3D "AIR" EQ - "Night EQ" PCB's Complete!

[frazzman]

Thanks Shot + Harpo for your insight, its greatly appreciated. I've taken all of this information on board in tweaking my build further...I'm starting to get my head around the intracacies of this design, its seemed to be deceptively simple... (at first...)

I have replaced the 5k62 series resistors at each band with a 20k multi-turn trimmer. I set the initial value for each trimmer to the series value as calculated in Harpo's spreadsheet. I then fitted the trimmers onboard. I also replaced R30 with a 56k resistor + 20k trimmer, as I knew that R30 would have to be tweaked > 56k due to my R pot center values.

My aim has been to tweak each trimmer with the intention of flatening the frequency curve (all pots set to 0db center) as tested in RMAA. Adjustments have been made gradually on each band.

My freq response at 0db center is still not flat but it's better...I suppose I am aware that the approach i'm taking is very much trial and error as I still don't have my head around the design completely. If the curve indicates a band is > 0db then the resistance is increased on the trimmer, if its < 0db, resistance is decreased.

The worst is a peak on the 650hz band about +0.3db which looks to be a side effect of a cut occuring an octave lower. I've got my 650hz trimmer at its lowest resistance positon so I can't offset the peak any further...

Some questions...:

Shot > thanks for your photo, it certainly helped.
- In your photo - I can see that you fitted a trimmer to the hi-shelf 5k6 resistor position, as I understand it, there shouldn't be any tweaking required on this band, as its additive only, so the 0db at center is of no concern. Any reason why you fitted a trimmer? I have a bump around 7k but it is present in my sweeps during bypass so I am assuming this is normal...

Harpo > thanks for the detailed breakdown... still trying to compute it all. Just a question in regards to the R fb/R30 value... whats the implication of resistance being set too high or too low here? If the value is too low, it would limit the max level setting across all the bands? What if its too high?
Given that I have replaced R30 with a 56k + 20k trimmer, I won't be able to tell if its set to max(R pot. center) * 1.1 ... (As per your formula) without getting the trimmer out of the circuit ...how precise does this value need to be?

My current frequency plot is attached, I am starting to think this is as good as its going to get...

Sorry for the long winded reply and I am reinventing the wheel here, I looked through this thread many times, but this area of the build is very murky, so I am hoping that I can share what I learn with future builders ...

 

[shot]

- In your photo - I can see that you fitted a trimmer to the hi-shelf 5k6 resistor position, as I understand it, there shouldn't be any tweaking required on this band, as its additive only, so the 0db at center is of no concern. Any reason why you fitted a trimmer?

I was aware that it shouldn't be of any concern, but when I was about to unscrew and desolder, I thought "what the hell". Since I had enough I just went on it also. Just to be sure I won't have to do it all again in case this value also needs to be tweaked.

 

[DeathToneM@]

This is great information for me as I'm just starting the process of getting the correct value resistors..
Found this project to be challenging in a good way and learnt more than some of the more 'painting by numbers' type projects.. Anyway, thanks for sharing guys!

 

[frazzman]

- In your photo - I can see that you fitted a trimmer to the hi-shelf 5k6 resistor position, as I understand it, there shouldn't be any tweaking required on this band, as its additive only, so the 0db at center is of no concern. Any reason why you fitted a trimmer?

I was aware that it shouldn't be of any concern, but when I was about to unscrew and desolder, I thought "what the hell". Since I had enough I just went on it also. Just to be sure I won't have to do it all again in case this value also needs to be tweaked.

Hi Shot, thanks for the info. Did you end up having to trim on the hi shelf pot ? Did you approach the trimmer adjustments in a similar was to me ? I.e. Adjust in whatever way makes freq response flatter at 0db center. My concern is that I don't know what value R30 ends up being, how does your freq response compare to mine ? All my boosts and cuts are working properly at least

 

[Harpo]

Harpo > thanks for the detailed breakdown... still trying to compute it all. Just a question in regards to the R fb/R30 value... whats the implication of resistance being set too high or too low here? If the value is too low, it would limit the max level setting across all the bands? What if its too high?
Given that I have replaced R30 with a 56k + 20k trimmer, I won't be able to tell if its set to max(R pot. center) * 1.1 ... (As per your formula) without getting the trimmer out of the circuit ...how precise does this value need to be?

Value is not critical, only adjusting overall unity gain level (IE 0dB in/0dB out with level pots centered and air-band off). I'd not put a trimmer in series to R30, but YMMV. If the trimmer fails one day -most often only the wiper looses contact but I've seen them fail at the outer ends as well- this summing amp stage is running open loop, only limited by rail voltages. I'd probably increase R30 to maybe 120K and put a 200K trimmer in parallel with trimmers initial value about centered for a much wider + and - trim possibility and if this trimmer should fail, this would only be a +6.6dB boost in level instead of a show stopper.
Tweaking parts values, you seem to be starting at the wrong side. As repeatedly said, you won't get a flat response out, no matter how much you tweak the level pot center values/summing amp gain settings, as long as you don't have the center frequencies of the different bands in the correct spot. FI you could plot the responses of the 10,40,160,640Hz bands when measuring at the NE5532 pin 1 and pin 7 of IC2 and IC3. Some DAW software would give you exact numbers at what frequency each max.level is located, just to know where these center frequencies of your build are located exactly if you don't want to measure caps values and calculate the corresponding resistor values needed to get there. Maybe reread my last post.

 

[frazzman]

Harpo > thanks for the detailed breakdown... still trying to compute it all. Just a question in regards to the R fb/R30 value... whats the implication of resistance being set too high or too low here? If the value is too low, it would limit the max level setting across all the bands? What if its too high?
Given that I have replaced R30 with a 56k + 20k trimmer, I won't be able to tell if its set to max(R pot. center) * 1.1 ... (As per your formula) without getting the trimmer out of the circuit ...how precise does this value need to be?

Value is not critical, only adjusting overall unity gain level (IE 0dB in/0dB out with level pots centered and air-band off). I'd not put a trimmer in series to R30, but YMMV. If the trimmer fails one day -most often only the wiper looses contact but I've seen them fail at the outer ends as well- this summing amp stage is running open loop, only limited by rail voltages. I'd probably increase R30 to maybe 120K and put a 200K trimmer in parallel with trimmers initial value about centered for a much wider + and - trim possibility and if this trimmer should fail, this would only be a +6.6dB boost in level instead of a show stopper.
Tweaking parts values, you seem to be starting at the wrong side. As repeatedly said, you won't get a flat response out, no matter how much you tweak the level pot center values/summing amp gain settings, as long as you don't have the center frequencies of the different bands in the correct spot. FI you could plot the responses of the 10,40,160,640Hz bands when measuring at the NE5532 pin 1 and pin 7 of IC2 and IC3. Some DAW software would give you exact numbers at what frequency each max.level is located, just to know where these center frequencies of your build are located exactly if you don't want to measure caps values and calculate the corresponding resistor values needed to get there. Maybe reread my last post.

Thanks Harpo, I just re-read your previous posts on this subject - I think the penny finally dropped for me. Looking over this thread for many hours - What I am wondering is why there aren't more builders on this thread who have mentioned these issues. There are a few references to the issues but not in any great detail. In my mind, these 'problems' seem to be unavoidable as a consequence of the design... ?

I need to rework much of the board to encompass your suggestions. Looking over your previous posts in more detail, I am hoping I've finally got the gist of whats required.

This is my plan of attack:

1. Desolder paralled 2uf/0.2uf cap combos from each filter section

2. Measure 'real world' capacitance of paralled caps:
SUB-C55,C11 (2uf) & C12,C19 (0.2uf)
40hz-C54,C9 (2uf) & C10,C18 (0.2uf)
160Hz-C53,C7 (2uf) & C17,C8 (0.2uf)
650hz-C52,C5 (2uf) & C16,C6 (0.2uf)
2.5k-C51,C6 (2uf)

3. Calculate required resistor values using formula - Rx=1/(2*PI*frequency*measured capacitance in F)
Based upon actual measurements of paralled caps
SUB-R19,R21
40hz-R18,R20
60hz-R15,R17
650hz-R12,R14
2.5k-R9,R11

4. Solder fixed value resistor+trimmer combo to achieve calculated resistor values. Using DMM, measure each resistor+trimmer combo and preset to calculated values.
> Solder resistor combo to PCB @ required positions.

5. Adjust 5k62 series resistor(s) - (R44, R42, R40, R38, R36) based on 470k rev log pot center values (use Harpo's spreadsheet)
> Use 20k multiturn trimmers, using DMM preset each value to calculated values using Harpo's spreadsheet
> Fit trimmers in R44, R42, R40, R38, R36 positions

I know your recommendation was to fit resistor+trimmer(s) in parallel to prevent catastrophe if failed trimmer is incurred, but my thoughts were to replace the trimmers once the the correct values have been achieved.

Have I possibly missed anything?

Thanks for your on-going help, I'm hoping my learning can assist other builders in future. I have spent so much time building this EQ so I am keen to get it 100%.

 

[frazzman]

This is great information for me as I'm just starting the process of getting the correct value resistors..
Found this project to be challenging in a good way and learnt more than some of the more 'painting by numbers' type projects.. Anyway, thanks for sharing guys!

Hi DeathToneM,
Glad to see someone else still interested in this project... much credit to Harpo, the genius behind helping me refine my build!

What has become clear to me is that this a very easy build, from a 'component installation' perspective - there are no esoteric components or any difficult wiring/soldering....

However - the part that I completely over looked was how much work it would take to get the frequency behavior right... Having built many many other popular and apparently more complex projects, I have yet to incur one as fiddly and as time consuming as this....

 

[DeathToneM@]

However - the part that I completely over looked was how much work it would take to get the frequency behavior right... Having built many many other popular and apparently more complex projects, I have yet to incur one as fiddly and as time consuming as this...

Yep, my thoughts exactly.. But it's much easier to work out after reading yours and Harpo's explanations! And will be for new builders to come.. So thanks to all involved for documenting that..
I personally think this EQ is worth getting right as it sounds great!

 

[frazzman]

However - the part that I completely over looked was how much work it would take to get the frequency behavior right... Having built many many other popular and apparently more complex projects, I have yet to incur one as fiddly and as time consuming as this...

Yep, my thoughts exactly.. But it's much easier to work out after reading yours and Harpo's explanations! And will be for new builders to come.. So thanks to all involved for documenting that..
I personally think this EQ is worth getting right as it sounds great!

Completely agree... I wish I knew what I know now before having built this EQ... I have soldered and de-soldered components that many times ... Few busted PCB traces down the track...
With foresight, now knowing how particular this design is with component values... I'd measure my 2uf/0.2uf parallel caps and tailor the resistors accordingly as per Harpo's formula. Hopefully Harpo concurs that my plan of attack is a correct interpretation of what is required here... You will have he benefit of all of this knowledge upfront so hopefully you won't have to dissect your build like I have... I also wish of gone with stepped switches but my front panel design doesn't have the space for that many rotary switches... Fun and games

 

[Harpo]

This is my plan of attack:

1. Desolder paralled 2uf/0.2uf cap combos from each filter section

Why ?
Just pull IC2 and IC3 and lift one side of the frequency/gain setting resistors that would be lifted anyway to adjust to the required value, IE R21 where it connects to C11||C55 and either side of R23 for the sub band as an example.

5. Adjust 5k62 series resistor(s) - (R44, R42, R40, R38, R36) based on 470k rev log pot center values (use Harpo's spreadsheet)
> Use 20k multiturn trimmers, using DMM preset each value to calculated values using Harpo's spreadsheet
> Fit trimmers in R44, R42, R40, R38, R36 positions

substituting the 5K62 series resistors (ment for 500k rev.log pots with 10% value at 50% rotation) with -depending on value needed- 10K or 20K trimmers, trimmers pretrimmed to the calculated value and fitted in R45, R43, R41, R39, R37 positions would give you some spare 5k62 resistors and a cleaner layout. YMMV.

 

[frazzman]

This is my plan of attack:

1. Desolder paralled 2uf/0.2uf cap combos from each filter section

Why ?
Just pull IC2 and IC3 and lift one side of the frequency/gain setting resistors that would be lifted anyway to adjust to the required value, IE R21 where it connects to C11||C55 and either side of R23 for the sub band as an example.

5. Adjust 5k62 series resistor(s) - (R44, R42, R40, R38, R36) based on 470k rev log pot center values (use Harpo's spreadsheet)
> Use 20k multiturn trimmers, using DMM preset each value to calculated values using Harpo's spreadsheet
> Fit trimmers in R44, R42, R40, R38, R36 positions

substituting the 5K62 series resistors (ment for 500k rev.log pots with 10% value at 50% rotation) with -depending on value needed- 10K or 20K trimmers, trimmers pretrimmed to the calculated value and fitted in R45, R43, R41, R39, R37 positions would give you some spare 5k62 resistors and a cleaner layout. YMMV.

Harpo, thanks! Invaluable info and insight... Off to the local electronics supply store tomorrow to pickup some more trimmers, I'll report back on my progress... I'm hopeful that our exchanges on this subject will serve future builders well... I'm excited to hopefully finish this off finally

 

[frazzman]

This is my plan of attack:

1. Desolder paralled 2uf/0.2uf cap combos from each filter section

Why ?
Just pull IC2 and IC3 and lift one side of the frequency/gain setting resistors that would be lifted anyway to adjust to the required value, IE R21 where it connects to C11||C55 and either side of R23 for the sub band as an example.

5. Adjust 5k62 series resistor(s) - (R44, R42, R40, R38, R36) based on 470k rev log pot center values (use Harpo's spreadsheet)
> Use 20k multiturn trimmers, using DMM preset each value to calculated values using Harpo's spreadsheet
> Fit trimmers in R44, R42, R40, R38, R36 positions

substituting the 5K62 series resistors (ment for 500k rev.log pots with 10% value at 50% rotation) with -depending on value needed- 10K or 20K trimmers, trimmers pretrimmed to the calculated value and fitted in R45, R43, R41, R39, R37 positions would give you some spare 5k62 resistors and a cleaner layout. YMMV.

Hey Harpo,

I've spent the last few days reworking my left channel as per all the discussed recommendations, I had to use a lot of multiturn trimmers due to the 'non standard' resistor values required. 

I have calculated resistor values against real-world values for 2uf & 0.2uf paralleled caps. 5k6 series resistor values were also calculated and set with trimmers based upon your spreadsheet.

All trimmers were pretrimmed and checked with a DMM before fitting to the circuit.

These are my calculations:

Calculated resistor values adjusted for paralled caps (2uf/0.2uf):

10 - C55+C11 = 2.1 - R21 - Calculated Resistor Value = 7.58
10 - C12+C19 = 0.201 - R23 -    Calculated Resistor Value = 79.18
40 - C54+C9 = 1.991 - R18 - Calculated Resistor Value = 2.00
40 - C18+C10 = 0.201 - R20 - Calculated Resistor Value = 19.80
160 - C53+C7 = 1.975 - R15 - Calculated Resistor Value = 503.66R
160 - C17+C8 = 0.203 - R17 - Calculated Resistor Value = 4.90
640 - C52+C5 = 1.979 - R12 - Calculated Resistor Value = 125.66R
640 - C16+C16 = 0.201 - R14 - Calculated Resistor Value = 1.24
1280 - C51+C3 = 1.973 - R9 - Calculated Resistor Value = 63.02R


470k rev Log pot,  center indent adjusted values:

Sub - R44 - 445,000 - 56,800 - R45 = 10,300
40 Hz - R42 - 472,000 - 54,600 - R43 = 12,500
160 Hz - R40 - 483,000 - 55,600 - R41 = 11,500
640 Hz - R38 - 451,000 - 54,400 - R39 = 12,700
2.56 kHz - R36 - 481,000 - 61,000 - R37 = 6,100

R fb = 67,100

Attached is the new frequency plot. Still not flat, but the curve looks 'healthy'... Do you think adjusting the series resistors will get me closer, or more component tweaking? I've been meticulous about making sure every trimmer was set to the correct value.

The only thing I didn't touch in this whole process was the air band, is any tweaking required here? As you can see this is a bit of a boost up high

Any thoughts about where to go from here? This has been quite a learning process. The circuit board is starting to look a bit like a charred mess from soldering and desoldering so many times!  I'm really determined to get this right....

 

[Harpo]

Attached is the new frequency plot. Still not flat, but the curve looks 'healthy'... Do you think adjusting the series resistors will get me closer,..

Yepp (wild guessing better than +/-0.1dB). Reason is, tweaking for correct center frequencies also changed the gain setting ratio of the feedback und shunt resistors.

 

[frazzman]

Attached is the new frequency plot. Still not flat, but the curve looks 'healthy'... Do you think adjusting the series resistors will get me closer,..

Yepp (wild guessing better than +/-0.1dB). Reason is, tweaking for correct center frequencies also changed the gain setting ratio of the feedback und shunt resistors.

I think I follow you... I'm thinking some subtle adjusting of the  series resistor trimmers will get me where I want to be. Only thought is the +0.5 boost between ~5-10k looks a bit ominous ? Should I pay any attention to r35 on the high shelf band ?  I've left it at 5k62 as I figured that with the rotary switch in the off position it shouldn't be introducing a boost in this region when off ? This has been a good learning curve... Thanks again

 

[frazzman]

Hey Harpo,

Sorry to keep spamming this thread...

Spent some more time tweaking this EQ, I am not too sure if I completely understood when you said "tweaking for correct center frequencies also changed the gain setting ratio of the feedback und shunt resistors."

I'm still trying to correct the curve but since I don't fully understand this area I feel like i'm aimlessly adjusting trimmers, this is how the curve looks at the moment...

My frequencies must be slightly off (not exactly an octave apart, so high 40hz results in dip at 80hz etc) 

Is there a methodogical way you can suggest to further correct the frequencies? What has become very apparent with this design is that everything needs to be exact... as you know I already set the frequency resistor values precisely with the trimmers

Hope other builders can learn from this on going discussion...

Thanks

 

[frazzman]

This post post may fall into the void as I'm not sure if there are too many people still interested in this build. It's safe to say that I've become a little obsessed about getting this build 100%...

I just finished reading over every page of this thread... I can now see that many other builders have had issues with getting a flat response with pots centered. After days/months trying to achieve this, it might be time to move on..

After some more deliberation - What I'd like to know is, is there a significant impact running the rails at -/+ 16v rather than 18?

Reason I asked is that I can't get anymore than +6db boost on any band, I am running my rails a little low at 16v, could this be starving the circuit and impacting on max boost ? I know that r30 value could limit the total amount of gain possible across all bands but +6db boost max seems very low

 

[Harpo]

Reason I asked is that I can't get anymore than +6db boost on any band, I am running my rails a little low at 16v, could this be starving the circuit and impacting on max boost ? I know that r30 value could limit the total amount of gain possible across all bands but +6db boost max seems very low

Nothing to do with +/-16V or +/-18V. For +6dB max.boost your R30 parts value must have decreased or series resistors increased a lot. Usual boost is about +14.3dB with 5K62 series resistors and 56K2 Rfb. What is the amount of your RMS input voltage ? You might have the 'normalize the amplitude of test signals before analysis' checked in RMAA and a dBFS to dBV translation is converter specific, so might be opamps signals operating close to their supply boundaries.

 

[frazzman]

Reason I asked is that I can't get anymore than +6db boost on any band, I am running my rails a little low at 16v, could this be starving the circuit and impacting on max boost ? I know that r30 value could limit the total amount of gain possible across all bands but +6db boost max seems very low

Nothing to do with +/-16V or +/-18V. For +6dB max.boost your R30 parts value must have decreased or series resistors increased a lot. Usual boost is about +14.3dB with 5K62 series resistors and 56K2 Rfb. What is the amount of your RMS input voltage ? You might have the 'normalize the amplitude of test signals before analysis' checked in RMAA and a dBFS to dBV translation is converter specific, so might be opamps signals operating close to their supply boundaries.

Thanks Harpo, i'm not in the studio now to test.. but I think you're right, I checked in RMAA and the 'normalize...' checkbox is ticked by default on a fresh install.

Rfb is set to 67,100.
R Series values are set as follows = SUB: 10,300; 40hz: 13,100; 160hz: 11,500; 640hz: 12,700; 2.5k: 6,100
Each of these values + pot center value result in a gain value of exactly 1 so the pot center calibration should be spot on.

So to address your point, R30 has actually increased and the series values aren't excessively high.

Whilst I have you, in your previous post you mentioned that tweaking the center frequencies also changed the gain setting ratio of the feedback and shunt resistors - whats the best way to systematically address that? Or maybe its best to leave it with some -/+ 0.2db fluctuation. Looking over this entire thread from start to finish again last night, it looks like this issue has plagued many users and maybe ear is the best tool here ?

 

[frazzman]

The battle with this eq continues...
After fitting 14 trimmers to my left channel, many measurements and tweaks... My left channel is almost flat at pots centered, certainly as good as it will get.

I just completed the same surgery on my right channel.
Unfortunately after all these mods, I have run into an issue on the right channel. I have a very very low output, doesn't even register on my meters... With headphones on and the volume cranked I can hear a faint signal passing, you can hear changes in the audio as pots are adjusted, so I know it's working in principal

Could I have introduced a short somewhere that could result in this behavior ? All voltages are good throughout circuit, I can't seem to track down the fault. I suspected something had gone a miss around the feedback resistor but no such luck. Same output in bypass and out

Any help greatly appreciated, I can't rest until this project is complete

 

[Harpo]

You already excluded a broken right side XLR input or output connection ?
With input signal present, compare right side with left side channel to see where signal get lost might be the easiest approach, following from input to output, IE compare AC voltages at U4-pin1,...., U5-pin6, U4-pin7.