Ultra Low Noise Lab Power Supply Extender for your existing lab supply

[branadic]

Hi,

I wanted to have an "Ultra Low Noise Lab Power Supply" for experiments with voltage references and instead of having a bunch of batteries (remember your green finger print on this earth) and instead of always using a low noise regulator for each quick and dirty voltage reference setup I decided to build a power supply for this and similar purpose.

LT3042 and LT3045 are available for a while and page 23 of the datasheet gave me exactly what I needed. It took its time to grow this idea and meanwhile some guy on DIY audio came across with a ready to go pcb for a first setup.

The price of 12€ for an audiophile white looking board seemed okay to modify it to my needs. It's a two channel, fixed current board paralleling 3x LT3045 giving up to 15V and up to 1.5A per channel, more than enough. Instead of the 10way trimmpots on this board I added Bourns multiway pots in 100k and 100k with 100k in parallel for the voltage setting of up to 15V.
Further I bought an 50VA, 12Vac toroidal transformer and some panel meters for voltage and current. The panel meters are supplied by the rectified voltage at the input of the voltage regulators.

The case was planed using Schaeffer front panel designer, exported as dxf and lasered at a local company. I decided to use a proMa EG2 aluminium profile (103mmx56mmx168mm) based chassis, but with wider front and back panel instead of standard 103mm and thus wider top and bottom plate.

It works pretty well, even though I seem to have some minor issues with mains hum coupling somewhere into the output. I'm currently investigating on that and ordered some self-adhesiv mu-metal foil to wrap it around the transformer. Further I will replace cables by shielded ones.

One the other hand without the transformer it is a pretty nice extender for existing lab power supplies, forming them into a low noise supply for the bench. So this is the idea. Let me ask, is there any interest in a group buy of bare boards and the aluminium chassis parts? All other parts can be easily bought in the bay and at your local electronic store or distributor. I plan to redesign the board with additional current limit setting, which is currently missing.

Meanwhile the guy with the pcb cracked up and sells assembled boards with fixed output current to the audiophile community for big bucks. 

EDIT: Added a first picture of my build.

-branadic-

[Henrik_V]

branadic, if you put some effort into a clean powersource, how about a better decoupling and guarding from mains?
DMM power supplies transformers from Fluke/HP use two shieldings , one for PE and one for guard...

[mzzj]

branadic, if you put some effort into a clean powersource, how about a better decoupling and guarding from mains?
DMM power supplies transformers from Fluke/HP use two shieldings , one for PE and one for guard...

That what I was also thinking. Toroidal transformer is bad if you want to avoid coupling to mains borne noise and one without shield screens is extra bad.

18650 Li-ion battery pack and LT3045 would avoid all the ground loops and conducted noise..

[branadic]

That's an effort I can spend, but I'm sure most of the people here on the board are not willing to pay the resulting price. Thus the idea of an ultra low noise power supply extender for existing cheapo lab supplies
But noone said that I won't do both.

-branadic-

[prasimix]

How are you going to shield noise that comes from cheapo power supply? Did you already try to connect LT3045 and check result? I presume that you are well aware how ambitious is your title (i.e. ultra low noise).

[BravoV]

Whats the low noise target in numbers or measurable metrics ?

[branadic]

How are you going to shield noise that comes from cheapo power supply? Did you already try to connect LT3045 and check result? I presume that you are well aware how ambitious is your title (i.e. ultra low noise).

The LT3045 has pretty good PSRR, so no worries whatsoever. You just connect the extender with banana jacks to your common cheap power supply.
Yes, I already connected the board to my lab power supply (PPS5330) and measured noise (0.1Hz - 10Hz) in the order of 2µVpp @ 1V and 20µVpp @ 10Vdc. Yes, be sure that I got infected by metrology in almost every fashion

Whats the low noise target in numbers or measurable metrics ?

See answer before, take a look at the datasheet of LT3045 and remember that three are paralleled.

-branadic-

[prasimix]

Great, now I noticed in your signature that you're well equipped. I never been in position to measure anything below 440 uV that shows my "super-duper" Rigol DSO with input grounded .

[branadic]

Simply add a low noise amplifier to convert your Rigol into a low noise measurement system

Added a first picture to the first post.

-branadic-

[prasimix]

That make sense . First I have to find one which I can build (I believe there is a bunch of them someplace on this forum).

[T3sl4co1l]

I wonder how much emissions those LED displays give off.

I have a few, I should check.

In short: you can call it whatever you like, but it is not low noise unless you've MEASURED IT SO!

Tim

[branadic]

In short: you can call it whatever you like, but it is not low noise unless you've MEASURED IT SO!

I'm sure you can read and I already stated that I measured ~2µVpp (0.1 - 10Hz) @ 1V and ~20µVpp @ 10V in configuration as an lab extender. I didn' state that I measured it in configuration as a stand-alone lab supply.

I received some mu metal foil (0,1mm) today and wrapped it around the transformer (configuration as stand-alone lab suplly). I'm currently waiting for my low noise amplifier to load the input cap to 10V output.

-branadic-

[T3sl4co1l]

In short: you can call it whatever you like, but it is not low noise unless you've MEASURED IT SO!

I'm sure you can read and I already stated that I measured ~2µVpp (0.1 - 10Hz) in configuration as an lab extender. I didn' state that I measured it in configuration as a stand-alone lab supply.

Oh, in a reply lower down.  Hard to see at a scan.

What about 10Hz to 10MHz, say?

Tim

[branadic]

What about 10Hz to 10MHz, say?

Step by step. Currently I measure it in the stand-alone solution, first 0.1 - 10Hz, next 10Hz - 100kHz as I have an amplifier for that too.

-branadic-

[branadic]

First impression in stand-alone configuration (with transformer powered from 230V mains line) and 10V output voltage. One division equals 5µV (0.1x setting on the scope, 80dB noise amplifier).

-branadic-

[branadic]

Did some further measurements in both configurations and it seems that there is no difference between Lab Supply Extender and Stand-Alone version, low frequency noise (0.1 - 10Hz) is in the order of 2µVpp per volt.

Next step is to measure 10Hz - 100kHz region.

-branadic-

[branadic]

It turned out that the hum that I measured was due to the cable between DSO and ULNPS supply I used for measuring. Replaced RG58 coax by twisted teflon cable and picked up hum is now much smaller but not fully vanished.
Attached is a picture of the inside and a picture of the preliminary front panel (a piece of printed paper laminated and attached with double-sided adhesive tape).

-branadic-

[jpb]

An interesting project.

What is the position on the boards? I have more power supplies than I need really (an Agilent, a Hameg and a TTi) but I like the idea of being able to add an ultra low noise front end and perhaps use the same approach for dedicated power supplies to other projects (OTT but hobbyists don't need to comply with the constraints put on commercial engineers!)

Given that the meters display down to less than a mV and the current can be up to 1.5A a useful addition might be sense inputs on the front panel but I guess for your purposes you don't plan to use the higher currents and don't particularly need the accuracy.

[branadic]

The voltage meters measure the voltage at the output jacks and are just as accurate as they come from asia. Technical parameter are given as below:

Weight: 16g
Display mode: Five 0.36 LED digital tube
update rate: > 3s/time
Physical size: 48 x 29 x 22 mm (L*W* H)
Opening size: 45 x 26 mm (L*W)
Supply voltage: 3.5-30V
Lead wire length: 14 cm（4 lines）
Measurement range: DC 0-4.3000-33.000V
Operation temperature: -10 degrees~65 degrees
Measurement accuracy: (0.3‰+ 2words) {means the table worst case maximum error of less than 10mv normal within error 2mv}
Red line 1:Power Supply + Black line 1:Power Supply -
Red line 2:VIN + Black line2: VIN -

They are powered from the rectified transformer voltage. Similar for the ammeter:

Color: Black
Display: 5 digits 0.36" LED
LED color: Blue
Working voltage: DC 3.5-30V
Current measuring range: 0-3.0000A
Current measuring accuracy: ±(0.5‰+2digits)
Working temperature: -10°C~+65°C
Dimension: 48 x 29 x 22 mm
Cable Length: 140mm

It measures the current from the negative jack to the regulator board. The regulator board is the white one on the left side with the blue heat sink.

-branadic-

[Wolfgang]

Hi,

I have some suggestions for improvements of your design:

- Make shielded compartments, one for the input filter, transformer, rectifier and first filter caps, and seperate ones for the other stuff
- separate power supplies for the DMMs, also in a shielded box
- User a preregulator before the final ones, also in an shielded box
- Final regulators also in a shieled box, preferably with coaxial output
- Use feedthru caps for the connection between boxes

Am I paranoid ? Yes. 

I was trying to make absolutely quiet power supplies for noise measurement preamps. The problem I had is that most RFI is coming from outside, from LEDs, IOTs, Mobile phones, PLC, ... To kill all that is difficult and needs RF building techniques.
As a last resort, I made battery powered supplies as well, they are the best so far ...

https://electronicprojectsforfun.wordpress.com/power-supplies/battery-operated-power-supplies/

[Richard Crowley]

The photo of the insides of the enclosure look surprisingly conventional. I see no evidence of shielding or even wire routing that one would expect in a low-noise design.  Aren't the digital displays noisy?

[branadic]

Well, you can solve a problem from the direction of panic (top down), which ends up in an expensive piece of kit or you can solve a problem bottom up, build something and improve it until you reach your goal. I always choose the latter
Noise from the panel meters can be easily filtered as it is high frequency. Couldn't find a problem till now, but if it turns out to be a problem I will perform some required steps on that.
Up to now there was no need to use panic shielding. Well at least, as I'm not sure if the mu metal shield around the transformer was really necessary. But since this is a self-adhesive tape one I won't remove it anymore. Instead the classical way is:
- to twist all AC cables
- avoid ground loops, even shielded cables can create them, when shield is connected on both ends

If really necessary I will replace some cables by twisted shielded ones, but therefor I need to make further measurements to see if this is required here.

-branadic-

[Wolfgang]

Your approach is just as valid as mine. I have more effort, but your probably have more redesign steps.

Hint: If you want rocket science or extreme performance, try to exclude as many errors as you can beforehand. There will be enough pitfalls left for you to fix. Murphy never sleeps. 

[branadic]

You are right Wolfgang, but what is critical for the stand-alone supply is less of an issue for the lab extender, as you don't have the 50Hz hum and the related problems as a source inside you device. But obviously there is not much interest in such a lab supply extender. Would have thought the resonance is much bigger.

Finished the first measurements on my stand-alone solution in the 10Hz - 100kHz range and can now see the advantages that needs be addressed next.

-branadic-

[Wolfgang]

Hi Branadic,

if you just want to finish off the noise and ripple of an existing PSU, have you seen the tricks from Charles Wenzel with his active noise killer circuits ?

http://www.wenzel.com/documents/finesse.html

Maybe this helps, when mounted close to the circuit to be powered.

Maybe a comparison of voltage regulator chips is also interesting:

https://www.bartelsos.de/dk7jb.php/rauschen-von-spannungsreglern

Another great source of tricks and infos is the reverse engineering of old HP power supplies with top performance.
I am looking for one of those on eBay just to have a look inside. Manuals are probably available at the BAMA archive.

Regards
  Wolfgang