Unlike quartz, silicon is not piezoelectric. Thus, it is tricky to cause mechanical motion to interact with current or voltage. That is where MEMS is needed. Basically, mass on a spring is charged and placed between capacitor plates. An AC signal applied to the capacitor then excites the cantilever and the vibrating mass on a spring excites an AC signal across the capacitor. At this point the circuitry for both devices is quite similar. The capacitor and the vibrating mass in these parts are each complicated. MEMS is also needed to form these complex shapes that allow such high frequencies and improve quality factor, Q.
Thank you so much for this two part series. These Chip of the Day videos make new-tech and preloved-tech totally accessible. I'm now confident of tackling a number of projects needing specific frequencies where you just can't buy a suitable crystal. Your channel continues to be awesome, compelling viewing for enthusiasts the world over.😀
MEMS micromachining can be attained by etching Si in KOH [solutions] at moderate temperatures (~40-80C), protecting Si with a metal (Ni, for instance, which is resistant to KOH) and protecting SiO2 with metals and/or a (photo)polimer (also resistant to KOH). It's not difficult to obtain structures with an order of +/-1 micron.
A while ago I used a Microchip 25MHz MEMS oscillator to clock a Wiznet W5500 ethernet chip & found that with the MEMS osc I was losing 1-2% of received packets, presumably due to jitter. Swapping to a real xtal oscillator completely fixed the issue.
Been using SiTime MEMS oscillators for my project, including usage as a CPU clock source for microcontrollers as they're robust, cheap and surprisingly efficient. Also, there's a few reasons why handful of smartphones, iPhones and Android phones use MEMS oscillators is that they can keep on marching even when dropped onto solid surfaces that would easily crack the screen - only for the computer inside to not flinch at all. It would be interesting to see some mid-range digital watches using 32 kHz MEMS TCXOs as like I mentioned, they're electrically efficient so that way batteries would last for a good while.
The jitter spec looks quite low, which is important for things like Audio A to D converters. The frequency range is obviously much higher than necessary, though, for Audio. I wonder do they come in lower frequency ranges and retain that stability, or does it get notably worse if you divide the frequency down to say 192khz, 96khz or 48khz?
Yes, there's a few timekeeping MEMS oscillators that can be programmed for audio electronics, at any desired frequency rather than just the boring 32 kHz.
SI's app note: AN10050 I2C/SPI Programable Oscillators ( not going to do it with simple dipswitches - 👌) YES, some models are field reprogrammable; one of the great features is frequency tuning/hopping semi real time {register load time} You can change reference osc divider, fractional N synthesizer, output multiplier, and edge speed options. Fixed freq versions are dirt cheap; variable programming requires additional microprocessor etc (>10x more $$'s).
Defiantly one time programmable, found out the hard way when I messed some up with two digits transposed in the frequency 😪 Trying to program the module a second time with the correct frequency gives two red error lines: programming failure and verify failure. Luckily had enough spare blanks without having to wait for more to arrive.
I wonder if a zif socket could be fashioned ugly style out of pogo pins, pcb, and a latch... would be nice on a little breakout board, if somewhat specific in use...
I have used them alsready, eone for a 144Mhz.x clock for ham radioa applicaion which worked great. I also work on space projects, but there the MEMs technologie can have problems as I was told. Also the pahase noise has to be considered when using them as abase oscillator for higher frequencys. Bu reall perfect for all other type of appicatiosn where you quickly need a special frequency.
The last Wavetek video said it was part of a 14-episode series, so I'm sure there are more videos coming. If you want more Wavetek, check his back catalog. He worked on another Wavetek a year or two ago.
I have used the Micrel programmable silicon MEMS DSC8002 oscillator. It is an inexpensive way of getting a signal on a custom frequency. It works fine but has considerably more phase noise than a typical quartz crystal oscillator. If your application does not require a pure tone, it is a good choice. Review the specs carefully! I would estimate that mine, operating a bit above 100 MHz, occupies a bandwidth of a few hundred hertz.
Fully disagree with you. MEMS technology have a lot of different variations. Therefore you can choose the most suitable for you. If really want to test real MEMS oscillator - use SiTime products.
Actually, SiTime's oscillators are actually better than most of the other MEMS oscillators, so good that their manufacturing processes are patented. And yes, I have been using those SiTime MEMS oscillators as their garden variety version only cost between $1 to 6 a pop, still cheaper than most oscillators (quality quartz oscillators cost way more for basically the same performance as even cheaper SiT8008 oscillator), not to mention it can easily survive insane G force without missing a beat.
These parts became famous because because a hospital suffered a helium leak on their MRI machine and nearby iPhones failed. This was because the helium permeated the case and caused the tuning fork to vibrate far higher than the PLL could lock so it stopped working. (Same reason your voice gets squeaky when playing with helium). I think @Applied Science did a video on it where he wired up an SiTime oscillator and then exposed it to helium showing the frequency climbed until it failed. This is only a temporary effect - once the helium permeates out it works again, so Apple didn't get a bunch of iPhones for repairs. But it made the news. These devices are used because they are much smaller than their quartz counterparts.
Unlike quartz, silicon is not piezoelectric. Thus, it is tricky to cause mechanical motion to interact with current or voltage. That is where MEMS is needed. Basically, mass on a spring is charged and placed between capacitor plates. An AC signal applied to the capacitor then excites the cantilever and the vibrating mass on a spring excites an AC signal across the capacitor. At this point the circuitry for both devices is quite similar. The capacitor and the vibrating mass in these parts are each complicated. MEMS is also needed to form these complex shapes that allow such high frequencies and improve quality factor, Q.
I wouldn't call this "Space Age" technology... I would call it Next Millennium age technology
Thank you so much for this two part series.
These Chip of the Day videos make new-tech and preloved-tech totally accessible.
I'm now confident of tackling a number of projects needing specific frequencies where you just can't buy a suitable crystal.
Your channel continues to be awesome, compelling viewing for enthusiasts the world over.😀
MEMS micromachining can be attained by etching Si in KOH [solutions] at moderate temperatures (~40-80C), protecting Si with a metal (Ni, for instance, which is resistant to KOH) and protecting SiO2 with metals and/or a (photo)polimer (also resistant to KOH). It's not difficult to obtain structures with an order of +/-1 micron.
A while ago I used a Microchip 25MHz MEMS oscillator to clock a Wiznet W5500 ethernet chip & found that with the MEMS osc I was losing 1-2% of received packets, presumably due to jitter. Swapping to a real xtal oscillator completely fixed the issue.
That's really good to know.
Been using SiTime MEMS oscillators for my project, including usage as a CPU clock source for microcontrollers as they're robust, cheap and surprisingly efficient.
Also, there's a few reasons why handful of smartphones, iPhones and Android phones use MEMS oscillators is that they can keep on marching even when dropped onto solid surfaces that would easily crack the screen - only for the computer inside to not flinch at all. It would be interesting to see some mid-range digital watches using 32 kHz MEMS TCXOs as like I mentioned, they're electrically efficient so that way batteries would last for a good while.
The jitter spec looks quite low, which is important for things like Audio A to D converters. The frequency range is obviously much higher than necessary, though, for Audio. I wonder do they come in lower frequency ranges and retain that stability, or does it get notably worse if you divide the frequency down to say 192khz, 96khz or 48khz?
Yes, there's a few timekeeping MEMS oscillators that can be programmed for audio electronics, at any desired frequency rather than just the boring 32 kHz.
SI's app note: AN10050 I2C/SPI Programable Oscillators ( not going to do it with simple dipswitches - 👌) YES, some models are field reprogrammable; one of the great features is frequency tuning/hopping semi real time {register load time}
You can change reference osc divider, fractional N synthesizer, output multiplier, and edge speed options. Fixed freq versions are dirt cheap; variable programming requires additional microprocessor etc (>10x more $$'s).
Micro electromechanical systems, pretty fascinating. Any sufficiently advanced technology is indistinguishable from magic, as Arthur C. Clarke said.
If I’m not mistaken that company makes super neat MEMS microphones. 👍
Interesting. Thanks for sharing.
Defiantly one time programmable, found out the hard way when I messed some up with two digits transposed in the frequency 😪 Trying to program the module a second time with the correct frequency gives two red error lines: programming failure and verify failure. Luckily had enough spare blanks without having to wait for more to arrive.
I wonder if a zif socket could be fashioned ugly style out of pogo pins, pcb, and a latch... would be nice on a little breakout board, if somewhat specific in use...
I have used them alsready, eone for a 144Mhz.x clock for ham radioa applicaion which worked great. I also work on space projects, but there the MEMs technologie can have problems as I was told. Also the pahase noise has to be considered when using them as abase oscillator for higher frequencys. Bu reall perfect for all other type of appicatiosn where you quickly need a special frequency.
Hi from NZ as well :)
google map is amazing good job sir
Interesting video. Are you calling it quits with the Wavetek?
The last Wavetek video said it was part of a 14-episode series, so I'm sure there are more videos coming.
If you want more Wavetek, check his back catalog. He worked on another Wavetek a year or two ago.
Chip of the day and tool time interrupt the series continuity. Subscribe and take part in the journey😊
Crystal appears to be made with LIGA process.
I have used the Micrel programmable silicon MEMS DSC8002 oscillator. It is an inexpensive way of getting a signal on a custom frequency. It works fine but has considerably more phase noise than a typical quartz crystal oscillator. If your application does not require a pure tone, it is a good choice. Review the specs carefully! I would estimate that mine, operating a bit above 100 MHz, occupies a bandwidth of a few hundred hertz.
Fully disagree with you.
MEMS technology have a lot of different variations. Therefore you can choose the most suitable for you.
If really want to test real MEMS oscillator - use SiTime products.
Actually, SiTime's oscillators are actually better than most of the other MEMS oscillators, so good that their manufacturing processes are patented.
And yes, I have been using those SiTime MEMS oscillators as their garden variety version only cost between $1 to 6 a pop, still cheaper than most oscillators (quality quartz oscillators cost way more for basically the same performance as even cheaper SiT8008 oscillator), not to mention it can easily survive insane G force without missing a beat.
Hmm.. When I imagine a silicon tuning fork, I imagine a nice miniature crystalline structure. Not IMSAI Guy's big hairy fingers as shown at 2:04 Lol.
These parts became famous because because a hospital suffered a helium leak on their MRI machine and nearby iPhones failed. This was because the helium permeated the case and caused the tuning fork to vibrate far higher than the PLL could lock so it stopped working. (Same reason your voice gets squeaky when playing with helium). I think @Applied Science did a video on it where he wired up an SiTime oscillator and then exposed it to helium showing the frequency climbed until it failed. This is only a temporary effect - once the helium permeates out it works again, so Apple didn't get a bunch of iPhones for repairs. But it made the news. These devices are used because they are much smaller than their quartz counterparts.
cursory