Condor Cab Sim IC clipping

[bigfinger]

Hi folks, this is the first post. I'm Italian, excuse for my English.

I'm not able to solve this problem by myself.

I built a Condor Cab Sim by ROG. The project sound very good, it do his job fine, but after the first stage (J201) the ICs clip too early, in a very unpleasant way.

I made the PCB with eagle, so the circuit is triple checked.
I built two copies, without reuse the same components.

I usually test all my project first with "TINA". Everything seems to be correct.
I checked gain and frequency response with "Visual Analyser" (a very powerful tool). Everything is as expected, except headroom.

I try with a BF245A (instead of J201) as input stage to lower the gain to minimum (around 6 dB) but any humbucker PU is still capable of drive the emulator to clipping.
ICs are TL072.

Did anyone experienced a problem like this?

Thank in advance

ciao

bigfinger
_______________

[differo]

hmm no sorry, I've built couple of condors (layout found either on runoffgroove.com or here in forum don't recall) and no clipping problem?

[CynicalMan]

What voltage are you running it at? How hot are the humbuckers you're using?

[gtudoran]

I do have the same problem with a cabsim (in fact is a condor cabsim), i'm running it @ 9v and even with gutar signal i get clipping noise. I have Ibanez SA120 with HB SD SH 4 (not a very hot pickup).

I didn't study too much the problem but...if there is a solution i would like to hear it.

Best regards,
Gabriel Tudoran
Analog Sound

[bigfinger]

What voltage are you running it at? How hot are the humbuckers you're using?

As by ROG schematics, 9V (battery or external). The PU are: DiMarzio Dual Sound (bridge), DiMarzio PAF Pro (neck), DiMarzio Breed Neck, DiMarzio Bridge Evolution.

The signal after the first stage is still clean (at low gain) but the saturation, when occurs, is the usual JFET worm one (tested with a probe).

After this stage is easy to recognize true square clip, especially at the 3kHz boost.

Having read a lot of positive feedback about the Condor, I suppose the fault is mine.

Thanks again

bigfinger

[bigfinger]

http://www.radici.mi.it/ale/Condor SPK sim layout.pdf
http://www.radici.mi.it/ale/Condor SPK sim PCB.pdf
http://www.radici.mi.it/ale/Condor SPK sim schem.pdf

here the schematics I work with PCB and the layouts.

For people interested I have the TINA and Eagle files.
Hi

bigfinger

[MmmPedals]

mine also clips. I havent had time to debug but i can get a nasty distortion out of it.

[PRR]

What voltages at the FET?

Change the FET stage to a source follower.

[bigfinger]

What voltages at the FET?

Change the FET stage to a source follower.

I bias the FET of the Condor to obtain less gain, minimum distortion, good symmetry.
I usually test stage like this on a breadboard to select the best components and values.

After a lot of experiments, I developed this mods:
http://www.radici.mi.it/ale/Condor SPK sim 2.3.pdf

This sounds good, works at unity gain but suffers of great instability.

The first mod was change the JFET to buffer only (source-follower), but the condor drop to around -6 dB of total gain, so I changed the first IC stage to a +6 dB gain.
This works fine but some times don't work at all.

May be the solution is padding -10 dB the input and gain after the last IC, but I can't believe ROG published such a wrong project.

Thanks to all, ciao

bigfinger

[PRR]

You seem to have two ideas:

> I suppose the fault is mine.
> I can't believe ROG published such a wrong project.

I think any guitar through a boost on 9V supply is awful close to clipping. It may only be that your arm is stronger than ROG's testers. Gabriel and Mmmm report a similar issue.

Why don't you try your technical ideas "padding -10 dB the input and gain after the last IC", or a 18V-30V supply, and publish it as the "Improved No-Clip Condor"?

[PRR]

Not including the JFET gain-stage, nor your 27K+27K gain-stage, the filter network has no gain.

Gain is about -5dB (0.56) at 90Hz, -16dB (0.2) at 400Hz, -1dB at 2.8KHz. With such a wild variation of gain, "unity gain" can only be an approximation, and very dependent on what signal is going through.

Your plan without an input buffer, direct to the bridged-T network, has moderately low input impedance: 150K at the bottom of the guitar band, 60K at the top. This may tend to load some pickups.

You can take gain at the end without an additional amplifier. It is a "cheat": the gain-network impedance modifies the low-pass filter alignment. But with 1K, not enough to matter.

http://i48.tinypic.com/2yy14ki.jpg

[bigfinger]

Many many thanks, PRR, for your interest.

My poor know-how can't produce an idea like the your on the last stage. Simple, cheap, effective.

I was referring at the unity gain as a "no change in perceived level", which is a frequency response walking "around" the 0 dB line.

The speaker emulator is a non-standard stompbox: first it usually feed a mic/line input instead of a guitar amp, second it stay at the end of the of the FX chain, so it must accept a wide range of level, from the simple SC to boost+overdrive+distortion stomps.

It may be useful an input meter or a very "robust" circuit.

I will try with an input pad, emitter-follower and your output stage and if the noise floor don't rise too much, I will publish the results.

Thanks again PRR.

[PRR]

> it usually feed a mic/line input

OK. "Line level" is a wide target. In modern Hi-Fi and semi-Pro interfaces we usually reference 2V RMS at peak modulation. That's 2.8V peak or 5.6V peak-to-peak. You "can" get 5.6V p-p audio from a 9V battery, but it's tough; you can't apply 6dB gain at the input.

OTOH, the signal level is high enough that noise (hiss) is not a real problem.

Then one layout is to apply a fairly weak (<1V) signal, boost a little, take your filter losses, and accept a slightly low output.

Another layout is lossy filter first, followed by boost.

If you want HIGH (+4dBv nominal) level with NO clipping, you probably need to go to 30V (+/-15V) supplies, like many "Pro" signal processors. Converting the Condor to such power is fairly trivial. The op-amps are re-powered with +/-15V, inputs biased at zero V. The JFET boost at the input may conveniently be changed to a unity gain opamp. Some adjustable gain should be available after the lossy filter.

[Brymus]

> it usually feed a mic/line input

OK. "Line level" is a wide target. In modern Hi-Fi and semi-Pro interfaces we usually reference 2V RMS at peak modulation. That's 2.8V peak or 5.6V peak-to-peak. You "can" get 5.6V p-p audio from a 9V battery, but it's tough; you can't apply 6dB gain at the input.

OTOH, the signal level is high enough that noise (hiss) is not a real problem.

Then one layout is to apply a fairly weak (<1V) signal, boost a little, take your filter losses, and accept a slightly low output.

Another layout is lossy filter first, followed by boost.

If you want HIGH (+4dBv nominal) level with NO clipping, you probably need to go to 30V (+/-15V) supplies, like many "Pro" signal processors. Converting the Condor to such power is fairly trivial. The op-amps are re-powered with +/-15V, inputs biased at zero V. The JFET boost at the input may conveniently be changed to a unity gain opamp. Some adjustable gain should be available after the lossy filter.

That sounds like a really good idea.
Could this be done properly with a 9v battery and a MAX IC or other charge pump ? (555 maybe?)
I would really like a total cab sim box with more than one cab sim in it,there are so many,and I dont feel like building them all to find which one sounds best into my Tascam USB interface,it would be nice to have one that does several of the popular speaker emulations and can be boosted or cut for live or PC use.

[bigfinger]

Feeding PA is not a matter of level, because the input of the most of professional mixers accept from mic to line level on the same input, and without a switch.
Great difference is done by the balanced/unbalanced status of the line.
A DI box, which can drive long cable without inducted noise, is a must.

The speaker simulator is the perfect place to incorporate a phase inverter to drive balanced line.

For my band I buyed a Bheringer 8x DI, very cheap and very useful.

I would really like a total cab sim box with more than one cab sim in

The difference between all the spk-sim I studied is very subtle. They usually features three stages: a 80/100Hz hi-pass, a 400-600Hz dip and a 3k-5kHz lo-pass.
Some of them have only one or two stage.
The most important one is the low-pass and can't be omitted. This brings the buzzy saturation of our stomps to a pleasant state.
All the cutoff frequency can be modified. I will publish a comparation of all the schematics I found on-line. I made the frequency response with TINA (I need some time to assemble and convert).
You can easily reproduce the sound (frequency response) of a cam-sim with a plug-in equalizer on your PC. You can preset some of them and transfer the curve you  prefer in your stomp.

ciao

[stm]

Alessandro, i can offer to following advice:

Notice that the filters do not have gain except at the peak frequencies (around 90Hz for the hi-pass, and around 3.5kHz for the two lowpass).  The JFET stage was introduced to color the sound a bit and add some gain to compensate for the filter losses.  JFET selection should be based according to your guitar output level: single coils->J201, regular humbuckers->2N5457, hot humbuckers or active pickups->MPF102.  Using a J201 with hot humbuckers will certainly overload the following stages.

Anyway, one of the design goals was to use minimum parts, and in doing so it must be accepted that hot inputs must be attenuated in order to avoid clipping, but this also means that output will not reach unity gain, which is not necessarily something bad when you are not supposed to be switching the pedal on and off under normal playing conditions.

In summary, dealing with hot pickups requires the following:  choose a JFET that will produce low gain and a have higher dynamic range (like an MPF102) and then supply the circuit with +/-9V or up to +/-15V.  If you desire, you may use two additional opamps to provide balanced outputs and extra gain to reach or approximate line levels.

P.D.  Are you sure your Vref is at 4.5V?  Also, please make sure all four opamps have their outputs floating around 4.5V with no input signal.

[bigfinger]

Alessandro, i can offer to following advice:

Many thanks stm

JFET selection should be based according to your guitar output level

My last choice was a BF245A. It was a little too strong. I will check the MPF102.

Anyway, one of the design goals was to use minimum parts, and in doing so it must be accepted that hot inputs must be attenuated in order to avoid clipping, but this also means that output will not reach unity gain, which is not necessarily something bad when you are not supposed to be switching the pedal on and off under normal playing conditions.

Yes!!! (I told in previous post " Feeding PA is not a matter of level")

Are you sure your Vref is at 4.5V?

I will check tonight.

Thanks to all, ciao

[bigfinger]

Here some speaker emulators compared:

http://www.radici.mi.it/ale/SPK-emulators.pdf

In my opinion the condor cab (by stm of runoffgroove) is the best emulation of a guitar speaker frequency response.

ciao

bigfinger

[PRR]

> the filters do not have gain except at the peak frequencies

Actually (and mildly surprisingly), not counting any deliberage gain, there's no overall gain anywhere.

The hi-cut filters have +0.4dB bump near 2.85KHz, but the 400Hz notch is -1.5dB here, so the overall response is -0.6dB here. The 90Hz low-cut has 2dB bump but the notch is -7dB here so the result is -5dB. The worst-case is 10KHz and above where the buffer after the notch has -0.14dB signal.

However the notch-filter does ring. For a 1V peak square-wave excitation near 2.9KHz, I get 1.66V spikes after the notch-filter. If inputs might be similar to a square-wave (for example: fuzz), this suggests allowing 6dB headroom above the nominal peak input.

"Feeding PA is not a matter of level", true; but you do have to watch your levels all down the path.

[CynicalMan]

Here some speaker emulators compared:

http://www.radici.mi.it/ale/SPK-emulators.pdf

What's the V30 (russian) speaker emulator?

You also have to remember that different types of speakers have different responses. IIRC, American (read Jensen) style speakers have more of a mid scoop, a hump in the bass, and more low bass rolloff than British (read Celestion) style speakers, which have flatter mids and bass, and a lower bass rolloff frequency. Keeping that in mind, the Marshall and Mesa Boogie speaker sims have a more Celestion-ish character, whereas the Condor has a more Jensen-ish character. I've designed a sim that tries to hit different speaker styles:
sites.google.com/site/distorque/home/projects/speaker-sim