EA Tremolo revisited--rEAgenerated Tremolo

[TELEFUNKON]

The cold side of the level pot is hooked to a 47k & 10k voltage divider here.
There is a second equally dimensioned voltage divider with the bias trimpot.
Couldn`t you omit that first divider, and hook the pot to the hot side of the trim?
(Save parts count).

[Transmogrifox]

The cold side of the level pot is hooked to a 47k & 10k voltage divider here.
There is a second equally dimensioned voltage divider with the bias trimpot.
Couldn`t you omit that first divider, and hook the pot to the hot side of the trim?
(Save parts count).

I knew there was a reason I like this forum.  Yes, that does sound like a good way to get the parts count down.  We would just increase the 1uF decoupling cap to make the audio bleed-through to the bias as minimal as possible (really not a very big concern).

[Transmogrifox]

Since the original problem is lack of depth, would it be possible to simply subtract (that is, add out of phase) audio from the signal from the original EA circuit, cancelling the audio feeding thru in the 'low' part of the cycle?

That's what I thought I'd try next if this didn't work.  That actually may be a more simple way to do it, yet.  We could take a tap off the source (or emitter if BJT version) and mix it back in with the signal coming off the drain (or collector).  This way, the in-phase gain would be reduced as the out-of-phase gain was increased.  A nice complementary effect here...Though I can't think of a good way to do it with less added parts than the present design.

Would need the mixing resistors pretty large to prevent an interaction between source&drain as far as feedback goes.  Probably need to add an output buffer for this case....then we're into it the same # of parts. 

Here it is:  We can leave it hardwired at a set depth, and move the depth pot with A)Drain C)Source B) (wiper) = output to buffer (or output).  Twiddling with an extra resistor could help tune the maximum depth range.  Depth is controlled by panning between source and drain signal outputs. 

Hmm...I might have to breadboard this and see what happens.  This concept is more sensitive to variance in components, so a BJT for the main amplifier may be preferred over a JFET. 

[vhbram]

Is there already a pdf file of this project?

Thanks in advance,

Bram

[Transmogrifox]

Hey Bram--I haven't completed it yet.  I'm not satisfied with my PCB layout--there are a few details requiring hashing and hacking to make the battery fit in the box.  I'll start a new thread when I get the project PDF done.

Here's the original from GGG:
http://www.generalguitargadgets.com/index.php?option=content&task=view&id=108&Itemid=138

If you aren't looking to do the dual trem option, you can pretty well use the GGG layout with the CCS built into an epoxy ball and solder the leads down to the relevant PCB pads.  I know it's a jerry-rigged way of doing things, but it's workable.

Otherwise, keep holding your breath for a couple weeks and I'll get the project together as I have time. 

If you have a bread board, I encourage you to build it and give it a test drive.  The dual trem thing is really cool sounding.

[vhbram]

Hey,

when you have finished the pdf could you also place a post here?

Keep me in touch,

Bram

[Processaurus]

I got around to trying this today with a stereo version of the EA trem I hadn't worked on in a year because I managed to stuff most every transistor in backwards.  After fixing my mess I decided to see if Transmogrifox's mod would work on the GGG version of the EA (FET input buffer, BJT that does the amplifying).  Worked like a charm, after debugging a record number of errors in the original perfed up circuit (everything was wrong, backwards transistors, cold solder, missing wire, fluxy solder pads that would trick the scope, you name it).   Dialed in the 6v specified on the collector with the 10k trimpot and its all good.  I tried the 68K and 10K but it wasn't right, so I just threw the trim pot on there.

Thanks Transmogrifox!  This is, I would say, a indespensable mod for this circuit, if you have one, you'll pretty much have to be happy if you try it.  It can fit on a tiny piece of perf, and it took suprising little time to make.  I just clipped the 1.2K emitter resistor to ground and left it on there, and hooked up the emitter to the transmogrifox  BJT and 1K to ground. 

PS nice dual LFO circuit, very elegant, I did the same idea but the brute force way with an opamp panning circuit.

[thegnu]

Greetings from the future!

I'm about to build an EA Tremolo, and I've been looking at tweaks I may want to implement.  Is there anywhere this schematic still exists online, because I can't find it.  I'm interested in the depth mod, where exactly the volume pot is moved to, and where the CCS is inserted.

If anybody has the schem or could offer any insight, please let me know.  I'd post a new thread, but this thread kept coming up, so I figure this is the best place for googlers.

[Processaurus]

I uploaded a copy to my gallery, Transmogrifox, let me know if for any reason you'd like me to remove it, I assume it just got lost when geocities got closed down.

[thegnu]

I uploaded a copy to my gallery, Transmogrifox, let me know if for any reason you'd like me to remove it, I assume it just got lost when geocities got closed down.

muahahahahahahahaaaaaaaaaaa!!!! much obliged.  thank you.

[Transmogrifox]

Speaking of hello from the future...it's true this schematic along with everything else I had died with geocities and went underground while I went on a DSP software binge.

I have revisited this and done a little tweaking with LTSpice to improve headroom and also combined bias resistors for a slightly lower parts count.  I also added power supply reverse polarity protection as well as decoupling of the oscillator power supply from the audio supply.



I haven't made a pot model and symbol for LTSpice so here's the description of pots:
Volume = 250k pot, wiper is at the junction of RvolPotA and RvolPotB
Depth = 250k pot, wiper is at the junction of RdepthA adn RdepthB
Rate = 100k pot, wiper tied to ground.  Schematic shows this pot as RratePot

[bool]

Hi there, the conversion to a bjt gain cell looks great.

One question though: Did you think about converting the JFET "variable resistor" to a BJT? (or at least to a somewhat more consistent mosfet a-la bs170/2n7000 if high impedance is of any importance here since the gain element is capacitively connected?)

** but I suspect that R8 could be used to a good cause to tweak the BJT conversion ...

[Transmogrifox]

It's probably worthwhile experimenting with other ubiquitous component types.  If I'm lucky I might have a chance to tweak with this over the weekend and see if I can get a similar sound from a BJT as the variable R element. 

Part of what I love about the EA Tremolo is the effective LFO shape due to the transfer characteristic of the JFET gain element.  A BJT would be quite different, but maybe a MOSFET could come closer, as you have suggested.

[GGBB]

@Transmogrifox:

Hoping for some input here. I am working up a schematic for the ROG Modified EA Tremolo modified to use your current source idea. I want to implement the ROG circuit as closely as possible (for now) and just add the current source for the MOSFET. The schematic is below. I have a few questions - mainly does this look correct - but also:

I am uncertain about the values or necessity of R11 and R12. I expect they will factor in to the calculation you provided earlier, but are the actual values arbitrary? Or do they even need to be there?

Should the current source collector be tied (via R11) to the FET source directly (as drawn) or to the R3-C2 junction? Or does it matter as long as the bias is set correctly?

I know that the total value of R10 and the bias trim come from your calculation, but if biasing "manually" is there anything to listen for or any particular collector or emitter voltage to aim at?



http://i1190.photobucket.com/albums/z460/GGBB1/Miscellaneous/rEAgeneratedTremolo_Rev1A.png~original

[duck_arse]

ggbb - my still incomplete build of this has (your R10) 56k over (your R13) 8k2 for the bias resistors on the current source thingy. this from the breadboard measures for 2N5457, have you tested your 10k/10k/10k string for good volts, real world?

[Transmogrifox]

I can't view your schematic right now because I'm behind a proxy that is blocking whatever hosting site you're using -- but I did some simulations of this circuit, simplified things yet a little more, so I think the rEAgenerated Tremolo just got rEAgenerated.

In spite of my not being able to comment directly to your schematic at the moment I expect you will find something useful in the schematic below.

As for adjusting bias, all you're looking for is a place where you don't hear audible distortion.  If I were doing it by ear (without a DMM), I would crank input volume to max, then strum guitar, adjust bias until there is definitely bad distortion.  Then adjust the bias pot the other way until there is bad distortion.  Then set somewhere in between and if it sounds good -- it is good.

Here are the direct links because images below are rather large:
http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/rEAgenerated_Trem.PNG
http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/rEAgenerated_Trem_sym_plot.PNG


Here's how the simulation results turned out:


The beauty of using a JFET as the main amplifier in this bias setup is the JFET just self-biases and automatically rides along where it should.  The BJT sets the bias point and off we go.

The other advantage of this regeneration is I removed the AC coupling to the JFET gate which solved a lot of depth adjustment issues and removed some parts to boot.  Just 9V referenced the JFET variable resistor and max on is when the oscillator goes to 9V, then the depth pot is just a 10k. 

No more need for a boost switch because min depth is now boost.  I recommend a log pot for depth in this setup.

Anyway, I hope this gives some more ideas how to perfect the EA trem concept.

[GGBB]

ggbb - my still incomplete build of this has (your R10) 56k over (your R13) 8k2 for the bias resistors on the current source thingy. this from the breadboard measures for 2N5457, have you tested your 10k/10k/10k string for good volts, real world?

That's the thing - I don't know what voltage I'm supposed to look for (thinking 0.7 above source at gate). I'm still in the thinking/planning stage - no access to breadboard for a spell - basing my values on transmo's calculation earlier where he ended up with 68k/10k but that was set up a little differently. I was having trouble getting my head around the biasing so I deliberately separated the MOSFET and BJT biasing so I could hopefully understand it better.  I figured the 100k bias trimmer plus 10k over 10k should give me the ability to dial it in (my prelim calc gave me 56k/10k) but I'm not sure what role R11 and R12 play and if so how they fit in to the calculation.

I can't view your schematic right now because I'm behind a proxy that is blocking whatever hosting site you're using -- but I did some simulations of this circuit, simplified things yet a little more, so I think the rEAgenerated Tremolo just got rEAgenerated.

In spite of my not being able to comment directly to your schematic at the moment I expect you will find something useful in the schematic below.

As for adjusting bias, all you're looking for is a place where you don't hear audible distortion.  If I were doing it by ear (without a DMM), I would crank input volume to max, then strum guitar, adjust bias until there is definitely bad distortion.  Then adjust the bias pot the other way until there is bad distortion.  Then set somewhere in between and if it sounds good -- it is good.

The beauty of using a JFET as the main amplifier in this bias setup is the JFET just self-biases and automatically rides along where it should.  The BJT sets the bias point and off we go.

The other advantage of this regeneration is I removed the AC coupling to the JFET gate which solved a lot of depth adjustment issues and removed some parts to boot.  Just 9V referenced the JFET variable resistor and max on is when the oscillator goes to 9V, then the depth pot is just a 10k. 

No more need for a boost switch because min depth is now boost.  I recommend a log pot for depth in this setup.

Anyway, I hope this gives some more ideas how to perfect the EA trem concept.

Thanks, but even though I understand the words you are using, seeing the concepts behind them within the schematic are where I get lost. In particular the biasing. In this newest iteration, it looks like you have the FET gate and BJT base biased via (6.8k+150k)/22k. I think you mentioned before that you did that to keep parts count down - is that the only reason, or does it achieve another goal? They would therefore have close to the same voltage I assume - is that a goal? I'm hoping to have separate bias networks for the purposes of experimentation - and to be more faithful to the original ROG circuit. With Rog's version , the MOSFET gate bias is supposed to be at 4.5-5V, but I don't think this setup would be even close to that (not sure however). If you can just let me know if I am on the right track with a separate bias setup and MOSFET input when you see the schematic that would be great.

Also - 330µ from BJT base to ground - really? In other iterations of this circuit you used 1µ or 470n.

[Transmogrifox]

Now I understand the problem better.  If you're building without a breadboard you may be best off just building it with a trim pot and adjust bias by ear.  You will hear distortion if the bias is way bad.

If you want to just hard wire it correctly the first time, then I can't help until this evening when I get home and I can view the schematic.

For more fun and education, read on.

In this newest iteration, it looks like you have the FET gate and BJT base biased via (6.8k+150k)/22k. I think you mentioned before that you did that to keep parts count down - is that the only reason, or does it achieve another goal?

A JFET was selected to keep parts count down.  The JFET source is at a higher potential than its gate, so it's a happy coincidence that the combination of JFET amplifier with BJT current source here absolutely maximizes the headroom on this circuit.

They would therefore have close to the same voltage I assume - is that a goal?

Yes, the goal is to put the JFET gate at the same bias as the BJT.  This particular iteration of the design aims to maximize headroom and prevent clipping even with large input signals (such as the output of a booster or distortion pedal).

I'm hoping to have separate bias networks for the purposes of experimentation - and to be more faithful to the original ROG circuit. With Rog's version , the MOSFET gate bias is supposed to be at 4.5-5V... If you can just let me know if I am on the right track with a separate bias setup and MOSFET input when you see the schematic that would be great.

Separate bias on a MOSFET with gate at 4.5V is what you would want if using a MOSFET.

If the gate is at 4.5V, and you're keeping more faithful to the ROG schematic with 4.7k resistor in the drain, then you want to see that voltage somewhere around 6.75V.

To get 6.75V at the drain, this means the BJT should conduct 479 uA.  On the 1k resistor in the BJT emitter, this conveniently translates to 479 mV.  Add about 0.65V base-emitter drop and the base should be at about 1.12V.

With the old rEAgenerated schem reposted in this thread, we see a 47k resistor and 10k pot.  Let us choose to replace the 10k pot with a 10k resistor:
Current 10k resistor = 1.12/10k = 112 uA
Remaining voltage is 9-1.12 = 7.88V
R=V/I = 7.88/112uA = 70k

68k is a nice value near this, so working backward for sanity check:
9V*10k/(68k+10k) = 1.15V at the emitter.

1.15-0.65 (vbe) = 503 mV

Iemitter = 503mA/1k = 503 uA

Bias on MOSFET drain = 9 - 4.7k*503uA = 6.63V

Headroom high = 9-6.63 = 2.37V
Headroom low = 6.63-5 = 1.63V

I threw in that magic "5V" number:  4.5V bias + 500mV peak input signal.
A note about this:
This point where Vgate = Vdrain is only where the FET starts to go nonlinear. You don't get hard clipping until Vdrain = Vsource, so you actually have about another 1V to 1.5V of soft limiting before you hit the hard clipping threshold.  Setting on this end may allow to start compressing one polarity of the signal, but the reality is this is better to allow some asymmetrical soft clipping than to invite the onset of hard clipping sooner.  That said, this is actually a pretty happy balance for centering headroom.

This is part of why I went to reworking the EA trem.  The original ROG circuit leaves headroom pretty limited -- which is fine for signal levels associated with nominal pickup outputs.  Just not very good if it's being driven by something that amplifies.

Also - 330µ from BJT base to ground - really? In other iterations of this circuit you used 1µ or 470n.

The selection of this value was just a careless leftover from experimentally analyzing the amount of signal present here doing an AC sweep.  The 1u or 470n values will be perfectly fine.  They're just power supply noise filters, so if you have PS noise, then make it bigger.

[GGBB]

That's very helpful - fills in a lot of gaps for me. Thanks.

[Transmogrifox]

I had a look at your schematic now and it looks like it will work fine just the way it is. 

Some tips to make it so it "just works" right off:
You can eliminate R11 if you want because it doesn't do anything, but it won't hurt to leave it either.

To get into the values I was suggesting in my previous post just change R12 to 1k, and eliminate R11.

Then you could hardwire R10 as something between 68k to 72k, eliminate the bias pot, and probably come out ok.

If you do use the bias pot, the 100k value you have chosen will do nicely for dialing it into a range where the distortion is minimal.  If you have a meter and you can measure then aim for 6.75V on the drain.  Otherwise just adjust by ear.

Also, bias pot on BS170 isn't needed.  Just replace it with a pair of 47k resistors to get your V/2 reference.