# Completely different circuit response with 1000kΩ than with 1MΩ resistance

I have the following circuit (a band-pass filter) and I cannot understand what's the difference of a 1000kΩ resistance to a 1MΩ as seen here:

With R223 = 1000k, my circuit outputs the expected theoretical results (for input signal of 32kHz at amplitude of 500mV, it retains indeed 500mV almost) , but when I replace it with a 1M resistance it outputs 74mV amplitude instead, which is totally illogical! Maybe I have missed something obvious here? But aren't those two resistors exactly equal? Isn't 1000k = 1M? And the weirdest thing is that the real circuit (not simulation) also outputs the "illogical" result of 74mV!

• Have you had a look at the manual and made sure that m doesn't mean milli? Commented Dec 18, 2017 at 20:30
• I haven't realized that your comment about the real circuit is a question. To answer that I need the full schematic. The only thing which comes to mind with that sniplet is that 100pF isn't much and depending on your actual build (breadboard?) parasitics may affect your circuit (and no decoupling caps shown) Commented Dec 18, 2017 at 20:39
• @Arsenal Would a breadboard really have parasitics such that they can affect significantly an 100pF capacitor? And also, center frequency filter gain theoretically only depends on resistors, so how is this affecting the gain again? Maybe it's the gain bandwidth? But again, TL074 has a gain bandwidth product of 3MHz, which should allow much room for gain, especially in my unity-gain filter here... Also, I have decoupling caps of 10uF between the 12V supply.
– user172287
Commented Dec 18, 2017 at 21:18
• What do you mean with "replaced 1000k with 1M in real circuit? Did you replace what? Commented Dec 18, 2017 at 21:35
• @Jason: are you talking about a real circuit or a simulation? There is no difference between 1000kΩ and 1MΩ. They are just two ways to name the same thing, a resistor of $10^6$Ω. So if you mean reals resistors: please show a picture of them. If you are using a simulation (Spice): note what PlasmaHH said: Spice is case insensitive and 'M' means the same as 'm', i.e. 'milli' = $10^{-3}$, not $10^6$. 'Mega' = $10^6$ is abbreviated by 'Meg'.
– Curd
Commented Dec 18, 2017 at 21:57

In all SPICE programs I've used m and M are the same and are 0.001. As others have said, you have to use Meg, or write 1000K.

As for your 74mV on a real chip, you have not shown the entire circuit, but if the supply voltages are adequate it could be due to leakage on a dirty board (perhaps flux was the wrong type and not fully cleaned).

Offset voltage is +/-9mV max, +/-3mV typical, but a current from say V+ to pin 6 of the chip of 94nA will explain the offset. If your V+ is +15V, that's a resistance of 160M, which is pretty bad for a clean board, but plausible for a not-so-clean one.

In PSpice m and M are thousandths. You need to use Meg.

• Indeed, that's correct on the simulator, but still the second part of my question still holds... Why does the real circuitry also output 74mV amplitude? Is there some real constraint on the op-amp here that I might be missing?
– user172287
Commented Dec 18, 2017 at 20:36
• @Jason Most likely a poor simulation of non-ideal parameters, can't say without more information. Commented Dec 18, 2017 at 20:40

Not sure what kind of simulator that is, but a common pitfall is:

1M = 1 milli (Ohm, Volt, Ampere) and 1Meg = 1 Mega (Ohm, Volt, Ampere).

Because they don't care about capitalization they can't make out a difference between 1M and 1m so both end up being 1 milli.

Replacing R223 with a 1 ohm resistor will have an effect, but not nearly as much as you are seeing. It should cut the feedback resistance (and therefor the gain) by a factor of about 40%, not the 90%+ which you are seeing.

I very much suspect that you have inadvertantly connected the top of R231 (2.2k) to the junction of R21 and C21, and you have done it on the physical circuit as well.

If I had drawn that schematic, I'd have placed C21 and R231-R233 at the lower right corner of the circuit, so there would be no crossover and no mistaken connection.