as I’m still not that sure about what I’m doing, I’d love to hear your opinions on this schematic. I’d like to build an input module that allows me to amplify line-level/instrument signals to modular level. I’m specifically thinking about some drumcomputers of mine with stereo outputs, which is why the module should also be able to sum/mix the stereo signal to mono (hence the inverting opamp design). Furthermore I added LEDs as rough indicators of the correct level, but mainly because I like blinking lights
Is there something that could be improved or was overlooked? E.g. do you think a circuit like this needs ac-coupling? The instruments I tested this with seem to output only ac and the two LEDs (showing positive and negative peaks above 5v) light up pretty evenly, so I didn’t include it. Some designs I looked up seem to include it though…
Another thing I was wondering about: When I amplify a drumcomputer to 10Vpp at max, it is still considerably more quiet then, e.g. a steady oscillator-signal at 10Vpp. This makes sense because a drumtrack inevitably has some higher peaks (mostly the bassdrum), while the average volume will be more quiet. So should I go for a higher maximum amplification or should I attenuate all the other signals in the mix?
It includes some filtering and a stabilization cap, and, yes, AC coupling. It’s a non inverting design, so less suitable if you need to mix two inputs, but it may have some design features you’d want to incorporate.
Non inverting is more vulnerable to out of range input voltages so the AC coupling is more important. I’d be inclined to AC couple in any case, though, just to be safe. But I’d use nonpolarized electrolytics.
A few questions/comments:
Why are you using 10k input resistors? 100k is generally the standard input impedance for synth modules. Is there something about your inputs that necessitates drawing more current?
If I understand this diagram correctly (it’s Falstad, I loathe the Falstad diagrams) your maximum gain is 60k/10k = 6 on each channel. Stone’s is (1+10k/470) = 22. I suppose the drum machine has larger outputs than a guitar so maybe that’s all right. But if there’s a possibility to use this with a guitar or something similarly low level you might want to boost the gain. (For a guitar input you’d definitely want higher impedance, see the 1M resistor in Stone’s design.)
Why 100k input and feedback resistors on the second op amp stage? Here you don’t have input impedance to worry about, and using 10k will lower your noise floor.
For the LEDs, if you want them off below 5 V and on above, then that’s more or less the way to do it; if you want an indicator of how big the signal is, you could instead put the LED in the feedback of an op amp to make the brightness approximately proportional to the input voltage. I guess you used 10k and 14k to get a 5 V threshold, but 14k resistors are not E24 values so you’d have a harder time finding them and you’d pay too much for them. But 3.3k and 4.7k would give you 12(3.3/(3.3+4.7)) = 4.95 V, surely close enough and those values are cheap and easy to get.
If you need 2+ inputs, then perhaps have each of those (ac coupled and with a suitable input impedance) inputs buffered with an op amp, followed by a non inverting summing amplifier (with a suitable gain)?
Something else I thought of. For the LED comparators, when they’re below threshold, they’re reverse biasing the LEDs with -12 V. That’s not good; typically LEDs have a maximum reverse voltage spec around 6 V. I noticed a similar problem in a YuSynth design and addressed it by adding a 1n4148 antiparallel to the LED.
Either attenuate the rest or use a compressor on the drum sound.
I built an adaptation of the Ken Stone Stomp Box adapter because I wanted to be able to choose the amplification and attenuation factors. I’ve posted about this before, but weirdly at the moment I can not find the post. So here are the schematic and a pic once more. I added a 9V power output for an fx-pedal and used jumpers to set the amplification and attenuation factors.
Note that in the full Stone circuit there are two of those amplifiers — one for the tip and one for the ring. They go to tip and ring outputs, but they could also go to a non inverting summer. Though in that case I’d make the amplifier unity gain, and then boost the signal at the summer.
Am i right to assume, that the 10uF cap + the 150k to ground in the input section are the ac-coupling and biasing to 0V part? Could you explain what the 1M and 10pF to ground do?
I had a look at line-mixer schematics and 10k input resistors seem to be common. Is there a reason to use 100k resistores if these inputs won’t be used for modular signals?
They definitely have higher outputs, but maybe I will adjust this just to be more flexible with what I feed into it. I just don’t want the usable range of the pot to be too small before it goes into clipping so maybe I’ll add a boost switch or something like this.
Sam and Moritz Klein use this combination for the second op amp stage of their mixer designs. Since this amplifier is basically a mixer like this modded for a different input signal and adjustable gain, I figured that I could’t go wrong with these values.
Yes, that’s the way I want them to operate, but it’s good to learn how to set them up another way. But I will definitely follow your advice with the antiparallel diode you mentioned!
Other than this being the cleanest way to do it, would you say that for this application (summing stereo to mono) it’s particularly important to have the inputs buffered? I think otherwise I could do without the buffers, since I’m not aiming for high-end here
1M establishes the input impedance; as I mentioned above, for a guitar input you often see 1M impedance.
10 pF, some kind of filtering, I presume. 10 pF would be seen as a short to ground by high frequencies — I guess it’s an impedance in parallel with 150k so would filter frequencies above 1/(2πRC) = 106 kHz, so high frequency noise well above audio range.
The 10 µF similarly is a high impedance in series with 470R for low frequencies so filters below 34 Hz if I’ve done the calculation right.
Good point, this is not for synth inputs so doesn’t need synth standard impedances, and I don’t know much about line level inputs but it does look like 10k is not unusual.
Thanks a lot! The article you have linked seems very informative and I will surely refer to it more often in the future.
I’ve implemented some of the suggested improvements and think I’m almost ready to finally build this thing. One thing I’m still confused about is the appropriate values for the ac-coupling capacitor and resistor. According to the Doepfer DIY page the (minimum) capacitor-size can be calculated by this formula: „C ~ 1/(R*f) with f = lowest frequency, R = in/output resistance“.
How do I know my input impedance? Is this equal to the size of the input resistor, so 10k? If so, the formula would give me a capacitor size of about 2uF, assuming a targeted highpass cutoff at 50Hz. This seems a bit on the low side in comparison to some schematics I’ve looked at.
And I couldn’t find any information on how to choose the size of the resistor to ground. Does this matter at all?
The input impedance (resistance really, for our purposes) is the resistance between the input point and ground or other voltage source. If it’s connected to a non inverting amplifier input that’s a huge resistance to ground, so usually it’s best to have something like (for a synth signal input) 100k or (for guitar) 1M to ground before that. If it’s connected to an inverting amplifier input that’s a virtual ground, so you’d have the 100k or 1M or whatever input resistor between the input point and the op amp input.
For AC coupling the capacitor is a huge resistance to ground for DC signals, so again the impedance would be a resistor to ground before that. The 1M in the Stone circuit above, for instance.
For a non inverting configuration the AC coupling would be a capacitor in series on the input followed by a resistor to ground, forming a high pass filter with cutoff frequency 1/(2πRC). The 2π is important, if Doepfer left it out they shouldn’t have. You probably want this frequency to be below audio but not by too much. In the Stone circuit C = 10 µF and R = 150k, so the cutoff is 0.1 Hz. Note this is NOT the resistor that’s setting the input impedance in this case.
For an inverting configuration you’d need to add a resistor to ground before the capacitor, to set the input impedance. Then the capacitor and the input resistor to the op amp, and the values of those set the cutoff frequency. If you aim for a 1 Hz cutoff then, using a 10k resistor, C ~ 1/(2πfR) ~ 15 µF — or 10 µF would be fine.
50 Hz would be too high a cutoff frequency, I’d say — it’s not a sharp cutoff, so there would be some attenuation up to 100 Hz or so. I’d aim for more like 1 Hz to 10 Hz. Stone’s 0.1 Hz seems excessively low to me. The inverse of the cutoff frequency is about how long it takes to respond to a DC offset, so a 0.1 Hz cutoff means the DC offset isn’t eliminated until on the order of 10 seconds after it starts.
So the resistor to ground doesn’t figure in the equation and I can use whatever? So say a 100k resistor to ground followed by a 1,5uF cap and a 10k resistor would give me proper ac-coupling with a HPF cutoff frequency of around 10Hz?
Edit: All schematics for summing amplifiers that included ac-coupling I could find spare the resistor to ground, but usually have a variable resistance to ground via the input attenuator.
Of course After reading some more and doing some experiments I’m still somewhat confused about this stuff, but I think I’ll leave it for now and probably revisit this subject another time. Thanks a lot for your detailed explanations!