State of the Leaf? « LeafLabs Garden

LeafLabs Garden » Community » The Lounge

State of the Leaf?

(35 posts) (7 voices)

Started 4 years ago by robodude666
Latest reply from siy

« Previous 123 Next »

siy
Member

About drilling.
During preparation of image for mask I'm merging top and drill layers
(drill layer on top of tracks). This gives nice ideally centered marks after
etching which are extremely convenient for manual drilling. I guess this is very
common technique. Probably less common is that before merging layers I'm
patching drill layer Excellon file and replacing all drill definitions with
smaller one. Best results were achieved with 0.0197 drill diameter. For example,
beginning of drill layer file may look so:

...
%
M48
M72
T01C0.0354
T02C0.0400
...

After patching it looks so:

...
%
M48
M72
T01C0.0354
T02C0.0400
...

Couple days ago anton posted link to picture where you can see result of
applying this technique. Needless to say that mark greatly simplifies the
drilling. Also, I've found that bits in range 0.6-1.0 tend to "self-centering"
during drilling when these marks are present. To use this "self-centering" I'm
just slightly relax fixing of the board at the moment when bit reaches the
copper, allowing moves in plane perpendicular to drill bit (but still
disallowing vertical moves) and once bit goes through copper I'm fixing board
again. Necessary "hold-relax-hold" drilling skill is easy to develop once you
know about "self-centering".
This approach slightly relaxes requirement to initial positioning
precision and allows get ideally centered drills despite small initial
positioning errors (drill bit still should be at least partially inside the mark, so
"self-centering" is able to fix positioning errors in range about +-0.2mm).
And, finally, big 1.5-2x lens with shadow-less backlight greatly simplifies
board positioning during drilling.

About noise.
I did no measurements yet. In fact most my projects don't need an ADC so it was
not a concern for me. Nevertheless, I'll put ADC tests in my to-do list.

About USB signals.
Original Maple USB network is somewhat complicated by the means of number of
components and is inconvenient for single side routing. My board has only
two components in this part and both components were chosen to be convenient
for single side routing. They also should not be so hard to solder even
manually (one has SOT-23 package and other is about the same size but with 6 leads).

Merry Christmas and Happy New Year!

Posted 4 years ago #
gbulmer
Moderator

siy - very interesting tip on drilling. What PCB CAD do yo use?

Pete Harrison of http://www.micromouseonline.com/ recommends Eagle users try the drillaid ULP.
It reduces the centre hole in vias and pads to create that self-centred effect.

USB - I think Maple's USB circuitry is the same as several other boards, I don't think they designed it.
It is an amazingly large number of components, especially as it doesn't include ESD protection.

In theory the On Semi NUF2042XV6 is an ideal device, but the only distributer I know of which stocks it is Future Electronics :-(
If their footprint is used, it is as fine pitch as the 64pin STM32F (0.2mm space), and it is small, roughly 0604.
Not as compact as your approach.
I use a dumb MOSFET to activate the pull-up resistor, and so need a pull-up resistor to hold it off.

Merry Christmas and Happy New Year.

Posted 4 years ago #
siy
Member

I'm using Eagle, so thanks for tip, I'll look at that ULP.

My version of USB circuitry uses On Semi STF202 in TSOP-6 package which is about of the same size as SOT-23 but with 6 leads. I've purchased them twice - once at local store and once at Mouser (they were on stock, so I guess this is not a rare component). The main advantage of STF202 in comparison to vast majority of other similar ICs - different pinout which matches MCU pinout for USB signals and matches mini-USB socket pinout (no need to swap wires) for full-speed operation mode (under "matches" I mean compatibility with single side routing). I'm using (almost) standard (described in datasheet) connection with one exception - VBUS pin is connected to collector of the PNP bias resistor transistor (they also sometimes called "bipolar digital transistor"). These devices include necessary current limit and bias resistors and can be connected directly to the MCU IO pins. There are several such parts from different manufacturers which have SOT-23 package and identical pinout. For example, MUN2111 from On Semi or PDTA114ET from NXP to name a few (last one is used in my board). The emitter of the transistor is connected to VCC and base - to DISC net. Resulting equivalent circuit almost identical to Maple one (except it uses one PNP transistor instead of two NPN transistors connected to form an inverting switch). This circuit has one issue - it does not protect VBUS. But this is still better than no protection at all.

Posted 4 years ago #
gbulmer
Moderator

siy - Thank you for the pointer to the digital transistors.

When you write "The emitter of the transistor is connected to VCC", I assume you mean 3.3V.

RS components sell a range of digital transistors. Some have different values for the R1 and R2 resistors. I'd be tempted to use a 4.7K/10K or 2.2K/10K. Did you specifically choose the 10K/10K part, or did you go for it because it is common and easy to get?

There is an electronically 'perfect' USB interface IC, by Semtech
http://www.semtech.com/circuit-protection/esd-protection/

It is the EClamp2522P
http://www.semtech.com/images/datasheet/eclamp2522p.pdf
which has 7 connections, with the 1.5K pull-up resistor on an independent pin, i.e. not connected to Vcc. So the VBus power connection is always connected to the USB socket, and hence protected, *and* the 1.5K pull-up can be driven separately.

The big problem for me is the package is both small, and has a 7th pad under the chip. When we were experimenting last year, we tried soldering a DIY-PCB with something at this pitch, with a track underneath, and it got a solder bridge underneath the part. It was hard to diagnose and we could only repair it by removing the entire part, trying to clean the copper, and trying again until it worked. Frustrating. I think the spacing is too fine for solder mask to fit.
Have you any tips on how we might get this to work on a DIY PCB?

Posted 4 years ago #
siy
Member

Yes, I mean emitter is connected to 3.3V.

The one used in my design is just available at local stores, there is nothing special with it. If I remember correctly the requires current for USB pull up is only few milliamps and required base current is many times lower, so I think that almost any digital transistor will work fine in this circuit.

As for EClamp2522P, it looks interesting, but usually I'm trying to avoid leadless parts since they are rather inconvenient to solder manually. The soldering of such parts is much simpler with hot air gun or soldering oven, but if something went wrong, issues much harder to diagnose and fix. And as I can see from your post, you already know this. Solder mask significantly simplifies this task and its resolution is enough to fill gaps between pins even for such a small package (at least this is correct for solder mask used for my boards), although this requires very careful matching of the solder mask and layout. Probably most helpful tip here is proper design of landing pattern. Instead of using one suggested by manufacturer I'm usually prepare my own version with slightly longer pads, so they are not covered by package and reachable for manual soldering. Pin at bottom side can be soldered with hot air gun (or oven, of course), but since this part has pins only at two sides, you may try prepare pattern so pad for bottom pin will be visible at one side of the package. Beside making it much more convenient for soldering, this may help you fixing issues, because such form of the pad allows removing of excessive solder from bottom pin if necessary without removing part completely. I know that some people use via at the center of the pad for bottom pin. This makes pin reachable for solder from the other side of the board (and might help fixing issues too). But I've no experience with this technique and have no tips for it.

Posted 4 years ago #
siy
Member

By the way, USB protection/disconnect circuit as it is implemented in my design: http://we.easyelectronics.ru/uploads/images/00/22/11/2011/12/27/d14b13.png

Posted 4 years ago #
siy
Member

By the way, forgot to mention one important thing related to photoresist technology: the layout should be prepared so spacing between tracks is as uniform as possible and board space without copper should be minimal. Sometimes this requires changes in the track placement in order to allow background polygon fill remaining space. This serves several purposes: simplifies task for the printer, especially at finer resolutions (it prints mostly narrow black lines - most convenient pattern for the laser printers); great improvement in etching uniformity and quality - since etchant needs to remove almost constant amount of copper, time required to do this is also close to constant across the board; finally this extends life of the etching solution, since layout prepared this way usually requires removing of smaller amounts of copper.
Probably all of the above is well known approach, but I guess it worth to mention anyway.

Posted 4 years ago #
LarryP
Member

Postlathian wrote:
As it turns out, the stm32f2 and f4 chips have radically different peripheral
interfaces than the stm32f1, like we use on Maple. So a lot of work has been going
into bringing libmaple up to support stm32f4!

Greetings all,

From what Postlathian wrote above, it sounds like STmicro's overview of the F4 is more marketing speak than reality:

http://www.st.com/internet/mcu/subclass/1521.jsp says:
The STM32 F4 series remains pin-to-pin and software compatible with the STM32 F2
series and features more SRAM, peripheral improvements such as full duplex I²S,
improved RTC and faster ADCs.

Sigh...

Side note:
On a forum re NXP's Cortex chips, I posted that IMHO, ARM and their licensees should skip from Cortex M4 to Cortex M6. Anybody who's watched the original Star Trek[*] knows that naming any CPU M5 is just asking for trouble.

[*] Star Trek episode, "The Ultimate Computer" http://en.wikipedia.org/wiki/The_Ultimate_Computer

Posted 4 years ago #
gbulmer
Moderator

LarryP

From what Postlathian wrote above, it sounds like STmicro's overview of the F4 is more marketing speak than reality:

http://www.st.com/internet/mcu/subclass/1521.jsp says:
The STM32 F4 series remains pin-to-pin and software compatible with the STM32 F2
series and features more SRAM, peripheral improvements such as full duplex I²S,
improved RTC and faster ADCs.

I believe the STM32F2 and STM32F4 are very similar. The power circuitry and most peripherals look identical (I haven't looked at every peripherals so there may be some small differences).

The current Maple's are based on STM32F1. The functionality at the pins of STM32F1 are comparable to the STM32F2 and STM32F4, but the peripherals implementing the functionality in the STM32F1 are different.

I believe Poslathians point is they are different from STM32F1, and hence are quite different from the Maple library.
The text you reference doesn't claim anything different.

So I think that text you reference is accurate, and compatible with Poslathian, but you seem to be implying that it is misleading. Maybe you were assuming that the STM32F2 was very similar to the STM32F1?

(Full disclosure: I am not e member of LeafLabs or ST Micro staff.)

Posted 4 years ago #
robodude666
Member

Can we try keeping this thread to status updates from LeafLabs people? If you want to discuss new chips, boards, cad stuff, etc. make new threads for them please.

-robodude666

Posted 4 years ago #
gbulmer
Moderator

siy

By the way, forgot to mention one important thing related to photoresist technology: the layout should be prepared so spacing between tracks is as uniform as possible and board space without copper should be minimal. ...

I try to stick to a rigourous 10/10 track space, and use hatched polygon fills with the same ratio to try to get this effect. The hatched fill seems to work much better than a solid fill, both when printed and when etched.

I've been recommended put a 'sacrificial' border of copper around the outside edge. I've tried it, and it seemed to work. This seems to stop fine detail being over-etched.

Posted 4 years ago #
gbulmer
Moderator

robodude666

Can we try keeping this thread to status updates from LeafLabs people?

I think we went off your view of the topic twenty odd posts ago. Maybe LeafLabs should start a proper "status" thread? Or maybe use the blog a bit more?

Posted 4 years ago #
siy
Member

Hmm. Do you use dry film photoresist? For dry film photoresist mask is negative and polygons are not an issue at all because printer does not need to print anything at places filled by polygons.

For etching, recently I've tried method similar to one seen at this video http://www.youtube.com/watch?v=Uw0P0Mimbw8 (a vibro motor is used to generate waves in the etching solution). Results are very good, in particular in regard to etching uniformity.

I'm also adding a border around the main PCB space (just merge Dimension layer to the mask), but for different purpose - it prevents damaging of solder mask during final cutting/milling of board to the required size.

Posted 4 years ago #
gbulmer
Moderator

siy - We use presensitised PCB boards, e.g.
http://uk.farnell.com/cif/ab16/pcb-presensitsd-epoxy-2f-100x160/dp/1267741

So there is a positive film mask. We print the film with a laser printer. We believe the black is more even and dense across a hatched polygon area than a solid polygon area. A solid polygon, more than a few mm, is not such an even black when printed as a mask, and seemed pitted by the time the rest of the board was adequately etched.

I have access to a etching tank, which uses bubble agitation, and a heater, and the results are very even.

We haven't tried making a solder mask. Our boards only get etched and chemically tinned.

Our biggest problem is damaging the presentised resist layer when cutting the PCB.

Posted 4 years ago #
siy
Member

I see. Polygons is definitely an issue with positive photoresist. I'm experimenting with using chemical tin as resist during etching (this is a step of industrial-like PCB manufacturing process). For this approach photoresist is used only for opening areas of board which should be protected during etching. Then chemical tin is applied and it covers these areas. Finally photoresist is removed and board is etched. One of the advantages of this approach for positive photoresist is that negative mask is necessary (i.e. mask is inverted in comparison to regular use). In other words, with this approach there will be no issues with polygons. So far results are not good enough for actual use, but I just started my experiments and there is a lot of parameters which should be adjusted and then tuned for best performance.

As for mask, I'd suggest to use something like this: one http://www.ebay.com/itm/PCB-UV-Curable-Solder-Mask-Repairing-Paint-Green-100g-/370567399875?pt=LH_DefaultDomain_0&hash=item564789c9c3
It is extremely simple to use in non-industrial environment, although good contrast negative mask is necessary for best performance. If you decide to try it, following things should be keept in mind:
1. The thickness of the solder mask is critical, if it is too thick, it can't be properly polymerized regardless to exposure time;
2. Thin solder mask works just fine and close to industrial two component mask from the point of view of temperature resistance (experiments show that it can handle as much as 400C without any visible consequences) and reliability during soldering;
3. Mask thickness defines the color tint, so uniformity of the solder mask thickness actually defines the look of the PCB;
4. Contrary to description provided with the item mentioned above, it actually requires no chemicals (except washing detergent used to clean the board), no oil based solvent is actually necessary;
5. Actually very small amounts of solder mask are consumed during process, 100g is enough to cover, probably several square meters of PCB surface.
6. Since exposure time if high (usually 30-40 minutes for first exposure, when photo mask is used) requirements to photo mask quality are high;
7. Solder mask is very tolerant to processing times. For first exposure I've tried times from 30mins to 1hour 20minutes with almost equal results. For second exposure at least 40minutes are required, but increasing time at least to 2 hours does not cause harm (I've just not tried longer intervals).

Some time ago I wrote a detailed article of the process of applying solder mask with pictures taken during process, but the article is in russian. If you're interested, I may prepare English version of this article with recent changes in process taken into account (the original article describes early version of the process, where oil based solvent was present, later I've found that solvent is redundant).

By the way, have you tried to apply toner density spray to printed photo masks? In most cases it is able to convert print of average quality to virtually perfect photo mask. I'm using this one http://www.kruse.it/html/uk/density_uk.html , but I know that same results were achieved with this one too: http://www.power-tech.com.au/products/index.php?dispatch=categories.view&category_id=15

Posted 4 years ago #

RSS feed for this topic

« Previous 123 Next »

Reply »

You must log in to post.