Designing a PCB- question about CA3046

[somnif]

So, I've gotten bored enough to finally try my hand at making a PCB. I had a question about one component though. I want to use a ca3046/lm3046 transistor array to replace a handful of discrete parts (Its a 6 transistor fuzz with a matched pair octave function, place your bets). But I'm unsure of how to best deal with transistor 5.  I know the emitter has to be tied to the most negative point in the circuit (ground, in my case).

So I have some options:

1) I don't use that transistor. Can I tie pin 13 to ground alone and leave pins 12 and 14 floating? Or do I need the base and collector of that transistor in circuit as well? If so, what would be the best way to handle it without affecting the rest of the circuit.

2) If I do use that transistor, can I have a resistor between pin 13 and ground (as per the original circuit) without toasting the chip?

I know the chip sees a fair bit of use in the synth scene but haven't found specific answers, and my high school level EEE skills aren't cutting it for figuring out the best plan of action. Any opinions are appreciated.

[reddesert]

Superfuzz?
I have no experience with the CA3046 - https://www.intersil.com/en/products/amplifiers-and-buffers/transistor-arrays/transistor-arrays/CA-3046.html - but I see the datasheet says "The collector of each transistor of the CA3046 is isolated from the substrate by an integral diode. The substrate (Terminal 13) must be connected to the most negative point in the external circuit to maintain isolation between transistors and to provide for normal transistor action."

I take this to mean that the substrate is at the anode of these internal diodes and the collectors are at the cathode, that is the diodes are pointing "up" from substrate to collector. If you connected the emitter of Q5 at pin 13 to some voltage, and the collector of any of the transistors sank more than about 0.6 volts below that voltage, the internal diode would begin conducting, and now you'd have the emitter of Q5 trying to source current into the collector of the other transistor. So two parts of the circuit that are normally isolated would be talking to each other.

Could this actually happen?  I don't know, but a multi-stage transistor circuit (like the Superfuzz) often has several phase inversions so it's not totally crazy to imagine that an emitter might be going high while a collector is going low. You might be able to choose where to place Q5 in the circuit to keep this from happening, but it could be simpler to just not use it.

If you didn't use it, you might ground the base of Q5 as well as the emitter. That should keep Q5 from conducting under any circumstances, otherwise it could conceivably cause noise (may be far fetched, but grounding the base is easy).

[alanp]

If you don't mind a bit of SMD, then the BCM847 is a possibility instead.

[somnif]

Well, I've gotten the schematic built, that only took... far too long. Still unsure of what to do with transistor 5, but at least I have the rest of the crap on the page. Eagle is more temperamental than I had thought it would be 

Next step will be to figure out how to puzzle it onto a board layout. Wheeee. Maybe... maybe I should read some tutorials this time...

[Scruffie]

Use it as a diode for different clipping options.

[aion]

Just ground pin 13 (required for the other 4 to work properly) and leave pins 12 and 14 unconnected. You'll only be using 4 of the 5 transistors on the chip, so you'd need two other discrete transistors as well. I recommend using discrete for Q1 and Q2 and then using the array for Q3-Q6, as the discretes tend to be a bit higher gain (a 2N3904 is around 130-180 HFe) and the Superfuzz sounds best with higher gain up front and lower gain at the end.

Also, one practical bit of advice... I tried a NOS CA3046 from Small Bear, the DIP type, and the gains were all right around 40-50 - the bare minimum spec for the chip, but way too low for the Superfuzz. It was sputtery and terrible. Then I tried a SMD LM3046 from Mouser, new production, and the gains were all around 90-120, which was perfect. It tested flawlessly against a vintage Superfuzz I was A/Bing. So the 3046 isn't necessarily the silver bullet for the circuit - it depends on where it falls in the gain spectrum.

[somnif]

Just ground pin 13 (required for the other 4 to work properly) and leave pins 12 and 14 unconnected. You'll only be using 4 of the 5 transistors on the chip, so you'd need two other discrete transistors as well. I recommend using discrete for Q1 and Q2 and then using the array for Q3-Q6, as the discretes tend to be a bit higher gain (a 2N3904 is around 130-180 HFe) and the Superfuzz sounds best with higher gain up front and lower gain at the end.

Also, one practical bit of advice... I tried a NOS CA3046 from Small Bear, the DIP type, and the gains were all right around 40-50 - the bare minimum spec for the chip, but way too low for the Superfuzz. It was sputtery and terrible. Then I tried a SMD LM3046 from Mouser, new production, and the gains were all around 90-120, which was perfect. It tested flawlessly against a vintage Superfuzz I was A/Bing. So the 3046 isn't necessarily the silver bullet for the circuit - it depends on where it falls in the gain spectrum.

Excellent. It was the LM3046 I was building around (I want to start practicing with more SMD) and I used discrete 3904's for 1 and 2. Good to know I can leave 12 and 14 floating though, simplifies the rest of the madness a bit. Thank you!

Still unsure if I want to make clipping switchable, we shall see! (And yeah its a Super fuzz) At the moment I have bat46's+series resistance to emulate the germanium of the original per: http://rezzonics.blogspot.com/2014/01/germanium-diodes-vs-schottky-diodes-for.html . Decisions decisions.