Passive 6dB EQ

I mentioned in the Classic Solo thread that I had come across an old design for a simple 6dB EQ. The schematic (bottom left diagram of attached pic) shows just bass and treble controls. It is a simple technique that uses linear pots to obtain boost and cut from a single pot. The boost is 6dB shelving but an interesting aspect is that the cut is more like a HPF or LPF where there is no shelf and the cut continues forever. it is not too difficult to add a mid band with an LC circuit and the cut depth and breadth can be limited by a series resistor without adversely affecting the amount of boost that can be obtained. The design is easily scaled to work at different impedances and I see no reason in principle why a 600 ohm version that can be tacked into a line level signal should not be possible.

Thoughts?

Cheers

ian

I think with a switchable mid frequency from 400 to 16k like you have on your Helios this would be very useful as a shaping tool for tracking. Probably be perfect for things like drums, bass and guitar cab recording just to sweeten the sound, not to mention vocals. I would think a lorlin board like your pultec would be great.

> a simple 6dB EQ

I invented that.

A few years later.

I didn't use no coil though.

It is the 6dB-Lin hack of the 20dB-audio James control.

If you short the 1K the bass rolls-off forever instead of shelving.

PRR said:

> a simple 6dB EQ

I invented that.

A few years later.

I didn't use no coil though.

It is the 6dB-Lin hack of the 20dB-audio James control.

If you short the 1K the bass rolls-off forever instead of shelving.

Click to expand...

I always have trouble getting my head around the James topology. Never quite sure what value to use for the 50K between the controls in your example which seems to be necessary to reduce interaction between the two controls. If you short the two caps in the treble and replace with one up the slider and for the bass delete the two caps and replace with an inductor up the slider you have the beginnings of the design I posted but you still need to add the two 10Ks in the middle and you don't need the 50K any more because the L and C up the sliders now do that function. And I find it easy to see how to add further sections.

For my degree dissertation in 1973 I designed a FET compressor which used piecewise linear approximation in the side chain to set the compression law. My supervisor said it could probably be patented. Unfortunately I was an undergraduate apprentice with British Aerospace at the time and they 'owned' any IPR I created. It was not in their area of interest so they did not patent it. Ho hum.

Cheers

Ian

> Never quite sure what value to use for the 50K between the controls in your example which seems to be necessary to reduce interaction

Yes. Un-clear. It seems to be remarkably tolerant. No magic value. Fire-up Duncan TSC and see.

> could probably be patented. ...British Aerospace ... 'owned' any IPR I created. It was not in their area of interest so they did not patent it. Ho hum.

I've been in similar situation. Dad was also. Too late for us, but for any younger wage-slaves lurking:

Usually you go to a Company IP Boss and get an agreement that the future-tech Company is not interested in your buggy-whip invention, and ask them to release their rights to you. Of course that is a hassle, and a time-sink, and maybe stirs the Company into formally taking your invention (and sleeping on it). And as you know, audio is rarely a wealthy field, and most patents (or their wordings) can be worked-around if really worth the effort.

Of course you rarely work for a company with NO connection to your own interests. And interests can be broader than even the company realizes. CBS' comp-limiter patents originated as a tank intercom system, but were also the basis for the most-used US broadcast comp-limiters. Brit Aero wasn't just wings, they musta had intercoms and passenger entertainment. My employer was a university, so about everything was on the chalkboard, and they happened to have mass income from at least two patents in unexpected fields. So a patent release might be hard to get.

It occurred to me that the 6dB EQ works because linear pots have half their total resistance at the mid point. If we had a pot with a different proportion of its resistance at the mid point we could increase the maximum boost. Problem is that these pots do not match very well at all....  unless you make them switches. Once you use switched pots however, you can make their mid points track very closely. This would make a rather nice mastering EQ. The attached pdf shows in general terms how this might work using 12 positions switch pots. R40 and R41 set the base level attenuation and hence the maximum boost, so for example, if they were 40K and 10K respectively the boost could be up to 14dB.  The mid point of each switch would also need to be arranged so that the resistance from the wiper to the input equals R40 and from the wiper to 0V equalls R41 but this is easy with switch pots. I have shown variable capacitors and inductors in each of the three EQ sections just to indicate that these can also be variable via switches. There is also no reason why in principle you should not add another mid sections provided the two do not overlap.

The other good thing about this EQ is it is easy to scale it to operate an any impedance. So, for a 600 ohm system for example you would make the total resistance of each section and R40+R41 equal to 2400 ohms (or 3000 ohms it you have two mid sections). There is no reason you could not drive it from a balanced source and connect it to a 10K bridging input of a line input for the gain make up.

Cheers

ian

> It occurred to me that the 6dB EQ works because linear pots have half their total resistance at the mid point.

I would say: in Potentiometer mode they are half-voltage (-6dB) at half rotation.

Which is of course half resistance; but focuses on the potentiometric ratio. We hear intensity ratios, we do not hear resistance ratios.

> If we had a pot with a different proportion of its .... at the mid point we could increase the maximum boost.

The essence of the James is that at one time standard "Volume" pots were -20dB at half rotation. This was needed to get 30-35dB of useful adjustment on an AM radio, which (in the US) could pickup from 1 mile to 500 miles and be played at whisper or roar-- at least 35dB total gain-range required. (BBC-only radios might have a narrower input range, but pulling-in Paris or Rome was always possible with higher-gain radios.)

Hence the James is full of 10:1 ratios. (Some might better be 9:1, but nobody noticed.)

Tracking was not an issue until Stereo.

However with the death of AM radio, most "audio" pots today are 20% (-14dB) at center.

This and the stereo tracking issue makes the Bax a better plan today.

The "-6dB James-ish" is the same thought with 1:1 ratios, with some fiddles (or not) to get shelf (or slope) in extreme cut.

And constant reminder: James and Bax and nearly all other R-C EQs need a source impedance, usually low-ish, and often lower than you wish to jack-interface at.

I have done some more work on this to calculate resistor values for stepped pots. I used a brute force method for this by using a simulation and just trying different resistor values until I got the attenuation required. During this I noticed the not unexpected interaction between the mid and the other bands so I temporarily disable this for the purposes of this exercise. I have written this up in a short pdf document which can be found in the new 6dBEQ folder in the DIY section of my web site

http://www.customtubeconsoles.com/diy

Next I will put together a basic prototype to see how theory matches practice (thanks for the Lorlin switcj boards Bernd).

Cheers

Ian

I'm following this thread with big interest. I'm glad that I could help.
Bernd

Before I start building I need to decide how many mid bands to include in a mastering EQ. I am inclined to have only one because:

1. There is already interaction between the mid and the high and between the mid and the low. This would only become worse with two mid bands
2.  This EQ is similar in many ways to the REDD EQ which has a single mid band. I like the sound of the REDD EQ
3.  At the mastering stage I would expect it to be unnecessary to have to to use a lot of EQ

Cheers

Ian

If the curves are  broad, two or three mid bands with lots of interaction would be great for gentle sculpturing.
A bit like the Hendyamps Michelangelo eq.  It has only one low mid band, but two hi shelfs with differet frq.
I dont think there is any inductors in that design, so its proberly a bax/james design with a bit of tube saturation.
So it depends if you think of this design as a paintbox or a narrow frq. tool. I kind of like the simple broad idea with no frq selection . 

I would prefer a paintbox as well. ;D
Berns

OK, I will build it with two mid bands. The REDD EQ mid band spans from 2.7KHz to 10KHz which is about 4 octaves. So a lower band 4 octave wide would span 2.7KHz to about 500Hz. Does that sound about right?

Low shelving/cut  could be 20, 40,80, 150, 300 and 500. High shelving could be 8K, 10K, 12K, 14K 16K?

Thoughts?

Cheers

Ian

I think I would center the mid frqs a little further down. Mabye about 3k and 300. And the start point of the shelfs could be about the same or 2k/500....ca.

Joechris said:

I think I would center the mid frqs a little further down. Mabye about 3k and 300. And the start point of the shelfs could be about the same or 2k/500....ca.

Click to expand...

OK, so low mid from 300 to nearly 3K and high mid from 3K upwards.

It is always difficult to specify the frequency of shelving EQ. Do you specify the frequency at which is begins or the frequency at which is shelves?

Cheers

Ian

Well no and yes   The center of the low mids should be 300, so if its four ocktaves wide it should reach from 75 to 1k2 Hz.
The same with the high mids, center at 3k,  from 750  to 12k. The hi and low shelf should start a little higher and lower than the normal 1k point, at 500 for the low shelf, and 2k for the hi shelf.  This is just me thinking.. I hope somebody else chime in for some thougths about a broad paintbox eq.. I have trouble understanig what a 40 or 80 hz low shelf realy is.. Where does it start..?
How is it posible to restrict the curve to not touch anything above lets say 300hz?

Joechris said:

Well no and yes   The center of the low mids should be 300, so if its four octaves wide it should reach from 75 to 1k2 Hz.
The same with the high mids, center at 3k,  from 750  to 12k.

Click to expand...

OK, that is a good way of defining mid EQs and there is clearly come overlap between the two and also with the hi and lo shelving controls.

The hi and low shelf should start a little higher and lower than the normal 1k point, at 500 for the low shelf, and 2k for the hi shelf.  This is just me thinking.. I hope somebody else chime in for some thoughts about a broad paintbox eq.. I have trouble understanding what a 40 or 80 hz low shelf really is.. Where does it start..?

Click to expand...

The basic laws of physics mean a simple shelving RC  based EQ will have a slope of no more then 6dB per octave. Two time constants determine the start, end and frequency span of the sloping part. Typically a single control is used to set the boost/cut and this affects both time constants and  hence the start, end and span at that setting. So the answer to the question "where does a shelving EQ start" is that it depends on the setting of the control. I'll do a few simulations of the low shelving EQ to illustrate what happens and post them here.

How is it possible to restrict the curve to not touch anything above lets say 300hz?

Click to expand...

Shelving EQs have a relatively gentle change from flat response to boost/cut. One time constant determines the point at which 3dB of change has occurred.  The linked plot shows the 6dB EQ bass boost with a 10H inductor and 20K pot. I will do another for the 10dB version for comparison.

https://drive.google.com/file/d/0B_n67A1hN3qteDlEd3pNazBsdlk/view?usp=sharing

Cheers

Ian

Here is the 10dB boost version plot.

https://drive.google.com/file/d/0B_n67A1hN3qtd3RhYUZaN3RoSHc/view?usp=sharing

Edit: and the same EQ at the 5dB boost setting:

https://drive.google.com/file/d/0B_n67A1hN3qtZ2RYbWZRMkNiYWc/view?usp=sharing

Edit: and at 2dB boost:

https://drive.google.com/file/d/0B_n67A1hN3qtS0JwbUM0QmdRWHM/view?usp=sharing

Cheers

Ian

Ian, this looks exactly as the one I suggested back in your passive EQ topic. I used a log pot to experiment to get more than 6dB boost and that affected a few things. The shelving frequency shifted from boost to cut and the Q of the peak was much lower than the dips for instance. With stepper switches a lot could be done to correct all this, FWIW you could use a completely different filter for each setting (API550?)

To have a way to call the freq of the shelvings here we could use the frequency for which it's in the cut mode, -3dB for the HPF or LPF.  As you said mathematically speaking the points for where it starts and ends change with setting, as is hard to speak of -3dB point in a 2dB shelving setting. In the plots you show (speaking of the 3 for 10dB config) the knees are almost in the same spots all the time and the center freq for which the gain is the mid point is also (almost) constant, 200Hz in this case, a tiny bit higher for the 2dB.

I did build a prototype back then and made some plots, showing the interaction, difference in Q and in frequency. They should be in the older post somewhere...

JS

joaquins said:

Ian, this looks exactly as the one I suggested back in your passive EQ topic. I used a log pot to experiment to get more than 6dB boost and that affected a few things. The shelving frequency shifted from boost to cut and the Q of the peak was much lower than the dips for instance. With stepper switches a lot could be done to correct all this, FWIW you could use a completely different filter for each setting (API550?)

Click to expand...

These effects happen in a lot of EQs. How much they occur depends on the maximim boost/cut you want. With 6dB it does not happen at all. At 10dB there is a noticeable difference especially in the mid. With log lots, where the boost can be as much as 20dB, the effects will most likely be very exaggerated. The Helios for instance has much sharper mid cuts than the equivalent mid boost and its max boost is only 14dB. The only way to avoid that is to have separate switches and inductors for boost and cut so that the Q remains the same. This is what I did in the REDD EQ but it is much more complex and needs two pole switches which are expensive. A possible solution would be to have separate boost and cut controls but that doubles up the number of switches and inductors

To have a way to call the freq of the shelvings here we could use the frequency for which it's in the cut mode, -3dB for the HPF or LPF.  As you said mathematically speaking the points for where it starts and ends change with setting, as is hard to speak of -3dB point in a 2dB shelving setting. In the plots you show (speaking of the 3 for 10dB config) the knees are almost in the same spots all the time and the center freq for which the gain is the mid point is also (almost) constant, 200Hz in this case, a tiny bit higher for the 2dB.

I did build a prototype back then and made some plots, showing the interaction, difference in Q and in frequency. They should be in the older post somewhere...

JS

Click to expand...

I like the definition of shelving frequency being the mid point. Easy to measure and easy to understand. I will try and do some plots of the shelves in 2dB increments on one graph to see how well that works.

Cheers

Ian