0

What are the best terms for crystals/resonators/oscillators used with microcontrollers that cover each of the following groups:

  • clock sources that require an additional oscillator circuit (NOT gate) of the type found in many microcontrollers, and are therefore hooked up to both CLKIN and CLKOUT pins of a microcontroller
  • clock sources that provide a clock signal without additional circuitry, and can therefore be hooked up to a single pin on the device (typically CLKIN or OSCIN) without requiring additional pins
Jason S
  • 13,950
  • 3
  • 41
  • 68
  • related (but not identical) questions: https://electronics.stackexchange.com/questions/86676/what-is-crystal-resonator-crystal-resonator-crystal-oscillator and https://electronics.stackexchange.com/questions/36308/crystals-oscillators-and-resonators-what-the-difference – Jason S Aug 18 '21 at 17:05
  • 3
    To add to the confusion, there are times when you are using an external oscillator (rectangular can, for example) to drive an MCU where you are ***supposed*** to drive the XOUT and ***not*** the XIN! (Many Microchip MCUs, for example.) This is because driving the XIN forces the internal class-A inverter amplifier to *run hot*; but where driving the XOUT allows the internal class-A inverter amplifier to settle on a quiescent state, instead, and thereby greatly reducing power consumption as the inverter is vastly over-powered to deal with idiot-proofing. – jonk Aug 18 '21 at 17:09
  • @jonk interesting... that may be, but that doesn't affect how I would describe the clock source as one of these two groups of objects – Jason S Aug 18 '21 at 18:05
  • Just FYI. You get to decide how it matters. And sounds like you have. – jonk Aug 18 '21 at 18:10
  • @jonk do you have any reference on driving the output?. ASAIK all MCU’s use push-pull CMOS inverters with >=5M internal feedback so they draw current when enabled but not connected to any tank circuit or not oscillating at Vdd/2 from the shootthru limited current. – Tony Stewart EE75 Aug 18 '21 at 18:12
  • @TonyStewartEE75 Yes. An internal document I received from a Microchip FAE. It's about five pages in length. – jonk Aug 18 '21 at 18:13
  • Hmm this one does not support that. http://ww1.microchip.com/downloads/en/AppNotes/00849a.pdf – Tony Stewart EE75 Aug 18 '21 at 18:14
  • @TonyStewartEE75 Regardless, I do have the internal memo here. It covers my product of interest at the time, in the PIC16 family. – jonk Aug 18 '21 at 19:22
  • @jonk when does this internal memo date from? 1995? 2000? 2005? 2010? that may be helpful. I'm skeptical. (disclaimer: I work at Microchip but in the dsPIC division so don't work on 8-bit parts) – Jason S Aug 18 '21 at 19:24
  • @JasonS Late 1990's. If I want to spend some time, I could get the exact date. I used the technique and it worked as advertised. Power was about cut in half, as a result. (I was using a Harris oscillator IC at the time which was very very low power.) As Microchip literally builds their chips "forever" in order to support product lifecycles better than any other company I've used, I believe it will still apply to the same parts it addressed then. (The Harris part was the HA7210.) – jonk Aug 18 '21 at 19:28
  • OK, I'm just surprised, because if this were important to Microchip's currently-manufactured products, it should be in public documentation. – Jason S Aug 18 '21 at 19:30
  • 1
    @JasonS I thought it should have been in the public documentation ***back then***. I said as much to the FAE. But for reasons not entirely clear to either of us, it was available *only upon request.* (I didn't know to ask for it until the FAE mentioned it, as part of our discussion about reducing power requirements. It was then he said I could ask for it.) – jonk Aug 18 '21 at 19:33
  • If it were Class A there would be only a single FET and the internal feedback R would produce Vout = Vgs(th) which is lower than Vdd/2, so the regular usage might produce asymmetric square wave outputs. It should be a high RdsOn unbuffered gate. – Tony Stewart EE75 Aug 18 '21 at 19:33
  • @TonyStewartEE75 The document ***specially says*** that it is class-A. It's how it's written. I didn't make it up, despite your continued disagreement here. I'm just offering an experience of mine. That's all. How you take it is up to you. But you can probably just test the idea. It's not hard to do. What I'm saying is verifiable. – jonk Aug 18 '21 at 19:34
  • This could be **that** doc , but rather skimpy on details but it does say Class A https://onlinedocs.microchip.com/pr/GUID-72227D20-D1DB-4CD0-A965-095A250728AA-en-US-2/index.html – Tony Stewart EE75 Aug 18 '21 at 20:13

1 Answers1

1

Crystals are two-pin components and ceramic resonators are two-pin or three-pin components. Both are passive components in a sense that they don't have power supply inputs and thus don't oscillate by themselves, they need the oscillator circuitry that is typically provided inside the MCU. These can be used with a clock setting that requires "external crystal" or "external resonator" and must be connected with two pins to the MCU. If the MCU is capable of different clock settings, it must be configured to use this kind of clock source by turning on the internal oscillator circuitry.

If the MCU has a separate setting of accepting a square wave clock signal on a single input pin, this is typically just called "external clock". This selection also typically turns off the internal oscillator circuitry as it is not needed.

The square wave can however come from any source, like a "crystal oscillator" you can buy as a separate module that contains both the crystal and the oscillator circuitry and it takes in supply voltages and outputs square wave. Or any other circuit that produces square wave.

Some MCUs also have a setting for connecting external resistor and capacitor as an RC oscillator.

Justme
  • 127,425
  • 3
  • 97
  • 261