Two years ago I attempted to build a tremolo pedal for a friend. I tried various different designs that I found online, but found they distorted the signal and/or added a throbbing sound to the signal when there was no input signal. I ended up with various half working designs, but none that I was really happy with. Last weekend I decided to re-visit the problem, however I still couldn't find a design which sounded right. It may just be that I failed to get the designs to work correctly, but either way none of them gave the effect I wanted. So instead I sat down and worked out my own. It's a very simple circuit so it may well have been done before, but having googled around I couldn't see one the same it. It's based off the Trill Tremolo circuit, however uses a JFET to avoid distortion, and adjusts the input signal rather than the base current to give a more linear depth adjustment. Small changes, but I found they make a big difference to the sound.
(http://i.imgur.com/Mrhrw4t.png)
The depth goes from completely off to full cut, and the fact it isn't modulating the gain and that it's using a sine wave LFO means it gives a very smooth and natural sounding effect. The balancing amp is set so that when the depth is set to zero, the input is at the same volume as the output. It also has almost no throbbing sounds on the output.
I dug out the original pedal from two years ago and replaced the insides with the new circuit; it's looking a little worse for ware having been kicking around the bottom of a box in my uni room :L
(http://i.imgur.com/7McWA6m.jpg?1)
Below is the old circuit (left) compared to the new one on the right. I mostly use surface mount components these days so it was gonna be neater anyway, but even so it's a big improvement on size and sound.
(http://i.imgur.com/P6QOpmX.jpg?1)
But anyway, I hope someone finds this useful; I'll hopefully get some proper test audio from it on Sunday.
It seems similar to the Univox U-65RN amplifier that has a bipolar transistor in place of the JFET:
(http://www.eserviceinfo.com/preview/Univox_U65RN_guitamp.gif)
You can get a larger readable image here as a .pdf file:
http://www.schematicsunlimited.com/u/univox/univox-u-65rn-amplifier-schematic
It shows the tremolo oscillator on the lower left feeding a bipolar transistor that is connected directly to the input. I have one and it works.
looks interesting, I've just given my BB 2 sideways glances .... looks similar principle as the vico vibe. any reason for choosing a j201?
Quote from: duck_arse on December 20, 2014, 09:01:11 AM
Any reason for choosing a j201?
It was just the only JFET I had in my parts box at the time.
The reason for using a JFET instead of a bipolar transistor as I found that it distorted the signal as can be seen below (the shape and voltage varies with base bias voltage and input voltage, but this gives an idea) :
(http://i.imgur.com/9sIDaQ0.png)
The JFET by comparison didn't have any of that.
In reality, if it's set up right then the distortion only occurs when the signal amplitude is being pulled down by the transistor, so it's not very audible. But the JFET removes any problems with setup and just made for a cleaner effect. I had spent several hours attempting to get the bipolar version to work before swapping it out for a JFET.
Maybe a buffer in front to raise the input impedance?
Could do, though it might be easier just to increase the input resistor values proportionally; again, they were picked based on what I had in my parts box. It's not too low so is working fine atm, though it's possible that some of the finer detail of the signal is being lost.
It could also do with a 1 Meg pull down on the input on the schematic; it's there in the pedal, i just forgot to add it to the diagram.
hmmmmm .......
I've never used the fet as the modulator except with great gobs'o gate bias. I'll be interested in the envelope shape, any pics of yours?
It works because the coupling capacitor lets it bias itself, as can be seen in the simulation below:
(http://i.imgur.com/Cs1sNoa.png)
It says it takes 30 seconds in the simulation, but I've never actually noticed much change in the sound as it 'warms up'.
The image below is taken from an actual recording showing that it settles to about a 50:50 ratio:
(http://i.imgur.com/qj9MRTt.png)
mmmmm, I rekon it might be biasing on leakage from that 10uF. what happens when you put a 1uF poly there, trubble?
I was thinking about this trem and attempting to "harmonic-ise" it. interested? possible you think?
It won't be from capacitor leakage, as the capacitor used in the simulation is an ideal one so has no leakage. I believe it's actually gate leakage in the JFET, as can be shown in the image below; this is taken from the gate of the JFET (The previous simulation was showing the other side of the 330k resistor).
(http://i.imgur.com/nADq0Ts.png)
Definitely interested; how'd you suggest going about it? I've already given this pedal to the guy it was made for, but I can easily check out new designs in LTSpice..
well, except for having no idea, I do have an idea, which doesn't involve a dual pot. I'll have a swing at something tomorrow, maybe rough a circuit diagram. see what goes. I don't have any circuit sim stuff, I don't think I have the processor power/download speed/operating system/stomach/need for one, really, but could use all the help I can get.
Well I can simulate most discrete component (including vacuum tube) analog circuits and run audio samples through them to see how they sound with LTSpice, so if you can work out a rough circuit diagram I can draw it up and see how it'll sound :L
Ok, so I actually had a little go myself and this is what I came up with:
(http://i.imgur.com/pfm3lKk.png)
This is literally the very first draft, and the filter values were picked almost at random, so a lot of work is probably still needed, but this gives a decent skeleton with which to work (the type of JFETs in the diagram are only used as I don't have models for J201s).
It sounds pretty decent too; tremolo starts about 6 seconds in:
https://drive.google.com/file/d/0BzfEd3tVJdxRS2xONHpHRkxGWkU/view?usp=sharing
Quote from: Rapidrory on December 23, 2014, 07:15:23 PM
It sounds pretty decent too; tremolo starts about 6 seconds in:
Delicious!
You're saying you built with J201's though the schematic says 2N4117, right?
Is the track a sim or the actual circuit you built?
EDIT: I'm a dork. You built one, you sent it on, then you redesigned it for the harmonic trem and sim'ed that.
Reading comprehension FTW!
2nd EDIT: I would tend to agree with the comment above - you'll want to load a guitar with a higher impedance than the 50K or so you have there. Consensus seems to be that 1M is a better value for a guitar straight in. So either scaling the filter sections or a buffer would seem to be a good idea. The reason it's not showing up in the sim is probably that you aren't modeling the guitar pickup ahead of the input.
Take a look at the front end of Gus' sim here:
ASDF (http://www.diystompboxes.com/smfforum/index.php?topic=104029.0)
Everything to the left of the spot marked "input" is a simulation of the guitar pickup's parameters.
Yeah, I know input impedance should be ~1M, but I just hadn't scaled it for the sim; as I say it's a very rough first draft atm. I'm planning to increase the input resistance rather than add a buffer as I'm trying to keep parts count to a minimum with this one... though i guess it may add a lot of Johnson noise with the high resistance. Have to see how it goes.
I'm also gonna see if it's possible to switch back to 2N3904s instead of JFETs for the modulators just to try and make this build as simple as possible for other people.
no, no, no, stay with the jfets. you need a new name, though, it has about the steepest, squarest envelope on a trem I've ever breadboarded. real deep. you -can- round it up a bit, if you want to tune the fet to the osc signal with a voltage divider.
the phase splitter simplifies to C2-4u7, R22-56k, drop R23 and C11, R1 about right, R20 to 68k, and R2 and R4 to 10k. 1M gate resistors means C1 and C8 can be much smaller.
as the first diagram stands, the (my) balance amp distorts the bottom half of the signal. the input Z of this stage is around 38k, which is a bit low, but workable with C7 increased maybe. perhaps an R to ground like in the big muff will improve.
as for the modulator, the j201 works surprisingly well straight off. I tried a 2n5457 (yes) and mpf102 (NO!) because they were at hand, I'll run through a bunch of others as time permits. (and the depth pot seems to need either a reverse log taper, or a log taper wired backwards at the very least.)
[how big is your audio file?]
Yeah, tried it with bipolars and it was pretty terrible so definitely sticking to the JFETs :L
R22 and R23 were set up as a potential divider to lower the signal so as not to distort the output, but I'm guessing that you can combine the two functions using the bias resistors as the other half of the potential divider.
I've updated the sim to your suggestions and it works quite a bit better with those values :) Sounds roughly the same, but is better electronically
The balancing amp's gain needs adjusting down to match the input values a bit better, which may help with the distortion. Not much more I can do on this till after tomorrow unfortunately; apparently family want to see me at some point over Christmas :icon_rolleyes:
The output audio file is a 2MB .wav file. Also, if you have any better audio samples feel free to send them over; I only have ones I can find online :L
if you gaze upon RG's ersatz fender-pro circuits, or the pareidola (what now?) circuit, or the cardinal, you might get some more splitting/filtering ideas. the high pass/low pass aspects baffle me, they might come clear on a sim, tho.
[edit :] and merry xmas, everybloody!
Very good sounding sample ! ..and nice project too !!
some thought...
in Q3 could be used a 2N5089 like in some other design for compressor...
while for Q2 using 2N3565 should improve the sound contents.
Looking forward for updates.
cheers
I'm going to echo the suggestion for an input stage of some sort, though. A buffer should be sufficient in this design since you're using the make-up gain stage instead of a gain stage for each band.
I've had terrible luck getting FETs to work without added distortion, though. Is there something in this one that helps with that?
(Also, you may find that you want to lower R25 some to keep the low band from overpowering the treble band.)
Also: Can R23 and R22 be your depth control?
It might be possible to make active bass cut and treble cut filters to split the signal into the high and low parts more effectively, and also double them up as high impedance buffers. A JFET buffer would be pretty easy and only add one extra component, but if it could be done with two and get a better filter setup then that'd be worth the extra component. I'll have a look adding them after tomorrow..
I'm still attempting to come up with a solution that uses as few different components as possible to keep it easy to build, so I'm trying to stick to just J201s and 2N3904s if I can without it affecting the tone too much.
Might also be good to make R30 and R31 into a single potentiometer to adjust the mix between high and low tremolo tones.
I tried with yr last set of values and a fet input buffer, found that reducing R21 to 33k (just throwing resistor at it, really) made the effect MUCH more pronounced. the jfets without bias turn the lfo into square waves, so plenty of distortion in that respect.
I'd originally thought of running the low section with a single gang depth as per the original circuit, and the high section full depth, then mixing the high into the low after the balance amp, but my understanding of the workings of this circuit is not real good just yet.
un-balancing the upper and lower resistors in the phase splitter was an idea in a thread on the ersatz fender pro, might be worth a pot or fiddle.
> It works because the coupling capacitor lets it bias itself
The plan in post #1 (http://www.diystompboxes.com/smfforum/index.php?topic=109499.msg1001697#msg1001697) is "flawed" because there is *no* bias to the JFET gate, it does not "know" where to sit. Bad engineering to leave stuff floating.
In fact I am sure it works OK (JFET resistance goes up and down). The un-biased Gate is pretty sure to leak toward the S and D pins, which means toward ground. JFET resistance will be low. When LFO signal goes negative, JFET resistance goes high. When LFO signal goes positive, first JFET resistance gets a little less then JFET Gate starts conducting current. In this type of one-way rectifier, the *average* level on JFET Gate will tend to go negative, toward about half of the peak-peak LFO signal. If LFO level is well-choosen, this may be just where we want it.
Depending on LFO and JFET parameters, it may be "self-biasing".
If the fit is not so good, LFO amplitude adjustment might be wanted.
This is shown in post #9 (http://www.diystompboxes.com/smfforum/index.php?topic=109499.msg1002009#msg1002009). The positive LFO peaks are "clamped" to +0.55V (gate diode drop) and the negative peaks driven more negative.
This "should" produce a change of trem-depth in the first few seconds. In post #7 this shows as a change in trem step width.
I'm not sure the 330K to the Gate is critical or even needed. Zero added resistance (just the 10K of the LFO) would diode-bias 33 times faster, but we don't care, but it saves a resistor, but that resistor may be a helpful layout jumper.
If the LFO is left running all the time, all this re-biasing is done before you have the guitar in hand ready to play.
I do have a 1meg pull down resistor on the JFET gate in the physical version, however I forgot to add it to the diagram in post 1 (I tried to update it but for some reason wasn't able to edit the first post). added it as I didn't want to leave it floating; bad engineering, as you say.
You're right about the 330k being unnecessary, but it's the first time I've ever used JFETs so I wanted to err on the side of caution in regards to running positive voltages through the gate; I didn't know how much current it could handle so just went for basically non. In it's present setup the time taken for it to bias itself is pretty irrelevant by the time the pedal's in use.
In case anyone's interested; here's the modulated signals before mixing; (high frequency in red, low frequency in green)
(http://i.imgur.com/2lQHJH9.png)
Here's the latest update including the points from the last few posts. Missed anything?
(http://i.imgur.com/VQDh8oR.png)
EDIT: It should also work if R20 is put back to 68k, R22 is removed, and the Depth pot R27 is replaced with a 1 meg variable resistor. This'd create a potential divider between R27 and R20 without affecting the bias of the signal splitter.
What made you put in an input buffer? That's totally not necessary!
(Ducking, running)
:icon_twisted:
> wasn't able to edit the first post
Some forums allow edits "forever".
This can be abused, so some forums don't allow edits.
This forum gives you 24(?) hours. Enough to patch-up your typos.
For a lost resistor, just mention it later. If you are in a real jam (wrote something you regret) and missed the edit cut-off, use "report to Moderator" and ask that the distressing words/post be removed.
I tried the depth pot today with the j201's, and it didn't really work. the output level just followed the pot down. and I swapped the j201's for j210's, just for the grabbings, and it completely changed the tonal workings of the trem. so it seems different fets will sound different, a note for the schem.
haven't tried the mix pot yet, but I did try modified filter values, based on midwayfair's cardinal values. they worked quite well, but need fiddle.
Hmm, the simulation seemed to show the depth pot working, but then it's hard to test as you can't change it whilst the sim is running so direct comparisons are tricky... Could use a double potentiometer along side each fet (like in the original schematic for the basic tremolo), but that means you need a double pot.. alternatively, separate pots could be used, which would allow you to adjust the depth of the high and low tones separately which could be cool..
Something else I want to try is higher order or active filters to give a sharper roll off. Should make the effect more pronounced.
What component values did you end up with in the filter? I've just been going for matching values so it's not missing too many frequencies in the middle.
currently, the scribble shows 10k at R21, 5n6 at C13, 1n5 at C12, 1M at R24 and a 1M added across the channel of the other fet as well. my calcs suggest the low pass is 2800 and the high pass is 106Hz. I have no R13 or R14 fitted, or R16. and a source follower for the buffer.
I'm also switching like the cardinal between trem, vibe, and harmonic modes, just to hear. the pots under the fets means the trem shape will always be square, but it might end-up all the level control we get.
This is the bode plot for the filter values you gave; pretty good overlap actually. The red and green lines show the frequency response of a second order version of the filters (just stacked two filters in series).
(http://i.imgur.com/zph1pbv.png)
I ran some tests with both 1st and 2nd order filters using your values at maximum depth:
1st order: https://drive.google.com/file/d/0BzfEd3tVJdxRN0xWdXFnaklNQ3c/view?usp=sharing (https://drive.google.com/file/d/0BzfEd3tVJdxRN0xWdXFnaklNQ3c/view?usp=sharing)
2nd order: https://drive.google.com/file/d/0BzfEd3tVJdxRd2hqVnZIdEVGOU0/view?usp=sharing (https://drive.google.com/file/d/0BzfEd3tVJdxRd2hqVnZIdEVGOU0/view?usp=sharing)
I actually prefer the 1st order version; the 2nd order version is a bit too pronounced...
EDIT: I found a spice model for the J201 JFET online and installed it (previously I was just using a 2N4117). I see what you mean about it being a very square tremolo; very different from the subtle tones of the 2N4117. For comparison, here's the audio using a J201 (1st order filter): https://drive.google.com/file/d/0BzfEd3tVJdxRdGFFTlVrLV9Kcjg/view?usp=sharing (https://drive.google.com/file/d/0BzfEd3tVJdxRdGFFTlVrLV9Kcjg/view?usp=sharing)
I only listened to your first wav, due to googoo and download speeds, but I hear nothing like that sound when I "test". (any chance I can get those files without going to googoo.drive or whatever?) nothing near as subtle. I did more shuffling of values today, and added the mix, and swapped fets. prr has some curves in "another thread" that explain a lot of what I'm hearing.
C13 can vary greatly, I'm back at 10nF with R21 at 12k, f=1300Hz, and C12 at 560pF, R24 1M, f=290Hz. the fender originals are set to 1300Hz and 72Hz. I've got R30 back at 56k and R31 back at 47k, both following 100nF caps, and 100k as the mix pot. it works much like a tone control (see prr's curves), and helps cover the current lack of depth control. it goes with the trem-vibe-harm switch quite well.
fets. as paul noted above, some aren't going to work. the j201 doesn't care, it just goes square. as you get to higher Vgs's, the trem turns to smaller peaks and wider troughs. with mpf102's in, you get a very good helicopter impression. no guitar tones, just heli whoop. 2n5486, mpf106 also don't work. 2n5457 and 2sk30a-Y dropped straight in, the 5457's did a nice watery thing, and good depth. J210's were chop-chop, no middle fill, and often sounder like 2 trems and still a trough between. they also had a strange spot mid mix, where nothing seemed to be happening, but at a fast rate. your voltage levels from your phase splitter will make all my observations moot on your breadboard, with your fets.
and because of the mix control, you can use a choppy 5457 in the low pass stage, and a smoother j201 in the high pass, and mix between. or your choice. but only if you are prepared to do endless breadboarding. it is becoming quite versatile. [I really need to find someone who can play guitar, tell me if any of these sounds are useful.]
and people thought I'd waste my 3000'th. pah!
I ran into a trick with a modulator for a signal generator where a Wein bridge oscillator had to be stabilized by a JFET while minimizing distortion. The first schematic here:
http://www.ham-radio.com/k6sti/st-1000a.htm
shows a JFET level control that uses equal resistances going into the gate and connected from the gate to the drain. It is a different application, but it may be ideal for what you want - tremolo without much modification of the waveform and the resulting harmonic generation.