Ultra-G G100 Workalike

[George Giblet]

Something to play with:

http://www.geocities.com/george_giblet/effects/ultra_giblet_1.png

Based on the impulse response GFR posted the link to in this thread:
http://www.diystompboxes.com/smfforum/index.php?topic=44939.0

I don't own one so I cannot compare it.
I have not built this yet.

[GFR]

Nice!

May I ask you how did the "system identification"? Did you use some algorithm for pole/zero extraction from the impulse/frequency response or did you just "tweak" until it was close enough?

[George Giblet]

> or did you just "tweak" until it was close enough

I did a systematic tweak of the circuit based on 10 or more key frequencies.  It got to the point where I could change one component to the next value, then adjust other components to compensate and the response got no closer to the "best one so far".  I'd give it 0.5dB error on the whole,  but it creeps to 1dB at 10kHz.  If I tried to fix the 1dB at 10kHz it changes things elsewhere too much.  One option was to add another filter, but I argued that the impulse response has sound card roll-off and may be inaccurate anyway so I drew the line there.

People can tweak this as they please.  The key thing is it's very much in the ball-park (assuming we can trust the original impulse).

[Schematic updated to rev 1.1: A minor improvement is to add a 470ohm resistor in series with C3, to ground, and increase R6 to 3.3k]

[George Giblet]

Here's a plot comparing the V1.1 circuit with the measured impulse response.



The plot shows a fairly convincing match between the circuit and the impluse response.
The dB axis is in fine increments of 1dB.  The larest error is about 0.5dB around 800Hz. Changing the parts by one value increases this error more than 0.5dB which means it is difficult to tweak further.

There's no doubt some more work would provide an even better match.

The low frequencies of the impulse response show some indication of poor signal to noise of the original impulse (or perhaps quantization error in the sound card) - as seen in the rough region.  This is quite normal for directly measured impulse responses.

[12afael]

Great work George!!!

I wonder how it sound side to side with the original.
it look very simple , it is almost a legend because the schematic is not available.
the marshall se100 look better , the low pass filter have a more high order.