At the time of writing (17-Sep-2021) there is no end in sight to the current global semiconductor shortage which is affecting multiple parts across multiple industries. One of the affected parts is the Si5351A PLL Synthesizer IC used extensively in QRP Labs kits and by other manufacturers - the purchase of parts of SiLabs business (including clock generators) by SkyWorks may also be a factor. Either way, lead times are quoted at 12 months plus - in other words, nobody knows.
A substitute IC known as MS5351M is now available. There is no indication that in any way this is a clone or pirated or otherwise illegal device. Several correspondents recommended this IC to me and I know some vendors have already started using this IC in their products. However as yet, whilst it clearly works, I had not seen any published performance measurements. The datasheet of this IC is here (English version, from the manufacturer's page http://www.relmon.com/en/
The following results are the tests which I performed on several MS5351M samples to establish that the IC meets the required standards to continue to deliver the high performance QRP Labs products.
The results were favourable and in fact in my opinion, the MS5351M has slightly improved performance compared to Si5351A.
At the time of writing (17-Sep-2021) we still have substantial stock of Si5351A but to be able to continue production of various products we will soon start to use MS5351M. For all intents and purposes in QRP Labs kits and documentation, Si5351A and MS5351M will be considered interchangeable and equivalent.
I did not test the current consumption of the MS5351M (omission on my part).
Ken WA4MNT reported here that the current consumption of the MS5351M is about 8mA higher than the Si5351A. This increase may mean harder biasing of the internal circuits of the MS5351M which may account for the slightly higher performance (see below).
Appearance and pinout
Si5351A comes in a 10-pin MSOP package measuring 3 x 3mm. MS5351M has uses the same MSOP package and has an identical pinout. No changes are needed, when installing an MS5351M on a PCB designed for Si5351A.
Output amplitude vs Frequency
I measured output amplitude vs frequency (one sample of each of Si5351A and MS5351M) and they are very close to each other. In the charts below, measurements of output amplitude were made using an Advantest R3361C spectrum analyzer. In the first chart, the spectrum analyzer's 50-ohm input is connected directly to the Si5351A/MS5351M. In the second chart, via a 1K series resistor. The IC registers are in all cases configured for maximum output drive level.
The 50-ohms loaded output chart shows virtually identical output level of the Si5351A and MS5351M.
When the 1K series resistor is used, MS5351M shows a higher output amplitude above 100MHz, increasing to around a 2dB advantage at 200MHz.
Maximum operating frequency
The Si5351A datasheet specification for maximum output frequency is 200MHz. However, in practice the devices are found by me, and several others, to provide a stable output up to around 290MHz (though having increasingly less spectral purity as frequency increases). Three MS5351M samples were tested, and the maximum stable output frequencies were 313, 297 and 311 MHz. The MS5351M therefore appears to have approximately a 15MHz higher maximum output frequency than the SI5351A.
Additionally at high-end HF frequencies the output spectrum of an MS5351M is closer to a squarewave theoretical spectrum than the Si5351A, which leads me to conclude that perhaps the MS5351M is a slightly faster chip, maybe they use a smaller nm transistor size.
Crystal reference oscillator frequency
In our experience, QRP Labs kits using a 27MHz crystal as the reference oscillator for the Si5351A IC, connected directly to pins 2 and 3 without additional external capacitance, and with the Si5351A registers configured for default levels of load capacitance, the actual oscillation frequency of the 27MHz crystal is typically around 27.004 MHz +/- 1 kHz. In all QRP Labs kits the measured oscillation value is compensated for in firmware by the Si5351A register calculation algorithms.
Two samples of MS5351M tested showed reference frequencies (measured by calibration using QRP Labs QLG1 GPS) oscillation frequencies of 27,008,092 and 27,006,632. All other parameters were identical (identical PCB, power supply, load etc). Therefore the MS5351M is oscillating at around +7 or +8 kHz higher than the nominal 27MHz crystal frequency, compared to +4 kHz for the Si5351A. This should be borne in mind when setting up a piece of equipment using MS5351M and NOT doing a proper calibration - to get it in the right ballpark use a frequency of 27.008.
I tested phase noise at 9MHz which is the only frequency at which I am currently equipped to make phase noise measurements. I measure phase noise from 500 Hz to 100 kHz from the carrier. The MS5351M phase noise is a few dB lower than Si5351A across the measurement range.
I2C bus address
Si5351A-B-GT/GTR and MS5351M use the same I2C address.
Functionally I couldn't find any difference between MS5351M and Si5351A (all registers I tried are the same and do the same thing). I readily admit to not having tested every single register however I have tested everything used in the various QRP Labs kits which includes use of the three outputs on independent frequencies, use of two outputs on the SAME frequency but with a 90-degree phase shift (quadrature oscillator as used in the QCX-series CW transceivers). In all cases I was unable to find any difference in behaviour between MS5351M and Si5351A.
One strange finding with the MS5351M is that I found that when powered up initially, it produces a weak -52dBm signal output on the Clk0 pin, at 88.7 MHz. This signal disappears as soon as the device has been configured to produce your desired output frequency (s). It is therefore of no practical significance.
I swapped the Si5351A in two QCX-mini CW radio transceivers for MS5351M. One was built with a 27MHz crystal and one with our 25MHz TCXO option. In both cases the QCX-mini was indistinguishable from a "normal" Si5351A-based one in function and performance.
About the only thing I haven't been able to test for obvious reasons is service life and reliability. So it'll have to be an accepted risk on that one. But better than going out of business, since there's no other alternative to the Si5351A for the foreseeable future.
Review by Jim W0EB
Jim reported here on his tests of the MS5351M and also found it a good substitute for Si5351A
MS5351M does not appear to be available, at time of writing, from any of the usual European/US distributors such as Digikey, Mouser or Farnell. But it is available from Chip Mall, or from LCSC, or from AliExpress, and no doubt numerous other places. The price of MS5351M is typically a little less than Si5351A.
As far as I am able to tell, the MS5351M is a 100% suitable substitute part for the Si5351A, for both hobby projects and commercial products.
73 de Hans G0UPL
Disclaimer: results and conclusions above are presented in good faith, and measurements performed to the best of my ability. Neither QRP Labs nor Hans Summers G0UPL will be liable for any losses or damages arising from your use of the MS5351M - if unsure, perform your own measurements!