I am currently working on a project where we are trying to limit costs as much as possible and one of the results of this is trying to avoid using IMS(insulated metal substrate, in this case 0.125" aluminum instead of the typical FR4 in PCBs) in the power board design. We currently use an IMS in a large portion of our designs, however this particular job is more competitive and we are hoping to avoid the cost and slower procurement of the IMS.

The basic setup of the power board is a charge FET(s) and a discharge FET(s) as shown in the basic schematic below. There is a fair bit of control circuitry for the FETs that is not shown, as it is both complex and proprietary, and shouldn't be relevant for the scope of this question.


simulate this circuit – Schematic created using CircuitLab

For this project the current requirements are 120A charge for 3 min, followed by 4A discharge for 90 min, and repeated continuously for the life of the battery. The FETs we are planning on using are IPT015N.

The current design idea I am working with is a stamped copper bus bar with rounded square protrusions that fit in a slot in the FR4 allowing the drain tabs of the FETs to be soldered directly to the bar. This has the benefit of allowing us to put both control and power circuitry on the same board, as well as being much easier to have manufactured and prototyped.

My questions:

  1. Is this bus bar protrusion a reasonable approach to draw heat from the FETs in large-scale production? While the testing indicates it is a viable option, it seems to be rather unique approach.

  2. Are there any other alternatives to IMS that are cheaper and could be effective in this scenario?

  • \$\begingroup\$ Not having the faintest idea what "IMS" stands for in this context, I looked it up on Wikipedia and got anything from "Irish Marching Society" to "Irritable Male Syndrome", but I finally concluded that the closest fit was "Insulated Metal Substrate". We really shouldn't have to guess about these esoteric niche acronyms. If you define it in your question, you'll be much more likely to get useful answers. \$\endgroup\$
    – Dave Tweed
    May 23 '17 at 20:26
  • \$\begingroup\$ @DaveTweed Fixed(I hope). I was under the impression that metal substrate in high power designs was a fairly common concept, although it is apparently not. \$\endgroup\$
    – Redja
    May 23 '17 at 20:38
  • \$\begingroup\$ The concept is common, the acronym is not. When in doubt, define it. Remember, you're asking a bunch of strangers for a favor. You want to make it as easy for them as possible. \$\endgroup\$
    – Dave Tweed
    May 23 '17 at 20:40
  • 2
    \$\begingroup\$ Understood. Not realizing people have no clue what I'm talking about happens more times than I'd like to admit. \$\endgroup\$
    – Redja
    May 23 '17 at 20:47
  • \$\begingroup\$ Depending on your field ... IMS is well known in power electronics \$\endgroup\$
    – JonRB
    Oct 9 '19 at 7:36

You seem to be wanting to replace your IMS with a UMS (UNinsulated ...).

The IMS has already solved a bunch of related mechanical issues for you with regard to flexure, vibration and thermal expansion — issues that your ad-hoc solution will have to address separately. You'll need to have a good mechanical engineer evaluate your proposed solution for reliability issues.

  • \$\begingroup\$ Thanks for the input. Our mechanical engineers will definitely be involved before this goes to production. We have already had some discussions as to the thermal expansion aspect although you do bring up a good point with vibration. I will make sure to investigate further. \$\endgroup\$
    – Redja
    May 23 '17 at 21:13
  • 1
    \$\begingroup\$ You really haven't given a lot of details about how the PCB, the busbar and the FETs will be mechanically anchored to each other, but I'm just imagining a lot of stress on the FET leads and their solder joints, leading to fatigue and failure. \$\endgroup\$
    – Dave Tweed
    May 23 '17 at 21:52

The RF fraternity sometimes do something similar using a copper "Coin" built into an otherwise standard PCB, it is the sort of thing that is usually only interesting if you are making more then a few.

The other obvious thought is to increase the number of parallel fets in the charge leg, additional fets might be cheaper then complex boards....

  • \$\begingroup\$ One of the first tests was actually based on the coin idea, which gave rise to the protrusion idea as less parts = less cost (usually). The main issue with the multiple parallel fets is the amount of current that needs to flow through the board. 120A needs quite a large trace..... \$\endgroup\$
    – Redja
    May 23 '17 at 21:18
  • \$\begingroup\$ Some pcb houses have the ability to embed a chunky wire in the board for heavy current, I think wurth offers this. My other thought is to mount a solderable bus bar alongside the drain tabs and use multiple fets, tin plated copper with a pin every 20mm or so is an off the shelf part. \$\endgroup\$
    – Dan Mills
    May 23 '17 at 21:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.