File hc11osc.txt Motorola MCU Freeware- boxista downloadannut Eero A. Sarlin/ Field Oy Kasittelee MC68HC11 oskillaattoria, mutta koskee myos MC68HC04 MC68HC05 MC74HCXX MC74ACXX ja muita CMOS- ja HCMOS- prosessoreita ja invertterioskillaattoreita In the MC68HC11 the operating current (Idd) is the sum of the component currents Idd = Osc + CPU + Analog + STOP If we hold all variables (except frequency) constant, then for a given part..... A. The STOP (leakage) current is constant. B. The Analog current (A/D etc.) is constant if disabled or running on the internal RC oscillator. C. The CPU current is a linear function of frequency. D. The crystal oscillator current is a function of many variables including the external components, PC board layout, etc. and it is therefore impossible to state a general rule. If the external network is designed improperly, then it is quite possible for a low frequency circuit to use more current than a good high frequency design. The external network can (and should) be optimized. This is particularly important at low frequencies where the relitively high loss in the external network is high. If the amplitude of the signal at the input to the oscillator (EXTAL) is small, then the oscillator Idd goes up. The oscillator current has two components, the load current and the 'short' current. The load current is simply the current that is dissipated in the devices connected to the output of the oscillator (XTAL). The 'short' current is the Vdd to Vss current which is the result of both the P- and N-channel being turned on. This happens whenever the input signal is between a solid logic '1' and a solid logic '0'. This is roughly Vdd-0.6 and Vss+0.6 Volts. The lowest possible oscillator current occurs with no load with a clean Vdd to Vss square wave at the input. The current is a function of frequency because the load is simply the pin capacitance. This current is very repeatable. In fact, this is the way that the 'HC11 is tested (see the fine print at the bottom of the page in the data book). If the input signal spends a majority of each cycle in the high 'short' current mode, then the Idd can be very high. This is what happens in some low frequency designs. The feedback resistor provides the DC bias so that the quiesent point of the input signal is in the center of the linear operating range (approx. mid supply). If the loss in the network is high, the amplitude of the sinusodial signal will be small. The oscillator may always be in the high current range because it never reaches the point where one of the transistors is off. Beleive it or not, it is possible to put a resistor in series with the external network and actually improve the amplitude of the input signal. The theory for this takes is too long to include here. The following is a simple 'tune-up' proceedure for determining the optimal value for Rx through a very simple experiment. The 'Tune-Up' Procedure There is a very simple 'tune-up' procedure which can be used to optimize a Pierce oscillator design using any type of crystal, at any frequency, supply voltage, and temperature. The result is an 'optimal' circuit design which is stable, reliable, and low power. A. Construct an oscillator using the components (crystal, capacitors, inverter, resistors) that you intend to use. If possible, the final PC board should be used because the physical layout of the board, the type of sockets, etc. can affect the oscillator operation. B. Substitute a variable resistor for Rx. C. Insert a micro-ammeter in the supply line to monitor Idd. D. Tie all unused inputs to supply (Vdd) or ground (Vss). This prevents 'floating' inputs which can use relatively large currents that will affect the Idd measurement. In large chips, such as micro-processors, the chip should be held in reset. This prevents any internal clocking which would raise the Idd and affect the measurement. In Motorola MCU's, Reset will force all I/O ports into inputs. All of these should be tied off to avoid 'floating' inputs. E. Monitor Idd while varying Rx. There should be a very slight dip in Idd. This dip is very broad (low Q) and very shallow (only a few percent). The dip is the optimal point of operation. F. Remove the variable resistor and substitute the nearest fixed value. As the shape of the Idd curve suggests, the exact resistor value is not critical. Tolerances of up to 20% can generally be used. Note that the value of Rx also affects the operating frequency. Usually this is only a few ppm. If a precise frequency is needed, then it will be necessary to use the following sub-procedure... AA. Adjust Rx for 'optimal' Idd. BB. Adjust Cy to set the correct frequency. CC. Go back to AA and repeat until no changes are necessary. These are the definitions of the parameters.......... Crystal Parameter Definitions: R1 C1 L || O-----*--/\/\/\/----||----XXXXX---*-----O | || | | | | | | || | +-------------||------------+ || C0 External Component Definitions: |\ | \ | \ | \ -----| O----*------> | | / | | | / \ | | / / Rx | |/ \ | / | Rb | *----/\/\/\----* | | | _ | | | | | *----| | | |---* | |_| | | | Cx ===== ===== Cy | | | | === === - - Amplifier Parameter Definitions: | \ | || Cio \ | -----------||----------- \ | | || | \ | | | \ | | |\ | \ | | | \ | \ | | | \ | \ | | | \ Ro | \ -- >-----+---------*-----+ Av O---/\/\/\----*----------+| |------> | | | / | / -- | | | / | / | Cin ===== | / Cout ===== / | | |/ | / | | | / | === === / | - - / | / | / | / Some Practical Design Tips The following is a summary of some guidelines which may help in design. Rb: This resistor sets the bias point. It also affects the amplifier gain by introducing DC negative feedback. Therefore, it should be as large as practical. Usually the limiting factors are the input leakage current and the printed circuit board contamination (solder resin, finger prints, moisture, etc.). Rx: This resistor isolates the crystal network from the amplifier output and provides a needed phase shift. The R/C lowpass filter, formed by Rx and Cy, decrease the probability of spurious oscillation at high frequencies. The value of Rx should be approximately equal to the effective series resistance of the crystal at the operation frequency (if both Cx+Cin and Cy are equal). An increase in the value of Rx will decrease the amount of feedback and improve stability. Cx and Cy: These capacitors are part of the crystal load network. Typically, Cx = Cy = (2 x CL) I have simulated a couple of cases in the 1 to 2 MHz range that you mentioned in your bulliten board message. They are both for an 'HC11 in a plastic DIP with Vdd = 5.0 Volts at 25 deg C. 1.000 MHz R1 = 341 Ohms Rb = 10 MOhms CL = 10 pF Rx = 10 kOhms 2.000 MHz R1 = 154 Ohms Rb = 10 MOhms CL = 30 pF Rx = 1 kOhms This was done using an experimental crystal oscillator simulation program that I am working on. I would be interested in hearing the results of your work so that I can verify the accuracy of my program's results. Ken Burch 11309 April Drive Austin TX 78753 $