Every once in a while reference to my old (maybe 2006 era?) posts about a modification to the EA tremolo surface. My first revision was based upon FNJ598J FETs which are no longer in production, and honestly not the best for use as a variable resistance element due to a built-in protection diode between gate and source.
I have greatly refined this design, simulated the heck out of it and then breadboarded it the other night to make sure it's everything I think it is.
What is it? It's a revisitation of the EA tremolo with the main improvement being the ability to modulate to infinite depth. The resultant waveform in simulation at max depth is a pleasantly rounded square wave so it does this trick a lot better than some might have predicted.
This document explains it quite a bit better:
http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/rEAgenerated_Tremolo_Design_Rationale.pdf (http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/rEAgenerated_Tremolo_Design_Rationale.pdf)
And here is a link to the schematic. The image is so large it would be inconsiderate to post it directly until I have had time to make a reduced size image:
http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/rEAgenerated_schematic_img.PNG (http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/rEAgenerated_schematic_img.PNG)
Or PDF if you prefer:
http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/rEAgenerated_schematic_pdf.pdf (http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/rEAgenerated_schematic_pdf.pdf)
Please let me know if there are mistakes. Otherwise I think it's ready for the first daring soul who wants another take on this old tried-and-true friend.
As time goes on I will eventually post some simulation results comparing the 2 circuits and maybe make a web page for the rEAgenerated Trem. If somebody builds this and loves it...and makes a through-hole layout, I am happy to post contributed layouts on my site.
Another thing to be added is the current KiCAD project. I created an SMT layout, but the project files would make a quicker step-up to re-assigning footprints to create a TH layout if desired.
For now I'm happy if anybody can make use of what I have linked for now.
OK.. Now your "infinite depth" get me in trouble to build it with descrete items... :icon_evil:
(another weekend on my cracked breadboard..) :icon_biggrin:
I do like cake.
that depth pot is brilliant.
I will pre-qualify my statements with the disclaimer that I don't know what I'm talking about, just trying to understand.
The depth pot confuses me. Doesn't D1 half-wave rectify the LFO signal, thereby changing the LFO wave shape? I'd imagine swapping the depth pot and D1 order could be a simple fix for that if necessary, but would that throw off Q3? And why no decoupling cap between the LFO and Q3 gate?
First thing I'll mention is that I have breadboarded this and it does work as expected. Sounds very similar to the original EA on lighter settings, and is a good chopper on more extreme settings.
Quote from: GGBB on September 03, 2015, 06:08:57 PM
Doesn't D1 half-wave rectify the LFO signal, thereby changing the LFO wave shape?
D1 (which is the LED) has very little effect on the LFO shape. I tried different placements of this in simulation and found this configuration to be the best for keeping the waveform suitably clean.
The common emitter BJT amplifier as used in the oscillator is a current mode device. The LFO ranges from 9V down to about 3V by increasing current draw through the 10k resistor and D1 -- at which point the oscillator turns around and goes back up by decreasing current. The current never changes direction: In this oscillator it's either positive or less positive ranging down to zero. A BJT doesn't make current flow in the reverse direction (unless severely saturated) so the LED doesn't effect the expected current flow.
Quote
And why no decoupling cap between the LFO and Q3 gate?
The reason for moving Q3 drain to +9V was to do away with the need for AC coupling.
The LFO wave shape above the LED ranges from about 6V to 9V. When it is at 9V, the FET Vgs=0, so the channel is wide open (low resistance). As it decreases toward 6V, the gate becomes negatively biased vs the drain and source. This causes the FET to pinch off, or increase in resistance.
The only reason I can think for ground-referencing the FET and AC coupling to the gate is perhaps due to power supply noise coupling into the amplifier.
The rEA has solved the power supply noise concern by means of significant gain reduction from the original design combined with the series RC to decrease power supply noise.
I should have a layout for 1/8W resistors, 10x11 perf, mill bypass included, for perusal by tomorrow. it needs some massage at the mo.
4 pots and a switch, this has too many controls for a 1590LB build, going to have to look at 1590A. Breadboard time coming up- DA keeps me busy.
Quote from: Cozybuilder on September 04, 2015, 11:44:07 AM
4 pots and a switch...
1 pot is optional (P2,P3 connection) if you still want to shoot for the 1590LB...but for that matter all of the pots are optional so maybe I haven't made a relevant point ???
"When JFET Q3 is completely pinched off, the impedance at this node is very high. This translates to signal attenuation to a level that is inaudible (infinite depth)."
Can you explain this further? I'm trying to understand why this attenuates the signal below unity. It looks to me that C5 is still AC gain in parallel with the Q2 assembly.
Quote from: midwayfair on September 04, 2015, 01:10:59 PM
"When JFET Q3 is completely pinched off, the impedance at this node is very high. This translates to signal attenuation to a level that is inaudible (infinite depth)."
Can you explain this further? I'm trying to understand why this attenuates the signal below unity. It looks to me that C5 is still AC gain in parallel with the Q2 assembly.
(http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/Gain.PNG)
Q2 is acting as a constant current source, and the emitter degeneration multiplies the impedance looking into Q2 significantly such that it is more than 10x the impedance seen looking into the 2Meg parallel Q3. I am also assuming a fully pinched-off Q3 is more than 10x impedance compared to R9,R10.
So you are correct, there is still and AC-coupled path going through C2, R9,R10 but as shown in the illustration this puts the equivalent gain at something like -50dB. Unless you are going into a cranked amp or really high gain distortion, -50 dB is pretty much inaudible.
It will likely be audible going into a high gain distortion, but it will cut through enough that I predict a non-subtle chopper sound will result from using it to drive a cranked amp.
It's a good point you have brought up, because it shows that increasing the values of R9 and R10 will increase the depth even more if you intend to drive a high gain amp or distortion pedal with the tremolo on the front end.
Another thing I haven't computed, but expect is negligible is the parasitic capacitance across Q2 and Q3. Since this is on the order of pF I assume the charge needed to drive R5 at a current to have an appreciable level is going to be in the ultrasonic to RF range.
Now as to your question about why it attenuates below unity, let me try to guess what is causing the confusion:
The signal level seen at the source of Q1 will be very close to unity. However, because it is causing very small currents to flow in and out of C5 and is making extremely small changes in bias current in Q2, only ~-50dB worth of this gets translated to Q1 drain. So there's a party at Q1 source, but Q1 drain barely feels a wiggle because the current in R5 is holding steady at the bias set by Q2.
Q2 is the reason this works.
One other caveat: If you drive this with a high amplitude signal it will cause Q3 to turn itself on in splatty blats. It's a handy thing the volume control is on the input.
[EDIT]
I went back to LTSpice to see if my assumption about impedance on the pinched-off FET is correct. It is correct as long as R9 is removed. It turns out the effect of R9,R10 linearising resistors significantly reduce the effective impedance looking into Q3 to 10's of kOhms, and output gain is on the order of -32 dB.
Remember, this has about 12 dB gain, so the difference between max level and min level is 12 dB - (-32 dB) = 44 dB, so it's anything but subtle at max depth.
If driving high gain pedals from the front end, maybe the first mod is to remove R9.
I hope that has helped to clarify.
For most uses my ears tell me the inclusion of R9 is an improvement while it sounds very abrupt at max depth. The LTSpice FFT shows me distortion is a bit lower on the rEAgenerated Trem vs the EA stock trem.
[/EDIT]
That does clarify it, thanks! I have not played with transistors paired that way more than a little bit (just a couple microphone circuits that I don't know what I'm doing with). The explanation was very useful.
I do think I'd want an input buffer with that 1M pot at the input. Is there a compelling reason or need to leave Q1 as a FET if I do that (obviously with proper biasing resistors changed where needed)?
Thanks T-x - the explanations are very helpful.
A couple of questions:
What should the polarity of C5 be? My guess would be negative toward R8, but I can't be sure from merely looking at the schematic.
I'm fairly sure that the volume pot should be reversed, and I think the rate pot needs have pins 1&2 joined not 2&3. Can you confirm that the depth pot is drawn correctly, with pin 1 at +9V?
EDIT:
One more question - will the DC on the depth pot make noise in the audio path when depth is adjusted? I know this happens in some circuits, not sure if it will happen in this one.
If C5 is a polarized cap it should be + connected to R8 and - to Q1,Q2.
It is correct that the depth pot is connected to +9V on one side, but pin 1 vs pin 3 I don't know.
As for volume & rate pot pinouts -- I will as much as admit that I didn't check these with datasheets from a specific pot. It's off-board wiring so it's easy to fix it if it's wrong.
My guess is the depth pot won't make audible crackle like most DC biased pots in the audio path. The gain on this depends on the level of the signal through Q3 since this is effectively a multiplier. A cap across the pot should fix that if there is audible crackle.
Quote from: Transmogrifox on September 05, 2015, 07:57:53 PM
If C5 is a polarized cap it should be + connected to R8 and - to Q1,Q2.
It is correct that the depth pot is connected to +9V on one side, but pin 1 vs pin 3 I don't know.
As for volume & rate pot pinouts -- I will as much as admit that I didn't check these with datasheets from a specific pot. It's off-board wiring so it's easy to fix it if it's wrong.
My guess is the depth pot won't make audible crackle like most DC biased pots in the audio path. The gain on this depends on the level of the signal through Q3 since this is effectively a multiplier. A cap across the pot should fix that if there is audible crackle.
Can anyone who has this breadboarded verify those pot configurations? In particular the depth pot orientation - needs to work so that clockwise increases depth. I have completed a layout with on-board pot connections but will need to make a change if the depth pot is opposite from what's drawn in the schematic.
I feel so ashamed, 2 days late and a row over. it couldn't be done 10 x 11, and there is a link to boot. I feel dirty. and those empty holes .....
(http://i.imgur.com/wMfP84H.png)
this layout is as yet unchecked, please advise if you see any bad layout/errors [R8 and C5 is a series circuit]. cap polarity is marked for if electros are used. the millenium is layed for a 2N7000. part numbers and transistors as per transmogs dia. pot connections are from the dia as well.
[edit :] v0.2 now showing, see below, and above. sorry about the now-excessive colour.
It looks good as far as I can see. "D1" reference on the millenium causes a little confusion with my schematic which uses that designator for the LED. If you change that to D2 or something that will refer to a part that doesn't cause confusion that way. Maybe I can redraw a schematic that includes the Millenium Bypass to make it more plain.
Thanks a bunch. It may not meet your own standards but I don't think there is any shame in empty holes.
As an aside this layout is also valid with 1/4W resistors if you don't mind "hair-pinning" them.
That's one helluva game of Tetris, duck! ;)
hair-pinning. well, I've never heard them called that before. and now I'm all grown up and use perfboard -sniff- I don't do that stuff no more.
tetris, yes, but who's winning?
Quote from: duck_arse on September 08, 2015, 10:56:01 AM
hair-pinning. well, I've never heard them called that before....
Hair-pinning is when you stand a resistor on end and bend the leads in the shape of a hair-pin:
(http://www.wb5rvz.com/sdr/common/images/resistor_hairpinning3.gif)
Compare to:
(http://carlasosenko.files.wordpress.com/2012/05/hairpin-meetup.jpg)
All I was saying is on your perfboard layout if somebody wants to use 1/4 Watt resistors instead of 1/8 Watt they could hair-pin the 1/4 Watt to make it fit :icon_cool:
Quote from: GGBB on September 06, 2015, 06:02:56 PM
Quote from: Transmogrifox on September 05, 2015, 07:57:53 PM
It is correct that the depth pot is connected to +9V on one side, but pin 1 vs pin 3 I don't know.
Can anyone who has this breadboarded verify those pot configurations? In particular the depth pot orientation - needs to work so that clockwise increases depth. I have completed a layout with on-board pot connections but will need to make a change if the depth pot is opposite from what's drawn in the schematic.
Bump. Anyone? Will your sim tell you T-x?
My layout is ready for the etchers out there, but needs this confirmed. Board mounted pots and LEDs - 1590B (2.2"x1.8").
> In particular the depth pot orientation - needs to work so that clockwise increases depth.
Look at the schematic.
When wiper is at the "+9V" end, trem is zero. When wiper is close to the transistor, wobble is maximum.
I dunno how you number your pots' legs, but hold a pot near your layout and it should be figurable.
(If we assume the Level pot is clockwise=louder, then the numbers on the Depth pot seem wrong.)
The Rate is absolutely a hard problem. The 1-2-3 leg issue is moot: if it runs backward just saw the jumper wiper-end and jump wiper to the other end. (Or note that in an R-C network big resistance is low frequency and work it out.)
But either way, smoothest action with normal "faster = clockwise" rotation will want a Reverse-Audio pot. Maybe not for a 100K pot with a 50K stopper, but here we have 100K and 1K and do need w-i-d-e range of resistance. A Regular Audio or Linear will have all the action in the last bit of rotation.
And no, you can't fix this with a "bend resistor". (Here our "pot" is working as a rheostat, not a potentiometer.)
_________________________________
> Doesn't D1 half-wave rectify the LFO signal, thereby changing the LFO wave shape?
The LFO AC is riding on the average DC flowing through the LFO transistor. The transistor current never *reverses*. It may get near zero but only for a blip. There's one-way current nearly all the time.
Quote from: GGBB on September 09, 2015, 07:43:23 PM
Bump. Anyone? Will your sim tell you T-x?
Ah...maybe PRR better understood the question you were asking. (Thanks PRR for chiming in) He has correctly identified what needs to happen from the point of view of circuit operation:
1) Clockwise rotation of the depth pot should move the wiper toward the LED. Minimum depth is when the wiper is effectively shorted to +9V.
2) Clockwise rotation on the rate pot should decrease resistance. Maximum rate is when the series resistance is at its minimum.
3) If you add the optional P2,P3 Pot, then maximum resistance across those two points is maximum depth.
duck_arse layout has these correct from this higher-level point of view.
My simulation won't tell me whether I got my pinout right because it all depends on whether CW rotation moves Pin 2 toward Pin 1 or toward Pin 3 and those assignments are whatever I arbitrarily decided in SPICE, while the real the pinout depends on a specific pot. That's what I didn't know.
If you were asking for the more simple answer, then I will need to take out a pot and a DMM to verify what happens between pins 2&3 on CW rotation.
It looks like your initial question was right:
In my schematic the
rate and depth pots are backward from the volume pot. So whatever pinout works correctly for volume is the reverse of rate and depth. [EDIT] Upon closer examination Rate pot is correct if Volume pot is correct. Depth pot is the only one to break convention [/EDIT]
I've been digging around for a while trying find the most enticing tremolo to build.
Mind sharing a sound demo? :icon_biggrin:
Quote from: Kevin Mitchell on September 10, 2015, 09:08:18 AM
Mind sharing a sound demo? :icon_biggrin:
Maybe this weekend I'll have some time to plug it into the puker and record a clip or 2. I can post audio samples from the simulation, but somehow it doesn't capture the more fluid effect of tweaking knobs and dialing it in more organically.
I never call pots 1-2-3, so I just lucked it if my connections correspond to transmogs designations.
I have tried the bend resistor on the rate pot before, trying to get a bit more resistance range, and was less than impressed. I used a 'C100k' for the trololo I built, nise and smooth.
Quote from: PRR on September 09, 2015, 11:38:37 PM
> In particular the depth pot orientation - needs to work so that clockwise increases depth.
Look at the schematic.
When wiper is at the "+9V" end, trem is zero. When wiper is close to the transistor, wobble is maximum.
I dunno how you number your pots' legs, but hold a pot near your layout and it should be figurable.
(If we assume the Level pot is clockwise=louder, then the numbers on the Depth pot seem wrong.)
The Rate is absolutely a hard problem. The 1-2-3 leg issue is moot: if it runs backward just saw the jumper wiper-end and jump wiper to the other end. (Or note that in an R-C network big resistance is low frequency and work it out.)
But either way, smoothest action with normal "faster = clockwise" rotation will want a Reverse-Audio pot. Maybe not for a 100K pot with a 50K stopper, but here we have 100K and 1K and do need w-i-d-e range of resistance. A Regular Audio or Linear will have all the action in the last bit of rotation.
And no, you can't fix this with a "bend resistor". (Here our "pot" is working as a rheostat, not a potentiometer.)
_________________________________
> Doesn't D1 half-wave rectify the LFO signal, thereby changing the LFO wave shape?
The LFO AC is riding on the average DC flowing through the LFO transistor. The transistor current never *reverses*. It may get near zero but only for a blip. There's one-way current nearly all the time.
Thanks Paul. That helps. And Thanks T-x as well for your explanations and most importantly all the work that went into this circuit. I love my ROG EA Tremolo, but I want more depth, and it needs rebuilding since there is some dodgy solder work that is acting up - it was my first pedal and was on perf. Eager to replace it with this one.
Here is my contribution for the etchers in the crowd. Untested as of yet - proceed at your own risk. This thing is monstrous at 2.2"x1.8" compared to duck's but the pots and LEDs are all on-board (cause that's how I like em) and it should fit in a 1590B just fine. Note that the LFO caps need to be ceramics to fit on the board - unless you can find very small 1uF film caps or you go with a lower value.
(http://www.aronnelson.com/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=51935) (http://www.aronnelson.com/gallery/main.php/v/diyuser/GGBB/Tremolo/rEAgeneratedTremolo_Rev3C.png)
Thanks for doing that. It's really nice.
Do I have permission to host it on my site with the original schematic?
Also, I didn't add a reverse protection diode because I determined this circuit will not be damaged by reverse power supply connection...for whatever that is worth.
> I determined this circuit will not be damaged by reverse power supply connection...
I'd feel better if it was built and wrong-connected to 12V for test, but...
FWIW, as I see it, you are 99% correct. From "+9V" there are just two paths. Through the 10K Rate pot, and whatever is under that can only get 1mA, and worst-case 10mW in the pot, which is all safe. And through the "5.23K", which could flow 2mA, which still leads to all-safe.
The 1% concern: if C4 audio power filter is electrolytic (as it must be, though you don't note polarity), and power is backward, we have a backward-cap near-short, so nearly 9V across R7 470r. This will flow nearly 19mA and put 0.172 Watts in R7. If R7 is 1/8W, this is technically over-spec and it is allowed to die. (However 680r would supposedly live and hardly affect audio path bias.) A lesser chance is that C4 holds say 4V, flowing 10mA, dissipates 0.040 Watts... it probably won't burst, but may not recover (leading to unexplained excess power crap on the signal).
There may be some Akido way to cancel power crap without a large cap. Personally I'd use 1/4W parts (or 680r) and then not worry about it.
Higher voltage would lead to more failure modes, but the history of electricity (and Users) teaches that there is no 100% way to protect against overvoltage (or carelessness). However at a glance, the thing may be safe to over 30V (or to the limit of C4). It mightn't work right until re-biased for the high supply (or put back on 9V), but shouldn't die.
me> ....there are just two paths.
Idiot. I missed that Q3 is tied to 9V_SIG. However this third path can only return through C5 or R10, both large impedances, and no smoke-releasing current can flow.
The "+" side of C5 goes toward Q3, which was not obvious at a glance. I think Gord got this right.
a series schottky could be hairpinned into my layout, next to D2, by opening that orange trace. for the belts and braces types.
nittypicky detail typo: GGBB you wrote 1N5718 when I'm guessing you meant 1N5817. Might cause some minor confusion. :)
This looks really exciting. Gotta build this at some point with a buffer -> fixed voltage-divider in front of it. Not a fan of extra volume controls.
Quote from: knutolai on September 15, 2015, 02:58:02 PM
This looks really exciting. Gotta build this at some point with a buffer -> fixed voltage-divider in front of it.
It doesn't need a buffer. R3 in parallel with the pot are the components determining input impedance -- your guitar looks right into the JFET gate otherwise. Putting another in front is just duplication of what is already there. R3 can be increased if you feel 1Meg is not enough.
To change to a fixed gain you can simply omit the volume pot. To set the fixed gain, change the value of R8 to whatever suits you best. For unity gain change R8 to something close to 4.5k. This will avoid adding parts for a resistor divider, and especially an extra input buffer.
I prefer gain at a bit over unity so it doesn't sound like a volume drop when active. A gain of 2 usually sounds the most natural between on and off -- psychoacoustically averages to unity at max depth. For this R8 = 2.2k to 3.3k probably would be the most natural sounding between active and bypass.
In either case you can most easily make this a fixed-volume unit using either of the layouts posted in this thread: nothing more than a component value change is needed.
Typical use of the volume control on this circuit would be a set and forget until you change your rig. At the same time by setting depth to min, this serves as a clean +10 dB booster, so the volume will be useful for anybody who likes to tweak knobs.
FWIW: The gain is designed to yield approximately unity gain output when the pot is an audio taper pot turned to the 12 o'clock position.
For those concerned about the pot at the input vs input impedance -- there is nothing magical about the pot being used as a volume control to make this circuit need a buffer. With the pot in place it presents a minimum of 500k impedance to the guitar when turned all the way up (my target input Z), and increases to about 950k at 25% (this is where the attenuation of the pot and the gain of the tremolo gain stage yield unity gain output and also corresponds approximately to 12 o'clock position on an audio taper pot).
Quote from: Transmogrifox on September 15, 2015, 07:54:10 PM
Quote from: knutolai on September 15, 2015, 02:58:02 PM
This looks really exciting. Gotta build this at some point with a buffer -> fixed voltage-divider in front of it.
It doesn't need a buffer. R3 in parallel with the pot are the components determining input impedance -- your guitar looks right into the JFET gate otherwise. Putting another in front is just duplication of what is already there. R3 can be increased if you feel 1Meg is not enough.
To change to a fixed gain you can simply omit the volume pot. To set the fixed gain, change the value of R8 to whatever suits you best. For unity gain change R8 to something close to 4.5k. This will avoid adding parts for a resistor divider, and especially an extra input buffer.
I prefer gain at a bit over unity so it doesn't sound like a volume drop when active. A gain of 2 usually sounds the most natural between on and off -- psychoacoustically averages to unity at max depth. For this R8 = 2.2k to 3.3k probably would be the most natural sounding between active and bypass.
In either case you can most easily make this a fixed-volume unit using either of the layouts posted in this thread: nothing more than a component value change is needed.
Typical use of the volume control on this circuit would be a set and forget until you change your rig. At the same time by setting depth to min, this serves as a clean +10 dB booster, so the volume will be useful for anybody who likes to tweak knobs.
FWIW: The gain is designed to yield approximately unity gain output when the pot is an audio taper pot turned to the 12 o'clock position.
For those concerned about the pot at the input vs input impedance -- there is nothing magical about the pot being used as a volume control to make this circuit need a buffer. With the pot in place it presents a minimum of 500k impedance to the guitar when turned all the way up (my target input Z), and increases to about 950k at 25% (this is where the attenuation of the pot and the gain of the tremolo gain stage yield unity gain output and also corresponds approximately to 12 o'clock position on an audio taper pot).
I have never encountered a situation where a large pot at the input of a circuit, when turned down, did not result in a noticeable loss of treble (even just the guitar's volume pot does that), not even other JFET circuits. That was the reason for suggesting a buffer and a smaller value for the volume control. It's only a few parts in a pedal that can EASILY fit in a 1590A. Though I like the idea of just trimming R8 as well (and in fact I'm surprised you didn't go with that.
Quote from: midwayfair on September 15, 2015, 08:47:06 PM
I have never encountered a situation where a large pot at the input of a circuit, when turned down, did not result in a noticeable loss of treble (even just the guitar's volume pot does that), not even other JFET circuits.
As a general rule what you are saying has sound reasoning to support it.
What you're referring to is the Miller capacitance multiplier effect against parasitic capacitive impedances looking into a FET gate (or BJT base).
Here's my answer in this particular case:
[JFET_Miller_FX-0] $ crss=3p 2N5457 reverse transfer capacitance
crss = 3.0000p
[JFET_Miller_FX-1] $ ciss=7p 2N5457 input capacitance
ciss = 7.0000p
[JFET_Miller_FX-2] $ Av = 5 Volts/Volt max voltage gain
Av = 5.0000
[JFET_Miller_FX-3] $ cmiller = crss*(1+Av)
cmiller = 18.0000p
[JFET_Miller_FX-4] $ c=cmiller+ciss
c = 25.0000p
[JFET_Miller_FX-5] $ rpot = 1M
rpot = 1.0000M
[JFET_Miller_FX-6] $ rpotTheveninMax = [(rpot/2) | (rpot/2)]
rpotTheveninMax = 250.0000k
[JFET_Miller_FX-7] $ r=rpotTheveninMax
r = 250.0000k
[JFET_Miller_FX-8] $ f3dB=1/(2*pi*r*c)
f3dB = 25.4648k[Commandline Calculator: http://sourceforge.net/projects/crunchcalc/
(http://sourceforge.net/projects/crunchcalc/)]
3 dB cut-off will be something like 25 kHz, using maximum stated values in the datasheet. Other JFETs will be similar. Difference? Low gain configuration.
Now add guitar output volume pot to this:
[JFET_Miller_FX-9] $ r = r+r/2 Volume pot is 500k
r = 375.0000k
[JFET_Miller_FX-10] $ f3dB=1/(2*pi*r*c)
f3dB = 16.9765kThis is how I rationalize it to myself -- and also AC simulation in spice shows a -3dB cutoff of about 70 kHz with pot set to 50%, so this agrees my worst-case calcs are truly worst case.
In the case of removing the volume pot entirely and hard-wiring the gain into R8 -- you're now only looking back into the guitar output impedance, or at worst into the volume pot so the situation improves dramatically.
Because the drain-source capacitance is multiplied by the voltage gain of the gain stage you can see how this claim quickly falls apart as you increase the voltage gain. I think you will find most places you have experienced tone-sucking from this has happened when the gain of the amplifier stage is more than 5. A gain of 10 or more might bring this into a realm that may be considered audible, especially combined with filter formed by cable and guitar volume pot.
QuoteIt's only a few parts in a pedal that can EASILY fit in a 1590A.
I agree with this, so you definitely won't break the bank nor force yourself into the next size up on doing it with a resistor divider and additional buffer ;)
QuoteThough I like the idea of just trimming R8 as well (and in fact I'm surprised you didn't go with that).
I considered it but perhaps not closely enough. It may be a good design change/update.
As for what I was thinking, I tasked myself with the objective to be able to take output from a really hot input source and bring it down to a level where it won't clip yet still preserves the full headroom available in this configuration (~6Vpp output swing). Is there such a hot source as this with which the guitarist has no control over the output level? I don't know. It was just an arbitrary design objective.
Zvex SHO may be an example of this -- If you crank it to where it's coloring your tone the way you like it's not like you have another volume control. So if you switch from an SHO and drive into this trem, you would want to pull down the level at the trem input until the output is close to the pk-pk level of the SHO coming in. Where this example falls apart is -- why wouldn't you put the trem in front of the SHO? I don't have an answer.
Food for thought ;D
Fair enough! You know I can't do the math on this stuff, and you're right, the best examples gain is ... let's say WAAAAAAAAYYYYYYYYYYY above 5. Charitably. ;)
I appreciate your input. I think you have a lot more real hands-on experience with this stuff in terms of using it to make well-done music. I just like to play with my guitar -- I'm a "bedroom rock star".
I was fully prepared to say "you got me on that one" because you really did point out something I had failed to consider :P. In this instance it works out, but it's definitely noteworthy to point out that this kind of configuration can turn bad with a little more gain and demonstrates that the examples you have in mind are probably not subtle.
Quote from: knutolai on September 15, 2015, 02:58:02 PM
nittypicky detail typo: GGBB you wrote 1N5718 when I'm guessing you meant 1N5817. Might cause some minor confusion. :)
Good eye! I've made the correction. If anyone builds it, please let me/us know here so others will know its been verified.
Thanks.
A demo with clean guitar is now added:
http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/demo/cleantrem1.mp3 (http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/demo/cleantrem1.mp3)
I have a bunch (probably 20 minutes worth) of clean and distorted stuff recorded now, so as I have time to edit down the raw recordings to things that demonstrate unique qualities of this circuit I'll keep posting them here:
http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/demo (http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/demo)
I'll probably be putting stuff bit-by-bit over the next couple of weeks until I have weeded out the redundant and/or uninteresting noodling about.
Also started a project web page, now with link to a zip file containing the KiCAD project file and gerbers for my SMT design (unverified at this point).
http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/pages/index.html (http://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/pages/index.html)
You will notice on the page I also added a Rev B design which includes a capacitor in the FET gate linearizing resistor. In Rev A the LFO DC biases the FET through this resistor limiting how deep the depth can get. I didn't notice this as a deficiency until I put this on the front end of my amp's drive channel -- cranked. It still is a very non-subtle trem effect, but it doesn't effectively mute on the off cycle, just trems between clean and dirty (which is a pretty cool effect in its own right).
With the capacitor in place, LTSpice tells me it goes from 30 dB depth to about 58 dB depth by adding this cap (same if in Rev A you just remove R9).
Rev B is probably a good choice for anybody wanting to use this to drive a high gain distortion from the front end...downside is it comes after these good folks have created layouts for Rev A. OTOH, Rev A is good, so it's a worthwhile build for anybody wanting an EA trem with more depth.
Quote from: Transmogrifox on October 13, 2015, 07:09:15 PM
You will notice on the page I also added a Rev B design which includes a capacitor in the FET gate linearizing resistor.
I've added that change to my layout - stroke of luck that it was a simple change.
Nice work. Do I have permission to copy and serve that image on my site? At the very least I could link to it, but I like to have as much as possible hosted on the site so that if the site is up, all it's goodies are there too.
Thanks
Quote from: Transmogrifox on October 13, 2015, 10:04:24 PM
Nice work. Do I have permission to copy and serve that image on my site? At the very least I could link to it, but I like to have as much as possible hosted on the site so that if the site is up, all it's goodies are there too.
Thanks
Hosting it is fine (thanks!), but please also provide the link to the original: http://www.aronnelson.com/gallery/main.php/v/diyuser/GGBB/Tremolo/rEAgeneratedTremolo.png.html.
RevB - done - needs checking. wangle your own P2 and P3.
(http://i.imgur.com/4W3Jxem.png)
Very tidy layout DA, I note 3 differences to the schematic:
1) S and D are reversed for both Q1 and Q3 (OK)
2) Cap C9 has been inserted between R9 and junction Q3G and R10
3) C5 and R8 order reversed (OK)
Quote from: Cozybuilder on October 15, 2015, 05:52:08 PM
2) Cap C9 has been inserted between R9 and junction Q3G and R10
C9 looks correct in the layout to me. This is the Rev B update so maybe you are comparing to Rev A?
I was :icon_redface:
Long-time lurker, first-time poster. Giving this a 5-year necrobump to say that I just breadboarded this circuit, and it’s fantastic. I had to make some tweaks to get the depth control to play more nicely — as the pot approached maximum, the tremolo would die out, and then reappear at max, but very distorted. I also found it took a while for the travel of the pot to kick in the tremolo at the minimum. My solution to both issues was to swap the 10K pot for a 5K, and put resistors on either side (currently I have 2.2K going to +9V and 3.9K going to the junction of Q4C/C6/R11). That has improved the control and usability of the depth pot significantly — and plenty of chop still.
I’m also still tweaking the rate values. The stock 1µF configuration was too slow for me, so I doubled them up in series (I have a lot of 1µF lytics around) to drop each of them to .5µF, which has been great so far, but I have yet to alter the 1K resistor R13 to see what that accomplishes (I’m shooting for maximum rate range on the one pot). Going to play with that next. (Edit: tried messing around there, but couldn't improve on the configuration above.)
I’m splitting the difference on the gain/input issue — I initially bypassed the 1M volume pot entirely and put a 5K trimpot on R8, but my hottest guitar did clip it a bit at its loudest. So now I’m planning on padding the input with fixed resistors and using the trimpot to fine-tune it. It does indeed have plenty of extra gain. I prefer the “set-and-forget” approach to the volume control.
I had planned/hoped to omit the blinking LED as I’m not really a fan of them on pedals in general, but I put one in the circuit for testing purposes (and because it was helpful to see when it was actually oscillating). When I bypassed it and connected the depth pot/resistor combo directly to Q4C, the maximum rate increased ever so slightly, and since I want as much range as I can possibly get, that helps me feel just fine about ditching the LED.
This is by far the most functional and versatile transistor trem I’ve heard yet! Thanks to Transmogrifox for coming up with this one, and to everyone else for their input as well…
welcome to posting, pacealot.
changes to the collector current will change the osc range. there are different ways to add the flashing led, and I can't remember, and you haven't shown, how it is implemented in this inst. there are ways to get a much wider range from this type oscillator, and steady consistent output at both extremes - some involve staggered cap values, some w/ caps on range switches, some using dual-gang rate pots. and some people use a mosfet instead of a bjt.
Thanks, d_a — I'm already beginning to notice the interactive effects on both depth and rate in changes to Q4 collector current. Right at the moment I'm trying to settle on the most likely "usable" range of rates from a series of new experiments on the values of caps 6, 7, & 8. Interestingly, I'm getting better results with a slightly higher value for C7 than the other two — could be a quirk of the specific 5089 I have in there, not sure. This is actually the fun part for me — auditioning values until the "best" ones present themselves, and then trying to replicate it upon boxing up!
Regarding the LED, I'm honestly much happier omitting the LED, so I haven't really come to grips with how it's designed into this specific circuit. I know people seem to like having a blinking light on pedals with LFOs these days, but I'm not one of them!
I do have a dual anti-log 100K pot on order actually, so maybe when it arrives I can play with that as a means to really get absurd in the rate department!
if you stagger the values of the three phase-shift caps, you can extend the range of willing-to-oscillate. shoot me a pm with an email addy, and I can send you some lfo diagrams on the q.t.
Done — PM sent. Thanks so much d_a!!
A brief follow-up on my experiments in getting more range on the rate sweep:
I explored the dual-ganged pot situation by adding an extra cap/resistor stage between C6 and C8, and taking that resistor to the other side of the rate pot. I also followed duck's suggestion to play with staggering the cap values, and I also changed a number of resistor values as well. What I came up with was touchy to get working, but once I settled on some values, it became the largest and most versatile sweep of any tremolo I've experienced yet.
I was going to try subbing in a MOSFET as per duck's source materials, but this arrangement has worked out so well that I decided to leave well enough alone for the time being! (I do still have some extra 5458s and 5089s and a 2N7000 I've not used yet, so I could continue exploring further later....)
Here's where my version of the circuit ended up (I renumbered the components to try to make more sense with the new arrangement):
(https://i.postimg.cc/F1ftKHw5/REA-Schematic.png) (https://postimg.cc/F1ftKHw5)
I realised that I made a mistake in the previous schematic and connected Q3 source to the wrong voltage source. Here's the correct one:
(https://i.postimg.cc/T5KCnRS3/REA-Schematic-Corrected.png) (https://postimg.cc/T5KCnRS3)
After some, not very successful, experiments with the EA Tremolo, I came to think of rEAgenerated Tremolo instead. After having it on breadboard for some days and trying some modifications, I decided to make a permanent build and made a vero layout as I couldn't find one. Click image to go to the gallery for full size.
(https://www.aronnelson.com/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=54138&g2_serialNumber=2) (https://www.aronnelson.com/gallery/main.php/v/Cab42s-layouts/rEAgenerated_vero.png.html?g2_imageViewsIndex=1)
It was a pretty easy build other than it is a bit tight around C6, C7 and C8 if you have large capacitors. I used J201's instead of 5458's. Does that make much of a difference?
A couple things was a bit odd, though. On breadboard, much of the rotation of the depth pot had no effect both in start and end. I used the suggestion by Pacealot above and tagged some resistors in series with the pot and another across the pot to lower the value. However, on the final build I had no such issue. I also had issues with distortion on breadboard, that went away in the final build.
I tried different values for C6, C7 and C8 for rate. I liked 680n, but the ones I had were huge, so I built it with 1u's, but socketed them, for when I get something smaller.
I might also see if I can get away with ditching the volume pot.
Thanks to Transmogrifox and every one else that has contributed.
Nice work! I really love this circuit. So much so that I've built three versions of it so far (including a stereo one for my bass rig which I still need to populate onto a PCB). The only area that requires re-tweaking each time has been the resistor network at the depth pot which you mention. I think that's due to the variation between JFETs (I've used 5458s, 5457s, and most recently BF244Bs, all with equally good results), so if your J201s are working, then I'd assume they're just fine. But I have had to dink with the resistor values there with each iteration to get the optimal depth range. It does seem to prefer a total of 10K at that point, just portioned out slightly differently each time.
Also Duck's suggestion of staggering the LFO caps to increase overall range is a very good one, and tantalums would fit nicely size- and value-wise (and in my experience tend to make for a "cleaner" oscillation shape than regular aluminium 'lytics). Since you socketed those, it could be beneficial to play around with some there...
Quote from: pacealot on May 31, 2020, 01:38:20 PM
...but very distorted.
It's fun to see people improving upon the idea and tweaking further to their taste :) I'm happy to see it's getting some mileage.
Some notes to help others understand why the Rev B rEA ended up the way it did. I hope these notes will help other builders to decide whether to start with some of the modifications/improvements suggested in this thread or try the original first.
In addition to the design rationale doc, which summarizes the overall purpose of each piece (https://cackleberrypines.net/transmogrifox/rEAgeneratedTremolo/pages/projfiles/rEAgenerated_Tremolo_Design_Rationale_RevB.pdf), these thoughts may be useful:
With input volume maxed, that was a "feature" that it would become distorted, and also has significant gain boost. I used the volume knob to drive my amp into a pulsating distorted sound, and also by adding a switch disabling the oscillator it could be used as a booster for getting a good crunch out of my amp. For a clean tremolo, the volume knob was always dialed down to a level where the peak output was somewhere between unity to 6dB. With those input volume settings I did not notice the distortion as being prominent -- it was a purposeful decision the volume control was placed at the input.
For a transistor amplifier like this also using a JFET as a variable resistor the best way to reduce distortion is to decrease the amplitude of the signal subject to the devices (smaller gate-source voltage, or for the variable resistor, smaller drain-source voltage). Best to attenuate at the input and amplify by larger drain resistor (R5).
If going for a fixed configuration ( no volume pot) the best setting if you want the loudest part at about unity would be somewhere around 1/5 resistor divider ratio at the input (e.g replace 1M pot upper leg with 750k, lower leg with 250k).
As for the LFO depth control range, I agree it can become fairly extreme. The purpose in the rEA implementation is after a certain point it begins to change the duty cycle rather than the depth of the effect. This is a feature for those who like to exploit the chopper type of effect, but definitely limits the usable range for those who want to dial in more of the smooth pulsating tremolo effects.