Transformerless output 312 LED VU help?

People seem to speak wonders of these chips.. Since they are cheap (compared to output trafos) I wondered if this would this work ok or not?  

I recently pulled a working balance board from my SSL9Ks and connected it to the output of the 312 (minus output trafo, of course) and all I got was a high-pitched squealing noise. I have the DRV134 in these boards (have some 1646 coming to test).  But is there a reason why this wouldn't work? I'm hesitant to test it again since I don't know what the noise was... My PSU is +-18V, so I don't think I blew the chip (I haven't yet been able to swap it back to the SSL to test again).

balancing board info for those unfamiliar with it:
http://www.beatbazar.com/guests/ssltech/ob-1/index.htm

Has anyone done this already? Good/bad results? CMRR on these is not as high as on an output trafo (not sure how important this is on the output).  I'm wondering if there is a reason to do this besides saving a bit of $ on output trafos?  If I later wanted to incorporate a VU, for example, would I need a buffer circuit or something to drive it or is there an easier way?

thanks in advance!

Thanks didn't find that one. But wow.. what a weird thread. Seems like Syn is talking to himself? Did all the posts in between get deleted or something?

Anyway, could my problem be DC offset? I'm using a 330uF bipolar, with a film and 47R resistor in parallel on the output.  I admit I never bothered to check DC going to the trafo.  Anyway, I will try a 10K resistor to ground after my output cap and see if that fixes things..

Any further tips welcome... thanks again!

Old thread, I know, but in the end, I got this little thing working OK. Don't know what the problem was initially, never figured it out, but next couple times I tried it, it worked and I've since etched a nice, compact board for it and it was in service at my friend's studio for a couple months (beat out a millenia on lead vocals on his last project! woo hoo!) Anyway, I want to put a 3-led VU on this.  I understand a comparator is the better way to go.. BUT, I'm unsure of how to set it up.

PLUS, the LM339 is a quad comparator, is it OK to ground the input of the last one and not use it (see schemo below)? Or do I leave it unconnected?

And can I take the signal directly after the output cap (before 1646) or will that mess with what's going to the 1646?

Below is a schemo I found online and modified for 3 LEDs... why would one choose to do an inverting comparator vs a non-inverting? And is the +12V going to the + inputs supposed to be clipping?

TIA!

I am not familiar with the LM339, but I built a signal/peak LED into my SSL9k using a soundskulptor circuit. (links to their schematics in the white market). The best place to tap the signal was indeed just before the balancing board, not sure if this also applies to a 312.

I wonder if it would be possible (and less complicated/part intensive) to use a LM391x 10-LED driver and just leave out the unwanted LED's?

OK, I thought I replied, but it disappeared? weird. Anyway, thanks for that, I did read the post about your SSL9K. A while back someone had PM'd me about the sound-skulptor peak LED, was it you? 
I had considered the LM3915 (I think that's the log one) but I was hoping it could be done more simply using the LM339. I'm also not sure what to do with the unused LED positions of the LM3915? Does some type of diode have to go there for the circuit to work? Haven't been able to find anything online about this specifically. Gonna try to get some samples from National, and can test when they come..

Here is an image of a typical setup of the LM3915, which usually does 10 LEDS. Does anyone know if either this will work with less LEDs? What would one do with the rest of the LED positions?  Or would the LM339 work with 3 instead of 4 LEDs? Sample schemo in my post above.

thanks again!

The 3915 will work with less LEDs but why spend the money...?

For only 3 or 4 LEDs the LM339 is ideal.  You can place the LEDs in series and save current for a bar display. You also have a lot more  control over LED thresholds with a custom voltage divider.

This is a good starter circuit to learn on. The LM339 is what is called an open collector output, so it looks like a transistor switch to ground when output is low, and a turned off transistor when output high. It is also a comparitor vs, an opamp so there are subtle differences wrt it's input characteristics.

JR

thanks john, that helps a lot.
Looking at it like a transistor, the positive is the collector, the inverted is the base and the output is the emitter? Depends if it's inverted or not I guess? Why would one use it one way or the other?

I just saw that peterc used one in the green pre, but with a TL072 before it. Is this some sort of buffer, and would that be necessary?

http://1176neve.tripod.com/id12.html

thanks!

No.. maybe better to think of it like an opamp input with a grounded emitter transistor for the output.. the output is that transistor collector that can either pull down to ground or not. It decides whether to pull down or release based on which input pin is higher than the other.

In the schematic you linked the tl072 front end is first a gain trim, followed by a half wave rectifier, and peak hold circuit. Without the hold circuit the meter would be very dim because audio peaks are so brief.

The LEDs in that circuit could be rearranged all in series to ground, then reverse  the 339 polarity so it is shunting the current around the LEDs to turn them off, and releasing, to let them turn on. All in series would use less current, and put less noise into the power supply.  A consideration in audio design.

JR

Thanks for all that. I'm going to read up and try and figure things out and post back... it's a bit over my head at the moment.  :

I council you to use a LM339 or a TL074 for simple schematic.
I'm designing a schematic with TL074 or two TL072.

I didn't think to use a TL074.. That would be cooler cuz I have a few of those to test with.. hmm.  Would one be enough for 3 LEDs?

OK, I'm sketching up a rough schemo of what I'm hoping will work and I'll post it soon. Looking at PEterC's schemo (link above), I don't think I'd need the gain trim...is that what the pot is supposed to do in the LM339 schemo I posted above?

I don't think I'll be able to do the LEDs in series as John suggests, even though it may be the better way. I'll have a relay switching control of the LEDs between this circuit and another, so I think they all have to have their own path to ground. I will have a separate rail/ground (not the audio rail) for the LEDs/339.

Can the peak hold be done as in PeterC's (found a similar circuit in a National Semi paper) with the unused (4th) comp of the LM339 (since I'm only driving 3 LEDs, it would be rather convenient)? 

I'm having trouble seeing how I set the reference voltage, and I guess experimentation is in order.

Will this work if I connect the LEDs the other way around: Anode to output of the LM339, cathode to ground? Instead of anode to +V, cathode to LM339 output? I mean, if the comp produces a positive voltage and the series R is high enough to limit current, this should work, right? If it can't happen that way it kind of complicates my evil plans but that's how I had imagined doing it, so thought I'd ask. I got a couple of 339s coming soon, and I'll play with them to see what works, but if anyone has experience, I'd appreciate the input. 

mitsos said:

I don't think I'll be able to do the LEDs in series as John suggests, even though it may be the better way. I'll have a relay switching control of the LEDs between this circuit and another, so I think they all have to have their own path to ground. I will have a separate rail/ground (not the audio rail) for the LEDs/339.

Click to expand...

can you be a bit more specific about it? It seems you're embarking in a complex thing. Do you want to use the same LED's for two things just for saving front-panel real-estate? There may well be a more global answer if you let us know the Grand Scheme

Can the peak hold be done as in PeterC's (found a similar circuit in a National Semi paper) with the unused (4th) comp of the LM339 (since I'm only driving 3 LEDs, it would be rather convenient)? 

Click to expand...

No: the output of a 339 is an closed-or-open contact, not a voltage.

Will this work if I connect the LEDs the other way around: Anode to output of the LM339, cathode to ground? Instead of anode to +V, cathode to LM339 output?

Click to expand...

No: the outputs are NPN transistors, emitters of which being connected to -V.

I see why it won't work with a comp. I confused the comp with something I was reading about opamps as comps. So a comp is just like a switch that when on goes to ground? It was actually an opamp that puts out a voltage somewhere between the voltage rails. So could I replace the LM339 with the TL074 like in PPAs post? Then I would have a voltage output right?

My evil plan is to use the LED for the VU and as switch indicators. Basically to save real estate.  So I thought I would put a relay controlling where the LED anode is connected and leave the cathode/series R going to ground. Well actually the relay would control where the +12V goes (either to the switches that light the LEDs or to this circuit which would be connected to the LED anodes as well. Maybe I'd need diodes between the opamp outputs and the LED anode, so that when the 12V is going directly to the LED it doesn't also go to the opamp outputs? 

I think I should connect V- to ground so that I'd get either a positive or 0 output, right?  I don't yet know if it should be inverting or not, but let's say if non-inverting, I would send the +12V through a series of resistors as voltage dividers going to the each inverting input and have the signal going to the + inputs in parallel, with the outputs to the LED anodes? Sound OK so far?

Is there a problem with taking the signal between the opamp (2520,melcor, 990 etc) output and the 1646 input?

Using relays is not very astute; you could do what you want using comparators in the level display. You would turn it off by disconnecting the -V supply of the comparators; the outputs would become high impedance and yiu could drive the led's from another set of comparators. Switching from one mode to the other would use one SPST switch to ground, toggling between one set of comparators or the other.
I'm not saying it is trivial, but relays expensive, bulky and limited MTBF

Sometimes I am quite less than astute. But you've probably figured that out by now...  :

So I think I'd have to have the LED cathodes going to both comparator outputs, and have +12 volts permanently on the anode of the LEDs?  Im starting to get a picture of this, but it's gonna take me a bit to get it restructured in my mind.

The switch indicators I am refering to are for the pad,phase,phantom switches (all dual throw, and I'm only dealing with the LED pole here). I think it might work like this(non-inverting?):  I could use a a V divider to bring the +12V down to 5V at the -input, and leave the +input connected to the NO of the switch, with 12V going to the pole, so when the switch is actuated, that would put +12V at the +input and the comp would give the LED a path to ground and thus light it?

The "VU" part is more of a mystery still, I'm leaning toward a TL074 so I can use the fourth opamp in a peakhold circuit.

OK, small update I guess. I got an LM339 and breadboarded a circuit similar (almost 100%) to PeterC's circuit, at his 1176neve site linked in my post above. The only differences are I did not include the buffer/gain trim opamp (even though I used a TL072 in the test, I only used half of it. Maybe I should include the other half, but it seems to "work" without it, except for... well, I can only get the first LED to light speaking semi-loudly into an SM57 with full gain, (which clips the crap out of the DAW).

I used a +15V regulated wallwart as supply, and tied the V- of both the LM339 and TL072 to ground. I also think I will use a 100K pot cuz the 10K doesn't give enough "range."

So I think that I need to rework the voltage divider(??), but the thing is, when I tested the LED circuit not connected to the preamp, I ran a 7.75V signal into it and trimmed using the pot until the Red LED lit (from Green Pre setup docs). Is that +20dB? So, I figured it was all set, but when I hooked it up to the output of the 1646, I could only get the first LED to light and only when I turned the 10K to max (or min, don't really know which it would be, but what I mean is that there were 0 Ohms between the input and wiper). That is the only time the first LED would light.

Any suggestions? Thanks!

OK, since I was using 15V, I reworked the V-divider to give the same voltage drops as Peter's circuit with 12 V, and so swapped the resistors for 15K, 12K,3K2 and 1K2. The last would be in place of the 1K and 470R in the schematic in my last post (which was actually one 1K5 on my breadboard). On paper, this gives me about the same voltage drops as Peter's original circuit, but in practice it gives me copletely unlit LEDs...  ???

I'll double check for loose wires, but if someone could verify my resistance values, Id appreciate it.  Again, I'm using a 15V PSU, but if/when I get the thing on a PCB, I will go to 12V. It's just what I have on hand right now.

Thanks!