Feeler: Quad 8 444 EQ with M/S extension in 51x format.

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
ruairioflaherty said:
Well done on going from schematic to proto so quickly. 

As a mastering engineer the main reason I couldn't use this build is ergonomics.  For me the space between knobs and the 1dB steps would rule it out for me.  I realize that this is much more likely to be a bus EQ for a space challenged engineer/mixer rather than a purist mastering tool.

Again well done on dealing with the noise issue so quickly.  Absolute levels of noise are only part of the discussion, the character of the noise is equally important.  For example my Pendulum OCL2 is very noisey but mostly at very low frequencies which is very tolerable for my work.

Ergonomics is without a doubt the biggest issue.

The steps, actually not so much, let me explain. It is not featured in the prototype, but the step height is solely defined by the value of the resistor in R82. For the 500 version I plan in adding another control switch. That would allow to make the EQ steps either 0.5dB , 1dB or 2dB.

Coming back to ergonomics, my wife is apprenticed to a mastering engineer in the bay area and I designed this EQ for her. She is 5'1" and for her space matters and she has no problem in using the knobs.

For the version that I am building for her mentor. I am racking this bastard and using http://www.ebay.com/itm/New-2-Pole-23-Step-Rotary-Switch-Attenuator-Volume-Control-Pot-Potentiometer-DIY-/221398438804?pt=US_Amplifier_Parts_Components&hash=item338c601394 this attenuators to provide either +/-3.0, or +/- 6.0 with 0.25dB or 0.5dB steps.

Noise is very low now... it has a white noise floor of EIN -135dB with a flicker noise of <500nV/Hz^0.5.

I really enjoy that I have managed to put it in a 500 series envelope, but I agree that it is much more useful when it is racked. That is basically the reason the frequency switches are on the PCB while the gain control and other switches only have header connectors.

I wanted to make it fit in a single space so the entire pair would fit in 2 space on a 500 series. At the same time, you can make this fit on a 1U space as well. But if you have a 1U space you probably won't need to deal with diminutive SMT parts.
 
dmnieto said:
Ergonomics is without a doubt the biggest issue.

The steps, actually not so much, let me explain. It is not featured in the prototype, but the step height is solely defined by the value of the resistor in R82. For the 500 version I plan in adding another control switch. That would allow to make the EQ steps either 0.5dB , 1dB or 2dB.

Very nice.


Coming back to ergonomics, my wife is apprenticed to a mastering engineer in the bay area and I designed this EQ for her. She is 5'1" and for her space matters and she has no problem in using the knobs.

I think I know exactly who your wife's mentor is!  And I've been in her room when visiting the studio, and yes space matters :)

For the version that I am building for her mentor. I am racking this bastard and using http://www.ebay.com/itm/New-2-Pole-23-Step-Rotary-Switch-Attenuator-Volume-Control-Pot-Potentiometer-DIY-/221398438804?pt=US_Amplifier_Parts_Components&hash=item338c601394 this attenuators to provide either +/-3.0, or +/- 6.0 with 0.25dB or 0.5dB steps.

Noise is very low now... it has a white noise floor of EIN -135dB with a flicker noise of <500nV/Hz^0.5.

I really enjoy that I have managed to put it in a 500 series envelope, but I agree that it is much more useful when it is racked. That is basically the reason the frequency switches are on the PCB while the gain control and other switches only have header connectors.

I wanted to make it fit in a single space so the entire pair would fit in 2 space on a 500 series. At the same time, you can make this fit on a 1U space as well. But if you have a 1U space you probably won't need to deal with diminutive SMT parts.

I strongly advocate avoiding the cheaper Chinese switches all over ebay, I know too many people who've had to rip apart builds and start again with proper Elma/Grayhill/Electroswitch.  You've put a lot of thought into this whole process, I look forward to seeing the rack mounted version when I't done.

Cheers,
Ruairi
 
ruairioflaherty said:
I strongly advocate avoiding the cheaper Chinese switches all over ebay, I know too many people who've had to rip apart builds and start again with proper Elma/Grayhill/Electroswitch.  You've put a lot of thought into this whole process, I look forward to seeing the rack mounted version when I't done.

Cheers,
Ruairi

Prototyping is king, verify before breaking the bank. I have some ELMA single pole with 47 positions on sight for a surreal +/- 6dB on .125dB increments, I don't even know how you guys can make a difference between such a diminutive step. It is actually challenging to find the right values because you need to start messing with 0.01% resistors and logarithmic scales suck.

 
dmnieto said:
Prototyping is king, verify before breaking the bank. I have some ELMA single pole with 47 positions on sight for a surreal +/- 6dB on .125dB increments, I don't even know how you guys can make a difference between such a diminutive step. It is actually challenging to find the right values because you need to start messing with 0.01% resistors and logarithmic scales suck.

Sure, my version is that a good switch can always be harvested from a bad project and re-used!

As a full time mastering guy I have no use for anything less than 0.5dB steps, although and yes messing with super low value resistors is a PITA.

 
Well, it is going to take a bit more of time to solve this.

There only one think that I consider show-stopper: Instability.

And unfortunately this current circuit is oscillating, the balanced receiver (IC3), the first element of the circuit oscillates quite strongly after about 40 minutes of use. I first thought that the power rails were collapsing (they weren't) For a while I believed that it could be PSRR related, so I added additional filtering on the opamp power.... didn't work. I also tried increasing the compensation capacitance... didn't work either. Adding a bypass capacitor parallel to the gain trimmer? Nope.

With that all the simple solutions are out. Some part of it may be related with the curious choice of having the feedback path after the decoupling capacitor. But the reality is that it might need an addition of a small stopper resistor, which may mess with the frequency response of the EQ or increase distortion.

I need to give it a big though about what in the layout may be causing this oscillation. :(

I'll update when I have news.

If anybody wants to know why HF oscillation is bad, let me explain how it sounds like. Brutal distortion (>5%), independent of input level. That appears suddenly after some use. Power cycling of simply touching the inverting pad will make the distortion disappear.

I lack the equipment (oscilloscope, I only have a dynamic analyzer capped at 101KHz) to see it, I know that dropping the gain trim makes the circuit stable... But for me it is not enough, i have not seen in the manuals any mention of instability on the gain switch, so the issue must be in my layout
 
So it is not oscillation the actual problem... It is actually DC offset at the output.

In the current circuit on the first opamp the decoupling capacitor is placed before the feedback path... as result of this, DC starts charging the capacitor slowly... till finally the DC reaches a breaking point and the opamp starts to clip on the positive rails...

Other than changing the circuit, does anybody have an idea of how to solve this? I may use additional circuitry to cancel the offset. I just wonder why would they put the capacitor before the feedback path.

Any ideas?
 
So I tried with a LME49710 in the place of the NE5534 and the circuit is stable with no DC drift. I tried other NE5534 (like Siliconix) and after checking with simulation (where the distortion also occurs), and taking into account that no offset cancelation exists in-circuit I have to conclude that the schematic I have must be wrong, as the circuit with a NE5534 is unstable.

For that, I will modify the circuit by placing the feedback loop before the de-coupling capacitor and adding a small 50 ohm resistor between the output of the opamp and the feedback look. This should increase the stability and allow any opamp to work. I could add jumper traces to select the path, but that net is very sensible to capacitance and I'd rather not adding harmful effects.

 
dmnieto said:
So I tried with a LME49710 in the place of the NE5534 and the circuit is stable with no DC drift. I tried other NE5534 (like Siliconix) and after checking with simulation (where the distortion also occurs), and taking into account that no offset cancelation exists in-circuit I have to conclude that the schematic I have must be wrong, as the circuit with a NE5534 is unstable.

For that, I will modify the circuit by placing the feedback loop before the de-coupling capacitor and adding a small 50 ohm resistor between the output of the opamp and the feedback look. This should increase the stability and allow any opamp to work. I could add jumper traces to select the path, but that net is very sensible to capacitance and I'd rather not adding harmful effects.

Excellent!  I can't assist with this on a technical level I'm afraid, but taking a real interest in seeing how this project develops.  Great work!
 
This is the noise floor of the unit. The harmonics of 60Hz are probably due to the circuit basically being floating above my test rack and bathed in 60Hz radiation (I really need to get a gdiy51x rack "pronto")

Noise floor is now at -140dBFs with the LME49710 opamp (-135dBFs before) dropping almost 5dB...

That LME49710 is reaaaly quiet
 

Attachments

  • Q444_noise_floor.png
    Q444_noise_floor.png
    19.1 KB · Views: 45
This looks really well laid out for all it is.  Kudos to you dmnieto!  Put me down on the group-buy list for a pair of these.  Keep up the amazing work!  :D
 
Update.

I got back the second spin of the PCB... I changed the grounding scheme and added more decoupling to the opamps... I also added my divergences from the original design (feedback chain before the capacitor in the de-balancing opamp) and the most important... I used a common trick and referenced the power supplies (in AC) to the inputs of the de-balancing opamp. That makes that any hum and its harmonics are common mode to the opamp and they get massively attenuated. PSRR is increased by almost 50dB now...

You remember those peaks in the noise trace? they are gone now. 
 
rob_gould said:
Hi there,

Any update on this one?

Oh, I took some vacation :) i have been a bit distracted in the last couple of weeks. I am optimizing a bit the layout to make sure it is a bit easier to solder before I publish the final gerbers
 
i'm sorry to come on so late, but i'd like to do a pair w/ iron/switches/panel, and maybe another pair of pcb/switches.  i think if everyone contributed an extra 10% to offset your time spent, it's well worth doing. or those who want to - that's a lot of work!! 
 
31hz said:
i'm sorry to come on so late, but i'd like to do a pair w/ iron/switches/panel, and maybe another pair of pcb/switches.  i think if everyone contributed an extra 10% to offset your time spent, it's well worth doing. or those who want to - that's a lot of work!!

I took a small break during the summer... my apartment looks like a war zone with at least 4 projects pending... So you are not late. I just didnt have the time to properly finalize it... but stay tuned. I will make this happen :)
 
Looks great, I'd love in on this as well for at least 2.  I may rack mine and try and run them off of full 28v rails.
 
Back
Top