Capacitor selection for Power stage ripple filtering (TCE Spark Booster)

[matopotato]

Looking at the TCE Spark boost I found two schematics Ember and Arsonist. They both use the same capacitor filtering in the power section 100uF and 100nF. I assume this is to filter out DC ripple.
Are there any other reasons for these caps?

Then I found one layout of the pedal here. And the power caps are now 220uF x 2 and 100nF x3.
What is the hope to achieve by 4 times more on the electrolytic one and 3 times on the film cap compared with the schematics?

They way they are put together, you would have 440uF plus 300nF, which is 440,3uF.
What does the 300nF (or the 100nF vs 100uF in the schematics above) bring?
From capacitance perspective it would be "lost in the tolerance" I would have thought, so there must be some other feature or difference between electrolytic and film when it comes to power filter.

[R.G.]

There are some subtleties involved. The caps are probably there to suppress DC noise as you guess; the problem is that the pedal designer has no idea what power supply will be connected. Power supplies may be low frequency transformers with rectifiers but no filter caps, all the way to 1MHz switching devices, so the ripple may be big, and of any frequency. The big electrolytics have lots of capacitance, but also bigger self-inductance and internal resistance (ESR). The smaller capacitors are lower capacitance, but lower inductance and resistance, so they're better at filtering high frequencies, like from switching power supplies.
There is another reason for power filter caps is that circuit pull power in pulses themselves, and interact through the DC power supply impedance. Putting filter caps on the power supply lowers the AC impedance of the power supply to prevent this - although that's not a common issue in smaller pedals.

[Rob Strand]

They way they are put together, you would have 440uF plus 300nF, which is 440,3uF.
What does the 300nF (or the 100nF vs 100uF in the schematics above) bring?
From capacitance perspective it would be "lost in the tolerance" I would have thought, so there must be some other feature or difference between electrolytic and film when it comes to power filter.

The main point with bypass caps is: 3 x 100nF caps spread around the board is *not* always the same as replacing then with a single 300nF cap somewhere on the board.  Tolerances are not a consideration.  This is even more the case when the 100nF caps are required.  Some datasheets make a point of highlighting the need for such caps.  The chips require such caps to be located right at the (power) pins to be effective and placing them elsewhere completely defeats their purpose.

There's many reasons why.  For analog chips it is usually to help prevent oscillations and or signal integrity.  It can also help HF noise from the PSU getting into the audio but that's more a special case.   It's normally "better" to have the caps than not.

On professional boards some designers, or companies, use such caps whether they are required or not to avoid having to find and debug weird circuit issues related to poor supply bypassing.  However, on small boards you can often get away without them.   Look as Boss pedals they don't bother with 100nF caps on the analog circuits.  They only put them on the digital or switching devices.

Look at page 7 of this datasheet.  They are recommending caps but not in forceful manner: so it's a good idea but maybe if your bypass caps a good you won't have problems.  Note also it says the caps need to be adjacent to the supply pins.  Other chips pretty much say if you don't use bypass caps near the pins you will have problems.
https://au.mouser.com/datasheet/2/308/1/MC33171_D-2315863.pdf

As far as the circuit examples you gave I'd say the 100nF caps are arbitrarily used and placed.   Look at the TC Electronic Spark boost layout. The caps aren't located near the chips.   So it's hard to justify having 3x100nF arranged as they are.  There's also the possibility the original PCB had them correctly placed and that has been lost translating to the layout.  When the caps are kind of bundled together like that you could use a 300nF or simply use a single 100nF as a token measure.  If the caps were correctly located near the chips then you should be a lot more hesitant about changing them since maybe they are there for a reason.

[matopotato]

Thanks a lot for your replies! It clarifies quite a few things for me 

For the build I was not thinking of using the MC33171, instead I am considering OPA2134. It is a bit more expensive than the TL072 that are suggested in some BOMs, but might provide a cleaner sound.
In the datasheet
https://www.ti.com/lit/ds/symlink/opa2134.pdf
I see that there is a recommendation as you described for 10nF ceramic caps. Chapters 9 and 10.
Which is not implemented in the different schematics/layouts I have come across so far.
Not sure though how much effect it would have to add them or leave as is.

Interestingly enough, I notice R.G. is already there. In the datasheet.... (Ch 10.2 p 19) 

[amptramp]

You can find DIP sockets with internal capacitors here:

https://www.advanced.com/products/ic-socket-adapters/decoupling-capacitor-dip-sockets

and here:

https://www.e-tec.com/v5/products/precision-ic-socket/precision-dip-ic-sockets-tht-through-hole/precision-dip-ic-sockets-wired-capacitors/

I have also seen (but not found on the internet) thin capacitors in a DIP format that were meant to be installed under an IC socket.  These were metalized polyamide film with one capacitor plate on the top and the other on the bottom.  There was a covering layer that prevents anything from shorting out against the board.

This represents some "best effort" solutions to getting decoupling close enough to the power leads ot capture high frequencies without excessive series inductance.

[matopotato]

While on the topic of TCE Spark Booster and caps:
I notice on both the Ember and Arsonist schematics that the cap C3(ember)/C4(arsonist) (green circle) has flipped polarity in the Layout:
.
C9 (blue) has also reversed polarity.
How would that affect this type of build?

Also C9 (still blue) is marked 10uF in the layout along with C11 (red) also being 10uF, but 1uF in the schematics. I hold the schematics as being the "proper" values (and orientations).
How would 10x expected value play out in this circuit?

I can of course try this out myself when I have the parts and can breadboard it. I am just curious as apparently people have built according to the layout and were happy with the result.

[R.G.]

While on the topic of TCE Spark Booster and caps:
I notice on both the Ember and Arsonist schematics that the cap C3(ember)/C4(arsonist) (green circle) has flipped polarity in the Layout:
.
C9 (blue) has also reversed polarity.
How would that affect this type of build?

It's a bad idea that you would mostly get away with. They're using polarized electros in DC circuits, perhaps with the expectation that there won't be any noticeable DC offset. Aluminum electros can withstand a volt or so of reverse bias without immediately dying, but it's not good for them, and they're most stable if actually polarized by a good fraction of their rated voltage. They go bad faster and change value a lot if not polarized with a DC voltage across them. As I said, you would probably get away with it, perhaps for quite a while. Well, OK, long enough for the warranty to run out. 8-)
Better to use bipolar/non-polar types.

Also C9 (still blue) is marked 10uF in the layout along with C11 (red) also being 10uF, but 1uF in the schematics. I hold the schematics as being the "proper" values (and orientations).
How would 10x expected value play out in this circuit?

C9 drives a load of 22K up to 122K. With 1uF, the low frequency rolloff is no lower than 1/(2*pi*22K*1uF) = 7.2Hz. With a 10uF, the rolloff is 10x lower - 0.72Hz. Neither matters to guitar (with low E = 82Hz) or bass (low E = 41Hz). You might try replacing C9 with 0.22uf. That should be good down to 1/(2*pi*22K*0.22uF) = 33Hz.
C3 seems to be bootstrapping the bias to get higher input impedance. It's value won't be critical if it's bigger, or modestly smaller. It might be OK with a 1uF film cap, almost certainly with a 10uF bipolar.

[Rob Strand]

I started writing this before RG posted but I got a long phone call.

While on the topic of TCE Spark Booster and caps:
I notice on both the Ember and Arsonist schematics that the cap C3(ember)/C4(arsonist) (green circle) has flipped polarity in the Layout:


C9 (blue) has also reversed polarity.
How would that affect this type of build?

Also C9 (still blue) is marked 10uF in the layout along with C11 (red) also being 10uF, but 1uF in the schematics. I hold the schematics as being the "proper" values (and orientations).
How would 10x expected value play out in this circuit?

Everything seems consistent until we get to the Dirtbox Layouts layout.   That layout has no 1uF caps at all.  Tracing the layouts does my head in at the best of times never mind this time of the morning.    So I'm assuming they have all been changed to 10uF.   Based on the schematic the change would be minimal.

Those cap positions don't have much DC bias so you can swap the polarity.  With JFET input opamps there's little reason to argue either way.   The MC33172 has bipolar inputs so you can argue there is a correct polarity.  (I haven't checked if the polarity does follow that argument though, there's a chance they don't comply with that reasoning, even so the DC bias is usually low enough not to cause polarity issues.)

[amptramp]

It's not that difficult to find 10 µF film capacitors these days and values between 1 µF and 10 µF are even more common.  It may be true that an electrolytic capacitor can tolerate small amounts of reverse bias but do you really want to take the risk of designing a failure mode into your circuit?  You can get non-polar electrolytics in 10 µF values as well, so there is really no reason to use polarized electrolytics other than board space, availability and the few cents more cost of items that are suited to the design.  Most of us here are not manufacturing large quantities of anything, so saving a few cents per unit is not a consideration.

[matopotato]

... They go bad faster and change value a lot if not polarized with a DC voltage across them. As I said, you would probably get away with it, perhaps for quite a while. Well, OK, long enough for the warranty to run out. 8-)
Better to use bipolar/non-polar types.

I am a bit surprised that it seems that TCE use them there then. Both schematics seem to concur.

[matopotato]

I started writing this before RG posted but I got a long phone call.

While on the topic of TCE Spark Booster and caps:
I notice on both the Ember and Arsonist schematics that the cap C3(ember)/C4(arsonist) (green circle) has flipped polarity in the Layout:


C9 (blue) has also reversed polarity.
How would that affect this type of build?

Also C9 (still blue) is marked 10uF in the layout along with C11 (red) also being 10uF, but 1uF in the schematics. I hold the schematics as being the "proper" values (and orientations).
How would 10x expected value play out in this circuit?

Everything seems consistent until we get to the Dirtbox Layouts layout.   That layout has no 1uF caps at all.  Tracing the layouts does my head in at the best of times never mind this time of the morning.    So I'm assuming they have all been changed to 10uF.   Based on the schematic the change would be minimal.

Those cap positions don't have much DC bias so you can swap the polarity.  With JFET input opamps there's little reason to argue either way.   The MC33172 has bipolar inputs so you can argue there is a correct polarity.  (I haven't checked if the polarity does follow that argument though, there's a chance they don't comply with that reasoning, even so the DC bias is usually low enough not to cause polarity issues.)

Yes, the layout and the schematics it says to be based on do not match.
But since it was verified I am guessing the impact was not much in the end as you all have explained why.

The Dirtbox Layouts layout use OPA2134 though (The MC33172 are for the other layout with the extra caps)
the datasheet indicate that 10nF could be used close to the pins instead of any far off 100nF etc as in the other layout.

[matopotato]

It's not that difficult to find 10 µF film capacitors these days and values between 1 µF and 10 µF are even more common.  It may be true that an electrolytic capacitor can tolerate small amounts of reverse bias but do you really want to take the risk of designing a failure mode into your circuit?  You can get non-polar electrolytics in 10 µF values as well, so there is really no reason to use polarized electrolytics other than board space, availability and the few cents more cost of items that are suited to the design.  Most of us here are not manufacturing large quantities of anything, so saving a few cents per unit is not a consideration.

Thanks, I was more considering the impact given that the orientation was different in the Layout vs the Schematics.
pedalpcb made the Ember I assume based on tracing the Spark original, so I am hoping their orientation is ok to use.
Or I might dig up bipolar electrolytes or Film caps for those three locations. AS you say the cost is small compared with any agony of having to redo the work later.

[matopotato]

I noticed in both schematics that R3 is 10k2. I am trying to understand what difference and impact that 10k2 will have vs a 10k resistor?
If it would have any impact on RC filtering, then 2% would probably not be enough for me to tell. Or does it have other purpose(s)?

[antonis]

IMHO, due to a bunch of them sitting on factory warehouse shelves..
(same stands for R2 (4k3) resistor..)

I'd use any handy E12 value close to specified one (e.g. 3k9 or 4k7 for R2)

[ElectricDruid]

+1 agree with Antonis. I can't see any conceiveable reason why those values should be so strangely specific. There's no *need* for them to be exact.

[Rob Strand]

Some filter circuits use odd values because that's the right value and it isn't a big deal buying funky values these days. 

However, putting things in context if you plug a guitar directly into the input the variations in output impedance between guitars is going to make a mockery of such precise values.   Back off the volume control and then it's really "pot luck"  what the output impedance is.

[ElectricDruid]

What *is* that input stage, anyway? It's not something I recognize.

Initially I thought it was a MFB filter (it does look a bit like one). But the weird part is that it's got the +ve input biased to Vref, with the op-amp set up as a unity gain buffer, but the output is tied to ground with 10K. What's going on with that?

[matopotato]

What *is* that input stage, anyway? It's not something I recognize.

Initially I thought it was a MFB filter (it does look a bit like one). But the weird part is that it's got the +ve input biased to Vref, with the op-amp set up as a unity gain buffer, but the output is tied to ground with 10K. What's going on with that?

No idea myself. Just assumed the schematics being the same that it is ok, or as tce made the original..

[antonis]

What's going on with that?

Could it make more sense if you omit C2..??

[ElectricDruid]

What's going on with that?

Could it make more sense if you omit C2..??

Not really, no. It makes it less like a filter - that much is true. But there'd be no point having R2+R3 needlessly in series, and it doesn't fix the problem with the output being tied to ground when the input is tied to Vref.

Unless you're seeing something I'm missing?