and masking? Usually, if I know I'm keeping it for a while, I just spray on the acrylic. Otherwise I just let them tarnish, but even my oldest boards aren't that bad.

 

[DirtyLude]

DirtyLude,
My breadboards are setup 3 wide with a power bus between each. This processor module spans a power bus with pins in two BB's.

Ya, that's how I did that SAM7 board. Unfortunately I did not home etch it. It would just expand into a monster home etched, with too many vias.

Your board looks very nice. Good work. The tinning does look nice and like I said, my 0805's and 0603's don't look any better.

EDIT: With boards like this I've started using really long header pins that stick up above the board. It makes it easy to clip on a logic analyzer, or even just a DMM while working.

 

[nike6]

I have now ordered 1206 resistors.

on the photo you see one 0805 turned by 90 degree's-
but it's a perfect electrical connecition.

what is really messy are the half-size SMD diodes.

now the next revision will again use conventional parts, maybe retain pads for 100nF SMD capacitors, but trimmer resistor will be conventional.

the SMD parts requires a bit of exact size.

well uhm the PCB is not self-made, i got them for 23 cents each.