Designing that OP-Amp OR15-soundalike

[BluesBerto]

Dear Readers,
This will be my first post on the SSGuitar-forum. I've been happily reading many a thread from a visitor-perspective. Forgive me the audacity not to introduce myself in the appropriate subsection and jump straight to the matter. Firstly I'll list my build goals. Secondly I'll list the constraints I'll subject myself to. Thirdly, I'll propose a method of approach.

1. The goal is to design, prototype and build a solid-state preamp that sounds like the Orange OR15.

2.
-I'd like to limit myself to the use of op-amp gain stages. Although -- and ROG keeps proving this point-- JFETs have proven to work well in solid-state adaptations of valve topologies, J201's are hard to come by in my region. The project will hence serve as an attempted proof-of-concept that op-amps can be used functionally to mimic valve topologies.

-The preamp should preferably work off a single 9V supply, although if nescessary, a dual-rail supply can be constructed and implemented

3. I've spent some time analyzing the OR15-schematic. (I'll appendicize it) and used a calculator to calculate the gain of the first 4 preamp tubes. Negative feedback of the op-amp stages will be set accordingly. I presumed the filters between the stages could be copied   one-on-one. To incorporate soft-clipping characteristics , antiparallel diode pairs are placed in the feedback loops. To adjust the clipping threshold, a resistor has been placed in series with the diode pair. (These will require tweaking. Asymmetric clipping in one or more stages might be desirable)

Now I'd like to ask some questions and place some notes:
-In the feedback-loops are radio-frequency  pF-value caps. How would I choose appropriate values?
-There are DC-blocking capacitors from the inverting input to ground. How would I choose these values? (They should not have a filtering function, since filtering is done through coupling capacitors and roll-off capacitors)

Of course, feel free to vent your thoughts

Berto


I'll add both my proposed schematic and the OR15-schematic.

[Katoda]

Hey, I would say welcome to the forum, but your nice introduction makes me feel inadequate to do so, since I am usually mostly a visitor myself. So hi.
You transcribed the schematic nicely, but you have to take into account that you will be working with low voltage and different impedances, so you will have to scale either the op amps or the rest of the signal path accordingly. Also, antiparallel diodes will produse symmetrical soft clipping, while the tubes tend to have  a more asymmetrical character. If you want to get as close to the real deal as possible, you should add another inverting  op amp stage after every clipping stage, and only clip one half of the signal at a time. To determine how much to clip each stage, you will have to study tube datasheets and determine clipping points for each stage in relation to the signal. When all that is done, you will probably have the most sophisticated solid state distortion pedal, but it will probably also be incredibly close to the real deal.

Damn, I want to build something like that too. I'm going to if you're going to   

Cheers

EDIT: Forgot to answer the question  :
I think you should just appropriate the miller capacitance of the tubes in regards to the impedance driving it to form a low pass filter. In the op amp circuit, you would then calculate the cutoff frequency of the cap and the resistor to the VCC. And that cap isn't necessary, since you are decoupling the signal to the reference voltage. If you were to tie them to the ground, in your example, would create a different DC bias point for the op amp. Also, on the schematic, the input 1M resistor should be tied to the ground, as tou don't want half the supply voltage across your guitar. There should also be a buffer after the tone stack, then a coupling capacitor and a resistor to the ground. Again, you don't want any DC on the thing following your preamp.

Second EDIT: Just got an idea, but it would take quite some experimenting to get it right. To conserve some op amp stages while still retaining the asymmetrical clipping concept, one might experiment with using diodes with different threshold voltages, and/or diode biasing resistors. If you get the clipping right, you're 90% there. But it would still require a lot of research.
I'd start this way: Determine the clipping points for the tube stages,  they will most likely be around 1-2V. Then, compare this to the diode forward voltage drop, which is around 0.7V for silicon diodes, and adjust the gain of the op amps by this quotient. To achieve asymmetry when clipping, you might have to go through some different diodes, but you could also just connect two of them in series for a clipping threshold of 2*Vf. Or, you could calculate the signal current and add a resistor in series with each diode to adjust Vf ever so slightly. Once you have the gain adjusted to the diodes (so approximately 70% of the tube gain) and the diodes set to approximate tube clipping, you should scale the op amp feedback resistors to match the op amp output impedance to that of a tube gain stage, which is the plate resistor parallel to the tubes internal resistance.
Also consider using a higher supply voltage instead of the pedal-classic 9V. More voltage equals more headroom equals more precision when adjusting clipping points.
Good luck!

[Katoda]

So I did the calculations for the first gain stage, they might be a bit off, since i didn't print the load lines, but dragged a ruler over my laptop screen.  The DC bias point is -1.8V on the grid, so it will clip the signal 1.8V in one direction and 2.2V in the other. This renders a quotient of 2.57, if you choose the lower clipping threshold to be a standard silicon 0.7V diode. So for the first gain stage, the gain should be 2.57 times smaller, the op amp output impedance should be around 38.5k, like in a tube. With the antiparallel diodes you should aim for Vf1=0.7V in one direction and Vf2=0.85V in the other.

There ya go, repeat the process for the rest of the circuit.

Cheers

[BluesBerto]

-I corrected the few mistakes I made. Firstly, I redid the circuit for a dual-rail supply. Secondly, I added a volume pot and a buffer before the output.

-To implement the asymmetric clipping, I added differently valued resistors in series with the diodes. These reflect the asymmetrc ratio of your load line guesstimate.

-In the first stage the negative lobe gets clipped more, in the second the positive, etc. (since the signal itself is not inverted)

-As for the respective gain of the stages, I'm not really sure what to make of them. To go by the theoretical gains of the individual tube stages (50x or 34dB) would probably yield too high signals and too much clipping.

-As for now: I've finished the schematic and a layout with it. It's time to stop the work for today (I'm getting hungry). In the next days I'll go over it a few times to check for basic feasibility. If the basic topology is satisfactory, I'll fab a pcb next week and put sockets in place of the ambiguous components, e.g. feedback resistors, diode-resistors

As a closing note; I will wholeheartedly and directly admit that I'm not an electronic engineer, nor do I have extensive knowledge of tube amplifiers. Therefore I sometimes revert to the build-and-tweak approach. Something I can afford because our local hackerspace had an entire component library donated, and the local polytechnique grants me acces to their PCB-facilities just because.
I'll advance in Micro-Increments
 

[phatt]

Hi Berto,
I don't wish to put a damper on your enthusiasm but if it was that simple it would have been done years ago.
I urge you to get a bread board and test your ideas before you start making PCB's other wise it will take you years to perfect the idea.
I've spent the last 20 plus years trying to perfect ideas and I have worn out the holes in a half dozen breadboards.
As already hinted; opamps are PP output and low Z so do not respond the same as a triode. (transfer curves are quite different)
Just inserting valve values between a row of opamps will not yield the mojo.
Katoda has noted that diodes won't replace the transfer curves of triodes and I agree.

As to the output buffer; A BJT won't have the internal Z to cope with the HiZ tone control. Use a fet or better still opamp buffer and run the highest voltage supply you can build, i.e. +/- 15Volts. (Headroom)
Be very aware that the Hidden enemy is NOISE,,, it takes a lot of reading and testing before you learn the hidden gotchas in building higain circuits.

Frankly I don't prescribe to trying to copy exactly a particular Amp sound as just changing the speaker can change the outcome quite a lot.
I use several circuits that on there own are not instant mojo but as a collective reap a very pleasing sound that delivers the classic sounds. Most of these circuit ideas are on this forum, some may help some my not as the Orange amp is more distortion than tone. (IMHO)

Without doing a simulation I can't be sure but I'm guessing you will have very little bass response and a lot of distortion with the circuit shown.

Will keep watching and help when I have more time.
Cheers,
Phil.

[J M Fahey]

1) congratulations on what you are trying to achieve 

2) it *can*  be done 

3) that said, "the devil lies in the details".

As an example,let´s check the first gain stage, which tries to emulate the first gain stage in OR15:

* first triode has gain about 40 to 50X
Your first iteration also used 40X gain ... problem is that OR15 first triode is fed some 200/250V DC so it can put out some 70-80VAC without clipping , while an Op amp will clip above some 9V RMS.
So exact same guitar which is handled flawlessly by the Tube will horribly clip the Op Amp.

Of course, lowering gain (your second iteration was about 15X) will improve the outcome ... but as you see now we are getting away from original schematic operation so we´ll have to account for that somehow.

As said above, it can be done ... but not "literally", we´ll have to compensate, and that at every stage.

* clipping diodes clip, but:
a) they clip symmetrical (tubes do not)
b) they start clipping at 0.7V peak while Op Amp would only clip (on its own) at some 12V peak
This alone will also change your design even more.
Just as an idea: put a reverse Zener diode in series with each conventional one; say a 9V one on top and a 12V one on bottom.
Now signal will clip on +9.7V and -12.7V bottom.

Also use widely different resistors in series with each: low value on the 12V side so it sharply clips, a higher value one on the 9V side so it loses gain but does not actually *clip*  but sinewave gets rounded (like a tube reaching the saturation area).

Now you have a stage which mimics triode clipping way better.

Apply same kind of tweaking and correction to all tube stages and you´ll get quite closer to the original full tube one.

Just warning that power stages can get REAL complex ... and I mean it.

But congratulations again and keep experimenting 

[Loudthud]

Did you ever notice that the gain equation for the non-inverting opamp has that silly +1 in it? Because of the massive gain of an opamp, the input waveform on the non-inverting input is basically copied over to the inverting input. So what you say... It means that no matter how you distort the waveform in the RF part of the feedback network, a copy of the input signal (the plus one part) gets added to the output so you get a mixture of distorted and non distorted signal. Yes, the famous Tube Screamer does it that way. If you want to eliminate this, you need to use only inverting opamp gain stages.