Yeah it’s probably not the best example to show as it’s more of a follower than a summer in configuration. But if it’s good enough for sequential then I guess it’s good enough for anyone. But it was purely to show how a front panel maps to an actual schematic.
I just referred to Sonusus’ op-amp cheatsheet and I see that it is a follower. I was re-reading Ray Wilson’s make: analog synthesizers really late last night but I recall non-inverting being rarely useful. I love how the next morning I see how someone has made use of that. But I do see how it works practically in a schematic, especially in contrast to the other inverting summing amplifiers I thiiink I saw in some schematics.
Bearing in mind what analogoutput said about fixed architecture and having enough power with a non-inverting config, I’m wondering if there is an advantage to choosing that in different applications.
savt22, what are the odds you have the panel layout in any form or fashion for that poly 6? Either a pic or a sketch or anything?
If you just need to buffer a single signal, which is not going to be below -8 V, especially one that isn’t coming from an unknown source (like another module in a modular system), then a non inverting follower is much simpler than a pair of inverting stages.
But a non inverting topology exposes the op amp input to the input voltage, and if that can be outside the power supply rail range it can damage the op amp, and for a TL07x (with ±12 V power) if the input voltage goes below -8 V it can produce strange behavior (phase reversal).
And if you have several signals going through resistors to a summing node at the non inverting input, and you disconnect one of the signals (without reconnecting it to ground — which is done in the schematic above), then the other signals’ amplitude in the output jumps upward — because the output is proportional to the average of the signals present.
With an inverting topology the op amp inputs are at 0 V, so no overvoltage and no phase reversal. And with a summing node, if an input is disconnected it doesn’t change the amplitude of the other signals in the output, because the output is proportional to the sum of the signals. But if you need to preserve the phase of the input(s), you need to add a second inverting stage to restore that.
Both are useful, but I tend to prefer inverting topology for summing unless the inputs are fixed and their average is known to stay within the -5 to +10 V range.
Do you just mean the front panel? I went very DIY.
This is what I ended up with. It’s going to get a couple of extra knobs when I replace the LFO soon. The two knobs far left are the mode and tuning knobs for the Pico, they don’t quite match up as added them after.
The PW controls the amplitude of the PWM LFO modulation, which I’ve done incorrectly. The LFO switch changes between off, full or half amplitude for the LFO going to the filter.
The 1G switches the filter envelope between the OR gate output, no gate, and the first notes’ gate only. The latter option actually has quite a cool effect in the stacked voices mode, as the filter envelope triggers every third note which adds this funk sort of feel to fast sequences.
It’s an 84HP 3U panel.
I’m actually in the process of redoing this all on PCB with 6 filters as my brother wants one. I’m a way off getting all the designs done though.
Awesome and inspiring, thank you for sharing. I am probably going to spend an absurd amount of time on that 2nd photo, but both are helpful for me making sense of putting the pieces together. Many thanks
No worries at all. Just go one bit at a time, that’s what I did. Start with power, then the pico board. Build the oscillator board, connect to the pico, test it… if it works, build all the oscillators… then one envelope generator, test, etc.
That way it will feel less overwhelming and the connections will make sense.
I’ve just had a peek at the MFOS power supply design. 3x3300uF on each side strikes me as rather a lot of capacitance, especially through half-wave rectifiers. On switch-on the capacitors will charge for the first time, and that will result in a current spike that a wall-wart transformer may not be happy with. It depends on the transformer rating.
I typically use 4700uF and full wave bridge rectifiers and dual output transformer, and over-rate a fair bit for the anticipated load. But that’s just me, perhaps I’m too cautious. Decent heatsinks on the regulators help too. (But you can’t find 'em for 78Lxx’s)
The MFOS board dos not make adding heatsinks easy, unless the regulators are mounted vertically. And don’t let the heatsinks short together - they will be at different voltages unless insulated from the regulators.
MFOS has an interesting note on the 79xx regulators - they need a 3mA load before they start regulating. I’ve not looked into this but there is no harm in spending 2p on a resistor of about 3K9 or 3K6 across the -12V output - or maybe use a resistor in series with a LED which will also serve as a power indicator.
With LM78xx’s and LM79xx’s, make sure to add a 330nF ceramic across the input and 100nF on the output close to the regulators to avoid them oscillating. Oddly, MFOS recommend 1uF electrolytics, I think ceramics are better for the job.
My version uses heat sinks for the vertically mounted TO-220 regulators, and resistor+LED loads as indicators and to guarantee a load for the negative regulator.
The TI datasheets make different recommendations for LM7812 and LM7912:
LM7812:
LM7912:
I don’t think I’ve ever noticed the latter remark before.
I took the non-electrolytic values I gave from the STMicroelectronicsI and Fairchild datasheets, in their ‘typical applications’ examples.
There’s only one way to settle this - Jello wrestling between TI and Fairchild reps.
Looking at the PCB designs for synth PSU’s from several suppliers, I doubt the designers have read the regulator spec sheets properly either.
Well now you got me curious, so I went and looked also at the ST and OnSemi datasheets and tabulated their recommendations:
Regulator caps
Positive regulators
Input | Output | |
---|---|---|
TI LM7812 | 0.22 µF Required if the regulator is located far from the power supply filter. |
0.1 µF ceramic Although no output capacitor is needed for stability, it does help transient response. |
ST L7812A | 0.33 µF it is recommended that the regulator input be bypassed with capacitor if the regulator is connected to the power supply filter with long lengths, or if the output load capacitance is large. … A 0.33 μF or larger tantalum, mylar or other capacitor having low internal impedance at high frequencies should be chosen. |
0.1 µF Although no output capacitor is need for stability, it does improve transient response. |
OnSemi MC7812 | 0.33 µF it is recommended that the regulator input be bypassed with a capacitor if the regulator is connected to the power supply filter with long wire lengths, or if the output load capacitance is large. … A 0.33 mF or larger tantalum, mylar, or other capacitor having low internal impedance at high frequencies should be chosen. |
— |
Negative regulators
Input | Output | |
---|---|---|
TI LM7912 | 2.2 µF Required if regulator is separated from filter capacitor by more than 3′′. For value given, capacitor must be solid tantalum. 25μF aluminum electrolytic may be substituted. |
1 µF Required for stability. For value given, capacitor must be solid tantalum. 25μF aluminum electrolytic may be substituted. Values given may be increased without limit. … The bypass capacitors, (2.2μF on the input, 1.0μF on the output) should be ceramic or solid tantalum which have good high frequency characteristics. If aluminum electrolytics are used, their values should be 10μF or larger. |
ST L7912AC | 2.2 µF CI is required for stability. For value given, capacitor must be solid tantalum. If aluminium electrolytic are used, at least ten times value should be selected. |
1 µF CO is required if regulator is located an appreciable distance from power supply filter. To improve transient response. |
OnSemi MC7912 | 1.0 µF it is recommended that the regulator input be bypassed with a capacitor if the regulator is connected to the power supply filter with long wire lengths, or if the output load capacitance is large. … A 0.33 µF or larger tantalum, mylar, or other capacitor having low internal impedance at high frequencies should be chosen. The capacitor chosen should have an equivalent series resistance of less than 0.7 W. |
Bypassing the output is also recommended. |
I think it looks like ST’s comments about (but not values for) CI and CO are switched around.
The ST L79 datasheet specifically shows a dual supply with much different capacitors on the positive and negative sides:
Similarly from TI:
On the other hand OnSemi shows a more symmetric design:
I was about to do this sort of search of data sheets myself so you have saved me a lot of trouble, Many thanks.