EA Tremolo, mod to get it smoother?

[Rob Strand]

The basic EA design certainly works, but I wonder if it would work better and sound less choppy, or whatever, if a trimmer was used to provide a basic bias to the gate of Q3,

It's a very elegant design provided you live with the limits.   Once you start making more demands from it, the nice simple circuit starts to collapse.

It does seem possible to extend the smooth region.   

Replacing the 1.2k resistor with a current source would extend the smooth region, perhaps only by a small amount but better than nothing.   If the amount of time the LFO is positive can be reduced then that might give some small apparent extension to the smooth region.    In small doses diodes or a VBE multiplier might help re-shape the LFO waveform.   There will always be a point where you crank the depth and it hits the choppy zone.

[PRR]

.......isn't a phase shift oscillator supposed to produce a sine wave anyway?....

What a worker "is supposed" to do, and what he really does, are often different.

'Sine' oscillators generally (not function generators) must run at loop gain of 1.000000..... Any less, they die out. More, they build up to infinity (and clipping).

The canonical phase-shift has loss of 27(?) so the transistor must have gain of 27. The common form of Wein is 3.

We normally build a little excess gain (for quick starting) and some mechanism to lose gain at high level. Clipping works and many-many PSOs are set 5% hot, clip 5% of the wave.

H and P (after Mecham) claimed that an incandescent filament would do the right thing. (In fact it won't be stable on a perfectly clean amplifier, it has to have some distortion.) Later oscillators can have hyper-elaborate gain controllers.

[fiestared]

Thanks for all the input. From what i gather here, I could:
1:
Swap the q3 to a J112 and maybe try a "1M linearization resistor between G and D of Q3, so that the LFO waveform translates better to the actual modulation"?


2: "Replacing the 1.2k resistor with a current source?" This sounds interesting, can someone try to explain how this this is done in detail? It is fun just to try things and see what happens and maybe try to learn a thing or two on the way...

[antonis]

"Replacing the 1.2k resistor with a current source?"

can someone try to explain how this this is done in detail?

In the original circuit, R9 is sitting on about 1.3V so 1.08mA is the Emitter/Collector quiesecent current..
Now 560R sits on about 600mV (one diode voltage drop 'cause the other one is "eaten" by V_BE..) hence 1.07mA is the effective quiescent current..
As you can see, we managed to maintain same Q2 bias condition, but with the benefits of Q2 Emitter "active load"..

P.S.
6k8 - 10k resistor value isn't critical as long as diodes current is bigger enough than 2N5088 Base current (10 times say)..

[Rob Strand]

The thread regarding the current source is this one,

https://www.diystompboxes.com/smfforum/index.php?topic=43902.0

Keep in mind there might be other posts on the same topic.

Transmogrifox changed a few more things other than adding the current source.



@antonis
Notice Transmogrifox's circuit moved the 180 ohm out of the DC path.
That gives the current source more voltage to work with.

[duck_arse]

erm, well, I have to retract and correct and explain post #4. I have a pso circuit using a mosfet that works quite well with the 2 red leds across the centre phase cap only, but the bipolar version from your build docs didn't work the same on the breadboard today. but and so - if you add a cap and shift the leds across the feedback resistor, it does work, cleans the output wave and neatly constrains the output level across the range.

a small sample [of two parts] suggested a larger value of CX for a higher hFE transistor. value is not critical, but affects [minimally] the distortion at higher speeds. also, the BC548 showed a slightly higher output level. not important. using blue leds [didn't matter size or lens type] provided the maximum signal level without clipping on the wave-bottom, between about 4V5 and 6V across the range, depending. the clipping leds may flash visible, but they are not for indicating anything.

also, to the [ my ] C2 and C4 annoyance - I have to retract my previous. if you fit two caps in series, you end up with less total capacitance, in this inst, 215.3nF. so, the cap to remove is the 10uF [C2], which is leftover from the buffer circuit, and is big enough to drive an unknown load. the 220nF is correct for the input of Q2, so bigger doesn't make any sense, and might lead to blocking/gulping. might.

so it doesn't really matter, but the preferred remove-and-link-across is actually C2. sorry bout any confusions. I hope the diagram helps.

ohh, and I didn't try driving the tremming section with this osc, but the output load tested should be equivalent. a point to note is that R10 and R11 were originally meant to scale the oscillator output level to suit the tremming jfet requirements.

[Fancy Lime]

[Hollow voice from off stage]: Aren't we at the point now where re-designing the whole thing from the ground up would be easier than sticking ever more band-aids onto the hull of a sinking ship? Anyone know the story of the Vasa? Using a dual opamps for the amplifier and PSO stages would save a bunch of parts and solve some of the problems, no? And I am only suggesting that because I assume we want to keep the PSO idea and principle architecture for some reason. Otherwise I think a tremolo is the ideal beginners DSP project.

Andy

[antonis]

@antonis
Notice Transmogrifox's circuit moved the 180 ohm out of the DC path.
That gives the current source more voltage to work with.

Good catch, Rob.. 

[fiestared]

.

[fiestared]

erm, well, I have to retract and correct and explain post #4. I have a pso circuit using a mosfet that works quite well with the 2 red leds across the centre phase cap only, but the bipolar version from your build docs didn't work the same on the breadboard today. but and so - if you add a cap and shift the leds across the feedback resistor, it does work, cleans the output wave and neatly constrains the output level across the range.

a small sample [of two parts] suggested a larger value of CX for a higher hFE transistor. value is not critical, but affects [minimally] the distortion at higher speeds. also, the BC548 showed a slightly higher output level. not important. using blue leds [didn't matter size or lens type] provided the maximum signal level without clipping on the wave-bottom, between about 4V5 and 6V across the range, depending. the clipping leds may flash visible, but they are not for indicating anything.

also, to the [ my ] C2 and C4 annoyance - I have to retract my previous. if you fit two caps in series, you end up with less total capacitance, in this inst, 215.3nF. so, the cap to remove is the 10uF [C2], which is leftover from the buffer circuit, and is big enough to drive an unknown load. the 220nF is correct for the input of Q2, so bigger doesn't make any sense, and might lead to blocking/gulping. might.

so it doesn't really matter, but the preferred remove-and-link-across is actually C2. sorry bout any confusions. I hope the diagram helps.

ohh, and I didn't try driving the tremming section with this osc, but the output load tested should be equivalent. a point to note is that R10 and R11 were originally meant to scale the oscillator output level to suit the tremming jfet requirements.

-So you mean that what I did on my photo is unaffecting the circuit? I would really appreciate if you could go through more step by step how to do it the on my circuit!

[iainpunk]

[Hollow voice from off stage]: Aren't we at the point now where re-designing the whole thing from the ground up would be easier than sticking ever more band-aids onto the hull of a sinking ship? Anyone know the story of the Vasa? Using a dual opamps for the amplifier and PSO stages would save a bunch of parts and solve some of the problems, no? And I am only suggesting that because I assume we want to keep the PSO idea and principle architecture for some reason. Otherwise I think a tremolo is the ideal beginners DSP project.

Andy

haha, funny that you mention that, my first (and only audio related) DSP project was a tremolo/vibrato for a highschool project, the product worked great, but the report i wrote was ''too long''

cheers, Iain

[rockola]

[Hollow voice from off stage]: Aren't we at the point now where re-designing the whole thing from the ground up would be easier than sticking ever more band-aids onto the hull of a sinking ship? Anyone know the story of the Vasa?

Classic case of management hubris making engineering decisions. The museum in Stockholm is well worth a visit, in a world where travel is again possible.

[iainpunk]

[Hollow voice from off stage]: Aren't we at the point now where re-designing the whole thing from the ground up would be easier than sticking ever more band-aids onto the hull of a sinking ship? Anyone know the story of the Vasa?

Classic case of management hubris making engineering decisions. The museum in Stockholm is well worth a visit, in a world where travel is again possible.

explain?

[duck_arse]

-So you mean that what I did on my photo is unaffecting the circuit? I would really appreciate if you could go through more step by step how to do it the on my circuit!

erm well, you said it didn't work as tried, didn't you? I have tried my correction mod on the breadboard, with your circuit's values, and can confirm that it does clean-up the waveform produced by the oscillator. whether it's the cure for your complaint, I can't say.

take your two leds - I found blue best for his purpose - and join them back to back, like clipping diodes. to one end of the two leds assembly [it is now bi-directional, it doesn't matter which end is up] solder one leg of a cap - any type, plastic or ceramic - of between 100nF and 220nF. you now have a free cap leg at one end, and a free led//led leg at the other end of your assembly, which shows on my diagram as CX and LED1 and LED2. this is again and still a bi-directional unit, it has no plus or minus or up or down, works the same either way.

now you want to tack-solder your assy across the R13 feedback resistor in the oscillator circuit. under the board is fine for the first test. the idea is to restrict the voltage swing of the collector so that the transistor does not clip the waveform when it switches hard on - you get a slightly lower level of oscillator output, but it has less distortion. this can be compensated for with the R10 and R11 resistors if necessary.

try this mod, let us know if it works or troubles or causes confusion.

[Fancy Lime]

[Hollow voice from off stage]: Aren't we at the point now where re-designing the whole thing from the ground up would be easier than sticking ever more band-aids onto the hull of a sinking ship? Anyone know the story of the Vasa?

Classic case of management hubris making engineering decisions. The museum in Stockholm is well worth a visit, in a world where travel is again possible.

explain?

Basically, the king wanted more and more and bigger and bigger cannons to make his flagship the best armed in all the Baltic Sea. But the original plans for the hull were never meant to support that weight. Long story short: the best armed ship of its time barely made it out of the harbor before sinking. 1300m maiden voyage, that was it. And because the sediment in the Stockholm harbor is highly anoxic, the whole thing is beautifully preserved 400 years later. The museum is indeed great.

Andy

[fiestared]

-So you mean that what I did on my photo is unaffecting the circuit? I would really appreciate if you could go through more step by step how to do it the on my circuit!

erm well, you said it didn't work as tried, didn't you? I have tried my correction mod on the breadboard, with your circuit's values, and can confirm that it does clean-up the waveform produced by the oscillator. whether it's the cure for your complaint, I can't say.

take your two leds - I found blue best for his purpose - and join them back to back, like clipping diodes. to one end of the two leds assembly [it is now bi-directional, it doesn't matter which end is up] solder one leg of a cap - any type, plastic or ceramic - of between 100nF and 220nF. you now have a free cap leg at one end, and a free led//led leg at the other end of your assembly, which shows on my diagram as CX and LED1 and LED2. this is again and still a bi-directional unit, it has no plus or minus or up or down, works the same either way.

now you want to tack-solder your assy across the R13 feedback resistor in the oscillator circuit. under the board is fine for the first test. the idea is to restrict the voltage swing of the collector so that the transistor does not clip the waveform when it switches hard on - you get a slightly lower level of oscillator output, but it has less distortion. this can be compensated for with the R10 and R11 resistors if necessary.

try this mod, let us know if it works or troubles or causes confusion.

Thank you so much for your help. I did try this and soldered it together/over R13.  (attached a photo), but i get just clean signal, no tremolo effect at all.. as shown i connected the leds opposite ends and the 104 cap in the other end. Did I do something wrong?

[antonis]

It seems that you've shorted 2M2 resistor..
(Q4 works as BJT-Diode configuration.. to verify it, measure DC voltage on Collector..)

[fiestared]

It seems that you've shorted 2M2 resistor..
(Q4 works as BJT-Diode configuration.. to verify it, measure DC voltage on Collector..)

Ok! i have removed the diods/cap, if that makes any difference for the verification, (I am new to this, so tell mi if I'm wrong here) I put the negative pin from my DMM on E and B on Q4, and measure the collector with the Red pin from DMM and get 8,21V. The same as the wire coming from my power supply. 

[antonis]

That shouldn't happen..
E sits on GND (0V) but B should sit about 600 to 700 mV higher so you shouldn't measure the same C-E and C-B voltage difference..

It seems like you've B & E shorted someway and there isn't any current flowing through Q4..
(No voltage drop across 15k Collector resistor..)

[fiestared]

That shouldn't happen..
E sits on GND (0V) but B should sit about 600 to 700 mV higher so you shouldn't measure the same C-E and C-B voltage difference..

It seems like you've B & E shorted someway and there isn't any current flowing through Q4..
(No voltage drop across 15k Collector resistor..)

Ok, Im not really following. Could you specifiy exactly how to measure Q4 correctly? Complete beginner here. As stated i put multimeter negative on both pin 1 and 2 at the same time while the red on 3.