NOTE: This post was significantly updated on 8-8-18.I have had nothing but trouble mounting the EMDC-13-2 smd parts on a suitable breakout board in order to create a directional coupler. I don't think it would be appropriate to create a whole new thread to describe the issues I ran up against, since the EMDC-13-2 is a rare part that many may never run across. However, as of this writing they are still available on ebay (
Tyco E Series RF Directional Coupler) and I thought it would be a community service to let others know the difficulties I experienced working with them. Because I originally asked about them in this thread, I thought it would be useful to document the problems I had here. If others think I should create a whole new thread, I will do so and move this post there.
Before I begin, let me state for the record that I do not blame the seller for the problems I experienced. They were sold in good faith and probably will solder to a board without issue if you use solder paste and a reflow oven. I don't have a reflow oven and tried to solder them by hand using a microscope and techniques documented in many places, including a very good tutorial by Dave Jones on the EEVblog.
Naively thinking I could use a microscope, a standard smd breakout board and a steady hand, I blithely dove in and soldered the 3 EMDC-13-2 parts I purchased to some smd breakout boards I had in my parts bin. Figure 1 shows the result.
Figure 1 -
The first sign of a problem occured when I tested the pads of the breakout board for continuity. To my surprise, all of the pads connected to the device displayed a short to all other pads. I tested the other breakout boards and this was true for all of them. Later, after ordering some additional EMDC-13-2s (see below), I tested each for continuity before soldering them and found that all 5 legs of the part showed a short to the other legs. So, the first lesson others might find useful is you can't use continuity testing on these parts to see if a solder job has problems.
After testing the parts mounted on the breakout boards I discovered none of them worked. I attempted to desolder the parts from the breakout boards in order to analyze the problem, but only the top part came off - See Figure 2.
Figure 2 -
Underneath is a white square of insulating material the remained connected to the breakout board. It is a bit difficult to see, but if you look closely, you will see a solder bridge between the top pads on the left side of the insulating material. When I looked at the other boards, their pads also had solder bridges, some at different locations. So, the second lesson is trying to solder these parts using a flux pen and conventional solder is a risky strategy.
The liquid solder flux may damage the insulating material, which creates these solder bridges, but that is a guess. It is also possible that my smd soldering techique is the problem, but I have had success in the past. Anyway, caveat emptor - trying to solder these parts using the standard technique didn't work for me and may not work for you.
To support the following narrative, I photographed an unmounted part (from a new batch - see below) sitting on its side - Figure 3
Figure 3 -
As is evident, the legs of the device are really stumps. I think it is clear that these devices are designed for solder paste and reflow oven mounting, not flux pen and soldering iron mounting.
Having failed with the standard approach, I decided to try another. I had some small double sided copper clad boards, so I decided to use them. I ordered 3 more EMDC-13-2 parts and when they arrived cut away the copper from each board to form a mounting surface to which I superglued the parts. I soldered some 22 AWG ferrite core winding wire to the legs of the part and to some pads I cut out of the board - Figure 4.
Figure 4 -
I then soldered RG-174 coax to the pads, put them in an enclosure and connected the RG-174 to BNC connectors on the enclosure - Figure 5
Figure 5 -
This worked, but only partially. One of the boards failed testing, so I ended up with 2, rather than 3 directional coupler boards.
While this post is about soldering problems with the EMDC-13-2 parts, not about whether I succeded in this enterprise, I thought I might as well document the testing results in case anyone is interested. Also, there are two issues that someone with directional coupler experience might address (see the results for tests 2 and 3 below)
I ran 3 tests. For the first, I connected a 1 V
p-p 10 MHz signal from my function generator using RG-58 coax through a Tee at channel 1 of a Rigol 1104Z scope and then to the input of the directional coupler. I measured the signal at the output, coupled and return ports on channels 2, 3 and 4 respectively of the Rigol through Tees terminated by 50 ohms. (The EMDC-13-2 breaks out the return port, which normally I would terminate with a 50 ohm terminator.) The results for the two couplers were:
| Coupler | In | Out | Coupled | Return |
| 1 | 989 mv | 924 mv | 264 mv | 26.4 mv |
| 2 | 1.012 V | 916 mv | 236 mv | 23.2 mv |
The
spec for the EMDC-13-2 indicates a main line insertion loss (for 5-50 MHz) of 1.1 dB typical and 1.5 dB maximum. Coupler 1 displayed a main line loss of ~ -0.5 dB, which is better than spec. Coupler 2 displayed a main line loss of ~ -0.86 dB, again better than spec. The spec indicates a coupling loss of 13 +/- 1 dB. Coupler 1 displayed a coupling loss of ~ -11.4 dB, again better than spec. Coupler 2 displayed a coupling loss of ~ -12.5 dB, which is in the spec range. The spec indicates a typical return loss of 17 dB and a minimum of 12 dB. Coupler 1 displayed a return loss of ~ -31.4 dB, which is significantly better than spec. Coupler 2 displayed a return loss of ~ -32.8 dB, which is also significantly better than spec.
For the second test, I reversed the connections to the input and output ports of the couplers, i.e., the coupler "output port" was connected to the input signal and the coupler "input port" was connected to the output channel on the scope.
(Added 8-8-18: While attempting to investigate the issue I originally indicated - the high coupled port value - I noticed I had left off the 50 ohm terminator from the Tee at channel 4 of the scope. I reattached it and the coupled value decreased to something more expected. I have modified the table to indicate the correct values. Also reconnecting the 50 ohm terminator to the Tee at channel 4 of the scope affected the results for test 3, which I have also updated.)The results were:
| Coupler | In | Out | Coupled | Return |
1 | 932 mv | 1.024 | 216 mv | 464 mv |
| 1 | 932 mv | 1.024 | 3.96 mv | ~ 0 mv |
2 | 932 mv | 996 mv | 236 mv | 468 mv |
| 2 | 908 mv | 1.044 V | 4.52 mv | ~ 0 mv |
The results of this test surprised me. The high value at the return port is expected. But I expected the coupled port to have an order of magnitude lower value. Perhaps someone with experience with directional couplers might suggest an answer. Does this indicate the directional couplers are not working properly?For the third test I returned the input signal to the coupler input port and reconnected the coupler output port to the scope output channel. I then disconnected the 50 ohm terminator from the coax running to the scope output channel (channel 2) to create an impedance mismatch and see how that affected the signal at the return port. The results were:
| Coupler | In | Out | Coupled | Return |
1 | 1.36 V | 1.96 V | 324 mv | 412 mv |
| 1 | 1.40 V | 1.96 V | 272 mv | ~ 0 mv |
2 | 1.41 V | 1.91 V | 312 mv | 470 mv |
| 2 | 1.41 V | 1.91 V | 267 mv | ~ 0mv |
The high value at the output port is expected due to reflection, but why is the input voltage increased? Does this mean the reflections are leaking through the directional coupler back into the input port? Does this suggest a problem with the directional coupler?.
(added 8-8-18:) Serendipitously, I received in the mail a mini-circuits ZFDC-20-4-S+, which I had ordered after building and testing the EMDC-13-2 based directional couplers. Consequently, I could run some tests on it and compare the results with those I obtained on the EMDC-13-2s. Note the ZFDC-20-4-S+ does not expose the return channel, so there are no values for this parameter.| Test | In | Out | Coupled |
| 1 | 988 mv | 924 mv | 128 mv |
| 2 | 984 mv | 1.008 V | 1.48 mv |
| 3 | 1.42 V | 1.94 V | 112 mv |
Results of running tests 1, 2 and 3 on a Minicircuits ZFDC-20-4-S+
There are two things to notice. First, the ZFDC-20-4-S+ spec stipulates a .33 dB mainline loss and 19.42 dB coupling loss at 9 MHz. These compare with measured values (for 10 MHz) of 0.21 mainline loss and ~ 18 dB coupling loss, which are in rough agreement. But the most important result is that for the third test, viz., when the 50 ohm terminator is removed from the Tee at the scope for the coupler output channel, the input channel voltage rises. Assuming the ZFDC-20-4-S+ is behaving correctly, this suggests the EMDC-13-2 based directional couplers are also behaving correctly.ConclusionsThe EMDC-13-2 seems designed to only work with solder paste and a reflow oven. It doesn't behave well when using a soldering iron and liquid flux to attach its legs to breakout board pads. Attempting to do so resulted in failure. Furthermore, you can't use continuity testing to see if your solder job worked. Each leg of the part shows a short to all other legs.
Using 22 AWG wire to connect the device legs to external pads on a copper clad PC board worked better, but at this point the results from testing are inconclusive as to whether this technique was successful.
Home
Help
Search
Login
Register
