Signal still has effect when not connected to circuit

In real life, I have a 50 Hz sawtooth with some high-frequency noise. I used Analog Devices’ Wizard to design a filter and transcribed it into QSPICE. I used PWL to make a source for my sawtooth and used a sine voltage source for the noise. I just ‘tee-d’ them together into to the input, disconnecting one or the other to see the effect. I realize I should use a summing amp or mixer, and I’ll get to that later.

What’s interesting is that after a few iterations, when the sine and the sawtooth are connected, the simulation fails (Singular matrix). When only the sine is disconnected, the sawtooth has some junk imposed on it every cycle. When the sine voltage source is set to “do not stuff”, the sawtooth looks fine.

How can it have an effect when it’s not connected to the circuit? When it’s stuffed but not connected, it is in the netlist with ¥0. Does that give it some phantom connection to all those ¥ ¥ ¥ ¥ ¥ s on the OpAmps?

I can’t upload the file, this link might work:https://1drv.ms/u/s!AhcH0TLqhogLir4N8KRvLVSovpuB8Q?e=Lja3dK

John

You have probably connected sine an sawtooth source in parallel. That is a short circuit. If you connect them in series u have desired effect, also to mitigate oscillation problem u should add resistance in series with capacitors.

Best regards


3 Stage filter, two signals.qsch (15.4 KB)

Ivan,

Thank you, that tip helps with my original intent.

I still wonder how V2 could have an effect when it was not connected to the circuit.
John

Here I have original output from your filter and I can’t see 300 KHz oscillation.

Except that high frequency oscillation at a start of sawtooth. That is a product of low capacitance at a opamp output.

Best regards

Ivan,

Thank you very much for the reply. It helps towards my actual goal of cleaning up the sawtooth.

My question about QSPICE is that if, in my original schematic, I erase the wire joining V2 to the circuit, and run the simulation, I have that high frequency oscillation at the start of the sawtooth. If I mark V2 as “do not stuff”, the oscillation is gone. I wondered how V2 has any effect when not connected to the circuit.

(I must note that I am not having singular matrix errors now; perhaps the file is changed somehow.)

After reading some posts regarding differences between the results of QSPICE and other simulators, I understand that QSPICE may include some parasitic effects. Is it possible that even though Q2 is not explicitly connected to the circuit, it has some effect through coupling or through the ground?

Possibly the oscillation you are referring to is this one, correct?
Here is not an answer but an observation. This oscillation does not appear to originate from V2, but rather from the timestep that V2 provides. QSPICE utilizes a dynamic timestep approach, and with V2, the timestep must be smaller than factors of 1/300000 seconds. If V2 is disabled but .option maxstep is added, the same phenomenon can be observed. If no maxstep is added, this oscillation will disappear.

Thank you for the input. I am experimenting with MAXSTEP in simpler simulations and will return to this one.

08/06/2024 Better defaults are chosen for the RRO device for operation near the rails.

Mike has improved the RRopAmp, and the simulation used in my previous message will no longer produce oscillations. Here is the simulation file which I forgot to attach in previous post.

18432-6.qsch (15.6 KB)