[SOLVED] Differential matching

[jgraygoza]

Hello,

I have a situation in hand.
I need to test a differential device in a network analyzer.
I have used transformers and baluns,
but they are limited in bandwidth.
I did some research and some advice I received and
a suggestion was to use a differential amplifier for both input and output.

So, it would be something like this:

Network Analyzer Port 1 <- unbal 50 ohm -> differential amplifier <- bal 200 ohm -> DUT <- bal 200 ohm -> differential amplifier <- unbal 50 ohm -> Network Analyzer port 2.

I found some ICs from Analog devices for the amplifiers.
ADL5561, ADL5562, ADL5565.

I kind of have an idea on how to put them together,
but I've rarely worked with differential impedances.

Any help will be greatly appreciated

 

[BigBoss]

Modern 4-Ports VNAs can handle to measure differential impedances.For instance, consider an differential amplifier,each input/ouput is connected to one port of the VNAs inputs/ouputs using single ended cables and connectors.VNA calculate differential s-parameters from mixed mode s-parameters.
Wideband 1:1 or 1:4 transformers can also act baluns,think about them.

 

[Joel-Tang]

If you are limited to two ports then you could terminate the unused ports to 50 ohms and measure every combinations of ports and build the S4P file from the data.
eg, S11,s12,s13,s14,s21,s22,s23,s24, etc....

I think ADS has a tool set for handling these files to create the touchstone file for you but I've used this touchstone_spec11.pdf successfully in the past.

Building a single S4P file should allow you to understand to co-dependence between the ports quite well.

However, this would only be suitable for a linear device.

 

[jgraygoza]

I am testing a filter so maybe your option could work. And yes, I am limited to two ports.

 

[biff44]

I feel sorry for u, I had to go thru exactly that 1 year ago.

The easiest (and most accurate) way is to use a 4 port modern Network analyzer, and let the analyzer figure out the 2 port differential S parameters.

You can use baluns, but I have found them to have a lot of ripple (perhaps +/- 3 dB over .5 to 6 GHz for the best of them), and not to have a very good return loss

If you have a single ended to differential amp on one side, and a differential input to single ended amp on the other side, you could calibrate the network analyzer with a thru, and measure the gain and reverse isolation of the DUT. Of course, u will not have a clue what the input/output return losses of the DUT are though.

 

[jgraygoza]

Biff44, that's pretty much what I need. vswr measurement*I already took care of. The current setup where I use transformers provides all the passband parameters I need. It's the transmission part in higher frequency that is causing me issues. That's why I am trying to figure out what amplifiers I could use in the same configuration that you mention. Are the ones from analog devices good enough? As always, your help is very much appreciated.

 

[biff44]

the ADL5561 has a single ended input resistance of 307 ohms when set to a gain of 5.6 dB.

I would try to find something that has an input resistance of 50 ohms and a gain of 0 dB.

If I could not find a better unit than the ADL5561, I would probably put a 6 dB attenuator chip right near the ADL5561's input terminals, and hope for the best. Then your network analyzer will at least be seeing a load somewhat close to 50 ohms. A non-50 ohm load will cause ripple (from the network analyzer cable) that may be hard to calibrate out accurately.

 

[jgraygoza]

I thought about using the attenuator option as a last resort. But I guess I have no choice if I can't find anything else.
I was also thinking of using opamps, but they don't go up in frequency very much, do they?
Do you know of any other manufacturer that builds this kind of amplifiers?
As always, your help is very much appreciated.

 

[biff44]

http://www.euvis.com/products/ic/ds/LI370_DS_A2.pdf

 

[jgraygoza]

If I were to use this differential amplifier or any other in the single ended configuration,
I would only need to ground the INN or the OUTN connections and feed the signal in or out of the system from the INP or OUTP connections, correct?
Or is there something else to look into before using it in a single ended configuration?

Thank you very much. Your help is always very much appreciated.

 

[BigBoss]

If I were to use this differential amplifier or any other in the single ended configuration,
I would only need to ground the INN or the OUTN connections and feed the signal in or out of the system from the INP or OUTP connections, correct?
Or is there something else to look into before using it in a single ended configuration?

Thank you very much. Your help is always very much appreciated.

You can do that but you can not profit the advantage of differential configuration

 

[Joel-Tang]

Of course, u will not have a clue what the input/output return losses of the DUT are though.

Isn`t this a pretty big deal?
The matching is a big part of the transmission measurement. Why not measure all port combinations as I mentioned earlier? It provides you the accuracy of a network analyzer which is important when measuring filters.

 

[jgraygoza]

I think I have something to work with.
So I will be using the AD8351 to go from single ended to differential.
After that I will attach an attenuator that converts 150 ohm to 200 ohm,
which I think it shouldn't be a big deal except for the chip resistors limitation is bandwidth if there's any.

As for the output, I have a complicated situation.
How to go back from 200 ohm differential to 50 ohm single ended?

I have some ideas, but I feel they don't have enough weight.

 

[biff44]

there IS such a thing as a matching resistive attenuator. It will do 200 ohms at one end and 50 ohms at the other port, but have non-reflective loss. A basic T attenuator but the values are not symmetrical. U might try that.

 

[jgraygoza]

biff44... I did try that too. It worked ok.
I got some differential amplifiers to make the matching possible.
I was able to get the results I needed.
Thank you everyone for your help.