Droppin' some volts and oscillation

[John Lyons]

So I'm using this arrangement.



(outlined in smallbears Maestro FZ-1 project)
http://www.smallbearelec.com/Projects/FuzzE-One/FuzzE-One.htm
Third schematic down.

I'm using it in a Sunn Buzz circuit that takes 1.5v
The power supply oscillates pretty heavily, thump thump...
When I change the bypass cap to 320uf and the series resistor to 15k
I can pretty much stop the oscillation but I don't think this is the problem...
more the band aid.

So at the risk of showing my fundamental lack of knowledge here...
What's the issue? I've tried this in a couple 1.5v circuits with the same
result.

[R.G.]

That circuit can't oscillate. It has no gain. It requires the collusion of the load to provide the gain.

What is probably happening is that the load is going into oscillation from the non-zero impedance of the supply.  The diodes form a shunt regulator, and their voltage goes up when the circuit's load current goes down; their voltage drops when the circuit's current use goes up. If this drives the circuit into some shaking of it's own, they could jointly thump. Unusual, but not unheard of. How much current is involved?

Is it possible that the load is sucking too much current and dropping the diodes out of conduction?

Is it possible for you to convert to either a Vbe multiplier or to a series regulator instead of a diode shunt regulator?

[Gus]

John
here is a circuit idea.  I have not built it yet.  Select a LED with a greater voltage drop than what you want for a supply voltage.  C1 >= 10uf should work.  Maybe add a 100uf cap at the 9VDC supply.

[John Lyons]

RG
Haven't measures the current but it's basically the same 3 transistor circuit as in the Smallbear project.
It does cleat up when playing, then fades with the signal into Oscillation.
Can't imagine it pulls more current that the sample circuit but I guess so.
With the added resistance and more capacitance is there any downfall in that approach?

Gus
So this is basically a current buffer you have shown, yes?
Does the 1m Pot fine tune the voltage?

[Gus]

Yes the 1 meg adjusts the voltage to the opamp buffer.

Question have you tried the circuit with an 1.5VDC AA, C or D battery?

[John Lyons]

Gus
The circuit works well off a AA battery.
The current draw is .45 Ma

RG are you saying possibly use an ubalanced V ref like steve had in the
FZ-1 before the diode shunt...?

[R.G.]

The current draw is .45 Ma

That's 450 micro amps?

RG are you saying possibly use an ubalanced V ref like steve had in the
FZ-1 before the diode shunt...?

No.

Just thinking out loud now. If the load is zero, no current into the circuit at all, then the current that flows is 9V minus one shottky and three silicons divided by the resistor. (As an aside, the low drop of the shottky doesn't help any in this circuit. A standard silicon would be fine.)  That's (9 - 0.5 -3*(0.6))/10K = 670uA. The circuit uses 450uA, so all is cool.

Maybe. A silicon diode isn't always 0.6V, and it isn't always low resistance. At 220uA (roughly) a silicon diode is firmly in the region where it's voltage and internal resistance varies. Go here: http://www.vishay.com/docs/85622/1n914.pdf and look at figure 1, at about 200uA.

You could do several things. First, drop that 10K to 1K and see what happens. That would blap the diodes into a region where their current varies insignificantly with any current changes in the circuit, so they don't change their voltage much.

Second, try a Vbe multiplier. Hook up a high gain NPN with emitter to ground, collector to your series resistor, and make the series resistor be 4.7K. Now connect a resistor from collector to base and another from base to ground.  The transistor base-emitter forces the voltage on the base-ground resistor to be one Vbe, and the high gain enforces that the base eats very little of that current. So the current through the base-ground resistor is Vbe/Rbase. Since the base eats very little of the current, it all comes through the collector-base resistor, and the voltage across that resistor has to be (Vbe/Rbase)*Rcollector. So the total voltage across the two resistors in series is Vbe* (1+Rc/Rb), and the transistor forces this to be true by its gain. Not only that, but since the voltage is (relatively) constant, so is the current across the two resistors that set the voltage. You can get voltage drops referenced to a diode without having to put up with an integral number of diodes. The feedback action keeps it stable.

Third, use a series regulator. The opamp circuit works. So would an LM317L, which tries to make 1.25V on its output.  Two resistors or one diode with an LM317 gives you nearly the same voltage but with rock solid control and low output impedance. The LM317 also eats more current than you're using now, but you could also add one diode to the stack and use an NPN as a series regulator with its base at the top of the now four-diode stack. It puts out much the same as you're getting now, but has its gain to hold things steady. It's base-emitter voltage varies much less than the three diode stack with full circuit current change, and the diode stack voltage sees no change at all. So the output voltage is steadier.

But I do like the Vbe multiplier.

[Gus]

R.G. thinking out load about your post

With the Vbe multiplier you need to make sure the transistor is always shunting current to ground.  So one need to know what the circuit draws at say 1.5VDC and have the transistor conduct more current than the circuit draws when the circuit is disconnected from the Vbe mult circuit.

If I understand correctly this is needed because of the high output resistance of the collector(you do have C to B feedback with one of the resistors and you did post use a high gain transistor).  I am guessing this is why you posted use a 4.7k from 9VDC to the collector.
9VDC- 1.5VDC = 7.5VDC
7.5VDC/ 4.7k = about 1.6ma(plus a little more because of the resistors C to B and B to G) with the effect disconnected

So if the circuit draws 1ma .6ma is used in the Vbe regulator section keeping the transistor in conduction as shunt regulator.

The Vbe seems like one of the the simplest to build and I do like shunt regulators

[R.G.]

With the Vbe multiplier you need to make sure the transistor is always shunting current to ground.

Yep. All shunt regulators get very odd around the region where the shunt element gets starved for current and goes into non-conduction. It's essential for all shunt regulators to evaluate how much their min and max currents will every possibly be, what happens on a fault, and what the circuit load variations do to them.

So one need to know what the circuit draws at say 1.5VDC and have the transistor conduct more current than the circuit draws when the circuit is disconnected from the Vbe mult circuit.

Maybe. One needs to know the circuit min and max currents in operation, and ensure that the transistor can conduct the max current the resistor allows without burning up, and also that at the max circuit draw, there is still enough current available for the transistor to have some current AND that the two setting resistors have enough current to force the transistor to behave.

If I understand correctly this is needed because of the high output resistance of the collector(you do have C to B feedback with one of the resistors and you did post use a high gain transistor).  I am guessing this is why you posted use a 4.7k from 9VDC to the collector.

Yep. I winged it on that one. A good design would evaluate circuit min/max draw, transistor action and gain at low currents, shunt resistor draw, and then figure the series resistor to have enough current available for both the Vbe multiplier and the load in all conditions.

The Vbe seems like one of the the simplest to build and I do like shunt regulators

It's a handy trick to keep in your bag. I ... hate... shunt regulators. Sneaky, evil things.    Shunt regulators require good prior knowledge of the loading. Well, OK, so do series regulators. 

[Gus]

R.G.
I just simed a Vbe reg circuit based on your post.
27.73k C to B
22k B to ground
2N5089 LT spice transistor model
1.5k load resistor
4.7k 9VDC to C
1.5VDC at the collector
.57ma in the emitter leg

[R.G.]

R.G.
I just simed a Vbe reg circuit based on your post.
27.73k C to B
22k B to ground
2N5089 LT spice transistor model
1.5k load resistor
4.7k 9VDC to C
1.5VDC at the collector
.57ma in the emitter leg

Looks about right.

This is the same circuit used for biasing the output transistors in most power amplifiers. Making the two resistors variable lets you dial in whatever voltage you want, and also makes the voltage vary with the temperature of the transistor Vbe, which can be put on a heat sink for thermal compensation.

You might want to try it with a dynamic load on the output, and also with varying values of the input current limiting resistor.

The screaming advantage that shunt regulators have is that they present a mostly constant load to the raw supply.

[ORK]

Why not simply drop a few volts by an ohmic resistive divider (at half a milliamp as long as the current draw is fairly constant?)

[R.G.]

Why not simply drop a few volts by an ohmic resistive divider (at half a milliamp as long as the current draw is fairly constant?)

as long as the current draw is fairly constant

[kaycee]

I'm probably missing the point here, but I use an LM317T voltage regulator to run my TA-28 at the required 1.5 volts from a 9 volt powersupply and its fine. The regulator, a couple of resistors and caps and a 5k trim to dial it in.

[Gurner]

I'm probably missing the point here, but I use an LM317T voltage regulator to run my TA-28 at the required 1.5 volts from a 9 volt powersupply and its fine. The regulator, a couple of resistors and caps and a 5k trim to dial it in.

I don't think you're missing the point - but your parts count is too high...

http://docs-europe.electrocomponents.com/webdocs/0e0a/0900766b80e0ac60.pdf (there's a 1.5V version called the MCP1702-1502E/TO ...upto 13.5V input, 1.5V out...quiescent current draw of something like 2uA!)

....one small input cap, one small output cap.....so three parts including the regulator itself ...a rock solid 1.5V for 50 cents .....job done.

(I'm all for creative solutions to seemingly tricky problems, but sometimes the best solution is cheap & actually made for the job!)

[kaycee]

Gurner, nice little chip there, I'll try that next time I make up a TA. I just followed the data sheet for the regulator, I think it was only the second one I'd done so it was worth it for novelty value

[John Lyons]

I've been using a 317 regulator. I just figured I'd try the diode to ground shunt and it was
getting weird on me si I posted about it. ...and diodes are already on the bench...
I'll look into that MCP1702... regulator. Thanks Gurner

[Gus]

I posted some links in this thread for regulators.
http://www.diystompboxes.com/smfforum/index.php?topic=93935.0

I do think the Vbe regulator is a elegant idea for this effect.  I like shunt regulators for certain applications.

If it is OK with R.G. (it was his idea in this thread) I can post a sim screenshot of a Vbe req for this effect.

[R.G.]

Go ahead Gus. I certainly didn't think up the Vbe multiplier.

[Gus]

Sim of the Vbe regulator with a a 2N2222 fuzz somewhat like the Ge fuzz no idea what this would sound like.
I did this to have an active load V2 is the input signal to the fuzz
Also note the two different biasing methods used for Q3 and Q4.  R4 is in the sim to measure current.



Sim shows a sensitivity to cap C3 value and a low frequency decrease in the Vbe reg output Z.  I reduced it from 1uf to .1uf

John could you try a smaller coupling cap value at the input to the last gain stage of your circuit that has the oscillation issue?