Thermal barriers, tombstones and other delights of printed circuit boards

While cool designers are designing the right boards and ordering production at state-of-the-art US-European factories, let us turn to the experience of developing printed circuit boards for the urgent production of one of Moscow-based (actually Zelenograd - Moscow) plants with manual basement assembly they are sitting on the second floor, and in the basement there is a line at 60,000 components per hour).

Further text - the personal opinion of the author. This is not the ultimate truth, but only one of the possible cuts of that huge layer of information that is currently available to the ordinary designer.
')


Let's talk a little about CAD. Despite the fact that there are quite a few of them, almost all of them provide the designer with approximately the same basic functionality. Possessing the necessary knowledge, it is possible in Sprint-Layout (no offense!) To make a high-speed differential pair that will work normally. The only question is how long it will take to trace this pair.



At the same time, you can mess up the same PADS (“working with tracing” from electronics mixes with examples of interesting designs) that it will take a long time and tediously to clear up:



What is it for? And the fact that if the designer is an idiot, then no ultramodern CAD system will solve the design tasks for it. Yes, there will be a check of all sorts of rules and much more. But absolutely everything is impossible to consider. This is especially true of soldering and installation issues (although with the placement check it has recently gotten better and better).
Unfortunately, in the design of PP, there are often absolutely mutually exclusive paragraphs. A typical example is blocking capacitors for power. On the one hand, it is necessary to place them as close as possible to the power supply of the chip.



On the other hand, it is impossible to put the components flush with each other if automatic installation is planned (is it not possible, what are the limitations of a particular automaton that the components will arrange) and reflow soldering (the capacitor may be in the so-called “shadow” from the high component) heating the paste on its contact pads will not melt as it should).
And we must not forget about the frequencies at which everything works for us, and how the inductance of the paths from \ to the capacitor will affect its properties in a particular circuit (I strongly recommend reading this and this for a better understanding of the issue).

But back to our stones and barriers. If you don't know what to do, then what should you do? Read the documentation, of course! And she is, and a lot!
We will not touch our guests, some of which have not been updated for more than forty years. Let us turn our attention to the IPC standards, which are used by a large (than the unsubstantiated personal opinion of the author, since I do not know the exact figures) of the designers of the PP around the world.



The standard tree shows that first of all you can read something from 2220 and 7351.
Of the 2220, we are primarily interested in 2221 - Generic Standard on Printed Board Design . The link is the previous version of the standard, because the current is only for money (well, we know that everything can be found if you wish).
In particular, about thermal barriers are written in section 9.1.3. If we briefly translate what is written there, then it all comes down to the fact that if the component is output (and 2221 is dedicated exclusively to output components), then it is advisable to make thermal barriers, since their presence facilitates soldering.

Let's now see what a thermal barrier is:



And what does it help?
Well, firstly, when soldering a wave.



This type of soldering is mainly used in very large-scale production. At the same time for the solder wave there are a number of restrictions on the placement of components and a bunch of recommendations on the landing sites for the components.
Dave, as always, explains clearly and simply what is what:



Optimization, in a word, in order to reduce the number of defects.

Interesting process:


The problem here is that not all cars have a pre-heating zone, and it does not always help if the polygons are large. It turns out that if you drive the board more slowly, so that large polygons get warm, then there is a great chance to overheat small components. Thermal barriers help solve this problem. But the devil is in the details, as always.

Before moving on, let's talk a little more about manual , not machine soldering.

Imagine that you have a multilayer board, as in the cdpv at the beginning of the article, and you need to solder the connectors into it. But some problem is that the board is six-layered, with four layers stupidly covering the entire board. And half of the pins in the connectors is the ground. You can, of course, use preheating to 100-110 degrees and then solder. But this is not always possible. Yes, and tantalum once again do not want to heat (because at first all small things were soldered, and the connectors were the last). This is where thermal barriers help. A soldering iron heats only the connector pin, a glass of metallization and contact patches, and not all polygons in all layers.

And here the problems begin.
It is quite obvious that the thermal barrier reduces the total area of ​​copper through which the component and the landfill contact. That is, increases the physical resistance of the contact. The parasitic inductance of the compound also increases.

A rather cursory search showed that there are at least three articles that examine the issue of resistance, inductance, and the relationship of all of this.

Here they are:

• A Thermal Relief Pads: Tradeoff between Manufacturing and Cooling
• Investigation of Relief Shapes on Different Printed Circuit Boards
• Modeling the Low- and High-Frequency Impedance of Thermal Reliefs

All articles are published by IEEE , which suggests that there is no outright nonsense.
Again, I highly recommend reading.

Conclusions of all the articles, in short: yes, thermal barriers are cool and useful, but you need to use it wisely (who would doubt, right?). If in doubt, it is necessary to simulate at least. Basically, this all applies to components for surface mounting, since the output components already have sufficient parasitic inductance of the conclusions.
As for resistance, yes, it increases slightly at the point of contact. Approximately one milli. If this is not critical, then you can use. And, of course, it is necessary to take into account the current that will flow through this connection. If it is such that it will be heated, then you should look at what is more important: some local overheating or ease of installation.

For myself, I brought this rule. You can safely use thermal barriers on mock-ups, since anyway mock-ups are mostly soldered by hand. On serial products, you should consult with a specific production, where boards will be mounted (this is generally useful to do, in the sense of consulting with production), because the technologist knows better his assembly line and its capabilities.

But it was all about the lead components. But what about the surface mounting? And here it is still somewhat more complicated. The standard 7351 - Generic Requirements for Surface Mount Design and Land Pattern Standard does not mention thermal barriers at all.
There is an interesting document in which the author advises not to use thermal barriers due to an increase in the connection impedance. Indeed, this makes sense: when designing power sources and other power circuits.

A little distracted.
I hope what the components look like looks like, many people represent.
If not, this is a good illustration of the process:



After packing the board, it goes down the conveyor to the reflow oven.
It looks like this:



Unfortunately, sometimes this effect is possible:



There may be several reasons for its occurrence (you can familiarize yourself with the main ones, for example, here ).
One of them - the curve thermoprofile soldering. Next come (not in order of importance): expired (or improperly stored) paste, incorrect stencil for transferring paste, crookedly designed landing pads for components, errors of the placer.
If a thermoprofile, pasta, a player and sometimes a stencil are a headache of production, then the curves of the site are the immediate jamb of the designer.
The main idea is that the pads should be of the same geometry, so that when the paste spreads, the same surface tension forces occur. After all, the rise of a component arises precisely because of the various surface tension forces that act on the planar component. It also happens that the site is very large in size (connection to the earthen landfill, for example), there is no thermal barrier, and with a crooked thermal profile, the platform warms up more slowly than another conclusion. As a result, the paste will quickly melt on a non-polygonal pad, and the component can stand.

Not bad illustration and related article :


Why did the component get up? Because even if the cutout in the stencil for the paste was the same in both sites, then in the process of melting the paste at the output, where the large area open to the solder mask, not only the heating is slower, but the paste spreads over a larger area, which in general severely degrades solder joint performance (insufficient amount of solder) and may cause the component to rotate relative to the landing pad.
One Moscow Region, in the sense of Moscow, the office offers in this case a compromise option:


If there is a need for a large open area from the mask, so that the paste does not spread over the landfill, you can make a stopper bridge of a mask of small width, which will prevent the spreading of solder during melting.

All of the above mainly applies to light components with frame sizes less than 0603 (in inches), although there are exceptions.

What is the bottom line?
Thermal barriers are definitely needed. Apply them wherever necessary and not necessary - not worth it. This is especially true of microwave technology. In my daily design of simple microcontroller projects, in my opinion, they do more good than harm. But much depends on the technology of soldering and the subsequent need for maintainability of the product.
When soldering in a furnace, one should be aware of the possibility of lifting components due to improper design of the landing pads and the jambs of the technologist serving the line.

In any case, when tracing a printed circuit board, first of all, turn on the brain and analyze!

Thank you for reading!

All errors noted please report in a personal.