diyAudio (
-   miniDSP (
-   -   Any way to implement Parametric attenuator? (

trevmar 31st August 2011 04:18 AM

Any way to implement Parametric attenuator?
1 Attachment(s)
I have implemented a compensation scheme for tiny 2 inch, pipe-based speakers which needs more than 20dB of equalization at low frequencies. This leads to the graph of the Parametric EQ section 'clipping' at +20dB even though the EQ itself works fine. It is apparently a display limitation in the plugin.

One quick and dirty way out would be to implement an attenuator in the Parametric EQ common to outputs 1&3. Is this possible (perhaps with a Biquad?). Else it is possible to change the displayed range on the plugin graphs to -10dB to +30dB (for example)?

minidsp 31st August 2011 07:37 AM

Mmmh, well 20dB of EQ is actually a LOT.. :-)

Though it may sound fine, it's not impossible (depending on your gain structure) that you could saturate inside the DSP without knowing it... In general, applying that much EQ (on any speaker / DSP or not) is not good idea/typical and that's basically the reason most DSP platform you'll ever see would have a scale similar to the miniDSP (some even capped to +/-15dB).

Hope this clarifies,


trevmar 31st August 2011 01:04 PM

Dear MiniDSP Devteam.
I can't believe that any DSP design team would tell a user that the application they are implementing is "not a good idea." What I have done with the MiniDSP is groundbreaking, and I would hope you guys will be able to learn from it, and not look down your noses saying "we never intended the MiniDSP to be used in that way."

Did you react in the same way when the Biquad computational element was suggested to you?

Thanks for responding, in any case.. I guess I will have to keep working on the issue myself

minidsp 1st September 2011 01:13 AM

First of all, please let us apologize if our earlier message sounded offensive or like if we were looking down on you. That wasn't our intent at all. We were just speaking on a technical aspect (electrical effect of adding 20dB to an amp/DSP 28bit path) and that's all there is to read. :-) We're not here to judge how the kits are used as there are indeed many good use our there. We'd love to hear of your utilization for this setup! :-)

What we wanted to highlight is the reasoning behind the 20dB scale & why it was capped on the scale to prevent from getting complaints of saturation inside the DSP (e.g. adding many PEQ boost to the same frequency) . If you do a bit of research on fixed point DSP (28bit) and their internal gain structure, I think that it will help understand a bit more of the headroom inside a DSP.

Ultimately, as we said earlier on, it really depend on your gain structure. Providing you're careful at the input and leaving enough headroom you may be fine. You could look into building a gain scaling biquad (i.e. gain linear value in Biquad equations) if you want to scale the whole PEQ down. Not sure it would perform the same effect you're expecting though as it would just be equivalent to simply not boosting the signal that much for each band.


trevmar 1st September 2011 01:35 AM


Originally Posted by minidsp (
Not sure it would perform the same effect you're expecting though as it would just be equivalent to simply not boosting the signal that much for each band.

True, but lowering the gain would allow me to visualize better the effect of small changes I am making in the region which is saturated high at the moment.

The headroom is fine - I am using the Analog gain potentiometer as the master gain control, with 0dB out of the source driving the input close to 0dB. Then the 80dB attenuation in stage controlled by the gain pot.

What I was asking about the biquads is this - the last time I designed with a DSP was 1986, and that was a head positioning servo for one of the early 5inch hard drives. So I am not familiar with Biquads - I am still learning how to manipulate the coefficients. I just implemented an LR 24dB lowpass in Biquad, which, to my surprise, worked, and at that point it was easy to work out that a gain biquad would just have B0=gain_coefficient, with all other parameters zero.

So I figured it out :) Anyway, thanks for responding. Sorry if I snapped back too harsh a reply last time. I guess I am just too lazy to do everything myself :)

minidsp 2nd September 2011 07:50 AM

Cool, it makes sense now what you were after!
Glad to hear that you got it figured out and looking forward to seeing these in the miniDSP project gallery!

Have fun,


All times are GMT. The time now is 11:51 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2