Hi looking at the R3 UNO Schematic i'm a bit confused about the crystal layout
the new R3 shows a 1Meg Ohm resistor on both pins and is the the caps I see going from each pin to ground 100nF
would be good to know the answer
Thanks
The UNO use a ceramic resonator rather then a quartz crystal. They are cheaper and not as precise. The resonator has the load capacitors built in and that is what you see in the schematic.
You can substitute a quartz crystal. If you do, add the appropriate loading capacitors (typically 20 pF each).
johnwasser:
The UNO use a ceramic resonator rather then a quartz crystal. They are cheaper and not as precise. The resonator has the load capacitors built in and that is what you see in the schematic.You can substitute a quartz crystal. If you do, add the appropriate loading capacitors (typically 20 pF each).
I understand
so if I use a 16mhz quartz crystal and not a resonator I don't need the resistor but do need the caps (I ment to say in first post 18-22pF) I use 22pF normally
the resistor is whats confusing me as I've never seen it mentioned in hackduinos before
I don't know what the 1M resistor is there for. A reading of the CSTCE16M0V53-R0 datasheet shows no use of a parallel resistor.
The easy thing to do in a case like this is to look up the p/n, and check the datasheet,
Then you immediately see it's a ceramic resonator with built in caps, and not a "crystal".
I've never quite understood why some ckts have a 1M R and others don't, but I've seen it
used a lot of times. Some of the ckts in the d/s show it. It may help oscillator startup.
2 days ago, I just wired up a proto ckt with a mega328 chip using a 16 mhz ceramic
resonator, and no 1M R, and it works fine. Am using a standard UNO bootloader chip.
oric_dan:
The easy thing to do in a case like this is to look up the p/n, and check the datasheet,CSTCE16M0V53-R0 Murata Electronics | Crystals, Oscillators, Resonators | DigiKey
Then you immediately see it's a ceramic resonator with built in caps, and not a "crystal".
I've never quite understood why some ckts have a 1M R and others don't, but I've seen it
used a lot of times. Some of the ckts in the d/s show it. It may help oscillator startup.2 days ago, I just wired up a proto ckt with a mega328 chip using a 16 mhz ceramic
resonator, and no 1M R, and it works fine. Am using a standard UNO bootloader chip.
Its very confusing as i use no resistors either but thought it was a R3 update so wanted to add it if it helps but hey as its a bit of a mystery will leave it to what i know
after all they say if it aint broke don't fix it
The resistor is a 'bias' resistor necessary to increase the 'gain' of the oscillator and is necessary for the lower 'Q' resonators. It should never be used with a crystal.
Bob
Docedison:
The resistor is a 'bias' resistor necessary to increase the 'gain' of the oscillator and is necessary for the lower 'Q' resonators. It should never be used with a crystal.Bob
that's great info thanks
Its very confusing as i use no resistors either but thought it was a R3 update so wanted to add it if it helps but hey as its a bit of a mystery will leave it to what i know
I should have mentoned before, MUCH better to scrub the resonator and use an actual
crystal [with caps] instead, as John already mentioned. You get 20 ppm accuracy with
a xtal, but only 0.5% with the resonator. R3 may be an update, but use of a resonator
was not the best design decision.
oric_dan:
Its very confusing as i use no resistors either but thought it was a R3 update so wanted to add it if it helps but hey as its a bit of a mystery will leave it to what i know
I should have mentoned before, MUCH better to scrub the resonator and use an actual
crystal [with caps] instead, as John already mentioned. You get 20 ppm accuracy with
a xtal, but only 0.5% with the resonator. R3 may be an update, but use of a resonator
was not the best design decision.
I always use crystals as I believe they are better and 22pF caps
least I know now that the res is not needed
thanks
Hallo
I have done some reading into this topic and I will summarise my findings.
As Bob correctly mentioned, the feed back resistor forms part of of the gain of the circuit. What we need to do now is expand on this and add a bit more to get clarity as to what the designers are intending to do by putting it there in the first place. Please refer to the attached figures
Function 1: Reduction of Spurious Frequency Emissions
Opening figure 1 we can see what could happen if there was no feedback resistor. Here we have an open loop inverter which could have infinite gain. The inverter outputs an AC waveform at 180 degrees out of phase. There is a limitation to how big Vout can and therefore Vout starts to "clip" by levelling off. Under these conditions we say that the inverter is behaving in NON LINEAR MODE
The Vout inverted sine wave now begins to conform to a "sine wave with a square shape" so to speak. Square waves are problematic in that they contain higher harmonics of the fundamental frequency. For example, a square wave at fundamental frequency Fo =20MHz can have additional frequency components at 3Fo (60MHz), 5Fo (100MHz), 7Fo (140MHz) and so forth (look up literature on fourier transforms of square waves). The spurious radiated emissions at these higher harmonics can cause your design to fail if it has to go for Radiated Emissions Standards test at a certified laboratory
The goal therefore is to get the inverter to operate in a LINEAR mode whereby the Vout signal is more like a PURE sine wave. The pure sine wave will not have any higher harmonics associated with it. To do this we have to reduce the gain so that "clipping" does not occurr. We do this by providing a feed back resistor Rf as shown in figure 2. The Lower Rf, the lower the gain, the higher Rf the higher the gain. Once again, one has to consult the reference design datasheet to see what optimal values should be chosen for Rf and the microcontroller in question
Function 2: Assisting Oscillator Startup
Another bit of information I picked up is that feedback resistor Rf assists the oscillator to start up when power is applied to the microprocessor i.e. that little bit of feedback from Vout to Vin is sufficient to get the oscillator running The clue that gives this away is when you mentioned the use of a resonator as opposed to a crystal. I am suspecting that resonators behave differently to crystals particularly when it comes to oscillator startup stability and therefore feedback resistance is a requirement.
So to summarise:
When choosing Rf:
Make an informed decision as to what value should be chosen for the crystal/resonator in question. Most literature states that lower frequency crystals require higher values of Rf.
The gain of the circuit is proportional to Rf. The higher the value of Rf, the higher the gain
Rf should be chosen such that the Vout signal is not clipped i.e. in this scenario the inverter is behaving in LINEAR mode
Rf should not be chosen to be too small, otherwise the gain of the inverter will be small and therefore the oscillator may not be able to sustain an oscillation
If Rf is chosen to be too big, then the small feedback required from Vout to Vin to get the oscillator started on powerup may not be sufficient
One final point. If ever one has to go into mass production, the whole "startup stability" of crystal circuits needs to be treated with respect. One may build 10 preproduction prototypes and they all perform as expected. The proof of the pudding comes when thousands of your designs are coming off the production line. The goal would be to predict with ceritainty that the crystal circuit start up is indeed a low risk.
The purpose of the resistor between XTAL1 and XTAL2 is to bias the inverter that is connected between those pins into the linear region, so that it will work as an oscillator. But the atmega328p datasheet doesn't show an external resistor in any of the clock schematics. Elsewhere on the net I found mention that the atmega series provides an internal resistor between those pins when the fuses are configured to use a crystal or ceramic resonator. So I conclude that an external resistor is unnecessary with these chips. I've never used one in my atmega or attiny designs.
I've used other chips that don't provide an internal resistor, and for those chips you definitely do need to provide an external resistor, otherwise they won't oscillate.
Please refer to the attached figures
@DigitalRS, good comments, did you find those figures somewheres, or make them
up yourself for illustration purposes?
I would think the load capacitors on the xtal/resonator would also have a lot to do
with rounding out the edges of the oscillator waveform and preventing harmonic
generation.