Gibson SS tremolo circuit pedal design

[Toddy]

I got a Gibson G10 solid state amp last year. The only thing good about it was the tremolo. So I sold the amp. Then, I decided to design the tremolo circuit into a pedal and I need some help since it was made for an amp not a 9V separate box.

Here is the original design:


And, here is my chicken scratch:


Here are my questions:
Q1 is labeled as MPF161. Since it is a very hard to find FET and the voltages are different, what would be a good FET substitute there?

Would a BC109 still work with the voltages dropped?

Would any resistor values have to change becausse of the voltage bias?

Would the input/output/battery placement (ala Trem Face) work where I placed them?

Any help would be greatly appreciated. I would just like to get one working (just to see if can be done) before the year ends.

Thanks,
Toddy

[petemoore]

  I can't see the original schematic..
  Here's some questions...
  What is the supply voltage originally ?
  Is there more to the circuit [besides the PS] than that 1 transistor ? [seems too small a circuit to get tremolo going, but I didn't really study it...but it looks like this may be a fragment of what made tremulations in the amp.
  The 'three capacitor huddle' looks a similar configuration to the LFO portion of the EA Tremolo.

[George Giblet]

The JFET is a p-channel device, so that's really going to limit your choices.   Looking at the data sheet for the MPF161 the general device parameters have enormous spans implying you could pretty much use anything.  However the datasheet also shows the devices might be categorized, shown with colored dots, or a printed letter, on the actual device.  The exact part that Gibson bought isn't shown on the schematic, although you might be able to look at the actual device.

Putting that aside, I suspect you should be able to use a p-channel with a Vp of 1 to 3V.

You could easily re-jig the circuit using a BC559 and an easier to get N-channel mosfet.

As far as the bias resistor goes, you will probably have to reduce it, perhaps around 22k to 47k. I can't work out what the circuit is actually running at.  The 6k8 value looks a little small to me is that the right value?  not 680k or 6M8?

[teemuk]

Some suggestions:
- This circuit likely needs gain to raise the signal amplitude to a level that corresponds the signal's amplitude in the amp circuit at that point of the signal path. At least buffer the input, perhaps with just a simple source follower FET or Opamp voltage follower.
- It possibly makes more sense to convert the footswitching arrangement to bypass instead of one that just disables the tremolo oscillator.
- I'm also pretty sure the tremolo oscillotor doesn't bias correctly with Vcc=9V supply using the same component values as used in higher Vcc circuit.

[Toddy]

A cross-reference to a MPF161 is 2N5461/2. The Trem Face uses a 2N5485 and a BC184C for its simple circuit. Would that combination work in this one? And the earliest version of the Colorsound Tremolo, originally drawn by Steven Giles back in 2000, has 2 BC349s and 1 BC169 for its simple circuit.

You are correct about my error, George. The resistor value labelled 6K8 should be 6M8.

Leemuck, the original circuit had a TIS98  for Q1 as the power buffer. That seems to be a big part for a 9V circuit.

I no longer have the amp. But if someone has one, can they tear into it and read the actual parts?

Type in Google (or Bing)  Gibson G10 schematic, or Gibson G10 parts list to pull them up on the net. It's easier to read than my scan and copy of a copy.

Thanks,
Toddy

[George Giblet]

> The Trem Face uses a 2N5485 and a BC184C for its simple circuit.

There's many way to tweak the circuit, some will tweak it to something else.  If you tweaked it to use those parts, it would be close but not the same, and then you would probably be better off building the tremface.  (the Vc to 9V transition makes it not the same anyway).

You can build circuit with all four combinations,

BJT                 FET               
NPN                N-Ch                    tremface
NPN                P-Ch                    original G10
PNP                P-Ch                    "flipped"  temface
PNP                N-Ch                    "flipped" G10

In principle the last two should behave like their polarity flipped brothers.

So if you want to emulate the overall behaviour of the G10 *and* use an easier to get N-Ch JFET then the circuit to build would be a the "flipped" G10, which would use a PNP for the oscillator and an N-CH for the JFET.

To 'flip' the circuit you basically build the same circuit except and,
- replace the semiconductors with the opposite polarity
- Reverse the polarity of the power. In your case the + would feed the PNP emitter, and the FET would connect to the +rail (like the tremface).
- Keep the signal ground on the - rail
- Add a 100uF cap across the power.
- Tidy up the circuit to make sure there aren't any DC issues.

For the last point that means adding 1uF NP cap between the JFET, like you have in your circuit already.

Now the circuit at this point would need to be powered from the same supply voltage as the G10, which I suspect is in the 12 to 18V region.  The next step is to tweak it to work on 9V:

-  I'd drop the 6M8 a tad to 4M7.
-  Change the collector resistor to about 33k

Yet another tweak is to ensure the JFET bias is stable when the battery runs down.  The G10 runs from main and will have a relatively constant DC supply.  If you build an effect to run from battery the DC rail drops when the battery runs down.  If you look a the tremface it includes a zener regulator on the JFET bias.  You should copy that circuit.

Loose ends:   The oscillator on the original circuit will swing more then the 9V circuit.  The 9V circuit may or may not sound the same because of this.  I would err on lower Vp Jfets so thing scale down better.

[R.G.]

Some comments.

- The LFO starts and runs fine on 9V, as is. It's possible that tweaking the values could produce some advantage, but it does start and run. The voltage-feedback setup there is remarkably adaptable.
- BC109 probably works well. So would most NPNs with a gain over about 200. Again, the voltage-feedback setup is remarkably adaptable, if not all that predictable.
- It is very much like the EA trem done with a P-channel. I personally consider that an advantage, as the EA's use of an N-channel without bias is pretty contorted.
- Circuit operation is a phase-shift oscillator providing an LFO to a JFET biased modestly off. This changes the JFET channel resistance, very much like the EA trem. The bias sets  you up for different JFETs in the series. The circuit depends on some external resistance to form a voltage divider with the JFET channel resistance.
- In the suggested schematic, disconnect the bottom of the 150K depth pot from ground.
- Buffering before providing signal in for the JFET divider is a good idea. Increasing the signal is not, as the smaller the signal, the less the JFET will distort it as the signal approaches Vth in size. A 100mv (raw guitar size) signal works well in simulation.
- J175 works well in simulation. The bias change lets you use JFETs from the normal variation in one part number. However, the size of the LFO signal provided to the JFET, about 3V peak to peak, requires a JFET- with a Vgsoff of about 4-6V.
- The size of oscillation in a phase shift oscillator is always as large as the signal can swing before hitting some kind of limiter. It will self adjust its signal size to the limits. In this case, the voltage feedback setup on the NPN lets it swing to the power supply, so it limits at nearly the power supply, whatever that is.
- Stabilizing the JFET bias in the face of battery change is a good idea, like in the Trem Face. 6.8V for the zener is good. The reverse breakdown of many NPN transistors is about that, if you're being cheap.

So, with a buffer in front of it and the 150K pot leg loose from ground, it should run as scratched; at least that will let you adapt it. Should be a little smoother tremolo than the EA.

Circuit simulators are our friends.   

[Toddy]

Thanks guys for your tremendous help!! I didn't even think anyone would be interested in this %^&*amamey idea.

So basically, the circuit is good to go as it is, except for the addition of the 6.8V zener diode and maybe changing the 6M8 resistor to a 4M7 resistor and not grounding the depth (intensity) pot. If there is anything else please let me know before I draw up the new schematic with corrections. The big thing is that I can keep the transistors the same even with 9V. I also have to get my parts order list together...which is an important thing.

Wow RG, I didn't know you had a circuit simulator. Now that's cool stuff!

Overjoyed,
Toddy

[JKowalski]

Wow RG, I didn't know you had a circuit simulator. Now that's cool stuff!

You can too!

Look up LTspice. Spice in all of it's various iterations is the classic computer analog circuit simulator. In the case of LTspice, you draw out your circuit schematic using a library of components, then you can test it using different analysis methods. The most common method is basically like starting it up in real life and using a oscilloscope to measure parameters at various points over time. There's alot more you can do with it as well, such as measuring the frequency response of a circuit from signal in to out.

Incredibly useful program, and it's not a actually that hard to learn to use. And it's totally free.

I always design all my electronics project with this program - it's alot simpler then breadboarding, of course!

[George Giblet]

Assuming Pch JFET,
-  Leave the 6M8,
-  tweak the collector resistor to about 47k (a close match for old w/TIS98 @ 15V vs new w/BC109 @ 9V),
-  add stable JFET bias circuit say 10k to +V rail, 4.7V zener to ground, trimpot across zener, wiper to same point as show.
   With the Pch JFET the BIAS network goes to ground, not +V liek the tremface
- optional 10uF electro cap across zener.

[R.G.]

I wondered about tweaking stuff so I simmed it dead stock. Worked without tweaks. In any case, a working sim is just a starting point.

A note about simulators. They put out *exactly* what is in the models and the equations. Unless you know where this does not match the real world, you can paint some very pretty pictures that cannot be made in reality.

[JKowalski]

A note about simulators. They put out *exactly* what is in the models and the equations. Unless you know where this does not match the real world, you can paint some very pretty pictures that cannot be made in reality.

That's true.

Simulator -> Breadboard -> PCB is the way to go, in my book. I'll never skip the b-boarding step, that's the most important, both for tweaking and real world testing.

It's just easier to get the design stage done in spice - it's like collecting your ideas on paper, but with the added bonus of seeing it in action.

[George Giblet]

> Worked without tweaks.

The bias point with 82k produces a low collector voltage and the output swing is lower (about 4.5Vp-p).  With 47k you get about 6.1p-p and with 27k about 7.8Vp-p.

The ~15V original doesn't swing to the full rail voltage so to match the original non-linearity you do't want to push the output too close ot the rails.  On the other hand the original has a higher swing relative to the JFETs Vp, so you want a high output.

When you factor in the actual parts used for the build it's not worth splitting hairs, probably better off judging by ear; perhaps try 33k and 47k; in retrospect probably err on the 33k.

[George Giblet]

> I'll never skip the b-boarding step

It depends on how many years you have spent comparing simulators to actual circuits, and how critical it is to control the design.  You develop a feel for what to believe.

Building an actual circuit is very poor at dealing with part tolerances - you don't find the problems until you do the production run and that's the worse time to find problems!

[Toddy]

George and RG, you both have great points and ideas. Is it possible for both of you to draw up a schematic of what you think should be done and then post it?

My original intent was to tweak it as little as possible to get it working. But right now it seems that I am confused until I see it written down. Some people are audio/verbal learners, and others like me, are visual/hands-on learners. I need to see what is going on in your heads about the circuit.

Thanks so much guys! It'll be great once it's workin'!
Toddy

[George Giblet]

Check out
http://i476.photobucket.com/albums/rr130/ggeffects/Gibson_G10_trem_v1.png

You might find it hard to get 150k pots, you could sub 220k's instead.

(The JFET part shown doesn't have to be adhered to, it's just the default part from my simulator,
 use the part RG suggested or something you can get hold of.)

[George Giblet]

I lifted the ground side of RV2 to minimize the gain drop, however now there's no DC path on C7 -ve terminal.  Add a 100k to from C7's -ve terminal to ground.  Will fix schem soon.

[Toddy]

Thanks George! Great work!! Now I see where you're coming from.  I will put sockets in for transistor swapping to see if that makes a difference, although everyone doubts it. I KNEW it was possible to isolate this circuit!! Now I can put my parts order list together.

Thanks to everyone who helped!!

My wife says I like tremolo because I like to waiver on things. Yeah, I know...not that funny. But I still love her and have to live with her so I laughed at it. Concider the alternative.
Toddy

[George Giblet]

I'll update the schematic to include that 100k resistor!
Currently it's rev 1.1, will update to rev 1.2.

I'm thinking you might have trouble getting pots that are 150k and also pots that are C taper pots (reverse log).

-------------------
OK it's done.
http://i476.photobucket.com/albums/rr130/ggeffects/Gibson_G10_trem_v1-1.png

I recommended the J176 P-channel JFET over the J175 in attempt to mimic the original more closely.

Here the reasoning:

Original JFET MPF161,  Assume yellow sort category used,
     Vp = 1V to 6V, Assume Vp typically = 3.5V
     Yfs ~ 3250us (rds_on = 308)

Osc swing about 5.2Vp-p

Osc / Vp  = 5.2/3.5 = 1.5

For new version, 9V + 39k

Vosc swing is 3.0Vp-p

For the same depth of modulation require same ration of Osc to Vp of 1.5  ie want

Vp  =  3.3/1.5 = 2.2

A J176 has a Vp typ around 2.5V and a rds_on = 250  which is a close match.
The J175 would have less depth to the modulation.

[Toddy]

Sorry that it's been a couple months replying back. George, thanks so much for all your help!! Posting the revised schematic and letting me see your thoughts down in diagram worked wonders for me!!

I am finally starting to put my parts list together. For now, I am taking everyones advice and making the J176/BC559 NPN version. But, I have two people (and Santa Claus) trying to search for the original MPS161 FET and BC109C without having to spend a minimum of $50 on an order even though the MPS161s only cost 60 cents/piece!! I figure I am going to make the original PNP FET version after I get the new NPN version working when the time comes. But I have one last request to ask...

Is there a way for someone to make a circuit layout from the schematic you revised (version 1.2)?

Yes I know that the circuit is polarity reversed from the original. But that should be easily adapted afterwards.

Ever grateful,
Toddy