Mtm radio music

Hey, I am currently trying to make a KOSMO version of the radio music. I know, I know, teensy 3.2 is impossible to get at the moment, but I still have one here. :wink:
I have a question though: When looking at the schematic:

I see that there is a separate GND and DGND, where DGND. DGND is highlighted in the schematic. At the teensy DGND is connected to teensy-GND and GND is connected to teensy-AGND. The 5V regulator is connected to DGND and the CV-in protection diodes are also connected to DGND. It is not used anywhere else. Can anyone explain the reasoning behind this? My first guess would have been that the 5V regulator (78L05) should not even work when its gound is not connected to the same ground as the input voltage, but apparently I am wrong xD

The choice of 22k resistors for the CV inputs and 220 ohms at the output seems also a bit unusual and for the reset (trigger) I would rather use a transistor, but whatever…

I am mainly confused by the different grounds, maybe someone has an idea?

AGND is analogue ground for the analogue inputs, if you are reading pots or analogue signals connect the ground to here. They are internally connected, I recently did some work with Teensy 2++ and I found the pots were not read correctly until I connected the AGND to them.

1 Like

The CVs connect to the op amp non inverting inputs, so current normally will not flow through the 22k resistors, so their value normally won’t matter. I think they don’t do anything unless there’s an overvoltage on the input, then they help protect the op amp. The input impedance is set by the 47k resistors, and I don’t know why those aren’t the more usual 100k, but close enough.

220R for the output resistor is low. If the output jack is for instance shorted to 12 V while the op amp is putting out 0 V, then the current is 12/220 = 54 mA and the power is 12^2/220 = 0.65 W, which is more than twice the 1/4 W your resistor is likely to be rated for — it could burn out. With a 1k resistor the power is 0.14 W which is safe. I’d change it.

1 Like

Thanks! I will use 1k at the output then, and maybe switch 48k to 100k for input. I noticed that I can not use the usual transistor gate in, because then it inverts the hat signal.
I still don’t understand the digital ground thing, but I think I will just copy the original there…

1 Like

Normally the Analogue and Digital grounds are connected together at a single point in the circuit, possibly with some filtering between them to prevent the noise created by the digital circuit from being injected into the analogue circuitry.
In the Teensy 3.x we cans see in https://www.pjrc.com/teensy/schematic.html that the two grounds (GND=VSS and AGND=VSSA) are indeed connected together at a single point with a ferrite filter between them. (Green annotation.)

As the two grounds are connected internally on the Teensy board, the 78L05 will work as connected, but its GND pin really should be connected to the GND pin of the Teensy as the 5V it produces serves to power the whole Teensy, noisy digital circuit and all, through the Vin pin which is referenced to the global (AKA Digital, AKA VSS=GND) ground. (Red annotation.)

EDIT: I used the 3.0/3.1 schematic for my analysis above, but the 3.2 is similar.

1 Like

I watched a bit of this:

And I think I will just go with a single GND and GND plane and try to place the analog parts away from the SD card reader (or at least route the signals not too close to one another). :slight_smile:

I was also wondering about C2. This is apparently for AC coupling, but there is not resistor to GND after C2, so probably the input impedance of the opamp is deciding the cutoff frequency then? Any reason for not adding a resistor or just forgotten?

I’ve been looking at that and trying to figure it out… After C2 is an active low pass filter with a cutoff about 19 kHz, but C2 appears to interact with that filter and mess things up. In LTSpice if I put in a sine wave with an offset (blue curve) with the circuit as shown the output of the low pass filter tries to be a sine centered on the negative rail (green curve), but if I add a 100k resistor to ground after C2 I get a sine centered on 0 V (red curve). Either way the output is the same regardless of the offset on the input. So I’d say yeah, C2 without a resistor is a mistake and one should be added.

1 Like

Eventually I remembered this post from Matthew Skala and I think probably this is related to what’s going on:

If the op amp isn’t there then the capacitor without the resistor blocks DC, but put in the op amp (even just a simple voltage follower, eliminate R3, R4, C1, C4) and it misbehaves. The resistor to ground prevents the op amp bias current from charging up the capacitor.

2 Likes

This sounds like it’s related, but they talk about C4, not C2…

1 Like