Post Go back to editing

HMC292ALC3B conversion loss problem

Problems with power transfer are occurring with the HMC292ALC3B. The HMC292 has been mounted on a PCB not identical to the evaluation board using a CPW design and SMA launch which has been tested previously, as shown here:

The setup involves the HMC292 used as a down-converter with an LO of 24GHz at 16dBm, and an IF of 2GHz at -1dBm. The output of the RF port is:

M1 and M2 are the mixed signals, M3 is LO leakage, and the other components are mixing of harmonics of the LO/IF. The above figure is not normalized for cable loss between DUT and FieldFox which is 5.5dB at 24GHz. This is a conversion loss of 30.5dB, and an LO/RF isolation of 55.5dB, which are both inconsistent with the data sheet

Data points at 19, 21, and 25GHz for LO/RF, LO/IF, and RF/IF isolation were measured by exciting the LO port with 13dBm and the RF port with 0dBm while then looking at the other port with a spectrum analyzer (3rd port 50ohm terminated). The results are:

The frequency dependency of the data sheet is not consistent with the measured data points. Furthermore, the return losses were found by measuring the S11 of each port while the other terminals were 50ohm terminated. Cable loss is calibrated out. The following 3 figures are all inconsistent with the data sheet. LO return loss:

The IF return loss:

RF return loss:

HMC292ALC3B was mounted and then was removed and replaced again, so the only thing I can think of is that the heat damaged it. 


Any recommendations would be greatly appreciated!

  • Hi,

    We are looking into this. Please give us a day or two.


  • Hi,

    A couple of things to note regarding this:

    1)The HMC292ALC3B has not been released or sampled yet so it seems like you might be using the HMC292LC3B part.

    2) From your description and the spectrum analyzer plot it seems like you're doing up-conversion and not down-conversion. 

    3) The board image shown does not look like its an ADI/Hittite eval board for this part so its difficult for me to directly debug your issue. My suggestion is to use a similar board layout to the one on the product page if you need to design your own board or order one from our website.. Gerber files can be found here  :… 

    Look for the Eval-HMC292LC3B and not the Eval-HMC292LM3C.

    Our board has 2 rows of vias for each trace, vias on the ground paddle, and vias around the leads as well. We use CPWG traces and our dielectric substrate is Arlon 25FR. The losses at these frequencies are very high and difficult to calibrate out without proper board layout.

    4) These parts are GaAs MMIC parts with ESD HMB of class 0. Therefore it can be easily damaged, so adequate ESD protection is need while handling it.

    5) A comment regarding the return loss measurements. For RF and IF return loss the LO needs to be applied at +13dBm or the mixer is not 'turned on'. The LO frequency needs to be between 16-30GHz. Otherwise the ports will look like an open or short. For the LO return loss measurement the power level needs to be at least 13dBm so that the internal diodes have enough power to turn on. Most VNAs cannot source that much power so you will need an amplifier and use a coupler to measure the returned signals. There are a few explanations on the Keysight website for this.

    6) For assembly of these parts please follow the directions on the attached app note.

    Hope this helps and let us know on any progress


  • Thank you for the assistance. The IC was replaced with a new one and it now works!

  • Hi,

    I have this mixer built into a design and appear to be tracking multiple failures on multiple PCBs due to this mixer. There appears to be a high percentage that have very high conversion loss.

    I want to correct one point above - the data sheet for both the HMC292LC3B and HMC292ALC3B state the ESD rating as 1000V Class 1C (although I did see quality/reliability test data that appeared to show it only tested to 250V Class 1A from memory) - so what is correct?

    I have an evaluation board which has a conversion loss of ~11dB which seems reasonable given data sheet specifications and PCB losses.

    I just wanted to alert the community to the fact that the high conversion loss is not isolated to a single user.