2

I´m working on a new design with MPC5744 (LQFP) microcontroller.
As far as I know, the crystal unit drive level could be obtained from the following equation:
P=2*Rl*(pifreqVdd*(Cl+Co))2

Taking in to account MPC5744 evaluation board, we have following values:
Rl= crystal series resistance 50R
Freq=40 MHz
VDD=1,7V (from datasheet)
Cl=8 pF (stray capacitance 4 pF)
Co=6 pF (shunt capacitance)

Applying those values in the equation I obtain that the crystal unit requires 894uW Drive level.
I don’t understand how the eval board crystal unit could work, as it has a 500uW drive capability.

I think that I´m doing something wrong but I don´t know what.

brhans
  • 14,723
  • 3
  • 35
  • 51
Davidtxi
  • 21
  • 1
  • 2
  • What would help is a link to the documentation where you found that equation (sorry I'm too lazy to look it up myself). Also state where you found those drive levels. Crystal series R is often much less than 50 ohms, probably NXP only guarantees operation for R < 50 ohms. Lower R means less power is lost and less power is needed to make the crystal oscillator work. – Bimpelrekkie Sep 21 '16 at 13:06
  • Hi,I found at this application note http://www.nxp.com/files/microcontrollers/doc/app_note/AN3208.pdf (pag4). – Davidtxi Sep 21 '16 at 13:28
  • Regarding series resistance, if i´m not wrong, oscilates around 50Ohm. For instance, this reference belongs to eval board NX5032GA-40.000000MHZ-L N-CD-1. – Davidtxi Sep 21 '16 at 13:38
  • if i´m not wrong, oscilates around 50Ohm. Then I'm sorry to say that you do not understand how a crystal oscillator works. The 50 ohms is an (unwanted) property of the crystal. Ideally it should be 0 (zero) ohms. The oscillator does not "oscillate around 50 ohms" that is just nonsense. – Bimpelrekkie Sep 21 '16 at 13:42
  • Ok then what is series resistance value for NX5032GA-40.000000MHZ-L N-CD-1 reference? I was supossing 50 Ohm (Rmax at 40MHz). – Davidtxi Sep 21 '16 at 14:09
  • It depends on the crystal which is used. And even if you know that, this Rs is often not mentioned in the datasheet. It does not matter that much actually. On the eval board there will be a crystal which will just work. Why are you so concerned with the power of the crystal oscillator ? – Bimpelrekkie Sep 21 '16 at 14:39
  • I was thinking on AEC-Q200 version of eval board crystal, with extended temp. range (system requirements). But due to minimum order quantity, I must change it. The problem is that the most crystal units (besides some of TXC), with similar characteristics, has 200uW-100uW max drive capability. – Davidtxi Sep 21 '16 at 14:52
  • 100uW-50uW-10uW drive level restrictions are coming from recent miniaturization of crystals. Use old-style bigger beefier HC-49 type crystals. – Ale..chenski Sep 21 '16 at 16:14
  • microslice AT Xtals are now rated for 1uW in SMD packages. – Tony Stewart EE75 Sep 21 '16 at 17:03

3 Answers3

4

Let me correct your language a bit. The maximum drive capability of the crystal in the eval board is 500uW. It requires typically 50uW to work.

Your calculation indeed indicates that it is possible the crystal is being overdriven. That could result in drift or other problems. However note that the 50\$\Omega\$ is a maximum figure and the voltage across the crystal is probably less than the supply voltage (you would probably need an active FET oscilloscope probe to measure it directly with any accuracy).

Chances are in practice the crystal is operating within proper bounds, but you should certainly take care to use clock monitor features if it's a safety-critical application. There is always this sort of unsatisfactory guarantee of proper operation (drive level and starting) with crystals unless the manufacturer recommends a specific supplier and model (and stands behind their recommendation).

You can use an external oscillator module or a resonator if you want better guarantees.

Spehro Pefhany
  • 397,265
  • 22
  • 337
  • 893
0

I think your assumptions are based Vrms=Vdd and computing VAR not Pd

The voltage may be a Square wave @Vdd or Vdd rms but the current is a sine wave. This is what you see on the inverter input signal.

  • Converted using Vpk^-.5 to rms for sine wave and 50% of Vdd for a squarewave
  • This should result in the fundamental waveform rms Voltage amplitude.
  • Obtain reactive current from I=V/Z
  • Then compute Pd=I^2Rs in the series R is power loss.
  • we do not compute VAR = W, but reactive power at resonance induces conductive losses in series R.
  • (I believe) this is a molecular junction interface limited to 1uW (typ) for microslice crystals to >50uW for rugged series overtone crystals

This application would use 50uW NX5032GA-40.000000MHZ-LN-CD-1 ( but is rated for 500uW

Tony Stewart EE75
  • 1
  • 3
  • 54
  • 185
  • Thank you, I was thinking on NX5032GA-40.000000MHZ-STD-CSU-2 (AEC-Q200, -40 +125, system requirements) reference but due to minimum order quantity, I must change it. The problem is that the most crystal units has 200uW-100uW max drive capability. – Davidtxi Sep 21 '16 at 14:35
  • MOQ=1 http://www.digikey.com/product-detail/en/ndk/NX5032GA-40.000000MHZ-LN-CD-1/644-1045-1-ND/1128917 – Tony Stewart EE75 Sep 21 '16 at 14:50
  • Thank you but this reference is´t AEC-Q200 and extended temp. range (-40 +125). – Davidtxi Sep 21 '16 at 14:58
  • http://www.digikey.com/product-detail/en/ndk/NX5032GA-40.000M-STD-CSU-2/644-1137-2-ND/1788477 the problem is MOQ. – Davidtxi Sep 21 '16 at 15:01
  • read again .. Digi-Reel® ? : 644-1045-6-ND Minimum Quantity: 1 Quantity Available: 2,813 - Immediate Unit Price: Digi-Reel® – Tony Stewart EE75 Sep 21 '16 at 15:06
0

Power dissipated by XTAL can usually be reduced/tuned down by adding a series resistor on the output end of oscillator circuit, provided that this resistor (plus crystal ESR) is 3-5 times smaller than the negative impedance of the driver, so the oscillator can start reliably.

But when designing to automotive standard it is highly advisable to match crystal parameters with driver parameters, and limitations of "minimum order quantity" is usually not in vocabulary of automotive designers. Besides, there are many other sources for components other than Digi-Key.

Ale..chenski
  • 40,470
  • 3
  • 39
  • 109