Transistor snob question: BC550c vs. 2n5089 & BC109c for noise?

[Earthscum]

Hey y'all, long time, no see!

Got a question for the transistor snobs   

I'm getting ready to rework my buddy's NYC Big Muff to the specs I built mine. I opened it up and it has 4 BC550C's. You can see the schem for mine below. My question is how the BC550's compare to the 5089's as far as hiss at high gain. My custom DIY is extremely quiet compared to his stocker NYC, but I was wondering what your thoughts or experiences are. Should I just go ahead and swap them out for my spec'd transistors, or not bother and just leave the BC's (obviously, the first stage is getting the darlington no matter what).

Thanks for any input!

[R.G.]

I'm not really a transistor snob, more of a transistor agnostic, but I'll weigh in.

I'm getting ready to rework my buddy's NYC Big Muff to the specs I built mine. I opened it up and it has 4 BC550C's. You can see the schem for mine below. My question is how the BC550's compare to the 5089's as far as hiss at high gain. My custom DIY is extremely quiet compared to his stocker NYC, but I was wondering what your thoughts or experiences are. Should I just go ahead and swap them out for my spec'd transistors, or not bother and just leave the BC's (obviously, the first stage is getting the darlington no matter what).

Step 1: read the datasheets.
http://www.onsemi.com/pub/Collateral/2N5088-D.PDF shows noise figure of 5089 at 2.0db at some very specific conditions.
http://www.onsemi.com/pub/Collateral/BC549B-D.PDF shows noise figure of BC550 at 2.5db at somewhat similar conditions

Both datasheets show the noise for a variety of conditions, indicating that the noise is highly variable, depending on the source impedance and operating current. My conclusion: the differences between the devices are swamped out by the operating conditions.

Step 2: Think.
The noise performance of an amplifier is vastly determined by the first stage, and that's the one stage where you've determined not to change devices. Conclusion: it doesn't matter - squared!

Step 3: Think some more.
You have to get the first stage to be quiet first, then go solve the other issues. http://www.onsemi.com/pub/Collateral/MPSA13-D.PDF shows the noise performance of the MPSA13 and 14. They actually *can* be comparable to the 5089, which is remarkable. But to do so, they have to be used at the right conditions of source impedance and collector current.

So what's the operating current? A quick calculation makes it about 500uA. The charts on page 3 of the datasheet lets you relate the voltage noise and current noise to the collector current - kind of. There are four charts that are all involved. Without doing a lot of math, it looks like that the device is going to have the lowest overall noise figure at about 30-40k source resistance for a 500uA operating condition.

Uh - what's the source resistance? It's the equivalent resistance of the thing feeding the input. If that's a guitar, you're close-ish. Single coils are often about 4-6K resistance, and humbuckers are 10k-18K. A little lower than perfect, but hey, we're just estimating here. From this, I conclude that you're going to get a noise figure **for the transistor** of about 2db, and that's remarkably close to the BC550 and 2N5089.  Conclusion: Pick your device, and go tinker the operating conditions for lowest noise.

Step 3A: resistor noise.
The thermal noise from the resistors at the input adds to the noise contributed by the device itself, which is what we've been waffling about above. Thermal noise is fixed by the resistance. Poor choice of resistor type (i.e. carbon comp) adds excess noise, so don't do that; use metal film. Wirewound is a bit lower than metal film, but I'm assuming you aren't going there. Cooling the resistors also runs down the thermal noise, by the obvious method, but I'm guessing you're not going to dunk this in a tank of liquid nitrogen or put dry ice in the pedal to play it, so there's not much to do there. Conclusion: use metal film for the first stage *especially* the bias resistors touching the base.

Step 4: step back and think some more.
The MPSA18 was designed specifically as a low noise preamp transistor. Looking at http://www.onsemi.com/pub/Collateral/MPSA18-D.PDF for the noise charts indicates that it can have a better noise performance *for the transistor alone* than any of the devices above. Will it be better in the circuit? Who knows? Depends on the operating point and resistor noise.

Overall conclusion: buy an MPSA18 and try it instead of the MPSA13. It might be better. Listen.

Step 5. butbutbutbut the rest of the circuit?
The further to the left of the schematic you are, the more the device type and conditions affect noise. Work your way left to right. Or buy some 5089s and MPSA18s and start sticking them in from left to right, and replacing with metal film in the same direction.

[PRR]

Possibly the same notes R.G. hit but different rhythm.....

> BC550c vs. 2n5089

The '550 is a FAT part, good for higher current and lower source resistances.

'5089 is a skinny part, good for lower current and higher source resistances.

There is a range between where either will be close-enough to no-added-hiss.

As a guide: '550 is used in mike-preamps around 1mA and 200 Ohms. '5089 is used in phono preamps around 50uA and 10K source. I'd say you are closer to the later.

However, usually the *first* stage is hiss-critical. If it has almost any gain at all, its self-hiss amplified will overwhelm the hiss of later stages. Your first stage has gain of maybe 15, which makes all later stages moot. (Only the worst Ge has hiss 15X the hiss of clean Si.)

Darlingtons have special issues. A DIY Darl would allow you to optimize the first transistor's current; not sure what that optimum may be.

Inverter NFB has a real problem in low-low-hiss. 33K out front sets a high-ish thermal hiss level. Lower would load guitar.

You can't always get all that you want without mega-redesign and total tweaking.

And you do have a TON of gain stacked-up, with clipping. That's not a neat problem.

I endorse data-peeking. But for this case, sockets may be your friend.

I have a suspicion that any non-dirty non-sucky transistor will give essentially the same hiss performance.

[PRR]

OK, the MPSA is not wrong for this job.

We find it working near 0.46mA. We have data for 1mA and 0.1mA. I sketched a split-difference curve for about 0.5mA.

Source impedance is un-sure. But clearly not less than that 33K resistor out front. There's probably another 5K-250K in the guitar, say for most conditions 10K-50K.

So source tends to be 43K to 83K. Use a blunt charcoal to mark this area.



We see the Noise Figure is essentially 2dB (1.5dB-2.5dB) all over this area. This is Very Good. If you could find a No-Hiss transistor, hiss-drop would be hardly audible.

Taking out the 33K is "better" but the drop is hardly audible. We are still up against guitar impedance. And simply removing the 33K makes the guitar work into a "virtual earth" (actually ~~~15K) which will load-away the top end of the guitar. (Which may not be bad, except it "depends on guitar" so will be inconsistent as you go through your axe collection.)

If that's not quiet enough, PLAY LOUDER at the guitar. Vol pot on 11, high pups, extra arm-strength vitamins.

[R.G.]

To add a final fillip to this and to further flog an already dead horse, bipolars get noisier if their base emitter junctions get reverse broken even once. Many times is obviously worse.

It would be very interesting to see what happens to noise if only the input transistor (that is, the one most subject to random overvoltages) is replaced with a new one of identical type. Front end transistors on guitar pedals sometimes lead a hard life.

It's sometimes useful to put in a 1N4148 across the base-emitter junction so normal operation back biases it. This adds the diode capacitance to the junction, but in most cases that won't compromise the frequency response much if at all. And it's a slam dunk against reverse breakage.

[Earthscum]

Thanks a ton, R.G. and Paul! I've been out of the game for a while. I couldn't even remember why I picked the darlington for the first stage     I did look at the noise figure charts on the datasheets, but I'll be damned if I could remember how to figure them. I have some other pedals I've been wanting to build... looks like I'm gonna have to start doing my homework again!

I guess I'll start with the first stage and see how much it reduces the noise before I fiddle with the second and third. I do remember picking the BC109 for the last stage based solely on it's subtle change to the sound of the final distortion (inaudible to a majority, probably). I know my coupling cap selection made probably the most impact on overall sound, though, so that will be a definite swap. This NYC has all electros, except for 4 ceramics. Should be an interesting squeeze to get poly's in, lol.

[bool]

Just curious how the transistor swaps turned out - esp. how is the darlingtons noise performance.