Technical FAQs

Thanks for the question and the interest in our T3 mixers. In general we recommend against filtering the LO for two reasons:

1. Placing a filter between the T3 and the LO amplifier will create a non-50 ohm environment for the mixer, which can cause increased spur levels and decrease the predictability of the spurs.
2. Filtering the LO will eliminate the desirable odd harmonics (third, fifth, etc) and reduce the linearity of the T3, which will reduce the dynamic range of the system.

The one caveat to this is if you are concerned about the noise contribution from your LO source degrading your system noise figure. This can occur when you have an LO source that is not very clean. In general the mixer will reject LO noise according in a down-conversion to its LO-IF isolation and the LO-RF common mode conversion loss (typically around the LO-RF isolation plus the conversion loss). This means that if your LO noise is more than 30 dB below your signal path noise you probably don’t have to worry about it. Otherwise you can mitigate the impact of the LO noise by using a high pass filter on the LO up to the IF frequency that you care about. This will allow you to keep the odd harmonics that are beneficial, although it may still impact your spur performance.

Good luck!

Marki Microwave offers a comprehensive line of MMIC surface mount band pass filters from 1 GHz up to mmWave. Marki offers two types of bandpass filters: CAD-optimized microstrip filters built on a low-loss substrate offered in surface mount and connectorized packages, and GaAs MMIC filters with tight fabrication tolerances offered as surface mount QFN and bare die. Both offer low loss and extremely broad rejection bands.

In addition, we have low pass and high pass filter designs available in surface mount packages.

We actually do specify VSWR on all of our recommended, modern mixer families (MM1 and T3). The concept of VSWR is somewhat meaningless for mixers. A mixer is by definition a dynamic impedance and is not be sufficiently described through VSWR measurements—especially since VSWR varies depending on applied signal levels. One of the most common complaints occurs when a mixer is reactively terminated at one or more of its ports with a reflective filter and poor mixer “VSWR” is blamed for the observed amplitude fluctuations. In truth, the problem has nothing to do with mixer VSWR but instead that the mixer has not been terminated with a wideband 50W load. In this circumstance, the amplitude fluctuations are actually being caused by unwanted intermodulation products lying in the stopband of the filter which have been allowed to back-reflect back into mixer. The re-mixing of this back-reflected intermodulation product leads to some very unpredictable mixer behavior and ultimately to amplitude ripple in the conversion loss. It should always be remembered that the performance of all Marki Microwave mixers are specified for wideband 50 W  systems. We always recommend that a mixer be properly terminated with appropriate wideband 50 W attenuators so as to avoid problems arising from reactively terminated mixer ports.

Not as a standard catalog product. All products are offered as commercial off-the-shelf products, with no environmental guarantees. Marki does have the capability to support military programs with high reliability packaging and environmental qualification in special circumstances or for large programs, contact [email protected] for more info.

There shouldn’t be much of a bias current. The diodes in the limiter are reverse biased, so all that is needed is the reverse bias leakage current.

Not necessarily. In general, datasheets for certain mixers are not available on the website either because the product is obsolete or because the mixer is a custom, non-catalog design. The only way to know for sure is to contact the factory at (408) 778-4200 or [email protected]. If the mixer is obsolete, a suitable replacement model will be suggested. In the majority of cases, Marki Microwave will continue to support designed-in products.

Configuration A vs. B does not reflect a different option that you need to order when you specify the part number, it is two different ways to use the same mixer. The reason that you would use it is to get the very best spurious performance out of the mixer. If you are just looking for low conversion loss and good bandwidth, then just use Configuration A and don’t worry about it. If you are obsessed with getting the lowest spurs, however, then it may be worth looking into Configuration B.

Many Marki mixers (currently all Microlithic and MMIC non-IQ mixers) are passive, double-balanced, diode ring based mixers. That means that they can be used as up or downconverters. It also means that the ports are totally interchangeable: the LO can be driven into any port that supports that frequency, and the conversion will take place. Configuration A and B just reflect the option to drive the LO into either of the high frequency ports.

It is useful to think of a double balanced mixer as having a filter at each port; the LO and RF ports have bandpass filters (from 3 to 20 GHz or 10 to 40 GHz, for example), while the IF port has a low pass filter (DC to 5 GHz or DC to 15 GHz in this example). Since the ports are interchangeable, and the LO and RF ports have the same frequency range, this means that the LO and RF ports can be swapped. Configuration A vs. Configuration B refer to these two ways to operate the mixer. Lets look at the typical double balanced mixer structure:

As you can see, the LO side and the RF side are not identical. One side uses a balun (in this case the LO), and the other case uses a magic-tee, where the IF comes from. For the mixer user, this is not set in stone. You can swap the LO and the RF, and the mixer will still work, but with different performance.

But how different? When would I use one vs. the other? Lets look at the MM1-0320H datasheet to decide. The first thing you notice is that in the A configuration, the conversion loss is better, but the B configuration requires less LO drive. This is because the one balun is less lossy than the other balun, so it can be used to reduce the conversion loss or lower the LO drive requirements, depending on which configuration is used. You will also see that the RF-IF isolation becomes the LO-IF isolation if the ports are switched, while the LO-RF isolation is the same either way.

On nonlinear specs is where the real benefit can be had, but it depends on the specific frequency and power plan of your conversion. The B configuration has slightly better IP3, but only at specific frequencies. Look at the 2IF x 1 LO spurious suppression:

If you are operating between 12 and 14 GHz, then configuration A is better. If you are operating between 14 and 16 GHz, then configuration B is better. The best way to determine which is best is to plug each mixer in and try it. Alternatively you can use the Marki PDK to compare the mixer spur levels, although we do not guarantee the accuracy of spur levels predicted at this time. Since these are physics-based models we expect them to be very good.

So in conclusion Configuration A vs. B simply gives system designers an extra degree of freedom when it comes to getting the absolution best nonlinear performance out of the mixer.

Unfortunately, it sounds like the mixer may have been damaged by the application of DC current. Marki mixers are remarkably robust, with many of our mixers lasting in harsh conditions in the field for decades without failure or degradation. There are very few known degradation mechanisms for our mixers, and the two most common are thermal damage and DC current application.

To test if your mixer is still good perform the following tests:

– Take a handheld multimeter, and set it to the ‘diode’ setting. The diode setting will measure how much voltage is required to drive 10 mA of current from the positive to the negative lead. For a standard silicon Schottky diode this value would be 0.7 Volts. For most Marki low barrier diodes it will be around 200-300 millivolts. Measure from the center pin of the I to the grounded case then from the center pin of the Q port to ground. Next reverse the leads, with the positive lead on the ground and the negative lead on the I and Q center pins. Now compare the measurements. If they are roughly the same, then the diodes are ‘alive’, and if one is significantly different, the diodes are likely ‘dead’. If one voltage is higher than the others it indicates that one of the diodes in the ring is probably burned out.

– If possible, measure the conversion loss of each side of the IQ mixer. The conversion loss of each side should be within the datasheet spec for conversion loss. If one side creates a significantly lower output signal than the other, then it is likely damaged.

After you have these results you can send an email including your findings to [email protected]. Unfortunately when diodes are burned out due to the application of DC current, this is considered field damage and the parts are not returnable under the warranty, so please be extremely careful when using this technique for IQ compensation, and consider one of these alternative methods. Application of the means to apply DC voltage, even, can expose the unit to potential power surges and ESD discharge that it is normally protected from by the shielded connectors.