Issues When Using FR4 as PCB Material
It might not be surprising that much of the testing at the UNH-IOL is performed using custom software written in various languages/programs by our employees. However, what many don’t know is that we also design and make custom test hardware. We design platforms that streamline testing or offer unique capabilities, and we also design test jigs to connect our test equipment with devices under test. In this post, we’ll focus on one of our test jigs to provide background on compensating for loss introduced by FR4 material relevant to testing data communications products.
All of our test jigs are made of FR4 material, which can introduce considerable loss on technologies that operate near or above 1GHz. The importance of insertion loss can be seen even at relatively low speeds such as 1000BASE-T, which operates around 125MHz. For 1000BASE-T physical media attachment (PMA) testing (IEEE 802.3 Clause 40), we use a test jig that converts an Ethernet jack to SMA connectors, allowing us to test a 1000Base-T transmitter on an oscilloscope. In order to correlate our measurements of this signaling to those of implementers and other labs, we had to characterize the loss of the 1000Base-T PMA test jig. This loss was found by measuring the differential insertion loss using a network analyzer and despite the short length of trace used and good engineering practices held in the design, the loss was over 0.5dB. This amount of loss is more than enough to cause miscommunication between engineers at two different laboratories and slow any attempts to correct the underlying issues. In this example, the differential insertion loss was found between 300 KHz and 200 MHz and the loss we measured is now applied during the post-processing of measurements to eliminate the effects of the jig.
We have also begun experimenting with using Rogers (RO4350B) material in some of our test jigs as we need a better PCB material than FR4 to accommodate testing new high speed technologies. Rogers material offers a lower dielectric constant (3.66 versus FR4’s 4.5) and lower loss than FR4 which will allow us to continue to create high quality test jigs and boards here at the UNH-IOL.
Michael DeGaetano, Research and Development
Ned commented:
Using a test jig at that low cost with those results is brilliant. Great work!
JEFF commented:
Keep us updated as you progress thru Rogers ceramic type material to their Liquid Crystal Polymer type when you get to optical communications,
Phil commented:
interesting write up
Sam commented:
very interesting
William Ketel commented:
There is FR4, and then there is FR4. But if you need the material made for high speed signals, then probably you do need something that is different from the cheap stuff. Rogers is one company that is able to provide materials more suited to the high frequency operation required. But the purchase price is a bit more than the low quality, low performance material. Sometimes though, you do get what you pay for.
Joe commented:
interesting, great job
