Chips in "difficult" packaging

Thread Starter

BladeSabre

Joined Aug 11, 2005
105
Currently I'm researching the AVR USB microcontrollers. There are four to choose from, and they come in 64-QFN and 64-TQFP. What I'm wondering is whether there's any sane, affordable way for me to include chips like that in real circuits.

(Wikipedia tells me that surface-mount technology is designed to be used by machines, not humans; and my previous experience with a similar shape was unpromising.)
 

Nik

Joined May 20, 2006
55
Hi, can you make a 'daughter board' by mounting the SMD whatsit on a DIL header ??

I've not tried it with SMD, but I used this trick to replace an 'obsolete' audio-amp IC...
 

kubeek

Joined Sep 20, 2005
5,794
Originally posted by Nik@May 21 2006, 12:23 PM
Hi, can you make a 'daughter board' by mounting the SMD whatsit on a DIL header ??

I've not tried it with SMD, but I used this trick to replace an 'obsolete' audio-amp IC...
[post=17253]Quoted post[/post]​
Here we have DIL sockets for SMD quad-packages, but maybe it is cheaper and better to choose DIL packages, as they might be easier to get in piece quantities.
 

Thread Starter

BladeSabre

Joined Aug 11, 2005
105
I don't know how I'd go about making an adaptor- I've searched for ones I could buy, and they're out of my price range. Some kind of socket thing would be best, because the thought of trying to precision-solder all 64 of those tiny pins makes my mind boggle.

maybe it is cheaper and better to choose DIL packages, as they might be easier to get in piece quantities.
The AVR ones don't have that option. I started looking at PIC devices today, and they have a cheaper USB controller in 28-PDIP, so that would likely be a better choice. (PIC also has the advantage that it's a lot easier to get, here in the UK.) The question there is whether this item actually exists yet, because its description page still says it's a "future product". But there do appear to be some in their shop...

And I still can't figure out what I actually need, in order to program either of these devices. Since that affects the starting price drastically, it's rather important in deciding which of the two brands would be more suitable.
 

n9352527

Joined Oct 14, 2005
1,198
Originally posted by BladeSabre@May 21 2006, 01:23 PM
I don't know how I'd go about making an adaptor- I've searched for ones I could buy, and they're out of my price range. Some kind of socket thing would be best, because the thought of trying to precision-solder all 64 of those tiny pins makes my mind boggle.
The AVR ones don't have that option. I started looking at PIC devices today, and they have a cheaper USB controller in 28-PDIP, so that would likely be a better choice. (PIC also has the advantage that it's a lot easier to get, here in the UK.) The question there is whether this item actually exists yet, because its description page still says it's a "future product". But there do appear to be some in their shop...

And I still can't figure out what I actually need, in order to program either of these devices. Since that affects the starting price drastically, it's rather important in deciding which of the two brands would be more suitable.
[post=17256]Quoted post[/post]​
The 18F4550 USB PIC (40DIP) is available in the UK from Farnell (for around £5), I have no idea about the 28SDIP USB one (18F2455?). They are compatible though, only different IO number and flash memory size.

PIC programmer is quite cheap to obtain, or make yourself. And once you've got a programmer or access to one you could burn a small bootloader program so that further programming could be performed through USB link without any programmer (enhanced flash self programming).

I have a conical tip for soldering iron that is excellent for smd soldering. It uses the surface tension insode the cone to dispense just the right amount of molten solder and prevent bridging and too much solder. I'm pretty much able to solder most of the smd components/packages manually except for things like BGA packages where the pins are not accessible. It is not that expensive to buy, but you would also need additional flux to work with it (and probably flux cleaner too).
 

hgmjr

Joined Jan 28, 2005
9,027
There is a possible solution you may want to consider involving the use of a PIC or AVR microcontroller together with an off-the-shelf USB interface device.

This approach could provide several benefits.

1. You can chose the microcontroller that you would like in the package that suits you.

2. You can chose from a variety of USB interface devices also in the package that suits your need.

3, And lastly, you could off load some of the software complexity to the USB interface device you select.

I did a bit of searching at www.datasheetarchive.com and there are several available USB inteface devices available in dip type packages. You will be the best judge as to which one of these devices provides the features that best suit your needs.

hgmjr
 

Thread Starter

BladeSabre

Joined Aug 11, 2005
105
The 18F4550 USB PIC (40DIP) is available in the UK from Farnell (for around £5), I have no idea about the 28SDIP USB one (18F2455?). They are compatible though, only different IO number and flash memory size.
I was looking at the 18F2450 - they don't have that one though. (2nd edit: They have the 18F2550, which also seems to be the right thing.) There's not much difference in price between the 28 and 40 pin ones, but I find fewer pins is generally easier to use. (I'd want to get a programmer that can do the big ones though, since I might decide I need them later.)

Thanks so much for pointing out Farnell. Apparently Google doesn't crawl their product pages- all that time I spent searching for serial numbers, they were there all the time. Farnell have so much stuff, wheee... and free shipping too!

PIC programmer is quite cheap to obtain, or make yourself.
That's good to hear. I'm still confused about them though- I've spent several hours reading about this stuff, and I've still not figured out which programmer can do what. I know the £30 starter kit won't do the USB controllers. Though if I go for the two-chip option that hgmjr was talking about, I could get that one and use a low-end PIC. (I've not actually done any microcontroller programming yet, and I hear the starter kit is useful for learning.)

Then there's things like this, which specifically say they do the 18F4550. But I'm dubious of something that uses the USB and the parallel port at the same time. And that software looks primitive.

And once you've got a programmer or access to one you could burn a small bootloader program so that further programming could be performed through USB link without any programmer (enhanced flash self programming).
I've read about this. It said that if I use Microchip's bootloader, the item won't be able to do other USB functions. But it's possible to do both with a custom bootloader then?

I have a conical tip for soldering iron that is excellent for smd soldering. It uses the surface tension insode the cone to dispense just the right amount of molten solder and prevent bridging and too much solder ... It is not that expensive to buy, but you would also need additional flux to work with it (and probably flux cleaner too).
Ah, the "carry a blob of solder on the end of the iron" trick? (I use separate flux frequently, because trying to use the flux core in the solder seems to require 3 hands.)

I've been looking for a good soldering tip for the really small fiddly things, and I'd be grateful for more information on this one.



There is a possible solution you may want to consider involving the use of a PIC or AVR microcontroller together with an off-the-shelf USB interface device.
Ah, I never thought of that. Thanks for pointing it out. I'll do some more research on that and get back to this. So far I'm having the old problem finding any of these devices in the UK, but they've got to be here somewhere...


Edited to fix pasting the wrong serial number.
 

Thread Starter

BladeSabre

Joined Aug 11, 2005
105
Hi hgmjr,

I took some time to do some more reading and try to figure this out.

I did a bit of searching at www.datasheetarchive.com and there are several available USB inteface devices available in dip type packages. You will be the best judge as to which one of these devices provides the features that best suit your needs.
I found some devices that were very nice apart from their awkward packages. In DIP type packages, all I could find was the CY7C63XXX, which are not the right thing.

So, advantage number 2 isn't showing itself right now, and the combined USB PIC devices would probably be good enough. Though, I'm wondering what you meant by this:
3, And lastly, you could off load some of the software complexity to the USB interface device you select.
Like I said, I'm new to microcontrollers. Anything that makes the software easier to write could be significant. And, I don't know enough at this stage to understand what kind of difference this would make. I'd be grateful for any further information.




I'm going to download the PIC development software and have a look at it. I've pretty much decided on getting the PICKIT2 now - it's a full starter kit, and it can program the PIC18F2455/2550/4455/4550 if I end up using them.
 

n9352527

Joined Oct 14, 2005
1,198
OK, the following are my opinions, or experiences depending on how you look at them, on developing with the USB PIC:
1. Microchip provides example USB firmware (device interface, packet transfer and chapter 9 handler) for their devices. It's pretty general and you could probably adapt it to your own requirements (different endpoints number, interface, configuration, etc.). They also have different host interface (HID and serial) drivers, in case you don't want to spend that much time developing the host driver.
2. They also provide a generic host device driver (windows only), that you could use through the supplied dll. No source code for the device driver, but the codes for dll and sample apps are available.
3. You could use the PIC USB bootloader to develop your program (upload the code to PIC) instead of using development kit. I prefer this as it allows me to work with the real prototype instead of generic development kit and no need for a programmer (except for initial bootloader programming). The way to use this feature is to build the prototype of your design and then download and burn the USB bootloader from Microchip (you need access to a proper ICSP programmer for this). Then you could use picdemfs program to upload your code through USB and run and test the code in a few seconds. No need for ICSP programmer anymore.
4. You would need to obtain your own VID and PID. You could ask Microchip to give one of their PID for you (there is a form on their site that you need to fill and fax). Or you could find a third party vendor that's willing to sell you one for a reasonable price.

Is your appliccation generic enough? Is it falling within one of the preallocated windows USB class? Or do you need to develop custom driver and firmware?

Another popular USB chip is FTDI. Maybe you should also look at it. Very simple to use, interface to micro and access in windows (appears as a com port).
 

Thread Starter

BladeSabre

Joined Aug 11, 2005
105
Is your appliccation generic enough? Is it falling within one of the preallocated windows USB class? Or do you need to develop custom driver and firmware?
I think it would be pretty generic. A virtual COM port would probably be enough, though I'm not quite certain what those are capable of yet.

Another popular USB chip is FTDI. Maybe you should also look at it. Very simple to use, interface to micro and access in windows (appears as a com port).
Ah, those look very nice too, though as far as I can see, there's nothing in a DIP-type package. I'm back to the original topic, of "is there any sane, affordable way for me to include chips like that in real circuits". If I was able and willing to use SMD, that would give me a lot more choice.


If this gets too tough, there's still the fall-back idea I thought of in the last topic, of using a PIC with a real COM port (which doesn't look all that hard, though it's not easy to tell at this stage). For laptops, there are complete USB-to-serial converter units available for 5.50 GBP.


So, the USB PIC. Thanks for this useful first-hand information.

Microchip provides example USB firmware (device interface, packet transfer and chapter 9 handler) for their devices. It's pretty general and you could probably adapt it to your own requirements (different endpoints number, interface, configuration, etc.).
...
3. You could use the PIC USB bootloader to develop your program (upload the code to PIC) instead of using development kit.
The USB bootloader looks like a good feature. What I'm still not sure about is how easy it is to combine it with another USB function, since Microchip present the bootloader as one of several alternative firmware options.
Originally posted by "microchip.com"
These are the basic alternatives available:
*Human Interface Device (HID) class firmware
*Communication Device Class (CDC) firmware
*Mass Storage Class Firmware
*Microchip Bootloader ~ CONS: does not provide other communication capabilities
*Microchip Custom Driver ~ requires a deeper understanding of the internal workings of the bus as well as the development of custom interfaces (Windows code development) on the PC.
4. You would need to obtain your own VID and PID. You could ask Microchip to give one of their PID for you (there is a form on their site that you need to fill and fax). Or you could find a third party vendor that's willing to sell you one for a reasonable price.
Yes, that's another inconvenience, but it's not a huge problem; and since I'm a one-person team here, I don't think there's anything stopping me using a number that isn't mine during the development process. That would mean I only need to obtain my own one if I ever actually get this thing to work...
 

n9352527

Joined Oct 14, 2005
1,198
The virtual com port is easy to use, just like any com port. The speed is limited though. I would imagine it is not going to be possible to set a com port to FS 12Mbps. FTDI has modules with DIP pinouts as well, that might be a possible alternative (or it might be by a third party). I know someone that purchased one for 12GBP, parallel bus interface and very easy to use (according to him).

You could use either HID class (64kbps) or the FS mchpusb.sys (I think the name is) to get higher speed.

USB bootloader is just one firmware that Microchip provides. They also provide real FS firmware and windows driver (custom driver you quoted in your post). It is not that difficult to customise.
 

Thread Starter

BladeSabre

Joined Aug 11, 2005
105
I found the DIP ones under "Evaluation Kits". I think 12GBP each would be a bit much, since I'd need a microcontroller in addition to that. (And the complete USB-to-serial cables I mentioned for 5.50 GBP have the same purpose, they're just bulkier. If I decided a virtual COM port would be suitable, maybe I could even rip them up and use the parts- that would give me the USB connector and most of the wiring as well.)

I'm aware that a virtual COM port would not handle full speed. I just don't know anything about hooking a COM port up to a microcontroller, and what would be involved in that. A module with a parallel bus sounds like it would make things easier, but it may not be difficult to get round it. I've been trying to find out if there's any problems re: processing the serial signals directly.

There are USB-to-parallel converter cables too, though I'm looking at about 7.50GBP each for those.


The PIC18F4550 and family is still very much the cheapest option, if I can learn to use it. I'm glad to hear that the customisation is not difficult, though it may still be difficult for me =). Thanks for the information.

--------

Edited to add: I wrote, "maybe I could even rip them up and use the parts". Well, I just peeled my £5.50 USB-to-serial converter. It took over an hour, and I had a few near-misses with the knife that were quite disturbing. So not something I want to do again. (If I were to unsolder the serial connector and attach wires where it was, the module would be pretty small- but it's not worth the time or the risk of injury.)
 

windoze killa

Joined Feb 23, 2006
605
Originally posted by BladeSabre@May 21 2006, 08:38 PM
Currently I'm researching the AVR USB microcontrollers. There are four to choose from, and they come in 64-QFN and 64-TQFP. What I'm wondering is whether there's any sane, affordable way for me to include chips like that in real circuits.

(Wikipedia tells me that surface-mount technology is designed to be used by machines, not humans; and my previous experience with a similar shape was unpromising.)
[post=17252]Quoted post[/post]​
Firstly I passed over most of the other responses.

These chips are certainly designed for machines BUT it doesn't mean we people can't do it. It is part of my job to do these daily. You wil NEED and very fine tip temperature controlled iron. VERY good eye sight or a stero microscope. If you have access to these then you should have no problems at all.
 

Thread Starter

BladeSabre

Joined Aug 11, 2005
105
These chips are certainly designed for machines BUT it doesn't mean we people can't do it. It is part of my job to do these daily. You wil NEED and very fine tip temperature controlled iron. VERY good eye sight or a stero microscope. If you have access to these then you should have no problems at all.
Thanks for the info. I have relatively poor eyesight, and my general soldering skill is probably substandard. A good soldering iron may be a good investment, but I could not justify the cost of a good microscope.

Fortunately the PIC18F4550 (DIP) looks at least as suitable as the AVR chips I mentioned previously.



My current problem is Schottky diodes - the only ones meeting my requirements use SOT-23 package, which is pretty small.

I was wondering, what about conductive glue? Would that work? I guess it may depend on how much resistance I can afford to add in any particular situation. And, the regular epoxy glue I've tried is not terribly easy to aim either. The main point would be to avoid burning my components...
 

n9352527

Joined Oct 14, 2005
1,198
Conductive glue is a big no no. Anyway, SOT23 is not that hard to solder, the pins are quite far apart and bridging is usually not a problem. A big blop of solder may look unsightly, but as long as it works I don't think it's a problem...

Iron with controlled temperature is a must in my experience, not only high temperature iron would burn the component, in some cases it would peel the pads as well. And I've seen people crack the case of SOT23 by using ordinary iron. Just a quick dab (2 or 3 seconds) would make a good joint and wouldn't destroy the component.

If you can not afford a microscope, then a magnifier will do as well. I don't know how much it'd cost, but I think not that prohibitive. I have one with an adjustable leg and an integrated fluorescent light, a very useful item.
 

Thread Starter

BladeSabre

Joined Aug 11, 2005
105
Can you explain why glue is a "big no no"? I was just reading about it, and the Loctite one claims to be an alternative to solder for "temperature sensitive components". What's wrong with it?

Anyway, looks like I'm going to need a new iron whatever I do. The magnifier thing sounds neat. There's a lot of kit that could make my life easier I guess...
 

Dave

Joined Nov 17, 2003
6,969
Originally posted by n9352527@Jun 7 2006, 11:47 AM
Conductive glue is a big no no. Anyway, SOT23 is not that hard to solder, the pins are quite far apart and bridging is usually not a problem. A big blop of solder may look unsightly, but as long as it works I don't think it's a problem...

Iron with controlled temperature is a must in my experience, not only high temperature iron would burn the component, in some cases it would peel the pads as well. And I've seen people crack the case of SOT23 by using ordinary iron. Just a quick dab (2 or 3 seconds) would make a good joint and wouldn't destroy the component.

If you can not afford a microscope, then a magnifier will do as well. I don't know how much it'd cost, but I think not that prohibitive. I have one with an adjustable leg and an integrated fluorescent light, a very useful item.
[post=17592]Quoted post[/post]​
I would say manually soldering a SOT23 device is a fair challenge. The distance between the two pins on one side of the package is less than 2mm, and the package itself does not lend to easily stabilising the package whilst soldering. Could you be confusing SOT23 with SOT223?

Dave
 

Thread Starter

BladeSabre

Joined Aug 11, 2005
105
I'm not too worried about the distance between the pins this time. These diodes have only three pins; one side has only one, and the side with two has the central one missing, leaving a ~2mm gap. The real problem here is the length. The pins are less than 1mm long. Even with a temperature controlled iron, I don't like my chances of trying to manually solder that without destroying the component.

And like Dave said, "the package itself does not lend to easily stabilising the package whilst soldering". I bought a few already and they remind me of nail clippings. If I don't burn it I'll crush it, or ping it across the room and spend half my time crawling around on the floor with a torch. (It's happened with bigger things.) Anyway, that's why I asked about the glue idea.

What I'd like to do in the short term is to stick legs on a few for breadboarding.
 

n9352527

Joined Oct 14, 2005
1,198
What I usually do is tin the single side pad and then just tack solder that pin quickly (for SOT23-3), then I solder the other two pins, and then back to the other one. As long as soldering each pin doesn't take more than 2 or 3 seconds then the other joints would be intact. Longer time might melt the other joints or necessitate holding the package down or gluing it before soldering. For SOT23-6 tack one of the corner pin first, and work your way diagonally. Don't try to straigthened it out if it doesn't sit quite right, that might end in one of the pin comes loose or cracked case. A tweezer with a broad grippy head is very handy for initial tacking, the one that grips when released and not the other way around like normal one so you don't have to pinch it while soldering

I'm fairly sure it was a SOT23 and not that bigger SOT223 with flat tab. I used to work with them both in another life at a semiconductor manufacturer. The gap is not that bad, bigger than SOIC or quad flat which can still be carefully soldered manually.

I would suggest you to try soldering one and see how it goes. You might find it doable.

A light coloured antistatic mat would make it easier to find those small missing components, unless you fling it down on the floor (carpet floor is the worse) in which case it is time to cut your loss and leave them for the vacuum.

The glue doesn't really work for multi-pins devices and close pins, I guess it has to sit flat to work best. It also doesn't come off easily, if you have pins that close together and too much glue at one pad, when you sit the component down it would squeeze the excess glue all over the place and cause bridging. Or you might dab too much of it and cause bridging before placing the component. Or not enough and end up with a hanging pin. And I haven't found an easy to way to clean that yet or repair it without sacrificing the already glued joints. Messy. This is just my experience, I've never had a satisfying result with conductive glue. I'm sure Loctite doesn't agree with me :p
 

Thread Starter

BladeSabre

Joined Aug 11, 2005
105
OK, I'll try one. There's a problem with my order that I'd better sort out before soldering anything though, so not today. If it doesn't work, I can try the glue still (if I can find anywhere that sells it in small quantities).

Melting joints that I've done already when trying to solder the new ones seems to be a common problem for me. (As is losing things in the carpet- I think springs are the most common. At least they're shiny.)
 
Top