Marshall high impedance response simulator

[dsmnoisemaker]

Hi all:
I´m designing a tube power amp "approximaton", in which i simulate the trasnfer curve of tubes using a diode ladder, to get sucessive gain reduction versus output amplitude, like compression..

After that i added something similar to the marshall output impedance simulator (http://www.freepatentsonline.com/5467400.html), but instead of a opamp, i´m using a transistor as a gyrator, and some other values for components..

i tried to do the same thing with CMOS inverters and twin-t filters on feedback, but somehow i don´t get the same kind of response, looks like phase shifts don´t let me mix the treble boost with the bottom peak in a nice manner (like the marshall ckt does).

so, i decided to keed the marshall type simulator..but i´m worried about IP..maybe, eventually i could sell this things, and i wouldn´t like to get "cease and desist" letters from them..

a penny for your thoughs....

[teemuk]

The overall idea by itself is prior art since it's nothing but a frequency response simulation. I don't know the legal part but I figure you might get into trouble if you copy the circuit and your products start to seriously compete with Marshall and Blackstar (for whom Bruce Keir works now and who also uses the circuit in their amps).

On the other hand, most cabinet simulators and overdrive effects use almost similar circuits so few modifications to distinct your version from the patented one shouldn't be a major problem. For example, simply do not use the gyrator in the feedback loop or something like that. Another example: Randall uses a generic 4-band graphic EQ circuit in their G3 "Valve Dynamic" amps to achieve the sought after response. They have just replaced the potentiometers with fixed value resistors. The possibilities to replicate a certain frequency response are endless. The sole merit of the patent is that it happens to be a very simple circuit to achieve the results.

[dsmnoisemaker]

thanks for your response teemuk.. btw, most of my projects get a lot of help from your book  , amazing info there, it´s my bible now...

I don´t think i´ll be able to compete with marshall or other commercial brand out there, i´m still a boutique builder doing this for fun on my free time..although maybe someday it will be my main occupation (if i ever loose my day job i´d give this a chance).

i think i´m going to insist on developing my own approach to this, i have a nice 3 CMOS inverters cabsim, and maybe i can use a couple more on getting the response/compression/dynamic frequency response.....

[J M Fahey]

Hi dsmnoisem.
I agree fully with what was posted above and add: it's an un-enforceable patent, and thus useless.
Even if you receive a cease and desist letter, they can do nothing.
Really a patent is no "protection" itself but simply the "right to sue" which is a very different matter.
The key is the word "prior art".
If it can be proved (easy) that many engineers, when faced with that problem would have given the same solution, the "invention" loses all of its legal power.
For simple circuits you can use Cmos Inverters, but for more complicated stuff, Op Amps rule.
Good luck and keep experimenting.
J M Fahey.
PS: google and read Don Lancaster's "The case against patents" (or patenting), a real eye-opener. BTW, where do you wtite from?
Read his other stuff too, *that's* a real Guru.

[J M Fahey]

Just to save you some time: Don Lancaster's (genius) "why patents are worthless":
http://www.tinaja.com/patnt01.asp
or, specifically the part about "prior art":
http://www.tinaja.com/glib/bustpat.pdf

[dsmnoisemaker]

thanks for the info JM, i´m now reading it..
Anyway, i managed to get a very similar response from 2 CMOS inverters in paralell (one as a peak filter at 100hz, the other as a high pass filter , and mixed back)..I like how it turned out, but is very, very component sensitive (well i´m dealing with 8dB 100hz peak, and 8db boost on high frequs above 2k, so, no wonder)

I simulated the response with my behringer eq after a distortion, and it gives nice "thump" and clarity..but i like the marshall solution cause it´s easy to control de "mix" of the effect (replacing the 56k resistor for a pot on the feedback loop of the upper opamp) with the eq bypassed, it sounded dull in comparison..

PS: I´m writing from Santiago, Chile..regards!

[J M Fahey]

Hola vecino dsmnoisemaker.
I was always interested in linear biased Cmos Inverters (CmI for short). I guess they have something in the order of 30x gain, tops, while any decent Op Amp supplies at least 100x or more, at reasonable audio frequencies. Anyway CmI become *very* cost effective for many uses, because of their very low cost and availability in "sixpacks" (CD4049, etc.)
Please keep posting your experiments.
J M Fahey from Argentina.

[dsmnoisemaker]

OK..this is the version i did with cmos inverters:


upper inverter is a resonant peak filter tuned to ~110Hz with a gain of 10dB with a reasonably nice Q.
Lower inverter is a hi pass filter.
Both are mixed through resistors, balancing the response.. Probably will need a buffer after due to high output impedance (ironic, isn´t it?) of cmos inverters.
I like inverters because they are cheap, available, they auto bias, have nice clipping characteristics and need little extra components to make them work.

this is the frequency response:


i like it more than the marshall ckt, since the 110Hz peak is sharper, and nearer an actual speaker response..but i miss the "mix" capability of the marshall ckt..any idea how to implement it easily here?
also, i´d like to just use 1 inverter for this.. i tried using different feedback configurations, but the resonant peak eats everything else..

[J M Fahey]

Hi dsmnoisem, congratulations.
Very good project, and very usable. :tu:
Just don't insist on saving one inverter!!  , you already have 4 of them unused in the same pack.
I'd try, on the contrary, to use the others, at least as buffers (input and output) and even to drive a clipping indicator Led, if nothing else.
As for the mix solution, I'd just wire a B25k or B50k pot from one output to the other, and use the slider of said pot to drive a final output buffer.
Keep the good work flowing. :tu:
J M Fahey

[dsmnoisemaker]

Thanks JM (gracias vecino!)
The advantages of using this on an amp are:
- you get this feature of tube amps without using mixed mode feedback. This means it is usefull for BTL configs and others that make MMF difficult to implement.
- You get the response fixed regardless of the speaker used (different speakers have different peaks)

i have to breadboard this to check clipping behaviour (on simulation looks ok, but on reality may be different)

next step is to get variable freq response like saturated transformers do. i have two plans:
- easy cheap one: using diodes to change frequency response by "stages"
- better, harder one: using envelope detector and LDR´s....
any ideas?

[teemuk]

Here's one idea: Take a look at Eric Pritchard's US Patent 4,995,084. One of the discussed circuits wraps a band-pass filter inside a negative feedback loop. The negative feedback extends the response of the filter but when the signal is clipped the loop becomes less effective and the passband of the filter gets narrower.

[phatt]

Hello dsmnoisemaker,
                             Very teck and a lot of this is over my head but Have built similar stuff using 4049 and it's ok but maybe just use the 4049 for dist and use the cabsim circuit. Marshall sim Will work far better, IMExperience.

Also maybe Take a look at small Laney Amps for ways to impliment the 4049.

Here is a screen shot at the Cabsim circuit,, I've built about 4 of these and only have one left. This is just a slightly different circuit to the one found inside the Marshall JTM 30/60 Amplifiers.
The circuit you post is no where enough for what you will need to make it sound convincing.

The trick to most of this stuff is about *Maximising Loss* not Max boost
. Once you get the *Tone Shape* Then apply some gain If you get my drift

This circuit works well for Recording,, works well after guitar processors.
I don't like it as much for Live Use though. I use another setup for that.
Have fun, Phil.

[dsmnoisemaker]

thank phatt, but looks like you didn't get what´s the objective of the circuit. It is not a cabsim..its a replica of what a high impedance amplifier output does to speakers..a cabsim may be added after that to simulate an actual speaker response.

[phatt]

Hello again dsmnoisemaker ,,
Sorry if I'm missing something but just re read *United States Patent US5467400*

You said;  "It is not a cabsim..its a replica of what a high impedance amplifier output does to speakers..a cabsim may be added after that to simulate an actual speaker response."

So I'm scratching my head because from where I'm standing It's the same thing just coming at it from a different perspective.
effectivly achiving the same thing ,,except I assume you are trying to do it with the 4096 to add the distortion thing. But I might be missing something.

Best of luck with all. Cheers Phil.

[dsmnoisemaker]

well, it is partially the same thing..but not.
speaker frecuency response graphs show the typical response of a speaker under a low impedance power source, that is, using a very linear amp..
Tube amps, being transformer- coupled to the speaker, have high output impedance, so the impedance curve of the speaker generate nonlinearities (apart from the natural freq response of it)..

picture a SS very linear amp hooked to a celestion speaker..the celestion speaker will respond like the curve (steep high rolloff at 3khz, bass cut from 100hz down)..we add the higher impedance simulator, and we have a boosted 100hz and enhanced highs, sounding more punchy and open. Try putting a graphic eq on the fx loop with a similar curve. sounds great!