Basically I'm wondering if I can change the voltage reference for the LM3916 to have the uppermost LED (which lights at "+3 db"..I assume dbu? they say it's = 1.2V) read more around +16dbu/5v. I think the datasheet tells me this but I've given it a couple reads and I'm having a hard time figuring this out. It looks like the 1.2v is generated internally and then run through a voltage divider with the ground for calibration, but I doubt this is correct as then you could only adjust the calibration to make it more sensitive right? Any help with this question is appreciated, I think it's a simple answer, I just can't quite figure out the diagram. Thanks!
Calibrating LM3916
- Thread starter deep_bias
- Start date
Help Support GroupDIY Audio Forum:
> I think the datasheet tells me
http://www.national.com/ds/LM/LM3916.pdf -- don't make us look it up, my school's DNS is grumpy today.
> uppermost LED ... 1.2V.. read more around +16dbu/5v
You stick a hot mike on loud drums, get 2V signal. Your mike input overloads at 0.5V. What do you do? You stick some resistors between mike and preamp, a pad.
You tune a radio to a strong local station. The detected signal is 10V. You want to play soft, like 1V. What do you do? Turn down the Volume knob!
So stick a knob or pad between your 5V source and the signal input (pin 5) of the chip.
This input is very high Z. I would try a 100K pot or 100K+33K fixed divider.
If it follows a peak-catch cap, you may want to go 1Meg+330K, so the cap don't go down too fast.
Yes, they DO show a plan to scale the ref to 10V. First diagram. 1.2V is forced on R1 1.2K. Same current flows in R2 7.5K. R2 is 6.25 times bigger, might drop 1.2V*6.25= 7.5V. Actually R2 drops a bit more, there is another current leak. The math is right there. I'm sure it can be hacked for 5V.
But what can go wrong? This ref-scale is -also- LED current set. I hate cheap-tricks which do two jobs at once. And certainly it can't be scaled to infinity; it can't go over 35V, and I got sources bigger than that. And math is hard. Simultaneous equations is harder.
Nail it for 1.2V. If source is bigger, pot/pad it down. If source is less, amplify it up. This is just basic audio level management. Same as you do for +4dBu inputs eating -50dBu to +27dBu sources.
The drawback of this plan is that you need one extra resistor compared to a hard-math solution. That's 12 cents wasted. But already you've wasted 13 cents of brain-pain looking at their clever design. Their minimum-parts cleverness makes sense if you are making a million boxes. If you are making one, or eight, spend the extra 12 cents and be done with it.
http://www.national.com/ds/LM/LM3916.pdf -- don't make us look it up, my school's DNS is grumpy today.
> uppermost LED ... 1.2V.. read more around +16dbu/5v
You stick a hot mike on loud drums, get 2V signal. Your mike input overloads at 0.5V. What do you do? You stick some resistors between mike and preamp, a pad.
You tune a radio to a strong local station. The detected signal is 10V. You want to play soft, like 1V. What do you do? Turn down the Volume knob!
So stick a knob or pad between your 5V source and the signal input (pin 5) of the chip.
This input is very high Z. I would try a 100K pot or 100K+33K fixed divider.
If it follows a peak-catch cap, you may want to go 1Meg+330K, so the cap don't go down too fast.
Yes, they DO show a plan to scale the ref to 10V. First diagram. 1.2V is forced on R1 1.2K. Same current flows in R2 7.5K. R2 is 6.25 times bigger, might drop 1.2V*6.25= 7.5V. Actually R2 drops a bit more, there is another current leak. The math is right there. I'm sure it can be hacked for 5V.
But what can go wrong? This ref-scale is -also- LED current set. I hate cheap-tricks which do two jobs at once. And certainly it can't be scaled to infinity; it can't go over 35V, and I got sources bigger than that. And math is hard. Simultaneous equations is harder.
Nail it for 1.2V. If source is bigger, pot/pad it down. If source is less, amplify it up. This is just basic audio level management. Same as you do for +4dBu inputs eating -50dBu to +27dBu sources.
The drawback of this plan is that you need one extra resistor compared to a hard-math solution. That's 12 cents wasted. But already you've wasted 13 cents of brain-pain looking at their clever design. Their minimum-parts cleverness makes sense if you are making a million boxes. If you are making one, or eight, spend the extra 12 cents and be done with it.
