MUFF WIGGLER Forum Index

 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler Blog & NewsMW News   Muff Wiggler StoreMW Store 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

Benjolin filter output weak -- any thoughts?
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author Benjolin filter output weak -- any thoughts?

Dr. Sketch-n-Etch

I'm building some Benjolins for some folks based on the PCB layout and schematics published on the electro-music.com forum by Rob Hordyk. I've built two units so far, and everything seems to work perfectly, except that the filter output is very weak -- only a few hundred mV peak-to-peak. I really have to crank up my audio mixer to hear it. I've also simulated the filter section in Multisim, and the simulation is telling me the same thing.

I'm wondering if anyone here has encountered similar problems? I know that several people on this list have built Benjolins. Any advice would be most appreciated.


constantin3000

I once had the same problem troubleshooting a benjolin, in the end it was just the probe of my oscilloscope attached to the output, which caused the immense voltage drop.
Do the direct outs /pwm mix from the oscillators have a high output level (some volts peak-to-peak)?


inlifeindeath

the pulse out of OSC1 is very strong, something like +/-5v, but yeah, i get a much lower signal from the filtered output, especially with different COF and Res settings.


Hanz

Ah, wanted to point you to E-M Hordijk sub forum, but saw you already posted there...


decaying.sine

Doesn't Rob usually have VCAs on a lot of his modules as additional features? Maybe the Benjolin DIY is missing that.


fluxmonkey

it's as designed... filter out is line level, others are 5v. i dug up rob's confirmation and posted it in reply to your electromusic.com post


mono-poly

Filter out is indeed line level.


Dr. Sketch-n-Etch

Thanks guys, and sorry for the annoying cross-posting. In fact, I only registered at electro-music.com just to ask this question. I'm building a couple of these things for sungja, and wanted to make sure I hadn't fucked them up somehow. As long as they are working as designed, then I'm happy. I didn't offer to redesign them after all!

The dual oscillator/rungler/PWM circuits are very cool and clever. It would be cool to take the PWM output to a mixer and just send that to an external filter (or to multiple filters) as I don't perceive the filter on the Benjolin to be anything special. Alternatively, one could use the two 2164 VCAs to build a more conventional SVF filter on the Benjolin board. In fact, I believe that with a few passive component changes one could do just that (or at least a 12dB LPF) and get the output level back up.


xi-bot

Quote:
as I don't perceive the filter on the Benjolin to be anything special


i really like the filter
and believe rob uses his harmonic distortion on them like in the filters of the blippoo box
its described in this leonardo article http://leonardo.info/isast/journal/toclmj19.html

i build two benjolin's and both sound different
probably because i used different capacitor types (the resistors are 1%)
i swapped the "original" ssm ic's with coolaudio clones
what also makes an difference
funny that i prefer the sound off the clones ;-)

casperelectronics made an audio input mod for the filter
http://casperelectronics.com/finished-pieces/benjolin/modular-benjolin  /


Dr. Sketch-n-Etch

The filters sound fine, but I wouldn't hesitate to replace them with something more conventional in order to get robust output levels in my own build.

I used 5% resistors in my builds, because there is no need for precision. Also, I used mylar caps on the filter and ceramic caps for decoupling. I may replace the mylars with metallized polypropylenes.


rosch

actually the output should be like in any other gear too, i didn't have the impression that it's super low level by design.
but i had to troubleshoot each board i built (except for the workshop one) before i got them to work. they now sound very close to the original.
my problems were all little solder bridges, bad joints however. i vivisected them with the iron while i had them running.


numan7

@Dr. Etch-n-Sketch hmmm..... why not just pass the benjolin filter output through a preamplifer (modular, outboard or whatever) before the mixer, and turn up the gain until the signal peaks are around 5 volts?

cheers


Dr. Sketch-n-Etch

numan7 wrote:
@Dr. Etch-n-Sketch hmmm..... why not just pass the benjolin filter output through a preamplifer (modular, outboard or whatever) before the mixer, and turn up the gain until the signal peaks are around 5 volts?

Yeah, that's a good option: just put a gain cell at the end. I'm thinking of modularizing my board, so that's probably what I'll do. A little daughterboard with a TL072 and a few resistors will take care of it.

Incidentally, I don't know if there are any SDIYers reading this, but if you're thinking about adapting the Benjolin for modular use, just be careful not to emit the +/-9V power sources (78L09 and 79L09 -- these are NOT easy to find, but the 8 or 10V equivalents would also work fine). The reason is that the CMOS chip, CD4021 (or its equivalent, MC14021), really does not take too kindly to high supply voltages. The output currents increase from about 1.7 mA at 9V to almost 9 mA at 15V. I tested my boards with +/-15V while waiting for my 79L09s to come, and the 4021 got noticeably warm. Once I installed the 78L09 and 79L09, it did not.


CJ Miller

The SSM2164 circuits are very gain-sensitive, so it probably is easier to slap a preamp on the output rather than re-do the whole thing. Rob wrote that it took him a lot of time and effort to get the filter to work this way, and IMO he is much more clever than I am. I'll save figuring this out for designing my own SSM2164 projects.

Also the +/- 9V are because he likes to mix linear and CMOS. Most 4xxxB ics can be run from 18v. This way one can subject the logic chips to the full analog output without needing to worry about them accidentally getting voltages below ground.

The line-level output last night was convenient when I wanted to clip my headphones directly onto the header in my living room.


Dr. Sketch-n-Etch

CJ Miller wrote:
The SSM2164 circuits are very gain-sensitive, so it probably is easier to slap a preamp on the output rather than re-do the whole thing. Rob wrote that it took him a lot of time and effort to get the filter to work this way, and IMO he is much more clever than I am. I'll save figuring this out for designing my own SSM2164 projects.

The 2164 VCAs are being used as exponential converters in the tri-square oscillators, or as voltage-controlled resistors in the two-pole filter, so the sensitive gain issue is irrelevant. TL074 are the only gain cells in this circuit.

Quote:
Also the +/- 9V are because he likes to mix linear and CMOS. Most 4xxxB ics can be run from 18v. This way one can subject the logic chips to the full analog output without needing to worry about them accidentally getting voltages below ground.

Yes, this is a good idea. The only problem is that 79L09 are not something the electronics world really wants people to buy, evidently.

Quote:
The line-level output last night was convenient when I wanted to clip my headphones directly onto the header in my living room.

Just be careful not to nudge your alligator clip into an adjacent pin on that header, or you'll be in for a nasty, deafening surprise!


CJ Miller

Dr. Sketch-n-Etch wrote:
The 2164 VCAs are being used as exponential converters in the tri-square oscillators, or as voltage-controlled resistors in the two-pole filter, so the sensitive gain issue is irrelevant. TL074 are the only gain cells in this circuit.


It is true that they are not amplifying, but being VCAs they are operated through a bias setting, like an OTA. Hoshuyama and Dixon worked out that this reference voltage actually seems to set the temperature coeffiecient for the expo converters. IIRC about 0.78 volts dials it in to 3300ppm. And if for the filter its gain setting = transconductance it seems relevant to me.

Dr. Sketch-n-Etch wrote:
Yes, this is a good idea. The only problem is that 79L09 are not something the electronics world really wants people to buy, evidently.


This is probably why I scored a few hundred for $10 last month, nobody wanted them! OTOH 78L09 I have only a few of. The regular 1A versions are much easier to find. Fairchild still makes them, as I believe does On. The bigger versions should be fine if a Benjolin is in some kind of enclosure.

Dr. Sketch-n-Etch wrote:
Just be careful not to nudge your alligator clip into an adjacent pin on that header, or you'll be in for a nasty, deafening surprise!


Aye, that I would! Fortunately, my junk pile includes various header sockets and adaptors for industrial computers. I snipped off a two-pin jumper which was already wired to an in-line 1/8" jack. Took me a matter of seconds!


Dr. Sketch-n-Etch

CJ Miller wrote:
Dr. Sketch-n-Etch wrote:
The 2164 VCAs are being used as exponential converters in the tri-square oscillators, or as voltage-controlled resistors in the two-pole filter, so the sensitive gain issue is irrelevant. TL074 are the only gain cells in this circuit.


It is true that they are not amplifying, but being VCAs they are operated through a bias setting, like an OTA. Hoshuyama and Dixon worked out that this reference voltage actually seems to set the temperature coeffiecient for the expo converters. IIRC about 0.78 volts dials it in to 3300ppm. And if for the filter its gain setting = transconductance it seems relevant to me.

Well, first of all, I'm actually the Dixon to which you're referring! Secondly, just biasing a given 2164 at a certain VC voltage does nothing to cancel the temperature coefficient. In order for a 2164 VCA to auto-compensate its own TC in an exponential converter, the summed CV currents must first be processed through a separate 2164 VCA within the same package set to a certain gain setting prior to entering the CV summing inverter. For an expo converter, this gain setting is about 37%, which requires a VC voltage of about 275 mV. None of this is being done in the Benjolin circuit, where the 2164 TCs are all uncompensated.


CJ Miller

Dr. Sketch-n-Etch wrote:
Well, first of all, I'm actually the Dixon to which you're referring!


jawdrop Neat!

Dr. Sketch-n-Etch wrote:
Secondly, just biasing a given 2164 at a certain VC voltage does nothing to cancel the temperature coefficient. In order for a 2164 VCA to auto-compensate its own TC in an exponential converter, the summed CV currents must first be processed through a separate 2164 VCA within the same package set to a certain gain setting prior to entering the CV summing inverter. For an expo converter, this gain setting is about 37%, which requires a VC voltage of about 275 mV. None of this is being done in the Benjolin circuit, where the 2164 TCs are all uncompensated.


Thanks for setting me straight there, to what extent I understand what you are saying. So 275mV... 275 thousands should be .00275 Volts. Just off by a few zeros and a brain hihi

My point was only that there are 2164 gain settings of sorts. And I know that the Benjolin is not temperature compensated, it blows in the breeze!


Dr. Sketch-n-Etch

CJ Miller wrote:
Dr. Sketch-n-Etch wrote:
Well, first of all, I'm actually the Dixon to which you're referring!


jawdrop Neat!

I know; I can scarcely believe it myself sometimes. lol

CJ Miller wrote:
Thanks for setting me straight there, to what extent I understand what you are saying. So 275mV... 275 thousands should be .00275 Volts. Just off by a few zeros and a brain hihi

275 mV = 0.275 V

The original Intellijel Dixies (of my design, and actually named after me... see the connection? razz ) used this 2164 tempco scheme. However, it requires an additional trimpot and the tempco is not really that precise (but the VCO doesn't have to warm up -- that's a plus). Hence, in our last revision we did away with this tempco scheme and used a more conventional scheme with a tiny tempco resistor mounted underneath the 2164 chip. This freed up the tempco 2164 VCA, which we now use as the high-frequency corrective current source (in lieu of a dual transistor). The nice thing about this scheme is that now the high-frequency correction source is also temperature compensated (it wasn't before). Also, the new Dixies are easier to tune, and stay in tune better.

Ah, VCO design, so much fun! Dead Banana


Starkström

Dr. Sketch-n-Etch wrote:

I'm wondering if anyone here has encountered similar problems? I know that several people on this list have built Benjolins. Any advice would be most appreciated.


For what it is worth, I just got the Eurorack Benjolin by Epoche Modular, latest run (2015) and yes the filter outputs are (still) extremely low and always need extra amplification to mix with my other euroack modules. Interestingly the oscillator outputs of the Benjolin run on normal eurorack levels. I really wonder what triggered this design decision? hmmm.....

MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 1
Powered by phpBB © phpBB Group