goldenreference low voltage power supply

[dip16amp]

GRLV noise frequency graph.

 

 

Edited January 23 by dip16amp

[dip16amp]

I've been able to use my RME ADI-2 PRO & RMAA 6.4.5 to measure GRLV power supply output noise. Kevin asked about using my setup with a DC block. It started with a 10uf film cap and 10k resistor then an audio transformer to isolate the RME from the power supply. I discovered a few sources of noise and removed them to get the GRLV output noise down to -120 dB. The first source of noise was 50 Hz and it's harmonics from 7 LED volt meter displays. The second was from not having ground connected from the RME to the power supply. The third was 60 Hz and it's harmonics because the GRLV circuit was too close to the power transformers. 1" was too close and 4" was much better. The pics in the previous post shows the progress over the last 10 days. The final 2 pics show the positive and negative voltage outputs on left and right channels. Then the final pic shows before and after the GRLV circuit.

Here is the DC block circuit used to measure the output noise.

I used one for each channel.

 

 

 

 

 

Edited January 25 by dip16amp

[kevin gilmore]

measuring stuff like this correctly while avoiding ground loops and transformer radiation is very hard.

[Craig Sawyers]

Even if you eliminate those causes, you can be plagued with RF and other breakthrough into wiring, scope probes etc.

I chased my tail with this in attempting to make sensitive measurements to an active crossover containing a whole bunch of opamps. There was what looked like oscillation in the high tens of kHz. I was using a compact Tektronix oscilloscope. After well over an hour, I turned the scope on its its edge instead of on its feet. The "oscillation" disappeared. It was interference from the oscilloscope switched mode supply.

When making high sensitivity measurements I have to turn the heating off - the microprocessor based controller shoves out a whole bunch of interference (Honewell). And the LED ceiling lights - similar garbage from the rubbish SMPS's.

There is a guy who installed LED lights in his radio shack - and got a massive level of garbage into his gear. He ended up making a linear supply for the lights, which sorted the problem.

This and the next page http://www.chavfreezone.me.uk/2018/LED-Driver-QRM.html

[simmconn]

@dip16ampthe test results are all relative (in dB). Are you sure the 0dB reference is calibrated to 1V rms when you interpret -120dB as -120dBV? Even the most advanced Audio Precision APX555B is “only” spec’d at -117dB of range plus 1uV. Besides, the limited bandwidth of the line matching transformer and the high-ish 10k resistor in combination with the input capacitance of the ADC probably make the test result look quite a bit better than it actually is.

[dip16amp]

6 hours ago, simmconn said:

@dip16ampthe test results are all relative (in dB). Are you sure the 0dB reference is calibrated to 1V rms when you interpret -120dB as -120dBV? Even the most advanced Audio Precision APX555B is “only” spec’d at -117dB of range plus 1uV. Besides, the limited bandwidth of the line matching transformer and the high-ish 10k resistor in combination with the input capacitance of the ADC probably make the test result look quite a bit better than it actually is.

you're right in i have not calibrated it to 1v so i can't say 1uv, I'll strike that. from the post.

this does let me see noise on power supplies for comparison like this bk1735.

 

[audioguru1]

So I built a set and they work! I am however not seeing any adjustment with the 100k trim pots. Same for two boards and the positive and negative side of each. I've checked my work (and orientation), and all seems ok. I could just leave them (they run stable at +/-20v as expected) but i'd like to understand why i'm not seeing any adjustment. Looking at the circuit it seems pretty simple... I know I am probably missing something obvious.

[jamesmking]

3 hours ago, audioguru1 said:

So I built a set and they work! I am however not seeing any adjustment with the 100k trim pots. Same for two boards and the positive and negative side of each. I've checked my work (and orientation), and all seems ok. I could just leave them (they run stable at +/-20v as expected) but i'd like to understand why i'm not seeing any adjustment. Looking at the circuit it seems pretty simple... I know I am probably missing something obvious.

 

The output voltage is set by two resistors R8/R7 and R10/R9 (see schematic below).

the trim pots make only very small adjustments - a few mV for those who want spot on voltages - there is a trade-off between adjustment range and temperature stability.

If you want more adjustment range decrease the value of R24,R23,R18,R17. 810K instead of 1M for each gives about + and - 20mV adjustment range.

If you want reasonably accurate voltages AND lower temperature drift you can use 0.1% resistors for R8,R7,R10 and R9 with low temperature coefficients (stock resistors are usually about 100ppm, 0.1% are usually around 25ppm) and omit the 1M ohm resistors and trim pot.

 

 

Edited January 25 by jamesmking

[audioguru1]

4 hours ago, jamesmking said:

The output voltage is set by two resistors R8/R7 and R10/R9 (see schematic below).

the trim pots make only very small adjustments - a few mV for those who want spot on voltages - there is a trade-off between adjustment range and temperature stability.

If you want more adjustment range decrease the value of R24,R23,R18,R17. 810K instead of 1M for each gives about + and - 20mV adjustment range.

If you want reasonably accurate voltages AND lower temperature drift you can use 0.1% resistors for R8,R7,R10 and R9 with low temperature coefficients (stock resistors are usually about 100ppm, 0.1% are usually around 25ppm) and omit the 1M ohm resistors and trim pot.

 

I thought about making this value change (I did read every page of this thread- seems like one other person had the same issue, but just decided to live with it).
 

I’m not seeing any change in voltage at all even 1mv, which leads me to worry I have an actual issue. I am not measuring with a load, maybe that’s part of it? I’m at a loss - it’s a simple part of the circuit. Maybe there really isn’t enough range to see on my meter un loaded. 

Will do more testing, any suggestions appreciated.
 

[audiostar]

If you want to use the trimpots, you need to recalculate the 1M resistors depending on selected output voltage. Like for 15V, one of them per side would need to be 473k, so 1M is too hight for 20V output.

[dip16amp]

I used the bk1735 to replace the transformer as input to the grlv circuit.

here is the noise level before and after the grlv

then i tested an adjustable 0-48v 10a power supply i found on amazon for $40

DROK 48V Power Supply, AC 110V/220V to DC 0-48V 10A

and tested it before and after the grlv

 

 

Edited January 25 by dip16amp

[dip16amp]

i load tested the drok 48v 10a power supply and only got up to 27v at 10a

with a rigol dl3021a dc electronic load tester

here is the 1a steps

(edit: forgot to set the drok input voltage setting to 115 vice 230)

i was able to get 35 volts at 8 amps before reaching my load tester limit.

anps volts

1a 36v

2a 35v

3a 34v

4a 33v

5a 32.6v

6a 32v

7a 31v

8a 30v

9a 28v

10a 27v

then i connected a grlv with a 2k resistor to output 23.3v and load tested it

amps volts

1a 23.3v

2a 23.2v

3a 23.2v

4a 23.1v

5a 23.1v

6a 23.0v

7a 23.0v

8a 22.9v

9a 22.9v

10a 22.8v

 

Edited January 29 by dip16amp

[kevin gilmore]

golden reference was designed for only 5 amps.

[dip16amp]

On 1/25/2023 at 1:12 PM, kevin gilmore said:

golden reference was designed for only 5 amps.

but the mjw21196 is rated for 16 amps

what could go wrong?

do you need to redesign the grlv for 10 amps?

do i have to modify something or is it good as is?

i know the load tester is only rated for 200w and it survived

i can get a 60v 8a adjustable power supply for $47.

maybe it can do 35v at 8a

only 8 amps tho

 

 

 

Edited January 28 by dip16amp

[audioguru1]

9 hours ago, audiostar said:

If you want to use the trimpots, you need to recalculate the 1M resistors depending on selected output voltage. Like for 15V, one of them per side would need to be 473k, so 1M is too hight for 20V output.

Thanks a million - this makes perfect sense and I understand.

[kevin gilmore]

to do 10 amps you are going to need to boost the drive current. probably lowering the value of r1 to 120 ohms. same with r4.

[dip16amp]

ok, let me find some 120 ohm resistors.

 

[dip16amp]

found a couple 121 ohm resistors and here is what i get now

amps volts

1a 23.35v

2a 23.36v

3a 23.37v

4a 23.38v

5a 23.39v

6a 23.40v

7a 23.40v

8a 23.40v

9a 23.41v

10a 23.41v

 

Edited January 27 by dip16amp

[dip16amp]

i forgot to set the drok input voltage setting to 115 vice 230.

thought i was in the habit of checking that first before using.

now i can get 35 volts at 8 amps before reaching my load tester limit.

put a 3k resistor in my grlv to get 30 volts out.

now i get 30 volts at 8 amps out on a load test.

here is the noise level before and after the grlv regulator during the 8 amp load test.

 

Edited January 31 by dip16amp