Supply Voltage Differences and Bias Points

[Jolly Jimmy]

Greetings all,

How does one deal with the discrepancy between power supplies when biasing a circuit?

I will take the Fuzz-Face-ish thingy I currently have on the breadboard as an example. Q2C is at 5V when supplied with 9.5V. But powered with a variety of adapters and batteries, this bias point shifts by as much as 0.5V either way. This affects the sound somewhat.

Is this a common problem with other circuits? Is there some way of minimising the effect of different supply voltages? By way of limiting the max voltage? Is it down to better circuit design? Or should I pick an average reference and call it a day? And if so, is there a "Gold Standard" of sorts?

Thanks in advance.

[Rob Strand]

There's too many variables to be specific but I can give you a way to think about it.

How does one deal with the discrepancy between power supplies when biasing a circuit?

You try to minimize it by design, or you can use a regulator.  Unfortunately the answer isn't that simple.

If a circuit sounds good with 9V then you cannot expect it to sound *the same* when the battery goes flat say 6.5V.   You don't have 9V so something must change.  A regulator won't help here.   So one way is to design the circuit to work at 5V then regulate the supply to 5V and it will stay 5V until the battery goes flat.  It is rare to do this for most pedals but it is not rare when you design electronic test instruments.

So the next step is compromise.    You design the circuit to sound good at say 9V, where the battery spends most of it's time but you design the circuit so as to reduce variations around that point.   The "best" solution in this case is not obvious.   If your circuit biases at 4.5V (half supply) with a 9V supply you probably *don't* want it to be biased at 4.5V when the battery is flat because it is not half supply.  If the battery is flat at 6.5V then you might want the circuit to bias as 3.25V.  I'm just giving half supply as an example each circuit will have a point where it sounds good at 9V and a point where is sounds good at 6.5V.  The design goal would be to try to make the circuit follow the good bias points throughout the life of the battery.  It is probably unreasonable to expect the sound to be the same at 6.5V and 9V but you try your best to stay on the good bias points.

Adaptors are a different story because the voltage goes up.  If you only regulate when the adaptor is plugged in then you use a crude regulator to reduce the output to be 9V (maybe a bit higher for practical reasons).  Without the adpator it just runs straight off the battery.   Boss have a crude scheme like this.    The other option is to dictate that the adaptor must be regulated so then you don't have to do anything, the supply comes in at the right voltage - the later Boss pedals do this.

[antonis]

Very well said by Rob.!! 

But in real world:

I will take the Fuzz-Face-ish thingy I currently have on the breadboard as an example. Q2C is at 5V when supplied with 9.5V. But powered with a variety of adapters and batteries, this bias point shifts by as much as 0.5V either way. This affects the sound somewhat.

Any Collector resistor of 10% tolerance should result in exactly the same effect.. 

Fuzz-Face-ish thingies are famous/notorious (choice upon your taste..) for such "colourful instabilities"

IMHO, a Fuzz-Face-ish thingy with the world's stiffer PS & huge input impedance & carefully set adamant bias & many other precautions should be considered more closely to distortion amplifier of good design practice..

[Juan Wayne]

Rob summed it up perfectly, you design for as wide a range of variables it'll go through in it's life as you can, and you put a label by the DC plug just in case. That should have you covered 99% of the time.

That said, part of the Fuzz Face fun is that pretty none of that was considered, so it's still a wild little circuit that's incredibly versatile. And what the hell, some people even choose to starve it to get some particular sounds out of it. Others (myself included) choose to have an external bias control to simulate this, while using the most stable power source available.

[Jolly Jimmy]

Thank you all for your thoughts on this matter. I'm new to DIY effects and I've feel like there's so much to be learned from everyone here.

antonis, I didn't even know 10% tolerance resistors were a thing, but I'm using 1% so I don't think that will be a problem.

The main reason this bothered me vis à vis my Fuzzy-Face-ish Thingamabob is precisely because I have a bias control, and that I had carefully worked everything out so that the far extremes of the pot travel provided useable sounds just on the edge of crapping out in both directions. But now I plug in different power supplies and realise that one end isn't extreme enough and the other too much so.

[ElectricDruid]

Sounds to me like you've learned quite a bit about it on your own! Carry on!

T.

[antonis]

antonis, I didn't even know 10% tolerance resistors were a thing,

On Fuzz-Face-ish days, they were THE thing.. 

(the OTHER thing was Ge BJTs reverse leakage current - which you also didn't even know for modern Si BJTs..)

[Jolly Jimmy]

Sounds to me like you've learned quite a bit about it on your own! Carry on!

T.

Hey there, thanks for the encouragement!

antonis, I didn't even know 10% tolerance resistors were a thing,

On Fuzz-Face-ish days, they were THE thing.. 

(the OTHER thing was Ge BJTs reverse leakage current - which you also didn't even know for modern Si BJTs..)

Ah right, I see where you're coming from. When originally producing Fuzz Faces they would just plonk in the components of a certain values regardless of tolerance and leakage and so on, and that was that - hence the stories of people sorting through them to find the good ones.

I'm biasing the circuit "by hand" on a component per component basis so that's never an issue.

I have in fact already had the pleasure of encountering leakage in Ge transistors. A local store had some AC176s, so I purchased a sample of 5 and they were all totally unusable, 500µA + of leakage, complete runaways. Good news is they also had some Japanese Ge transistors, and they were much better. So I have some fab 2SB175s with good gains mostly in the 70 to 130 range, with leakages under 150µA, sometimes as low as 70µA.

In this particular Fooz-Facca-doobery-doo I have Si in Q1 and Ge in Q2 to get some of that Ge goodness, but so as to not have to deal with temperature variance messing around with the biasing.

[antonis]

with leakages under 150µA, sometimes as low as 70µA.

In this particular Fooz-Facca-doobery-doo I have Si in Q1 and Ge in Q2 to get some of that Ge goodness, but so as to not have to deal with temperature variance messing around with the biasing.

Glad to hear of no bias messing around when reverse leakage current doubles every 9-10 ^oC..!! 
(70μΑ at room temperature can easily reach 700μA at +55^oC..)

[Jolly Jimmy]

Yes, it's crazy to watch the leakage readings absolutely skyrocket when simply handling transistors during testing!

[Rob Strand]

Hairdryers are handy for checking thermal stability.  A heat gun is more risky unless it has got temperature control.

[Jolly Jimmy]

Hairdryers are handy for checking thermal stability.  A heat gun is more risky unless it has got temperature control.

I had a go at that when trying to use a reverse biased Ge diode in parallel with the BE junction of Q1 to counteract the effects of temperature variation (have I got that right?). I couldn't get it to work very well so perhaps not, anyway I ended up going the Si route for Q1. Topic for a new thread perhaps.

[Rob Strand]

I had a go at that when trying to use a reverse biased Ge diode in parallel with the BE junction of Q1 to counteract the effects of temperature variation (have I got that right?).

It's a tedious job matching leakage.  You would have to measure each contribution first to have any idea that  you were close.  Just plopping in components is about at hit and miss as you could get.

[antonis]

There was "somewhere" an essay of an Englishman at mid '60s dealing (among other things..) with the amount of Collector-Base reverse leakage current going through Base-Emitter junction, hence amplified by a factor of "β" (although those days they use "a" for Collector/Emitter current ratio..) but I can't recall neither author's name nor essay title/publish..