How to correctly solder RG316 cable to D-Sub solder cup ?

[electronic_guy]

Hi,

I need to solder 4 coaxial cables of RG316 type to D-Sub connector(DA15P064TXLF) with solder cups. The frequency of operation is under 1 GHz. How to accomplish this without any reflections due to mismatch at the solder joints ? Just soldering the centre conductor and shield to the solder cups (pair of pins, making one pin the signal and other ground), it seems like will not be properly impedance matched. The application would be to transmit IQ signals in base band frequencies such as 170 MHz or so.

I have attached a photo of the D-sub connector.

[Roehrenonkel]

Hi,
 
consider "combination"-Sub-Ds (3W3 or 5W5).

Good luck

[Phil1977]

Let´s assume you need a bandwidth of 300 MHz because it corresponds to a vacuum wavelength of 1m. Even with dielectrica it won't be less than 50cm.

As a rough rule of thumb non-coaxial distances that are less than 10% of the wavelength do not not distort the signal. That would mean to keep the distance of splitting up the coax to recombining it smaller than 5cm.

This rule of thumb may be valid for signal transmission. If you have higher precision requirements and need e.g. to prevent even small reflections you need coaxial plugs like the mentioned 5W5 or 9W4.

[electronic_guy]

Actually I cannot use 5W5 or 9W4 connectors. The D-Sub female connector is already fixed. This is on an Anritsu MG3700 Vector signal generator. The only way to get the IQ signals out is through a Female D-Sub connector 15 pin. So I need to solder a Male D-Sub to connect to that.

If I just solder the centre conductor and shield for ground, would that totally mismatch ? Aren't these D-Subs made to handle such frequencies ?

[Phil1977]

I just had a glimpse into the manual of this device: The D-Sub has 4 Signal pins and everything else is GND. So just use pins next to each other for Signal and shield. Try to keep the "separated" length of the coax short, but not so short that soldering gets a pain and too much temperature and stress are induced. Do the connection of symmetrical signals symmetrical in geometry.

Of course there is some impedance mismatch. But even the best test equipment can not use APC-7 everywhere - and I do not know the class of instrument you have but I assume it has some compensation and/or self test function so that you can calibrate it to the connected cables and test their function.

In summary: There is no special standard to solder coax to D-sub I know of - but it´s always good to keep symmetrical things symmetric.

[radiolistener]

Actually I cannot use 5W5 or 9W4 connectors. The D-Sub female connector is already fixed. This is on an Anritsu MG3700 Vector signal generator. The only way to get the IQ signals out is through a Female D-Sub connector 15 pin. So I need to solder a Male D-Sub to connect to that.

It uses differential output, so I think it needs to use balun transformer to convert differential output to unbalanced coax cable...

So, you're needs to solder balun transformer wires to D-Sub connector and then solder coax cable to transformer output. And all wires needs to be short enough to not add impedance mismatch.

If you have broken Anritsu Dsub to BNC IQ converter, you can disassemble it and see how it works. I bet it uses transformers, because it's the cheapest solution, so you can take transformers from broken converter and use it in your converter.

For low frequency it may need to use active circuit with operational amplifier to convert differential line to unbalanced coax, because transformers don't works good on low frequency.

[David Hess]

D-sub connectors were never intended for RF, but are commonly used to 100s of MHz anyway, like with the VGA standard.

Besides surrounding each signal with ground connections, I have seen the coax shield split into two so that it can attach to two adjacent pins on either side of the signal pin, which should halve the impedance through the connector helping considerably.