Digital SMD (surface-mount device) mod - upgrade or downgrade?

[Nada]

In another thread there was some side discussion around the use of surface mounted components in digital design. I found the discussion really interesting but it was way off topic so here's a dedicated thread.

 

I find the idea that SMD's are bad and should be replaced pretty terrifying as my soldering skills are rubbish.

 

So does anyone know of trials of SMD vs macro through-hole-components in digital boards or have DIY experience to share please?

[hochopeper]

The issue is often very miss-understood, because the side effects of doing it wrong are sometimes subtle. Another contributor to the miss-understanding is that the concepts are sometimes abstract and described with generalisations that shouldn't necessarily be taken too literally. Just because you can change something by adding some through hole components that you had sitting around and it still works, doesn't mean that the circuit is operating optimally.

 

For high speed circuits, SMD is the only way to go, though there are many examples of people doing SMD designs incorrectly too. Many ebay DIY DACs are a great example of how not to design digital circuits and are a personal pet hate of mine.

 

High frequency/high speed is defined by the rise time, formula on wikipedia shows that time for a signal to get from  - http://en.wikipedia.org/wiki/Rise_time - BW = 0.34/tr . For rise time specifications, consult the transmitting components datasheet and also understand the circuit, terminating resistors will control this. This is why a scope image of a fast rising signal is not necessarily what you want, though it may be good, if the circuit is well designed.

 

It is important to understand that the signal integrity of the each digital signal is important as well as minimising the digital signal's impact on the surrounding electronics (typically an audio circuit will have a mix of digital and analogue parts).

 

As an example, forcing high frequency signals (those whose rise time is fast, not just their square wave switching frequency) to traverse a larger loop than necessary creates antennas that will transmit and pick up magnetic and electrical noise. It is most important to consider also the return current. On a circuit board, given a contiguous ground plane the return current will tend to be mostly directly below the signal trace. Sending the signal via an extra loop in a through hole component is unnecessary and provides no benefit. The same issue can be caused in poorly designed SMD electronics, where the designer thinks that just filling spare space with copper connected to ground will be 'correct'. If the return current through the ground plane that is wanting to flow directly beneath the signal trace, is interrupted by a signal that is routed in the ground plane, then the return current for the signal must go the long way around to complete it's circuit. This creates effectively a dipole antenna. 2 vias could have been added to reduce this problem significantly.

 

 

I have thought that since this stuff is miss-understood  and sometimes hard to understand via posts on a forum, it may be worth having a small gathering at some point for technical discussions like this. I wonder if there would be some interest? I have a few books that discuss the topics, though I am still learning. I would consider hosting something like that once I have a few more DIY bits and bobs to a point of operating so that I could get out the circuit boards and some designs that I have, look at the details of the circuit board layout, component sizes in reality and discuss them. Then follow up by having a listen!

 

EDIT: I don't think I was involved in the other thread that you mention, could you point me to that thread so that I can reply with some context?

Edited May 26, 2013 by hochopeper

[Guest fordgtlover]

While I have read some articles on the matter, and I am aware of the dogma, I have no real knowledge of the issues in play. So, I watch this thread with interest.

 

Nada, thanks for starting the thread.

[3G_]

interesting thread. On a side note I by far prefer SMD than thru hole as far as neatness and ease of assembly.

[Addicted to music]

I have been exposed to SMD since I came out of high school. While I was doing COT electronics the firm I was with moved from through the hole to LSI and SMD. The benefits of SMD are reduction in size, lower power consumption, enable high speed possibilities that cannot be done successfully with through the hole component. reducing size was one of the biggest disadvantage to physically handle the pcb as connectors got smaller the care of removing connectors was a challenge as they can be easily physically removed. I remember bringing a pcb that was no bigger than an iPhone to show off in class at COT where most were Telecom techs and they couldn't believe the sizes of those LSI chips, it was the 1st time they have seen technology at this size. I remember at that time the size of a Commordore 64 or the 1st 286. I also remember the 2nd CD players from Yamaha were the 1st to have used LSI.

The benefits of SMD is reduction in parasitic inductance and capacitance, therefore enabling high speed capabilities, if it wasn't for SMD and LSI we will not have the items such as iPads or iPhone.

In terms of quality and use of SMD, it is a lot harder to tell what quality of passive component used aspecially when you have Asian copy cats reducing the cost even further. And from a DIYer or modifier, its an absolute nightmare to tell whether the manufacturer used a Sanyo Oscon cap, or something that look like it! Some of the identification on SMDs are difficult to read. SMD are machine mounted where solder is applied within the specified time frame before the integrity of the device is effected, when they have to be replaced, care in applied heat is a huge consideration, that's why in most cases the pcb ends up in land fill.

In a clock circuit, I would never attempt to replaced passive SMD with through the hole components as this this is where parasitic inductance and capacitance can compromised the accuracy of that oscillator. Nor will I replace SMDs with through the hole in a PC application. So in a digital domain, to me there is no benefit in replacing SMD with through the hole. Not to mentioned the fact that certain SMD components are strategically placed on that real estate for optimisation. So replacing it with a through the hole with wires hanging out of the pcb and using components that are 5-10% tolerance can servely effect the optimal operation of that circuit.

In an analog circuit, huge benefits and rewards if you replace audio decoupling caps with MKP. MKP can never be the size of SMD with the same value, and of course this is where the benefits out weigh the effective parasitic electrical nastie.

SMD are here to stay whether we like it or not. If you walk into Jaycar or Altronics, some of the passive components that are through the hole, are now being replaced by SMD on the shelf. There tolerances and purity are second to none, they are getting better in power handling as well and in some cases they exceed all expectations. A typical example is the hardware that I support, A well known manufacturer that hasn't gone full SMD have a huge failure rate, maybe that's because the electronics are not Twainese but made in Mexico, where a different manufacturer that have fully intergrated with SMD circuits have a reduction in failure rate that is thousands times more reliable!

[fetischizm]

nice posts hochopepper and pchan, yes the point where passives stop being resistors/caps etc. and start to become inductive is within the audio band as well. when you are placing a part, its useful to think of everything as a lumped element. inductance is one of the largest enemies of proper digital layout and SMD makes it much easier to minimise, as well as keeping loop area as small as possible. as things progress, I think we'll find that SMD thin film caps of larger capacity become available and when they do, they will compete with the very best film/foil caps in smaller sizes as far as linearity. I find the idea of random replacement of SMD parts with large PTH parts is in most cases clueless, the only motivation is normally based on false preconceptions that manufacturers and designers are somehow all out to save money as the only motivation, when in truth, high quality SMD is superior in just about every way

[Addicted to music]

Here's a pic of a video processing pcb, we rarely have any issues with these and the LSI chips are employed are similar to our I/O pcb but with driver chips on them. Note the LSI chips that the legs do not stick out making soldering almost impossible to do. All our products that are 32A, do not have any AC mains driven or AC supplied. All heavy duty motors are DC driven there are no AC components apart from the heater lamps that are supplied by a triac.

Edited July 4, 2013 by pchan

[fetischizm]

resistor network termination city!!

[Nada]

Looking at those video boards makes me want to stay an ignorant consumer.

[fetischizm]

they dont solder them by hand reflow FTW