Using 3 trimpots is very practical if you want flexibility. But I specifically chose the Bourns TC33X pot footprint because it very conveniently accommodates a 0805 resistor. You could even sandwich two of them. But the Vout one does have two 0805 footprints in parallel that are in series with the pot, so you could tweak Vout from resistors but it can become a pain if you have to do it for various Vouts.
This addon already has the CCS diode and its 2.2uF cap footprint. But don't think it's required (the 2.2uF cap). Also has the original protection resistor footprint, on the base of denoiser BJT, 10-47R. And I highly recommend installing it since this sets Vout of your circuit. You don't want that BJT to go bad.
CCS LED resistor has standard value, CCS current resistor doesn't have to be precise, standard values are fine.
Even the compensation pot could potentially be replaced with a standard value resistor. MPSAx6/BC3x7 on my discrete design needed some 4-5K to get within 50mV 35C-75C, so I'd just use a 4.7K.
BC8x7 from the LM317 design needed 13.38K so I'd just use a 13K resistor instead.
But tweaking compensation value does take a few tries. You can use LTSpice to estimate LED and CCS resistors based on common sense values for currents, maybe guesstimate the compensation resistor, then set Vout/bake/check how much you're out and what way, up or down. Each time you tweak the compensation pot Vout changes, that's why at least Vout pot is pretty useful.
Playing with the compensation pot you can get past the inflection point where the circuit starts exhibiting positive tempco, you need to back off a bit and you're fine. Normally with the values I used the circuit exhibits negative tempco and you trim with the pot to get towards that inflection point where you're pretty spot on across the operating temp span.
Also not sure how much compensation value varies. Maybe some random value like 6.8K would make most BJT/LED combos be somewhere 100-200mV out (up or down, at 12Vout) and might work fine for most applications, no need to tweak it precisely. Don't have enough experience with this so take everything with a grain of salt.
This addon already has the CCS diode and its 2.2uF cap footprint. But don't think it's required (the 2.2uF cap). Also has the original protection resistor footprint, on the base of denoiser BJT, 10-47R. And I highly recommend installing it since this sets Vout of your circuit. You don't want that BJT to go bad.
CCS LED resistor has standard value, CCS current resistor doesn't have to be precise, standard values are fine.
Even the compensation pot could potentially be replaced with a standard value resistor. MPSAx6/BC3x7 on my discrete design needed some 4-5K to get within 50mV 35C-75C, so I'd just use a 4.7K.
BC8x7 from the LM317 design needed 13.38K so I'd just use a 13K resistor instead.
But tweaking compensation value does take a few tries. You can use LTSpice to estimate LED and CCS resistors based on common sense values for currents, maybe guesstimate the compensation resistor, then set Vout/bake/check how much you're out and what way, up or down. Each time you tweak the compensation pot Vout changes, that's why at least Vout pot is pretty useful.
Playing with the compensation pot you can get past the inflection point where the circuit starts exhibiting positive tempco, you need to back off a bit and you're fine. Normally with the values I used the circuit exhibits negative tempco and you trim with the pot to get towards that inflection point where you're pretty spot on across the operating temp span.
Also not sure how much compensation value varies. Maybe some random value like 6.8K would make most BJT/LED combos be somewhere 100-200mV out (up or down, at 12Vout) and might work fine for most applications, no need to tweak it precisely. Don't have enough experience with this so take everything with a grain of salt.
Last edited:
Adjusting 3 trimpots in four Dnoisators I'd have to use on a preamp circuit certainly is far from practical. By any means.
I designed them so they're versatile for me, having three pots allows me to adjust them to any Vout.
If I were to sell them, which I might consider, I would probably adjust them myself.
If I were to sell them, which I might consider, I would probably adjust them myself.
Yes you were right, pretty bad when changing load.
But at least I can now use a red LED for the CCS in Denoisator. And maybe temperature compensate the denoiser but that shouldn't matter much.
Still, this version of the circuit worked fine for a long time, in my headphone amp.
This isn't new, in post #2,497 (more than two years ago) I showed how I installed it in my headphone amp using addon boards. I didn't use trimpots then, but it needs two of them, one to set Vout and one to set denoiser Vce. I just used resistors instead.
If you don't need precise Vout trimming you should be able to use resistors to set Vout, but denoiser Vce varies with BJTs so at least one pot for it should be used (R8 in previous post schematic). Collector of U4 needs to be similar to base of U2.
For no trimpots you should use Elvee's original Denoisator. The CCS versions need at last one pot if you don't want to experiment with resistor values.
I attached the CCS NoNoiser version but I used LT1117 regulator model, I recommend you get LM317-N spice model from TI.
If you don't need precise Vout trimming you should be able to use resistors to set Vout, but denoiser Vce varies with BJTs so at least one pot for it should be used (R8 in previous post schematic). Collector of U4 needs to be similar to base of U2.
For no trimpots you should use Elvee's original Denoisator. The CCS versions need at last one pot if you don't want to experiment with resistor values.
I attached the CCS NoNoiser version but I used LT1117 regulator model, I recommend you get LM317-N spice model from TI.
Using one multiturn pot to adjust the denoiser, and use the resistor that sets U4 and U2 voltages right, seems reasonable. The question is then to find that resistor. What is not acceptable is variations that might need further adjustments in time.
What about using Diego's original two transistor Dienoiser instead of Elvee's?
What about using Diego's original two transistor Dienoiser instead of Elvee's?
The whole issue with CCS NoNoiser is the denoiser BJT Vce, which has to be tweaked because of the 3x1N4148 (at 12Vout) output voltage swing limiting diodes. If I can find another arrangement then that resistor can be some standard value and no need for a pot.
I'm playing with this version in LTSpice, seems like startup swing is limited but there's a slight AC hit. Replace the three (at 12Vout) diodes with a resistor and small cap.
Another benefit seems to be faster startup. Couldn't find any other performance hits in simulation. Would this work?
I'm playing with this version in LTSpice, seems like startup swing is limited but there's a slight AC hit. Replace the three (at 12Vout) diodes with a resistor and small cap.
Another benefit seems to be faster startup. Couldn't find any other performance hits in simulation. Would this work?
Or maybe a better compromise. After reading TI's datasheet for regular LM317 (non -N) I noticed they do have a low programming current application:
With R1 at 1.2K we get more performance out of the CCS. Adding the latest arrangement to remove those startup swing limiting diodes, there's some 0.5V overshoot (which could maybe be lowered even more) and it should reach some 120dB of PSRR.
This is with a regular red LED, no extra diode in CCS, denoiser resistor can be some standard value, and Vout is set like regular LM317, two resistors in parallel should offer good enough precision. No trimpots, and stock LM317 tempco. And Vce of denoiser BJT is still high.
edit: Output impedance also gets a nice boost in performance:
With R1 at 1.2K we get more performance out of the CCS. Adding the latest arrangement to remove those startup swing limiting diodes, there's some 0.5V overshoot (which could maybe be lowered even more) and it should reach some 120dB of PSRR.
This is with a regular red LED, no extra diode in CCS, denoiser resistor can be some standard value, and Vout is set like regular LM317, two resistors in parallel should offer good enough precision. No trimpots, and stock LM317 tempco. And Vce of denoiser BJT is still high.
edit: Output impedance also gets a nice boost in performance:
Last edited:
OK, I simmed the circuit on the right, for 15v, and seemed to have got similar results to you. How do you name or describe that circuit so I can upload it?
That's still Elvee's Denoisator with CCS but I modified the startup swing limiter. CCS v2 or something I don't know.
For me it seems like a good compromise. If it reaches 120dB of PSRR then it's pretty nice, since it has the original LM3x7 tempco rating. And no pots required.
An addon board would look like attached photo, has everything required and would need standard values. You'd still need to replace the two original resistors with 1.2K and whatever is required for your desired Vout.
For me it seems like a good compromise. If it reaches 120dB of PSRR then it's pretty nice, since it has the original LM3x7 tempco rating. And no pots required.
An addon board would look like attached photo, has everything required and would need standard values. You'd still need to replace the two original resistors with 1.2K and whatever is required for your desired Vout.
Attachments
This was my take:
Post in thread 'D-Noizator: a magic active noise canceller to retrofit & upgrade any 317-based VReg'
Have a look at the following answers.
In the Nonoiser, the adj. pin is driven by a low-impedance emitter follower, but here the collectors have a much higher impedance
The filter for the startup spike needs different values for other Vouts and LT317 model is quite optimistic.
What voltage do you need for your application?
@Elvee
At 2.7K there's about 10x current in the divider compared to ADJ current. But 1.2K is a datasheet value so I suppose that should be a safe max value?