Hex inverter oscillator, where is the current draw going?

[jfiresto]

The problem was much worse for the old 4000 CMOS series, especially when run off higher supply voltages than 5V. These could draw several mA of quiescent supply current per gate if the input voltage was in the intermediate range between the standard input voltage levels.

The typical shoot through current of a CD4093 is no more than a milliamp, even at a 15V V_DD



(from the Intersil CD4093BMS datasheet, file number 3330, after an original from RCA)

[srb1954]

The problem was much worse for the old 4000 CMOS series, especially when run off higher supply voltages than 5V. These could draw several mA of quiescent supply current per gate if the input voltage was in the intermediate range between the standard input voltage levels.

The typical shoot through current of a CD4093 is no more than a milliamp, even at a 15V V_DD



(from the Intersil CD4093BMS datasheet, file number 3330, after an original from RCA)

Reading from an original CD4093B RCA datasheet (File 836 1974 databook) the peak current per gate is specified at 1.5mA. Intersil may have made some later improvements on their design to reduce the shoot through current.

Other gates in the earlier 4000A series were even worse with peak shoot through currents sometimes exceeding 10mA when operating off a 15V supply.

[jfiresto]

Sorry, I read ca. 1.4mA and managed to forget that while I searched for and found Intersil's much cleaner, redrawn (and rescaled) figure. I see that Intersil did not transfer the peak currents at 5V, which if I squint at the points the right way appear to be around 300µA, a bit less than the peak currents at 4.5V for an Nexperia 74HC132, a part the OP might be considering.

EDIT: After further squinting, what looked to me like current dots for 5V could be severely smudged arrows. I seem to remember a plot with shoot through currents at 5V, but where? I need to get more sleep.

[Infraviolet]

I did a few breadboard experiments, here's what I found:

74AC14, schmitt hex inverter, single gate oscillator with one extra gate as output buffer, powering a 1K ohm load, other 4 inputs grounded: 8mA, reducing to 6mA drawn when resistor is changed to reduce output frequency from 3.2MHz to 450KHz.

74AC14, schmitt hex inverter, 3 gate oscillator with one extra gate as a buffer, 1K load, 2 spare inputs grounded: 10mA at 3.2MHz

74HCU04, plain inverter, 3 gate osc with an extra gate as buffer, 1K load, 3.1MHz, 16mA, and much cleaner edges with less rining than any of the setups with the schmitt inverter (singly or in triple gate arrangement). Also found this drew 8mA if resistor varied to give 250KHz output, and 28mA if resistor made smaller to give 20MHz out. If the resistor was entirely removed so oscillation stopped (R ≈ "infinity" from the third gate's output to the first's input) then 5mA was still drawn, guess this is from putting current in to the 1K load at the buffering gate's output.

All types had the frequency vary somewhat if the suply voltage were canged, for example from 2.9Mhz at 4.6V to 3.15MHz at 5.3V.

Frequencies differed between the tests as I kept the same cap in place and just varied the timing controlling resistor to change frequency, picking whatever was closest to 3MHz from resistor values to hand, plus doing some slower and faster tests to get a feel for current consumption vs frequency.

So it seems the single gate inverter can be more efficient, and schmitt triple is more efficient than doing the triple with normal inverters, and any inverter gets more current hungry at higher frequencies. Still, the lack of ringing on the edges of the triple gate plain inverter version is pretty desirable, guess I'll be putting up with 16mA or so of current as the price of this over the schmitt inverter's ringing.

[BillyO]

and any inverter gets more current hungry at higher frequencies.

Nice experiment and good results.

The reduction in power consumption with frequency is to be expected as the predominant current draw is while the gate output is changing state.

A lot of that ringing is due to how fast an AC gate switches.  They can change state in around 1ns.  Keeping the traces coming off the output as short and wide as possible will go along way to helping with this.  As will having some resistance in series between the oscillator output and whatever it's driving.

However, all that said the ringing you are seeing may well be the result of a long ground lead in your probing or a high impedance ground in the power supply to the chip.

The image below is a buffered 74AC14 oscillator.  It has a rise time of < 1.1ns when the scope rise time is taken into account.  The overshoot is minimal and the ringing lasts about 2ns.

Edit (again) : BTW, for this circuit all 6 inverters are used.  1 for the oscillator, 1 as a buffer and 4 as drivers to get a 50 ohm output.  The total current draw at this frequency and driving into a 50 ohm load is just under 12mA.