LM1496 Ring Modulator (from Roland 100m)

[frequencycentral]

This is my redraw of the LM1496 based ring modulator found in the Roland System 100m modular synthesiser. The original (and unreadable) schematic can be found here: http://fa.utfs.org/diy/roland100m/schematicsshow18.html  I am fortunate enough to also have a slightly better quality hard copy of the schematic, on which the component values and chip pinout are readable. I own two of these ring modulators, one original Roland, and one clone that I built myself. I have always found them excellent once well set up, proper adjustment of the Signal Balance and Mod Balance trimmers results in zero modulation bleed if a carrier is not present.



I've been thinking of building an aliaser/bit crusher/ring mod pedal for a while, but the issue with all the designs I'm aware of is bleed. So I thought I'd go back to my synth roots and do a pedal based on this RM, with the addition of a modulation oscillator and clean blend.

Now for the question. The above schematic was designed to run from a +/-15 volt power supply. So I have two options in order to 'pedal power' it, either a charge pump bipolar supply, or a vref. This is what I'd like some advice with please!

[Jarno]

I just came across these on eBay: LM2577, or you could use the MAX1044. If you're really keen on DIY, you could try the 555 chargepump design.

[panterafanatic]

what do the filled in black arrows represent? i haven't seen that before

you could use that layout for the charge pump you linked to me, could you adapt it to give out 15v rather than 17v?

edit: with the charge pump you made, could you switch out the diodes for something that'll lose a little more voltage?

[frequencycentral]

what do the filled in black arrows represent? i haven't seen that before

The upward pointing arrows are +15 volts, the downward pointing arrows are -15 volts.

I should have said that the concern I have with using a charge pump is the risk of hetrodyning between the charge pump's oscillator and the ring mod's modulation oscillator. What do you guys think? Would it just be simpler to use a vref?

[panterafanatic]

it might be easier to do Vref so you don't have to worry about it heterodyning. do you have an ida as to how much it would affect the modulation if you did charge pump? i haven't build a ring modulator yet.

[Cliff Schecht]

Rick, I can show you how to bias the MC1496 for single rail operation. I don't like how the datasheet does some stuff but you have to consider that everything they're doing is tailored for RF frequencies. I've got a 1496 based pedal design kicking around somewhere, perhaps I can dig that up.

I would avoid the charge pump (regulation not great) and maybe look into a cheapo bipolar supply. I find the bipolar supply is indespensible for my bench and usually have more than one on hand. For powering circuits, I don't use adjustable supplies unless I have to though, I prefer the whole quick connect power connectors (PAiA uses a Molex part). I've fried too many parts with flying power leads shorting where they aren't supposed to, so a simple connector system on a solid power supply is a blessing. Speaking of which, PAiA sells a fairly inexpensive regualted +/-15V supply .

[frequencycentral]

Rick, I can show you how to bias the MC1496 for single rail operation.

That would be cool Cliff. Wouldn't it just be as simple as connecting the -ve points to ground, and connecting the ground points to a vref?

[Cliff Schecht]

Something like that. You have to bias the bases of the Gilbert cell mixer and the current sinks as well, but that is shown in the datasheet like I've done it. I thought they did it differently in the datasheet but comparing theirs to mine shows that they are similar. Here's what I have in my notes:



I never got to test this but I think it'll work. Everything is biased for 1 mA (datasheets recommended value) and I tried to choose only common 5% values for everything. Take the output either differentially or single-ended. IMO single-ended is simpler and saves you some hassle when compared to a differential output. Roland's method above would also work well because you could directly couple the outputs to an op amp stage without having to deal with blocking caps and 1/2 VCC references (output is already biased to 1/2 VCC ). A couple of notes as well: The resistors after the input blocking caps should be more like 10k, or whatever is enough to cut the input signals down to about 50 mV; also the resistors coming off the null pots have their values switched, the bottom should be 470k while the top one is 100k.

[frequencycentral]

It will be pretty easy for me to remove the one I built into my modular synth and hook it up to my breadboard to test it with a single rail supply with a vref.

I'll use this vref circuit: http://s210.photobucket.com/albums/bb292/frequencycentral/16d2b230.jpg

I'll try an audio range variation on the oscillator in this schematic for modulation: http://s210.photobucket.com/albums/bb292/frequencycentral/DeliriumTremolo.jpg

I'll also try it with a Buff n Blend added.

[Cliff Schecht]

Just remember that pins 1, 4, 8 and 10 have to get a positive voltage (8 and 10 at a higher potential than 1 and 4). My bias values could use a little tweaking to get more headroom but the idea is still solid. However, if you already have the Roland style one built then I'd just modify that as well. The input buffer on the y input is probably unnecessary and from what I can tell, they are shoving a lot of signal into the y input with such little attenuation (1k series, 100k to ground - weak voltage divider). That's surely going to turn everything into a square wave which would limit the sound of the balanced modulator to fat sounding square waves. Not that I dislike square waves in any way, but sometimes I want ramp waves damnit!


Something else that I like to do is to bring out the null controls to the front panel, when designed right you can get some very cool waveshaping going on. When you starting blending perfect 5th's with octaves and "unbalance" the modulator the bleedthrough can really beef up the sound.

[ricothetroll]

Hi !

I bought some LM1496 a few days ago, in order to play around with them. I first tried the Roland circuit (thr original one with +/- 15V psu, but it was VERY noisy (mostly carrier bleedthrough, actually more bleeding than modulated signal), so I decided to try the one found in the datasheet. That one works well, I got rid of the carrier bleed but I still had a slight but annoying bleed, that looks like a full wave rectified version of the carrier. I then left the breadboard and did a pcb, following Paul Perry's advices : I isolated the carrier's path as much as I could (carrier path surrounded by ground path create a kind of shielding and prevent capacitive coupling, no audio path too close of it, star grounding, carrier coming from an external source - my MOTU souncard). And no improvement compared to the breadboard version!!!!   

Do you guys have a clue of where that rectified version comes from ? And what about getting rig of it ?

Best regards.

Eric

PS : while doing my tests I found out that one IC of the two ones I bought is a lot worse than the other for bleed suppression, I can't even get rid of the "original carrier" with it ! Maybe that's the one I tested Roland's circuit with...

[John Lyons]

Rick
Did you ever do anything more with this?

[frequencycentral]

No I didn't John. I love ring mods with synths, where you can do a lot with the modulator, but for guitar use I find it somewhat frustrating to have a fixed modulator. Sometimes I'm seduced by the idea of a ring mod pedal, but whenever I put a guitar through a ring mod in my synths it's just atonal noise to my ears!

[Cliff Schecht]

Something I always do with my ring mods is to run a Whammy pedal through them. If you get the wet and dry signals in the right inputs (don't remember which goes where), you get an extremely cool way to blend the signal with octaved versions of itself. On top of that, because you have the ability to shift pitch in real time, you can blend pretty much any interval that you can find on the Whammy. Just sweeping the Whammy at an octave down or up gives you some really interesting ring mod sounds that I haven't gotten anywhere else. My favorite sound is the cool Jamaican drum sound you get when the rocker is set just right. I used to have some soundclips up somewhere but god knows where those are now..