Is there an error in the usual Okko Diablo schematic?

[bordonbert]

Hi guys.  I've been working with what is usually considered to be the Okko Diablo schematic to come up with my own take on it.  I have found this same version in a number of places on the web.  The circuit is pretty standard stuff that has been tweaked fairly well.  2 x Mu-amp stages and a source follower.  It should offer good performance for classic overdrive rather than metal stuff, at least that's what I'm hoping.

Understanding the action of the SRPP stages is not too hard but the schematic doesn't seem to have things as I expected.  I've attached a pic of one stage of what normally passes as the circuit for reference.

Usually the bottom end of C4 would be expected to be connected to the lower end of R6 not the upper end as it is here.  Surely R6 is having no influence in this circuit?  The intention is for C4 and the upper JFET follower to bootstrap R6, (put the same AC signal at both ends so there is only a constant DC current through it with no changing AC current).  This makes it appear to the lower JFET as an extremely high impedance load.

Can anyone else who knows a bit about this circuit shed any light?  I do wonder if the original schematic has been drawn out incorrectly.  This same oddity occurs in both gain stages of the Diablo and in the Boost portion too.  If there is anyone out there who has one of these pedals and can check this out I would appreciate it.

[drolo]

Here is a snippet from Geofex describing what the resistor does:

The 1K resistor between the source of the upper JFET and the drain of the lower JFET works some magic. The upper JFET is perfectly capable of being a source follower with an output impedance of maybe 200 ohms. However, it can't do that and keep the drain of the lower JFET happy at the same time. Putting a 1K resistor in series lets it do both. The 1K resistor isolates the upper source from the lower drain so that the upper source can drive loads down to about 10K ohms with only trivial loss of gain. It doesn't change the voltage/current the lower drain sees by much.

taken from here:

http://www.geofex.com/Article_Folders/modmuamp/modmuamp.htm

[bordonbert]

I found that posting in my travels too drolo.  Unfortunately it confirms the same observation I have made. I understand what the resistor does in his setup but the Diablo is not wired like that.

To do what he describes, the connection between the cap and resistor must be made at the bottom end!  Look at the pic next to that section and you will see what I mean, it is not the same as the schematic I posted above.  The Okko Diablo schematic shows its upper JFET components wired differently.

Without spanning the resistor with the capacitor it does nothing!  The whole point is to give the resistor the same AC signal voltage at each end, then it effectively acts as a constant current source and looks like a very high impedance load to the JFET below.  The top JFET is just a source follower which gives the same voltage at its source as it sees at its gate, and of course that is the signal fed to it by C4 from the drain of the lower JFET.  So the same voltage appears at each end of R6.  No voltage change means no current change, hence a constant current.  And constant current sources have very high impedances.  The lower JFET likes that and can give high gain because of it.

I think we may need someone with an original Diablo to chase this part of the circuit through and see which is right.  Any takers out there?

[drolo]

Oh I see indeed,
And also I did not read your post carefully as you already mentioned that...
No idea what effect the cap will have connected like this. Goes a bit beyond my understanding of the circuit :-(

Did you try to breadboard it to see how it behaves in the 2 different configurations?

[bordonbert]

Not fully yet, I noticed the oddity and wanted to see what others thought about it first.  I have set one stage up as a Clean Boost and got about 4x gain out of it but this was less than I was anticipating.  You're right though, testing it out fully is the sensible approach from here.

What we basically have is a choice of two circuits to choose from both shown in your Geofex article.  The way it is currently wired in my post is as the JFET Muamp at the start of the article, (look at the C4 connections).   This has no drain resistor for the bottom JFET, and putting one in actually does nothing in real terms so R6 could basically be removed.  Just putting in R6 certainly does not turn it into the other alternative, the SRPP amp shown at the end, that depends on the resistor being in the source of the upper JFET and spanned by C4.

Interestingly, one of the differences between the two is that the Muamp circuit cannot drive output loads very well with its high output impedance, while the proper SRPP can as it has a much lower output impedance.  Looking at what is hung onto the outputs of each stage of the Diablo you can see relatively low impedances there, and they are all frequency variable too.  It would be essential to have a low output impedance in each stage to cope with that.

My gut really tells me that we need to shift C4 down to the drain of the other JFET!   

[drolo]

Go for it !
The worst thing that could happen is that it could cause a chain reaction that would unravel the very fabric of the space-time continuum and destroy the entire universe :-D

[drolo]

I got curious about this one and tried to find the origin of the schematic that is around, freestompboxes as usual.
In the thread people seem to be complaining about their build having less gain than what they heard in youtube demos. Someone actually mentioned the cap weirdness you noticed but no real conclusion came out of that...
I guess the lack of gain could well have to do with the cap being connected wrong. That said, would they trace the cap wrong for all gain stages? Or just made assumptions ... ??

Sorry, I'm not really helping I guess ...
Maybe someone who knows these circuits well will be able to explain :-)

[bordonbert]

But you ARE helping Drolo.  Between us we have fixed the problem!!!   

I've just cut the tracks of my Okko Boost clone PCB and rerouted the circuitry so the wiper of the Bass control is connected to the bottom of the resistor/drain of the JFET.  Hold your breath.  No unravelling of the space/time continuum.  It works!

Yes, we have much more gain, the controls work as you would expect, the tone is as it should, the whole thing just comes to life.  The original circuit is WRONG!  Now to revise my layouts and etch more PCBs to match for both the boost and overdrive setups.

Oh, I'll post on the Freestompboxes site to let them know.  What a partnership, eh? 

[drolo]

Glad it's working :-)

What a partnership, eh? 

Yes, a smart observer (you) and a genuine ignorant babbling about without any logic (me) ;-)

[cooder]

Wow this is cool, thanks for puzzling that out! I've built one previously and was underwhelmed and thought it didn't work quite right without having a clue as to what's the story...
I'm still wrapping my head around this, if you have at some point a corrected full schematic I'd be keen to see the light... I'm just wondering also if I stuffed up elsewhere too apart from this.
You guys rock!

[bordonbert]

Just for the record here is what I reckon each stage should look like:

https://dl.dropboxusercontent.com/u/21235584/Diablo_Stage_Corrected.jpg

The difference is around C4/R6.  Yes, such a small change does make a big difference.

[cooder]

Ok, according to this I've drawn a schematic, would this be a correct version then?
I did the changes at parts C9 and C13 refering to my schematic.
I might try at some point to make a board out of it if this is the working version.
Also, there was another post saying that C20 (47uF at charge pump) was obsolete, is that correct? Does it matter to have it there?

[bordonbert]

Hi Cooder.  Great work on the schematic.  Basically you have everything I have spoken about in place.  The following is a bit cold and pernickety but it should make it clear and easy to check:

1. You have corrected the connection of C9 to R9.
2. You have corrected the connection of C13 to R13.
3. You have made sure the redundant 47uF capacitor from pin 4 to 5 of IC1 is removed.  (This one only seen in some versions of the schematic).

I also made a couple of additions.

1. As I experienced some slight oscillation in the second stage at the highest settings of the gain control with my original layout, when I designed a new PCB I added additional components between the wiper of the Gain pot and the gate of Q3 as a precaution.
   a) I added a 2M2 resistor between the wiper and ground.  This simply means the JFET still has bias as the pot is used and maybe loses contact with the track occasionally as it ages.  Crackly pot protection!  Notice this has already been done to the first stage Q1 and the source follower Q5.  I wasn't sure if the values used for R6 and R18 were meant to affect the controls in any way so I chose 3M3 across the Gain setup to try to prevent that.  (Overkill I think but it costs nothing.)
   b) A series resistor between the Gain wiper and Q3 gate and a capacitor from the gate to ground.  This acts as a low pass filter removing any high frequency signals from the second stage input.  The input filter at Q1 is 47k/470p which has a 3dB point at 7.2kHz.  This is lower than I was expecting and must be part of the overall tone shaping design which is flawless to my ears.  I chose a higher frequency of 18.5kHz and used 39k and 220pF for the filter at Q3 in order to not affect the sound in any way.

2. I run it along with my own Boost (Okkoish) pedal and a Behringer EQ pedal from a shared power supply block supplied with the Behringer. It gives me a reasonable steady nominal 9V, actually 8.75V.  This showed an internal supply line of about 8.4V and a generated supply line of around 15.5V.  However, looking at the supply line when the unit was in use and the LED was on it dropped to 7.4V and 13.4V, too low for my liking.  This was actually due to the extra 10mA LED current passing through R1 on the input to the power supply.  And this was without the main Pedal ON/OFF LED in place adding another 10mA.  R1 is there to work with D1 and protect the unit against connecting reverse polarity power.  If that happens, R1 originally set the current through the 1N4001 to around 80mA.  I decided to drop the value of R1 to 27R.  Fault current would then become about 320mA.  Remember this is only through the protection diode which can handle 1A, no current passes through any other part of the circuit.  In this way the voltage drop across R1 when in use is divided by 4.

3. I added a little bit of circuitry to control a LED according to the 9V/18V selection for the power line.  This is simply a couple of transistors used as saturated switches.  I designed a couple of cleverer designs like Schmitt triggers and comparators but they were more complicated than I needed.  This is simple and works well as there is no switching speed issue here.
   a) It simply turns off the first transistor when the voltage rises to 18V and a second transistor senses this and turns on passing current through the LED.  PNPs make the level sensing easier to switch.
   b) I was a little worried about the amount of current available from the generated 18V line and tried to be careful to not add to the load on it in any major way.
   c) The LED will draw about 10mA so the control circuit must be powered from the 9V input line.
   d) ANY general purpose PNP transistors will do!  Take my advice, there are NO magic sweet sounding types.  In 99% of reported cases that issue is a myth and down to pure placebo effect.  Any small signal diode will do for D40 too, it is only shifting the DC voltage level of Q41 to make the action more secure.  1N914, 1N4148 are both common types for this and dirt cheap.  Use anything you have lying around for this whole area.

At some time I will put up a full schematic for this but mine is in scribbled bits and very messy at the moment as the project is not completed.

[bordonbert]

I've attached the working schematics I used.  There are some slight changes.

1)  Stage 2 Q4 has components added to its grid.  These are a bias resistor to help when pots go crackly and a filter set at a high enough frequency to not affect the signal and which will help with stability which I had a problem with in a previous layout.

2)  R14 was 43k and is now 47k, a more common E12 value.  It's in series with a pot for God's sake, why choose a rarer E24 value? The extra 4k represents less than 10% of its value at the low extreme and nothing worth even calculating at the upper extreme!  The difference won't even be heard.  The control works in a balanced and full way in my system with this value.

3)  Likewise the C12 8n2 has been substituted with 10nF as it is in everyone's box.  This also makes no difference in my own unit.

It sounds great for what I want, not out and out fuzz just lovely classic blues and rock tones and every control acts in a gradual progressive and balanced way.

[cooder]

Great work, thanks for posting! When time allows I will incorporate that into my schematic and try to get a layout together.
I can't make sense of the last file yet, the 'diablo voltage selector'.... how and where does that come into it, what am I missing there...?