1

I have a little habit of socketing all DIP chips used in my circuit (so at least if the design failed I can recycle the chips easily). Now since I am trying my hands at multimeter DIY'ing (with a all through-hole construction hence all the DIP chips) I am seriously considering side effects of placing chips in sockets.

The chipset include the ICL7135 4.5-digit ADC. On the digital side there are CD4511B/ULN2003 LED display driver chipset and a pair of GAL16V8s implementing the autoranging logic (without those there would be a few more 4000-series CMOS chips.) On the analog side there are the AD736 RMS-to-DC converter, REF03 voltage reference and a few op amps (TLC227x and OPx177 primarily.) There is also a power supply section with two MC34063s.

I think socketing digital chips, MC34063s as well as the primary ADC won't hurt (especially the GALs which required programming) but what about the analog chips? Will socketing those chips bite me? (If I can socket the REF03 maybe I can individual test and label them before plugging them into the PCB)

Maxthon Chan
  • 2,873
  • 1
  • 16
  • 34
  • Logic dictates that a socket adds nothing but a few micro ohms of resistance, the same as an extra millimeter of copper traces on a board, but I'll let someone more in the know answer properly. – Passerby Nov 12 '15 at 11:42
  • @Passerby My fear here is that sockets may introduce leakage which can be a problem for a 4.5-digit multimeter. – Maxthon Chan Nov 12 '15 at 11:45
  • Define "Leakage", or what you think may be happening. – Passerby Nov 12 '15 at 11:47
  • @Passerby I think this is about the fact that PCB and socket resistance isn't infinite, just very large, and OP is rightly worried that trying to get 4.5 accurate digits requires care. – pjc50 Nov 12 '15 at 11:57
  • If the chip is susceptible to damage or abuse (blown interface chips) the socket makes repair much easier. If it's very expensive, sockets let you pay large batch prices for your boards, and only pay for chips when you need to... –  Nov 12 '15 at 12:58
  • @BrianDrummond The few ADI chips onboard are not cheap... REF03 (or the pin-compatible ADR03 on an adapter board) costs somewhere from US$4-6 a pop and AD736 costs $8. Also I have two GALs that require programming. – Maxthon Chan Nov 12 '15 at 13:05

1 Answers1

4

So what are some things that adding a socket for a TTH DIP part actually does, circuit wise?

  • It adds a bit of wire length to your traces (all equally)
  • It adds a bit of resistance (mΩ at most, unless corroded), which depending on the socket quality changes more or less over time
  • It adds a tiny bit of inductance and capacitance to your traces
  • It adds more space where dirt can accumulate and form GΩish resistance paths

Mechanically it does

  • add the possibility of your ICs falling out due to extreme vibration or temperature cycling or PCB flexing
  • reduce the possibility of damaging the IC due to PCB flexing

So what you have to do is to think if these are a problem for your design, which we can not really do, since we have no idea about your design. But since you are using TTH parts and with a bit of guessing, what we can tell already is:

  • Your bandwidth requirements are far below the case where length and inductance makes a difference
  • The tiny changes in resistance will likely be much lower than the tolerances of the parts you are using already

The mechanical parts are up to your liking, and then whats only left is the dirt that can accumulate, which is also preventable in other ways.

So all in all, we can say that its pretty safe. Actually if you look at quite some older good multimeter teardowns on youtube et. al. you will see that there are quite some ICs that are socket by people who know how to design even 6-7 digit multimeters.

TL;DR

You should be fine with socketing.

PlasmaHH
  • 6,528
  • 5
  • 38
  • 49
  • Much more in depth than the post I was half way through writing - +1! – Nick Johnson Nov 12 '15 at 11:54
  • @NickJohnson: Still worth checking if something important was left out that you would have written – PlasmaHH Nov 12 '15 at 12:00
  • It's a dual-slope multimeter so anything that should have a bit of "frequency" is the integration capacitor (a few hundreds of kHz at most for ICL7135) and the rest have a designed maximum frequency response of 100Hz. I am worried that the leakage current may throw my measurement out of the window. My reference voltage is 1V generated by dividing the 2.5V from REF03 using a ten-turn pot. – Maxthon Chan Nov 12 '15 at 12:13
  • @MaxthonChan: The sockets themselves do not produce any leakage currents (or at least that you could measure in any way), it is the dirt that accumulates between the legs. You have the same problem with every part, the sockets just give more room for that dirt. Solder flux residues will have orders of magnitude higher leakage current possibilities, so clean everything and you are fine. It is only 4 1/2 digits, not 14 1/2. – PlasmaHH Nov 12 '15 at 12:23
  • I like the sockets for their flexibility: I am still undecided which op amp to use so if I can socket the op amp I can try different chips by swapping them in and out of the board. LM324, TL084, TLC2274, LT1014, AD713 and more quad op amp chips all have the same DIP-14 pinout. With socketed chips each part can be replaced with a compatible one if so desired. (not just op amps, ADR03 in place of REF03 too) – Maxthon Chan Nov 12 '15 at 12:26
  • @MaxthonChan: To add some anecdotic to this: whenever I come across soldered in parts that I have to repair, I replace them with sockets. Not a single negative result with this so far. – PlasmaHH Nov 12 '15 at 12:28
  • @PlasmaHH Nope, you covered it all quite comprehensively. – Nick Johnson Nov 12 '15 at 15:04