Filterbox 500

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

jensenmann

Well-known member
Joined
Oct 10, 2004
Messages
2,850
Location
Karlsruhe, Germany
Thanks to christmas, lockdown, weather and my wife´s mood I had time to realize a project which will hopefully end as a 500 series PCB available to the group. I´ve breadboarded the circuits roughly and everything seems to work as it should. Schematics are attached. It´s basically a Tilt EQ with HP/LP filters, polarity reverse and an output circuit switchable between electronically balanced and a choice of two transformers. Here´s some thoughts why it is what it is:

First of all I haven´t seen such a thingy before and I don´t believe that anybody really needs it. That´s a good reason for me to make it and hopefully no manufacturer will clone it and make money out of my work :)

The input circuit is pretty much standard, filtering HF garbage out before entering the circuit (R1,R3,C2 and R2, R4, C1).
The second stage is reversing polarity, which can be bypassed if wanted. This can create some funny results if used in parallel to a given signal. The input impedance of the following stage can be really low hence a chip is needed which is able to drive 600 Ohm easily.

Next is a Tilt EQ, right out of Douglas Selfs book. I was not too thrilled from my breadboard version and thought it needs some tweaking to fit my taste. That´s why I´ve added optional trimmers (P2-P5). So everybody can modify the endstop resistance for both LF boost/cut and HF boost/cut to own taste. I´ll add space for trimmers and resistors on the PCB. Since the circuit inverts phase another phase reversal stage is necessary.
Next is a high- and lowpass filter. It´s our Harrison Ford Filter project with some modifications to the filter frequencies which I posted in the according project thread.
The output stage relies on a DRV134 output driver. I´ve had not so good results with direct coupling these. That´s why I´v added coupling caps. It´s output signal is sent either directly out of the box or into a switch to determine which of the two transformers is used. Both trannies indeed sound different and are the result of listening tests with several other contenders. The good thing is that they are on the cheap side. I wanted the transformer´s artefacts to be obviously audible. So they will both bring their own character to the party. The Edcor PC600/600 has a steel core while the OEP is 50/50 Ni/steel. That gives them different distortion characteristics, mostly depending on the level sent through them. Unfortunately I have no clue if a Zobel network is needed, so I´ll add space on the PCB for that purpose. Any values will have to be determined. Same applies to a load resistor.
To supply the LEDs with constant current I´m using a LM317 in current limiting mode. It´s 1,25V reference voltage across R30 will force roughly 10mA through the LEDs, no matter how many are on or off. The LED supply voltage is decoupled from audio rails by R28/Cpsu.

Please feel free to add comments, hints, suggestions and let me know if there are errors.
 

Attachments

  • Filterbox 500 V1.JPG
    Filterbox 500 V1.JPG
    298.4 KB · Views: 223
The output switching seems a little strange.  Without transformers +Out is AC coupled while -Out is not.

Also what transformers were you planning on using?
 
john12ax7 said:
The output switching seems a little strange.  Without transformers +Out is AC coupled while -Out is not.

Also what transformers were you planning on using?

Good catch, changed...
Transformers are Edcor PC600/600 and OEP A262A2E.
 

Attachments

  • Filterbox 500 V1.1.JPG
    Filterbox 500 V1.1.JPG
    298.1 KB · Views: 150
One thing I've wondered is when switching transformers like this does there need to to be a load on the primary and secondary? A discharge path for any currents.
 
I don´t think that there will occur problems at such a low power. Anyway, R26/27 are load resistors for the trannies, they should be good enough to prevent any inductive peak while switching. Their values need to be determined together with the Zobel network. I´ll do that when the prototype PCB is here.
 
I modified the schematic for a commonly available bicolor LED to toggle between the two transformer options.
 

Attachments

  • Filterbox 500 V1.2.JPG
    Filterbox 500 V1.2.JPG
    298.5 KB · Views: 63
Here´s a picture of the layout. Green is the bottom layer, blue is the top layer. For ease of reading the bottom layer is on top, but view is from top.
 

Attachments

  • LayoutV1.2 bot.JPG
    LayoutV1.2 bot.JPG
    2 MB · Views: 66
Same layout but now the top layer is on top.

I´d be really happy if someone could take the time and crosscheck everything.

LEDs are not on the PCB, they will be wired to the solderlugs on top of the switches.
 

Attachments

  • LayoutV1.2 top.JPG
    LayoutV1.2 top.JPG
    589.1 KB · Views: 58
Some theory about the tilt circuit, thanks to Wayne Kirkwood of ka-electronics.com:
https://www.waynekirkwood.com/images/pdf/Implement_An_Audio_Frequency_Tilt_Equalizer_Balena_EDN_February_2_2012.pdf

Attached is a suggestion for values giving +/-3db @100Hz/10kHz by Wayne.
 

Attachments

  • Ambler_Tilt_EQ_BW.jpg
    Ambler_Tilt_EQ_BW.jpg
    36.1 KB · Views: 76
Finally some news. The prototype PCBs arrived today and the first one is stuffed. Upto now only a few minor tweaks will be necessary to the final version. Some drillings need to be enlarged, the silkscreen can be improved in some spots and two switches need a hair more distance. But I haven´t tested anything yet. Too tired, I need to sleep now. Over the weekend I should have plenty time to test everything.
 

Attachments

  • Filterbox 500 Proto V1.jpg
    Filterbox 500 Proto V1.jpg
    951.7 KB · Views: 113
I took me exactly 40 years to start making my own PCB layouts. Everything I did before was perfboard because breadboarding a circuit is faster than doing a PCB layout, at least for the stuff I was doing. For a single PCB it just made no sense to me to have a PCB fabbed.
 
A few observations from my fist test:

The Tilt circuit works as it should. I started with a trimmers-only version to find out how tweakable the circuit is. Unfortunately it didn´t respond very well. Tweaking only one trimmer really bends the transfer function in a way that it doesn´t sound linear any more with the pot in the middle position.

The amount of boost and cut available with the original values is too much to my ears. The get that lower P2 and P3 need to have larger values than in my V3 schematic suggested. Wayne Kirkwood´s schematic uses 24k in these positions. So if one would want to tweak in that spot it´d be better to use 50k trimmers. In a rackmount device with more real estate on the frontpanel than in a 500 module it might be a cool feature to replace P2 + P3 with a dual pot to be able to adjust the max. boost cut as a feature.

After fiddling with the trimmers I removed them and installed the resistor values suggested by Wayne. It´s +/-3dB "only" but it sounds good to me and still is more boost/cut than I´d want to use. I´m fine with that at the moment. For a more detailed opinion I have to use it in a real world application for a while.

Both phase reversal and output sections are working properly. Switching between the output option really changes the character of the unit. It´s doing all I wanted it to do.

Unfortunately the HPF/LPF section has a bug. While the lowpass works fine, the highpass is motorboating in the CCW endposition (lowest cutoff frequency). Additionally the circuit has a 6dB boost when engaged which is wrong anyway. I´ll have to look into this further.
 
Next update:

The above mentioned problems are solved. I had to cut a trace and add a bridge to P6b because both ends of the pot were connected the wrong direction. No more motorboating and the HPF works as expected. To get rid of the 6dB boost in the Harrison filter circuit I had to replace R10 with a bridge to lower gain in this particular spot by 6dB. C9 is unnecessary now.

Overall there still was 6dB more gain than in hard bypass. I fixed that by changing R7 + R8 to 4k99. That´ll need to change C3+C4 to 68pF. I´m going to update the schematic the next few days as time permits. (Edit: done, see attachment) Note to self: check the gain structure next time earlier...

A quick and dirty distortion measurement with my NTI measurement-gameboys gives 0,012% distortion @ +6dBu and 1kHz, all circuits + transformer in.
@+6dBu the Edcor has slightly worse distortion than OEP, probably because of the steel core. This changes with frequencies below 60Hz where the low inductance of the OEP seems to dominate distortion.
Increasing the level to +8dBu changes distortion completely. Below 100Hz the OEP has increasingly worse figures. It hits 1% at 31Hz while the Edcor shows 0,16%.
 

Attachments

  • Filterbox 500 V4.GIF
    Filterbox 500 V4.GIF
    63.2 KB · Views: 111

Latest posts

Back
Top