What does this transistor do? Small Clone

[Scruffie]

I've long wondered what the purpose of Q3 is in the Small Clone circuit, it seems to buffer and filter the 4047 voltage but to what benefit? I use 4047 clocks a lot and if i'm missing a trick it'd be nice to know, especially as EHX weren't known for using extra parts they didn't have to if it saved a few cents.

http://tonepad.com/getFile.asp?id=97

Anyone shed some light on it?

[dwmorrin]

You're totally correct in it's function as an active filter and poor man's voltage regulator to the MN3007 BBD and the CD4047 clock.
The amount of small current in the base is easily filtered by the base's 10µF cap, resulting in a large amount of filtered current in the emitter-collector circuit.  The base's cap value "looks like" it's been multiplied by beta in how it will filter the emitter-collector circuit.

These clocks run at high frequency, and I think almost every schematic for BBD chips calls for some kind of voltage regulation.  I've never tried running a BBD from non-regulated voltage, but I'd imagine the result is lots more noise.  And the clock noise canceling scheme of the BBD probably needs everything to be very precise to get good cancelation.

[duck_arse]

the 10k//10uF on the base will mean the transistor turns on slowly. this might allow other parts of the circuit to reach stable operation levels before the 4047 and mn3007 need stuff from them.

[Scruffie]

You're totally correct in it's function as an active filter and poor man's voltage regulator to the MN3007 BBD and the CD4047 clock.
The amount of small current in the base is easily filtered by the base's 10µF cap, resulting in a large amount of filtered current in the emitter-collector circuit.  The base's cap value "looks like" it's been multiplied by beta in how it will filter the emitter-collector circuit.

These clocks run at high frequency, and I think almost every schematic for BBD chips calls for some kind of voltage regulation.  I've never tried running a BBD from non-regulated voltage, but I'd imagine the result is lots more noise.  And the clock noise canceling scheme of the BBD probably needs everything to be very precise to get good cancelation.

Usually the most seen is a small resistor and cap to ground to filter them a bit, except for when the 3207 is used which seems to have worse bias drift issues than the 3007 so uses regulation generally (that and a 10V max input).

I did ask this question once before a long time back and was given this answer;

The third transistor, between the HF clock and the BBD is to allow a better range of modulation of the clock. It's just a small improvement to the basic circuit, and will also help with providing a more equal mark / space ratio over the clock's range.

An explanation as to why or how this is the case though wasn't given so not sure how much to trust it... if it is true though, it may very well be worth using elsewhere.

[greaser_au]

I'd suggest DA is right,  It's a delayed power-on/reset for the 3007 & 4047.

david

[teemuk]

Should that link open some kind of file to download? I don't see it.  


Edit: Nah, forget it, I found it another way.

[dwmorrin]

This guy did a really good explanation and comparison of the BBD chips: http://www.electrosmash.com/mn3007-bucket-brigade-devices
He has links at the bottom with great BBD stuff if you want to do further research on these circuits.

Usually the most seen is a small resistor and cap to ground to filter them a bit, except for when the 3207 is used which seems to have worse bias drift issues than the 3007 so uses regulation generally (that and a 10V max input).

I think we're talking about two different things.  I think you're referring to the audio input, and the Vbias max, and I'm talking about the main power supply.  The MN3007 can be powered by up to 15V, while the Vbias max is 10V.
If you compare with the Boss CE-2, which does use a small resistor (33Ω) and cap for the power supply, the cap is much bigger: 220µF.
I am still in favor of the idea that the purpose of Q3 is to provide active filtering, so a much smaller cap, 10µF in the small clone's case, can provide similar filtering results as a much bigger cap.  (Cap's value gets multiplied by beta for the emitter circuit)
EDIT: I do see that the clone also has a 220µF cap.

The mark to space ratio of the two clock signals needs to be very precise for the clock noise cancellation effect to work properly.

I would be interested to understand why a time delay would be needed.  There is no mention for providing this in any of the notes I've read for BBD, but I'm here to learn, so I would be interested in anyone's thoughts on this.
I see similar looking "transistor time delay" circuits online, but they usually have a resistor feeding the base of the transistor in series if it is a delay circuit.  In this circuit, I see Q3 turning on as soon as there is 0.6V on the base, which would be very fast....
What would the harm be in starting the BBD and the clock quickly?

[dwmorrin]

Finally figured out the name of this circuit: http://en.wikipedia.org/wiki/Capacitance_multiplier
Capacitance multiplier.
It has both the effect of smoothing power supply ripple and delays how fast the power reaches the the load.

The article points out it does not assist in voltage regulation, just ripple smoothing.

Nothing in the 4047 data sheet suggests it requires any special treatment in this way, so I would guess that EH was designing for worst case, very noisy AC adapters to still work with the clock and BBD.

[Scruffie]

This guy did a really good explanation and comparison of the BBD chips: http://www.electrosmash.com/mn3007-bucket-brigade-devices
He has links at the bottom with great BBD stuff if you want to do further research on these circuits.

Usually the most seen is a small resistor and cap to ground to filter them a bit, except for when the 3207 is used which seems to have worse bias drift issues than the 3007 so uses regulation generally (that and a 10V max input).

I think we're talking about two different things.  I think you're referring to the audio input, and the Vbias max, and I'm talking about the main power supply.  The MN3007 can be powered by up to 15V, while the Vbias max is 10V.
If you compare with the Boss CE-2, which does use a small resistor (33Ω) and cap for the power supply, the cap is much bigger: 220µF.
I am still in favor of the idea that the purpose of Q3 is to provide active filtering, so a much smaller cap, 10µF in the small clone's case, can provide similar filtering results as a much bigger cap.  (Cap's value gets multiplied by beta for the emitter circuit)
EDIT: I do see that the clone also has a 220µF cap.

The mark to space ratio of the two clock signals needs to be very precise for the clock noise cancellation effect to work properly.

I would be interested to understand why a time delay would be needed.  There is no mention for providing this in any of the notes I've read for BBD, but I'm here to learn, so I would be interested in anyone's thoughts on this.
I see similar looking "transistor time delay" circuits online, but they usually have a resistor feeding the base of the transistor in series if it is a delay circuit.  In this circuit, I see Q3 turning on as soon as there is 0.6V on the base, which would be very fast....
What would the harm be in starting the BBD and the clock quickly?

Nope, I was referring to power too, the 3000 series is 15v max, the 3207 is 10v, I was talking about the 3207 needing voltage regulation. The 3007 just needs filtering (especially at only 9V supply in the small clone).

So this statement;

The third transistor, between the HF clock and the BBD is to allow a better range of modulation of the clock. It's just a small improvement to the basic circuit, and will also help with providing a more equal mark / space ratio over the clock's range.

How is it giving a better range of modulation... just by giving it a smoother voltage?

[dwmorrin]

Now I see what you were saying!

The quote say's the transistor is between the clock and the BBD... but I don't see Q3 as being between them.  Q3 appears to simply supply filtered power to both chips, at the same time.  The clock outputs are connected directly to the clock inputs.

Check out the schematic for the small clone in this thread: http://www.diystompboxes.com/smfforum/index.php?topic=36077.msg254654;topicseen
The original circuit called for a Reticon SAD1024, and in the lower right shows how to mod the circuit to accept a MN3007.
Perhaps the need for Q3 isn't for the MN3007, but for the SAD1024... although nothing on the notes for the SAD1024 indicate it either.
Going with the original theory - that EH would cut corners if possible - I don't think they always did this.

I think your quote's "small improvement" of the circuit is just the extra filtering provided by Q3.

Check out this info from that electrosmash analysis:
Following the datasheet recommendations, BBDs guaranteed a correct behaviour under a certain clk frequency (100~200KHz). It is because the input capacitance of the clock pins causes progressive degradation of clock pulses above the frequency limit: The clock pulses change from square to a trapezoidal/sawtooth waveform. When that happens the switching action is not perfectly synchronized resulting in a signal degradation.
The clock frequency limitation is the reason why the Reticon SAD1024 chip was a largely preferred choice for flangers effects which involve short delay and fast clock: the input capacitance on the clock pins was much lower (110pF) than on the MN3007 (700pF), allowing it to be clocked at much higher frequencies.
When suitably buffered, though, there are reports of it clocking MN3xxx chips up well past 1MHz, which takes the device close the zero-delay zone, being convenient for fine flanger effects."

To me, that's saying we need our clock pulses to look really good to avoid a "noisy" chorus.  I don't know how ripply the 4047's supply has to be to get to that point... that would be my next question.

[Scruffie]

The clock not being square wont lead to noise as such, it'll lead to a messed up delay basically, it needs to be a proper square wave for the chip to function and I don't think the 4047 is the issue, it's used in many places without such a set up, including other EHX products (the clone theory for example, which also switched from SAD to MN, this was due to the MN being more reliable and having better noise characteristics according to EHX themselves) which were designed by the same designer.

I think the quote saying between them is just an error and he meant between the power and the chips.

I'm not saying EHX would cut corners, but if a part wasn't necessary, they wouldn't use it.

[dwmorrin]

Clone theory schematic shows a 78L15 voltage regulator.
Memory man uses a discrete transistor power supply to regulate/filter the voltage.
Small clone has no proper regulator, but they still at least stick on a "ripple reducer" transistor.
That makes sense, right?  The small clone is the budget model, so it gets the least power regulation.

If you were making this pedal today, and you were feeding the 4047 and 3007 well regulated dc, I don't think you'd need Q3.
I think the only trick here is that a cheap, unregulated dc wall wart will work OK with Q3 cleaning it up.

[armdnrdy]

There's one thing that catches my eye about this circuit. The VGG going to ground. I don't believe that I've ever seen that.

It's my feeling that the transistor is creating a separate supply rail to keep the audio path free from clock/BBD noise. The transistor will isolate the power.

I've seen the same type of thing in the WEM Hyperflanger but it was used to create a separate ground rail.

[Scruffie]

Clone theory schematic shows a 78L15 voltage regulator.
Memory man uses a discrete transistor power supply to regulate/filter the voltage.
Small clone has no proper regulator, but they still at least stick on a "ripple reducer" transistor.
That makes sense, right?  The small clone is the budget model, so it gets the least power regulation.

If you were making this pedal today, and you were feeding the 4047 and 3007 well regulated dc, I don't think you'd need Q3.
I think the only trick here is that a cheap, unregulated dc wall wart will work OK with Q3 cleaning it up.

Forgot that they used regulators.

Fair enough, with what armdnrdy says too I concede it's just there to regulate, the person I quoted was either mistaken or worded badly.

I think the datasheet says you can just ground Vgg if you want if I remember correctly.

[gena_p1]

dwmorrin ist right, it's not a slow reset, it's "electronic choke".

4047 and 3007 produces high "binary" noise, and this part protecting audio part from 4047 clock whistle, so they are powering from stable filtered separate power source.

Boss metallzone and megadrive contain same kind of circuit, but for protecting high gain from PSU hum.