Hi @holger
Thanks a lot for this series of tutorials.
I fear that there is something wrong in the implementation of the output impedance of the generic opamp (Kicad 8.0.3 on Linux, Ngspice v42). Indeed, when I connect the output to a load with the same impedance, the voltage is not halved. If I take even lower impedance loads, the output voltage exhibits saturations that I do not expect.
Am I missing something ?
Please open a new thread for your question, do not use this thread on introductory videos.
I split it off for you both.
The entire function of an op amp is to make its inputs equal.
Let’s look at the simplest configuration of an opamp, the voltage follower. This is constructed by connecting the output to the - input, as shown here:
If you drive the + input to +1V, the output will rise to +1V to make the inputs equal.
Now let’s say the output impedance of the opamp is 10 ohms. If you connect a load of 10 ohms, the output would be halved as you expect. However, the opamp will sense the decrease in output via the feedback connection, and will boost its output until the - input matches the + input of +1V.
Internally, the output driver will be putting +2V into the output impedance. You’ll drop 1V across the output impedance of 10 ohms, and the other 1V across your 10 ohm load.
Does this help?
This really has nothing to do with KiCad, though. It’s basic opamp theory, best discussed in an electronics forum.
1 Like
Why would you want to do this?
The opamp output is a voltage source, just like, say, a car battery. Would you “impedance match” a car battery to the load? Better warn the Fire Brigade first.
The ideal opamp has an output impedance of 0 ohms (try matching that), a real one a few ohms plus a complex frequency dependent impedance rising with frequency (that’s the simple model).
I think you’ve somewhere heard/read about maximum power transfer, where indeed impedance matching is relevant, but that’s for specialised applications (eg, RF circuits), certainly not for opamps.
@RRPollack Thanks for taking time to explain my misconception. Indeed, the output voltage is not expected to be halved. However, there is still an occurrence of saturation, related to both Power supply and Rout, that I cannot reproduce with LTSpice and UniversalOpAmp1.
@ML9104 I was simply trying to build up an experiment where the effect of this output impedance can be observed.
The “simple” way of doing “impedance matching”, which I have used, and works well for audio, is to put a resistor in series with the output of the last stage. In my case, I want to drive audio down a ‘telephone wire’ which is a nominal 600 ohms. So, an op-amp with a series 600 ohms (270+330 works nicely!) into the 600:600 ohm isolating transformer, and to the line.
At the receiving end, a 600 ohm resistor in parallel with the (relatively high input impedance) receiver input terminates it nicely. (You can always tweak things by adjusting the output resistance to give perfect voltage halving!)
Normally, for short runs of audio (under say 20 ft) we would drive low impedance into high-impedance, but for long runs, it’s easy to impedance match.
Opamp models are usually very simple, lacking output current limit and no interaction between load and supply current. Sometimes even lacking output voltage clipping near the rails.
Fair enough. You’ll see the effect in active filters (especially “Sallen Key” types), where a low-pass suddenly turns into a high-pass due to the amp output impedance.
Standard procedure for 75-ohm video. But you forgot to mention that a x2 amplifier is needed before the line.
1 Like
@davidsrsb I think that your comment drove me in the right direction. First, LTSpice’s UniversalOpAmp1 does not implement neither voltage clipping nor current limitation. Second, my experiment was requiring an unrealistic output current. The saturation that I observed with Kicad’s OpAmp is related to the intrinsic output current limitation of the model.
Yes, you need the x2 amplification. One issue with doing it PRE-transmission is that, if you are driving the wires at a high line level, you simply might not have the needed voltage headroom… so you just recover the levels at the receiver.
An advantage of doing it at the receiver is that you can use gain pots to take out the effect of losses in the wire, which can be noticeable over longer distances.
This topic was automatically closed after 89 days. New replies are no longer allowed.