Neumann W492 equalizer DIY project (PCB docs added)

[5v333]

I would love to take that one for a ride.
Great job!

 

[delaymix]

I do not have words...! an awesome job! Congratulations!

 

[weiss]

Hi all,

I recently finished my W492, and wanted to take a few minutes to document the build. I've modified the core EQ pretty heavily with a view to using primarily for 2-bus / mastering work, and also built a number of other functions in, so hopefully this will be useful to some folk down the line. Huge thanks to everyone who's posted on this project across the site, and of course to Audiox - I'm definitely perching on the shoulders of giants here!

wow, awesome work andy!!! 

 

[TwentyTrees]

Thanks Weiss, and everyone else for the very welcome feedback! Think I'll tackle something more straightforward for my next build - a new Hadron Collider or something... 

 

[erquere]

Hi ! Could someone tell me which kind of available pots I can buy for this project ? Without spending 50 bucks for one ? I've been searching and bought wrong ones at Mouser...

 

[erquere]

Do these ones fit, or not ?

http://capi-gear.com/catalog/product_info.php?cPath=50_131&products_id=446

 

[erquere]

I finally found some Vishay-Sfernice pots... now the question is : if I just buy two dual-gand pots (P4 and P6 ?) and six mono pots, is it ok ?

 

[erquere]

Could wirewound pots fit too ? Much much cheapier than Sfernice ones !

 

[alhe]

Some really nice builds in this thread. How is the W492 for tracking?

I see most people build this without transformers. In my experienece, most, if not all Neumann modules have had Haufes installed - is this not the case for the W492?


Edit:
So to answer my own question, the original W492 uses a Haufe T4344/90485 input transformer and a electronically balanced output.

pri: dBu 10V
2:1
200:4,7k
20-30kHz +/- 0,5dB

 

[pollomartian]

Can I use 47k potentiometers? I have some plastic conductive potentiometers, but they are 47k...

 

[Whoops]

Can I use 47k potentiometers? I have some plastic conductive potentiometers, but they are 47k...

The pots are dual gang 10K LIN.

Use the correct ones

 

[pollomartian]

Hi all,

I recently finished my W492, and wanted to take a few minutes to document the build. I've modified the core EQ pretty heavily with a view to using primarily for 2-bus / mastering work, and also built a number of other functions in, so hopefully this will be useful to some folk down the line. Huge thanks to everyone who's posted on this project across the site, and of course to Audiox - I'm definitely perching on the shoulders of giants here!

Features:

• Dual mono Audiox W492 4-band EQ with fully parametric mid bands, plus Harrison Ford Filters HPF / LPF
• Rotary switches for frequency (24 steps/band) & gain (12 steps/band, switchable +/-)
• Channel components matched to high tolerance (less than 0.1dB difference between channels at all settings - except when boosting or cutting 15dB on the high shelf, when there's 0.2dB difference!)
• WIMA filter caps for low shelf and low mid bands, Styroflex for high mid and high shelf
• Switchable mid/side matrix (a la Wayne Kirkwood)
• Per-channel bypass (inside mid/side matrix) and master bypass
• Vintage Jensen output transformers (switchable per channel - relaxes the sound very slightly)
• Switchable gain range per channel (x1 / x0.5)
• Switchable frequency range for low mid (x1 / x0.5) and high shelf (x1 / x3) bands
• Variable bandwidth for low mid and high mid bands (Q = 0.45 to 3.6)
• THAT1246/1646/LT1058 ICs used in HPF/LPF
• Backlit plexiglass logo badge 8)
• Panels and NRG case by Frank

It sounds superb - very clean and precise, low distortion, and quieter than I expected given the amount of wiring involved (sub -94dBFS). There are 13 PCBs including filters, M/S matrix, power supplies, relays etc, and all the front panel controls are wired off-board, so hooking everything up took a good few days... The combination of fully parametric mid bands and shelves plus filters makes it really flexible, and I'm very happy with the styroflex caps for the high end in particular.

I spent a long time planning, listening and testing with this one - a good couple of years of reading and planning, and dozens(!) of iterations of front panel design. With all the potential for issues with such a complex build I was more than usually nervous when firing it up for the first time, but debugging was mercifully straightforward this time - the biggest issues were a bad solder joint in a SMT frequency switch, and popping on some switches (see below). In terms of mods to the W492, here's what I did following a ton of research, breadboarding and measuring (all component references are to the Audiox schematic, and I used REW for measurements):

Variable bandwidth
10K pot (RP) with 3k3 series on CCW (RB2) across U3A (low mid) and U5A (high mid), wiper to ground via 330R series (RB3). This involved running wires to one side of each of R16, R17, R26 (low mid) and R31, R32 and R41 (high mid) - I left little loops in the resistor leads to allow this when populating the boards. One caveat - the centre frequency wanders a little as the bandwidth is changed from wide to narrow (e.g. a few Hz at +15dB at 1k), so there may be a better way to implement variable bandwidth on this circuit, but after using it extensively on a variety of material it doesn't feel like a major issue. And ideally I would have preferred rotary switches to pots for stereo matching, but front panel space didn't permit that so I matched the pots instead as best I could.

Gain switches
5k resistor ladder with 5k resistor from CCW switch position. The +/- (boost/cut) switch swaps the CW and CCW poles going back to the EQ board, so the CCW position on the switch is always flat. Note that the shelves popped loudly at first when switching between boost and cut - putting a 1M resistor from each end of the switch to the wiper reduced this to manageable levels.

Gain range
For low & high shelves:
R8, R45 = 620R = +/-18dB gain range
R8, R45 = 1k2 = +/-8.5dB

For the low and high mid bands:
R15, R30 = 220R= +/-15dB
R15, R30 = 4k7 = +/-7.5dB

Frequency range
High shelf x3: swap C16 & C17 28.8nF for 10nF (lovely air boost)
Low mid x0.5: use empty C8 and C10 pads to switch in an extra 100nF at each point

Here's some photos from the build:

Testing bandwidth mod (lots of wires coming off the PCB!):
http://ringlefinch.co.uk/wp-content/uploads/2018/11/4-W492-bw-testing.jpg

Populating the internal control panel:
http://ringlefinch.co.uk/wp-content/uploads/2018/11/5-W492-control-panel.jpg

Control panel done:
http://ringlefinch.co.uk/wp-content/uploads/2018/11/6-W492-control-panel.jpg

Assembling and wiring it all up:
http://ringlefinch.co.uk/wp-content/uploads/2018/11/7-W492-populating.jpg

And all racked up!
http://ringlefinch.co.uk/wp-content/uploads/2018/11/8-W492-racked.jpg

I'm really pleased with how my W492 has turned out - it's been a major labour of love, and a great exercise in learning about state variable filters (and pushing the bounds of established designs...). Hope some of this helps someone further down the road!

Andy

This is the most impressive DIY project I've ever seen.  Thank you very much for showing it. Wonderful...

Where have you bought these rotary switches?

 

[TwentyTrees]

This is the most impressive DIY project I've ever seen.  Thank you very much for showing it. Wonderful...

Where have you bought these rotary switches?

Thanks Pollo! Appreciate the kind words.  ;D I got the frequency switches off eBay, pre-assembled with SMD resistors - seller called “tortoise6688”. I’ve got a few off them in the past with good results - the switches don’t have the nicest feel in the world, but they weren’t expensive (<£20 each I think, assembled) and have been very well matched. Their stock is very variable though, so I don’t know if they still offer them.

 

[mikeyB]

been reading on this for a while. Wondering the possibility of adding in bell/shelf switch on the LF and HF bands.

Is this something that can be done easily? How would this be implemented? Just experimenting with the circuit at the moment,
May even consider re-doing the PCB if I like it enough with all mods tried!

 

[mikeyB]

Slightly off topic but relevant!
I've not had time to study the schem yet, but is the topology akin to the calrec eq? If so, then that too must be DC coupled and conductive plastic pots are also needed for that eq as well.
It just so happens that the project plumped for the omegs from the start so it was not an apparent issue.
Could someone confirm if this is the case as i would like to build the calrecs at some point as well!! It would be good for others to know if this is the case and be noted in the Calrec thread.
Gyraf(Jakob-sorry!), maybe you could chime in if youre around?

Thanks in advance!!

 

[winner]

I want to know, I want to use multiple resistors to select the fixed frequency band, how to calculate the value of P4 because I want to change the value on the panel; including the values ​​of C8, C10, how to calculate, is there a formula, thank you

 

[TwentyTrees]

Hi Winner,

It’s definitely possible to calculate the filter frequencies - just google “state variable filter calculators” and go from there.  If you have specific frequencies in mind for your switch then you may also find it worth testing with a pot to find the right values before you make up the frequency switch (assuming you’re looking at a rotary switch or similar). You can dial the pot into the frequency points you’re looking for and measure the resistance at each point, then work up your own resistor ladder. It’s a super-flexible circuit!

I want to know, I want to use multiple resistors to select the fixed frequency band, how to calculate the value of P4 because I want to change the value on the panel; including the values ​​of C8, C10, how to calculate, is there a formula, thank you

 

[winner]

Hi Winner,

It’s definitely possible to calculate the filter frequencies - just google “state variable filter calculators” and go from there.  If you have specific frequencies in mind for your switch then you may also find it worth testing with a pot to find the right values before you make up the frequency switch (assuming you’re looking at a rotary switch or similar). You can dial the pot into the frequency points you’re looking for and measure the resistance at each point, then work up your own resistor ladder. It’s a super-flexible circuit!

Thank you for your patience. I understand the use of adjustable resistors to determine the resistance value, but I don't quite understand the capacitance value in schematic 2/3. Forgive me for my ignorance. I am a newbie. The description says the

presence/absence filter +/- 15dB, 0,063-1 kHz
- presence/absence of filter +/- 15dB, 1-16kHz,

why in schematic 2, C7 + C8 and C9 + C10 determine the filtering range 0.063-1KHZ, in schematic 3, C12 + C13 and C14 + C15 Decide on the filtering range 1K-16KHZ,

I saw that schematic 2/3 is identical except for the four capacitor values. I want to know their way, how to determine this value, I only know the RC filter, but the corresponding, thank you for your reply

 

[TwentyTrees]

... Forgive me for my ignorance. I am a newbie...

There’s nothing to forgive, we all start at the beginning! 

The W492 uses a pretty straightforward state variable filter topology. The choice of capacitor value at, eg, C7+8 and C9+10 effectively sets the frequency range that the 10k pot can deliver, and it’s an inverse linear relationship - frequency goes down as capacitance goes up. As you can see from the schematic, the capacitance in the High Mid section is about 16 times lower than the Low Mid section (100nF vs 6.2nF), and the frequency range is about 16 times higher.

With that in mind, if you’re looking for a different overall range for a given band, it should be pretty straightforward to just extrapolate. So for example, if you wanted a range of 125Hz to 2kHz (double the Low Mid values), you should be able to get there by using half the capacitance at C7+8 and C9+10 - so 50nF (which you could get quite neatly using 47nF at C7 and C9, and 3.3nF at C8 and C10). I used exactly this principle in my build when I put an additional 100nF switchable at C8 and C10, to give the option of halving the frequency range in the Low Mids (from 60-1000 to 30-500). I did the same with the High Shelf too, adding a switch to reduce the capacitance from around 28nF to 10nF to triple the frequency range (lovely air boosts!).

Hope that helps. If I’ve misunderstood your question and you’re actually looking to understand the theory of how it all works, I’d point you to this really useful overview of the SVF topology (it gets heavy on the math, but that’s not essential to understanding the high level picture):
https://www.electronics-tutorials.ws/filter/state-variable-filter.html

Andy

 

[winner]

There’s nothing to forgive, we all start at the beginning! 

The W492 uses a pretty straightforward state variable filter topology. The choice of capacitor value at, eg, C7+8 and C9+10 effectively sets the frequency range that the 10k pot can deliver, and it’s an inverse linear relationship - frequency goes down as capacitance goes up. As you can see from the schematic, the capacitance in the High Mid section is about 16 times lower than the Low Mid section (100nF vs 6.2nF), and the frequency range is about 16 times higher.

With that in mind, if you’re looking for a different overall range for a given band, it should be pretty straightforward to just extrapolate. So for example, if you wanted a range of 125Hz to 2kHz (double the Low Mid values), you should be able to get there by using half the capacitance at C7+8 and C9+10 - so 50nF (which you could get quite neatly using 47nF at C7 and C9, and 3.3nF at C8 and C10). I used exactly this principle in my build when I put an additional 100nF switchable at C8 and C10, to give the option of halving the frequency range in the Low Mids (from 60-1000 to 30-500). I did the same with the High Shelf too, adding a switch to reduce the capacitance from around 28nF to 10nF to triple the frequency range (lovely air boosts!).

Hope that helps. If I’ve misunderstood your question and you’re actually looking to understand the theory of how it all works, I’d point you to this really useful overview of the SVF topology (it gets heavy on the math, but that’s not essential to understanding the high level picture):
https://www.electronics-tutorials.ws/filter/state-variable-filter.html

Andy

Thank you very much for your reply, I fully understand. If I want to add another band, I just need to add (copy) a schematic 2 or 3, change its capacitance value to the frequency I need, and say that they are completely connected in series. Is that right? thank you