Reducing noise in audio circuits

I

iamjackslackof

Well-known member
Howdy esteemed forumites!

I am planning on jumping back into laying out my own PCBs, specifically an all-in-one Jimmy Page kinda pedal, which will include an EP-3 preamp clone, delays, Phase 90, etc. I am hoping to replace my Chase Tone Secret and Clinch EP-PRE and several delays with it, and as a hopefully permanent board pedal, I want to reduce as much as possible.

To begin I will only work on the EP-3. I'll probably just use Robert's schematic, but since I have full control over the PCB and parts, I can hopefully bring the noise down as much as is practical.

Initially I am only concerned about noise from components, lead dress, assembly, etc. I'll worry about the PCB design and layout noise side of thing later.

Long-ass preamble to my question done, my question is, what are my options for reducing noise, if money is no obstacle?

I know a few things already, like quality parts, shielded cable runs, etc. My plan is to use high quality Dale resistors. I don't mean money is no obstacle completely literally, but for a one-off pedal that I plan on using for a long time, so I will spring for Dales and other fancy parts if they make an appreciable difference in noise. I'm have a Cioks DC7 and a Walrus Audio Canvas for high quality, hopefully quiet power.

Specific questions:
  • My understanding is Johnson noise is less in larger wattage resistors, so should I use 1/2 or even full watt resistors?
  • Lower values resistors have less noise too as I understand it, should I look at tweaking some EP-3 component values (like voltage dividers maybe?) to reduce resistance wherever it makes sense and won't change the circuit operation?
  • Is there a better option than Dale? I was looking at the CMF60 line I believe.
  • What caps should I use?
    • I always use nice Nichicon audio electros in my builds already, should I use film or some other fancy material?
    • Will swapping electros outside the signal path to a different material matter at all?
    • Is there a material for signal path caps that is best for low noise?
  • How can I get the lowest noise from the charge pump?
    • Should I audition the TC1044SCPA?
    • Maybe use a different charge pump IC altogether?
  • What about noise in the power filter section?
    • The EP-3 does some less-common stuff with diodes, can the diodes be optimized for noise?
    • Will adding more filtering help? Maybe a breakout board to add more caps or something
  • The 2N5457 already needs to be within pretty specific parameters, should I audition those for noise too?
  • Would auditioning pots or any other parts help?
  • Perhaps most importantly, how can I measure noise, so I know my mods are helping? Do I need a 'scope or something?
Anything else I'm not considering? I will be combining multiple pedals eventually, some with LFOs and some with charge pumps, so I'm sure I will have more questions later, but this is too long already :)

Thanks everyone!!

EDIT: Jimmy briefly used a Univox Unidrive, which I built the AionFX version of, but I found it too noisy. I'd love to take anything I learn from this thread and apply it to that circuit as well!
 
Last edited:
There's a couple factors here.

For one, low noise isn't really a huge focus in guitar pedals. Guitars are noisy (especially single coils), and a lot of pedals are fuzz which will for a lot of them be noisy no matter what (unless they're gated), so it's not usually worth it to chase low noise.

Second, due to that first point there's not a lot of knowledge on what exactly matters for noise, and how much. In general terms I have the impression that you probably don't need to focus on every point you mentioned, since the noise levels don't just keep adding up that easily, the more noisy things (and earlier they happen in the circuit) will matter much more than what kind of resistors you use later on. But I don't really know.

There are some real gurus around who could probably answer at least most of your questions, but since you're in this deep I think the best course is just to learn the actual science behind the noise. Where it comes from, how can you calculate, simulate or measure it, and start testing from there.

Or in other terms, what makes you think the EP-3 preamp designer and Chase Tone haven't already brought the "noise down as much as is practical"? If you're going to wade into the less practical nitty gritty details, you need to become an expert yourself.
 
I'll bite. I'm not an expert here in the least, but -

Using honkin' 1/2W or 1W resistors is gonna run at odds with your design goals of unifying your pedals in some big uberpedal. Even laying out with SMD components, a Phase90 is gonna fill up a lot of a 125B enclosure, let alone all of the other stuff you mentioned in there.

If you're laying out your own thing only for yourself, why bother with a charge pump? Just power with 18V. Or if you want a bipolar supply, then use one of those fancy traco modules, and maybe even put a LDO on either side of it. (e.g. if you want +/- 9V, go get a traco module with +/- 12V, an 7809 and a 7909 for a super stable 9V). Or build your own fancy power brick.

Charge pumps are pretty noisy (even the ones that run at 45khz or whatever). If noise reduction is your overall goal, getting that away from your audio path seems like a good idea.

You can go with super fancy components if you want, though at the end of the day, physics doesn't GAF if your resistor is from Dale or from UNI-ROYAL (the resistor of kings!), your noise floor is still affected by the resistance itself. Thin film resistors are allegedly superior from a noise standpoint, though. I dunno if the fancy mil-spec ones from Dale are thick film/thin film or what, but they have good marketing.

If you're really chasing the dragon here and trying to shave µV off of your noise floor, you'll probably need an oscilloscope and do measurements rather than qualitative ear tests. Buy a couple of fancy dale resistors and less fancy resistors and cheap resistors and measure them!

Outside the signal path, I don't think any changes are going to help much, certainly not to the point of trying to put 100uf worth of film caps in an offboard box or something. You could look into the hybrid polymer electros which are pretty nice.

You said you didn't want to deal with routing and stuff but honestly bad routing, ground loops, etc are going to inject more noise than anything else. (And honestly this will be super hard to iterate on, since you can't really measure until it's totally built). It's really easy to make
 
+1 pretty much agree with everything @z2amiller mentioned.

diminishing returns etc. on the watt rating for resistors.
you’re not running high currents in a pedal circuit to warrant anything higher than 1/2W or even 1/4W.
i’m not qualified to comment if it’s worth it, but i guess you could get 1/2W metal films in a 1/4W size.

theoretically you could avoid a chargepump with the use of a power transformer with something like an 18v or 15v AC secondary wiring and rectify it to DC, filter it, regulate it, and supply the secret preamp that way, but that’s insane lol. would be quiet though.

another thing to consider is appropriate grounds routing.
you want to avoid loops, and (ime) a star ground scheme is a successful way to go.
 
Last edited:
Plug the whole circuit (or just a section, for starters) into a Spice simulator (I use Tina-TI). Do a noise analysis. Tweak, repeat. This is how I learned about noise optimisation. Just make sure you have good Spice models for your active parts, that take their noise into account.

Here's a litle circuit that could get you started in Tina-TI. CTRL-Alt-N starts a noise analysis.
(you'd have to download the OPAx205 TINA model from TI to get it to work). Note how much the value of Rs affects the output noise.

1774371661283.png
 
Last edited:
It's vaguely related so I'll share an anecdote.

I built the Effects Layouts Sunnbeam a little back, and it was pretty noisy. I thought it might be the random RC4558 chips that were from different places and I wasn't sure if they are all legit. I replaced them with RC4558's from LCSC first (no change), and then with TL072's (no change again).

However, I noticed it was also oscillating at high gain levels, and when I fixed that by moving the different in/out wires further away from each other, the noise level significantly dropped, and it fixed my noise issue completely.

Not really that relevant or useful, but one more point for "routing is probably more important than part choices", at least if you mess up the routing. For part choices, I would definitely listen to JTEX.
 
You must log in or register to reply here.
Back
Top