Phantom power drop resistor

[Spencerleehorton]

its seems we have a difference of opinion!!!?

Abbey says its inadequate for this purpose and squarewave says it works well?

could someone explain why a SMPS over a linear is quieter?

 

[Bo Deadly]

It all comes down to implementation. It is possible to make either very quiet. It is easy to make either noisy. Linear rectifies AC, filters and then regulates which is pretty crude if you think about it. Linear gets noisier with more load whereas SMPS generally does not (in fact SMPS performance is better under load). Linear regulator ICs can be pretty noisy actually. SMPS OTOH starts out with the same crude rectifying and filtering but then converts that rough DC into very rapidly switched AC pulses. That goes through a transformer to isolate but because it also makes a high frequency (50kHz+) resonant circuit, the switching frequency can be eliminated almost entirely with relatively small LC filtering. It then modulates the pulse width to supply more or less current to the output. The problem with SMPS is that if there are imperfections in the circuit, components or layout, you can get LF noise. So SMPS have a bad history for being noisy. But modern SMPS designs (within the last 10 years or so), ICs and transformers have been perfected over the years to the point where the LF noise is basically non-existant. Anything left can be crushed with a choke and cap although only down to a certain point (1kHz or so). And because the technique of modulating switching is so efficient, they are a fraction of the size and put out less heat.

There are a few problems to watch out for though. One is, as I have mentioned, SMPS ICs usually have a low-load mode where they actually turn off switching entirely for brief periods. This creates low frequency noise. But the solution is to just make sure you load it enough such as by adding a load resistor. And of course if the size of the SMPS is correct, that may not be necessary at all or the resistor can be larger and consume less power. Another issue is that the SMPS module can emit EMI at low frequencies. Meaning the output is clean but there is electromagnetic fields emitted that can be picked up by nearby high impedance / high gain circuitry. So it can be important to position the SMPS far away from any such circuitry. I have one unit where I made the entire supply external because of this. But I did another mic pre where the SMPS was within may 10cm of the mic circuitry and in the end it was super quiet. Another issue that I have heard of by not seen myself is that if you have multiple SMPS stacked together, small differences in switching frequencies can cause "beating" noise at LF. I have not seen this myself and I have a unit with 3 SMPS in the same enclosure (2 are the same). The solution in this case would be to make an SMPS that generates all of the desired voltages from one IC.

 

[moamps]

could someone explain why a SMPS over a linear is quieter?

It's not. At all. The specs say 350mVpp of ripple. It is over audio band but it is still there.
And you, most probably,  can't fix it if it fails. And you should waste some power in order to stabilize it. That's one of the reasons why Abby says it isn't a good choice.
Your NEVE deserves a good linear power supply .

 

[Bo Deadly]

Fight! 

It's not. At all. The specs say 350mVpp of ripple. It is over audio band but it is still there.

At 67kHz. Even without a choke and cap it will never be heard.

Your NEVE deserves a good linear power supply

So what about Neve's Neves? Every new product from Neve and just about everyone else use SMPS.

 

[moamps]

At 67kHz. Even without a choke and cap it will never be heard.

You are wrong here. Imagine a high quality stereo setup with two DPA or B&K microphones in a church, a preamp with your 67kHz garbage all over the place and a 192k (or DSD) recording sistem set to 96kHz. What the result be?   

So what about Neve's Neves? Every new product from Neve and just about everyone else use SMPS.

I hope they don't use SMPS LED drivers.

 

[Brian Roth]

FWIW, I've done some work on a few Neve Capricorn (digital) desks which use a bunch of SMPS units, even in the mic preamp/ADC units.  Exception:  the 48V phantom supplies are linears.

Of course, that desk was a 1990's design.....

Bri

 

[abbey road d enfer]

its seems we have a difference of opinion!!!?

Abbey says its inadequate for this purpose and squarewave says it works well?

could someone explain why a SMPS over a linear is quieter?

My main concern was with the fact that this type of smps is designed to power LED's, which means the primary regulated parameter is output current, because LED's want controlled current. However, the regulation performance is definitely not a s strict as it is for phantom voltage. An LED string does not require 1% accuracy, and noise is a non-concern (within limits).
Typically, I don't know how accurate is the output voltage when such a unit used for phantom powering (i.e. way below it's current-regulated zone), and most of all I don't know how noisy it is in this case, because it is a non specified parameter.
Using a bleeder resistor would only ensure it is not in choke mode, but would not guarantee any kind of voltage /noise performance.
That's the reason why I would not recommend this type of smps for phantom. I'm not saying it can't work, but it's kind of a lottery there.
It would take a very bad designer to build an unusably noisy linear phantom psu, but I've seen it happen; it's not proof that the concept doesn't work, just that a viable concept needs a viable implementation.


There are  many smps that are designed for voltage mode, and their voltage regulation and noise performance are well documented. It is a much safer bet. Indeed, adding complementary filtering in order to get rid of HF noise components is wise. One should make sure HF noise do not enter the system. There is usually additional filtering on phantom distribution, so conducted emissions are not a serious concern, but radiated emissions should be avoided, by containing them outside the powered unit.

 

[Spencerleehorton]

Ok a linear regulator design it is.
I’ll post up a proposed schematic just for a once over in case anyone want to suggest improvements.

My main concern is when I put a Neve 1290 on the bench to put in a 5k log output pot I then tested the 48v and lots of noise then appeared in the scope signal!!
With the unit in the patch bay I couldn’t hear any noise being introduced but wanted a fail safe noise free (if possible) solution.

Great discussion btw.

 

[Spencerleehorton]

Also have some other questions:
1. Can a TL83 be used as a drop in replacement for LM317,  drop in making sure correct pins are used.
2. Will the TL783 be ok for 4 channels of 48v on Neve 1290, do I need to isolate each 48v
3. If I do have to isolate each 48v what’s the best way?

 

[Spencerleehorton]

This is what I’m thinking I should do.

 

[abbey road d enfer]

This is what I’m thinking I should do.

negative of 220uF cap should be permanently connected to Pin 1.

 

[mhelin]

How about the 48V  / 380 mA IP Phone supplies like this CISCO (for some reason those CISCO's are all refurbished, something wrong in the design?):

https://www.ebay.co.uk/itm/CISCO-48V-Mains-Adapter-Power-Supply-IP-Phone-7960-7940-7941-7942-7962-7911-UK/303039072664

I've used one like above to power a dual Neve pre , with linear 24V regulation for the preamp modules, but  without additional filtering for phantom. Should probably add that choke and maybe an extra cap and resistor too, it's not too noisy but could be better.

There are now also cheap 48V PoE (power over ethernet) supplies and splitters available:

https://www.ebay.co.uk/itm/48V-2A-96Watt-For-PoE-Switch-Injector-100-240V-AC-to-DC-Power-Supply-Adapter/252991086335

https://www.ebay.co.uk/itm/Nice-Power-over-Ethernet-POE-Splitter-10-100mbps-48V-to-12V-POE-Adapter-I-ksJ7/264450213583

https://www.ebay.co.uk/itm/POE-Splitter-Adapter-Cable-Power-Over-Ethernet-48V-12V-Compliant-IEEE802-3af/372275206425

Maybe you could just put PoE connector into Neve rack unit and use the PoE "power injector" to power the pre.

 

[abbey road d enfer]

All these ads provide no specs, so it's hard to say. I would not consider using them without any additional filtering.

 

[tk@halmi]

Also have some other questions:
1. Can a TL83 be used as a drop in replacement for LM317,  drop in making sure correct pins are used.
2. Will the TL783 be ok for 4 channels of 48v on Neve 1290, do I need to isolate each 48v
3. If I do have to isolate each 48v what’s the best way?

I am not sure if anyone mentioned that an LM317HV part exists that takes 60V between its terminals and is a drop in fee the regular LM317.

 

[Spencerleehorton]

My next question is what is better in this diagram and why?

 

[abbey road d enfer]

My next question is what is better in this diagram and why?

The right one is better because it has additional filtering on the phantom voltage. However, these two have different resistors to power the LED, why? Indeed, with 3.9k, luminosity will be higher than with 10k, however it also depends very much on the type of LED. You'll need to experiment. Typically, LED's operate at 10-20mA current, but it may be too bright for comfort. With 10k, current is 4.6mA, with 3.9k it's 11mA. that may be too much... or not.

 

[Spencerleehorton]

Ah ok thanks, I was thinking 3.9k or 4.7k rather than 10k, just took that off another schematic.

Right for the next check.

What values are the best for this one?

 

[Spencerleehorton]

Another couple of questions spring to mind on the config 2:
It has a 220R on the + side and a 10k or 3.9k let’s say on the - side.
Does it need the 220R?
Does it not matter which side the current resistor is for the LED?
I’ve got a red led in there at the moment with 220R and 10k.
To my mind the 3.9k needs to be where the 220R is, then I ask myself why put in the 220R?
Someone obviously thought it should be there for a purpose?

 

[abbey road d enfer]

What values are the best for this one?

How much current do you want to actually draw?

 

[Spencerleehorton]

Ideally I want to build one Psu feeding 4 channels for 48v

At the minute I have 2 Psu feeding 2 channels each