Type 32C Filters

[Bobby Baird]

I have been in the process of designing some 32C cut filters for some time. My motivation was out of the need to have some quality analog cut filters for tracking or live mixing. Other available circuits having hard to source parts or limited functionality. I have two versions. A variable card that will fit 8 channels into single 1ru case featuring true bypass, LPF bypass, LED visual feedback, THAT 1206 input, THAT 1646 Output, and a separate 3-rail power supply. Then I was thinking I needed fixed cut filters for stereo channels or for repeatable cuts, so I loaded up LTspice and found the resistances needed for the desired cuts. This unit is in 500 series format with grayhill switches.

 

[Ricardus]

I'd love to build a couple of those.

 

[Bobby Baird]

Boards should arrive today.

 

[Bobby Baird]

Time to build and test.

 

[Bobby Baird]

Parts arrived. I'll see if I can get both versions assembled this weekend.

 

[Bobby Baird]

500 Series unit works like a charm. I'll have to run sweeps with REW but just running pink noise all the cuts look to be exact!

 

[Bobby Baird]

I try and solder up the card version & PSU tomorrow.

 

[gyraf]

You lost me: What is a "32C" filter?

 

[JMan]

Harrison 32C high pass and low pass filters

 

[Bobby Baird]

Didn't have time to wire it up tonight. Have to wait until tomorrow.

 

[Bobby Baird]

I got the card working although, I had the pot pinout in the schematic and pcb backwards for the reverse log 50k pot. I just used some botch wires to correct that. Being that this card version is in the prototype phase it was not expected to be perfect first attempt. All the other features work as expected. With the .1uf and 50k combination I believe the sweep is way to wide for a tracking filter. I want to buy some 10k rev log pots and scale the caps to .68uF to get a 20hz to 500Hz HPF and 20kHz to 900Hz. The circuit is quite customizable by just altering 2 caps and pots while retaining the 1.2db bump.

 

[Bobby Baird]

I made sweeps of my 500 series unit at -18dbfs. -3db cutoff point is at -21dbfs for each position.

 

[thomasdf]

I am not sure I understand what the two versions are ? The 500 format, ok, but the other one ? Cheaper alternative without the rotary switches and pots instead ?
Beautiful work ! I have always been intrigued by these filters and their "musical" reputation. Yet I see on the graph they only boost about a dB and a half before cutting frequency in the low end. Can it still be heard ? Is there a way to make them resonate a bit more ?

Cheers !

 

[Bobby Baird]

I am not sure I understand what the two versions are ? The 500 format, ok, but the other one ? Cheaper alternative without the rotary switches and pots instead ?
Beautiful work ! I have always been intrigued by these filters and their "musical" reputation. Yet I see on the graph they only boost about a dB and a half before cutting frequency in the low end. Can it still be heard ? Is there a way to make them resonate a bit more ?

Cheers !

Yes, the variable version is cheaper and can fit 8 into a 1ru rack unit with onboard power supply. The bump is there to create a small presence boost near the cutoff slope. Take your favorite digital eq and A/B between a butterworth, and a resonate cut of a 1.25dB. With the butterworth, there is a sense of loss in level. The resonate cut filter seems to me, to be more seamless as far as perceived level. Yes, the resonance can be changed by altering the values of C312 & C313 in the HPF to values that aren't similar. Example C312 .1uf & C313 .22uf. Alternatively, you could adjust one gang of the pot with series resistance to change the resonance with a switch. This could be acheived with a resonant pot like the Little Labs VOG. Same with the low pass adjust C315 & C316 to be different. Its just a sallen-key based filter. Analog devices and Texas Instrument have simulators on there websites you could play with, or I can let you experiment with my LTspice circuit. Just download LTspice and install. Unzip this file into My Documents/LTspiceXVII on windows. Open spice, open the .asc file. Hit run. Left click the plot window and hit spacebar it will show you the response.

 

[Bobby Baird]

I didn't like the Rev. Log taper response. I bought some linear pots to experiment with this weekend. I want to get this project wrapped up. The 500 unit is tested and is fully functional. I made sure to balance relays and LED's between the rails to pull current from both rails as opposed to positive to ground, I see in other designs. This is the wrong approach in a 500 series unit because you want to keep the PSU balanced as to not make one rail work to hard. This could become a problem in say a 10 space rack with all units pulling from the positive rail. IMHO it should look something like this. An added bonus it keeps the noise out of the ground plane. My unit features seprate ground planes for GNDPWR & GNDA. Chassis GND connected to Chassis mounting hole or XLR connector.

 

[thomasdf]

Thanks for your detailled answer !

 

[MidnightArrakis]

Unsolicited comments:

1) Unless you are running "Controlled-Impedance" routing, if you would separate some of your tracking here and there on your PCB, doing so would then allow your "Copper Pour" to fill-in between those tracks and possibly improve your crosstalk readings.

2) It is typically recommended to have your "Copper Pour" at least 50-mils away from your board-edge. This prevents accidentally creating any unintended "Ground Loops" with closely-spaced adjacent boards or to a front-panel.

3) With that amount of large "Copper Pour" on this PCB, it would prove to be beneficial to have a bunch of "stitching-vias" placed around the layout to not only connect the "Copper Pours" of both layers together, but also to significantly lower the capacitance, inductance and resistance between the two "Copper Pour" layers.

4) Do you have a complete and finalized schematic of each of your circuits available? Just wondering.....

5) Did you do your PCB-layout using KiCAD?

/

 

[richiyobs]

I made sure to balance relays and LED's between the rails to pull current from both rails as opposed to positive to ground, I see in other designs. This is the wrong approach in a 500 series unit because you want to keep the PSU balanced as to not make one rail work to hard. This could become a problem in say a 10 space rack with all units pulling from the positive rail. IMHO it should look something like this. An added bonus it keeps the noise out of the ground plane. My unit features seprate ground planes for GNDPWR & GNDA. Chassis GND connected to Chassis mounting hole or XLR connector.

attached a relay bypass i used in my diodeBridge comp 500x proto....i also avoided GND to power relay coil.... at beginning i had clicks due to dc offset.... luckily i could fix it

can you please tell me where to find grayhill footprint? did you draw it whit kicad footprint editor?

at moment i'm redrawing harrison eq but omitted HP and LP section, to safe space in a 500 board... so your project could be cool to expand what i'm going to build

best

 

[Bobby Baird]

Here's what I have for KiCad. Did you see this https://groupdiy.com/threads/500-series-yannlu-audio-eq-500-harrison-mr-replica.67383/

 

[richiyobs]

Here's what I have for KiCad. Did you see this https://groupdiy.com/threads/500-series-yannlu-audio-eq-500-harrison-mr-replica.67383/

Great! thanks for sharing Grayhill footprint....
yes i saw Yannlu eq, it's based on MR series.... with variable Q.....
i preferred redrawing 3232, first for learning purpose and then for fun... for me the occasion to explor Wien Bridge configuration