• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Power Supply Mod Testing for Schiit Magni 3+

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,395
Location
Seattle Area
A member sent me a modified version of Schiit Magni 3+ for testing. The modification is based on the fact that the power supply design violates the specification from the IC company (low drop out regulator). Here is the schematic and modifications made:
Schiit Magni 3+ Mod Power Supply Schematic.png

Per note from the member, specs for LM 317/337 specifically forbit using large or low ESR capacitors in the output. I have not looked at the reason but it may be that the surge current will be too high. Using an oscilloscope, member found the best, compliant solution to be removal of caps 5 and 6 and replacing C3 and C4 with 10 microfarad tantalum caps. Result was slightly improved noise on the output of the power supply:
Schiit Magni 3+ Mod Scope PS ripple.png


Question is whether this improves performance any so let's measure it starting with our dashboard:
Schiit Magni 3+ Mod Measurements.png


Overall SINAD is the same, dominated by the high distortion. Power supply components are more or less the same to my eye. And there are enough variations here given that the old test is 3 years old as to not be able to make precise comparisons anyway.

SNR is identical to stock unit:
Schiit Magni 3+ Mod SNR Measurements.png


Stock:
index.php


Frequency response is the same:
Schiit Magni 3+ Mod frequency response Measurements.png


Lastly I ran a power sweep at 300 ohm. Results were the same in low gain but high gain produced slightly more power:
Schiit Magni 3+ Mod Power into 300 ohm Measurements.png


Can't imagine the power supply mod improving level of power. Maybe there was a running design change here.

Conclusions
I have not looked at the datasheet to see what it says about output loading. Trusting the member's notes, it seems to be a good change to improve long term reliability. It may have some negative impact on dynamic power but without specific testing, it is hard to say if there is any. Continuous power is at least as good as stock unit so all is well there.

I am relieved to see a mod be in service of bettering the design based on proper knowledge than vudu. :)

--------
-----------
As always, questions, comments, recommendations, etc. are welcome.

Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
 

Jimster480

Major Contributor
Joined
Jan 26, 2018
Messages
2,880
Likes
2,032
Location
Tampa Bay
Nice little test here, while it doesn't improve performance they eliminated parts which we know always makes the design better. They should sell it back to Schiit so they can bring the improvement to everyone!
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
@amirm Can you measure low imp. loads ?
When there will be differences, most likely the largest differences will be there when the power supply section has to deliver current.

Personally I would not have used the recommended 1uF tantalum in C3,4 position but rather a 10uF electrolytic.
The LM317 also likes to have a 100nF X7R in parallel to input pin.
Directly on the input pin of the reg a 1uF X7R or tantalum may also be a good idea.
Not that it would matter that much here.

Using large caps on the output of the regulator lowers the speed with which the regulator can compensate the output voltage (transient output voltage response).
You can still use large caps there but has to be fed using an inductor between the 10uF and 2200uF caps.
LM317 is quite tolerant for larger value output caps, the LM337 not so much.

When using large capacitors on the output of the regulator a reverse diode between output and input is needed. Wonder why Schiit did not do that or they were just not drawn in the schematic.
 
Last edited:

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,386
Likes
24,749
Location
Alfred, NY
The PS noise is, as you have seen, not terribly relevant- modern circuits have high PSR.

There's an art to getting the most out of the 317/337 family, including failure-proofing them which was largely not done here.

Best trick- you can reduce noise and source impedance by sacrificing efficiency. Heat sink them and load them to put out a lot of current. None of this matters in this application, but for uses where power supply noise is important, pulling 0.5-1.0 amp out of the chips really ups the performance. Erroll Dietz did a great app note on this.
 

Ken Tajalli

Major Contributor
Forum Donor
Joined
Sep 8, 2021
Messages
1,995
Likes
1,773
Location
London UK
I second what @SIY is saying.
But my 2 pence:
- from oscilloscope pictures, I believe there is a rectifier switching noise, that could cause EMI into the following circuits.
- I have had good results paying attention to shielding and quality of R5 R6 C11 C12, keep them noise free, the device goes quieter.
- take the transformer and the rectifiers out of the box.
 

GWolfman

Addicted to Fun and Learning
Forum Donor
Joined
Nov 3, 2020
Messages
624
Likes
1,041
Nice to see something a bit different on the site. What’s great is that he (or she) was measuring (what he could) and wanted to validate the rest of the system for possible improvement/deterioration.
 

PeteL

Major Contributor
Joined
Jun 1, 2020
Messages
3,303
Likes
3,838
@amirm Can you measure low imp. loads ?
When there will be differences, most likely the largest differences will be there when the power supply section has to deliver current.

Personally I would not have used the recommended 1uF tantalum in C3,4 position but rather a 10uF electrolytic.
The LM317 also likes to have a 100nF X7R in parallel to input pin.
Directly on the input pin of the reg a 1uF X7R or tantalum may also be a good idea.
Not that it would matter that much here.

Using large caps on the output of the regulator lowers the speed with which the regulator can compensate the output voltage (transient output voltage response).
You can still use large caps there but has to be fed using an inductor between the 10uF and 2200uF caps.
LM317 is quite tolerant for larger value output caps, the LM337 not so much.

When using large capacitors on the output of the regulator a reverse diode between output and input is needed. Wonder why Schiit did not do that or they were just not drawn in the schematic.
In your opinion, do you feel it can explain a 20% Increase in power or do you think it be like in Amir's opinion a running design change? It's not an insignificant benefit.
1684415720349.png
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
In 300ohm only a 10% higher power supply rail voltage would give an increase in output voltage.
I assume the voltage determining resistors have not been changed in value or they have been changed by Schiit during production.
A 20% increase in power = 0.8dB and would not be audible.
 
Last edited:

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
I second what @SIY is saying.
But my 2 pence:
- from oscilloscope pictures, I believe there is a rectifier switching noise, that could cause EMI into the following circuits.
- I have had good results paying attention to shielding and quality of R5 R6 C11 C12, keep them noise free, the device goes quieter.
- take the transformer and the rectifiers out of the box.
There is no transformer in the box, it is external. 15VAC 24VA
On the scope pictures the time-base is not visible so the probe might as well be picking up some common mode crap 100KHz switching crap from somewhere.
 

lc6

Active Member
Joined
Jul 19, 2022
Messages
183
Likes
276
See section 9.2 Typical Application on page 11 of the LM317 datasheet. Caps should still be resized and diodes added.
I fail to see the purpose of resistors R2 and R3 which add 1R resistance to the power supply rails after the LMs present a virtually nil resistance (i.e. LM's output voltage is virtually invariant with load current within operational limits).
 

Ken Tajalli

Major Contributor
Forum Donor
Joined
Sep 8, 2021
Messages
1,995
Likes
1,773
Location
London UK
There is no transformer in the box, it is external. 15VAC 24VA
Thanx, didn't know that.
On the scope pictures the time-base is not visible so the probe might as well be picking up some common mode crap 100KHz switching crap from somewhere.
I doubt Amir in his wisdom, would put timebase at 100kHz, trying to figure out PS noise!
Perhaps a choke and a resistor then.
 

Ken Tajalli

Major Contributor
Forum Donor
Joined
Sep 8, 2021
Messages
1,995
Likes
1,773
Location
London UK
I fail to see the purpose of resistors R2 and R3 which add 1R resistance to the power supply rails after the LMs present a virtually nil resistance (i.e. LM's output voltage is virtually invariant with load current within operational limits).
There could be local smoothing capacitors after the PSU on the board. The 1R resistor, helps with surges and noise.
Choosing the correct wattage, can also act as fuse.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Thanx, didn't know that.

I doubt Amir in his wisdom, would put timebase at 100kHz, trying to figure out PS noise!
Perhaps a choke and a resistor then.

I think the scope pictures are not taken by Amir but the one doing the mods.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
See section 9.2 Typical Application on page 11 of the LM317 datasheet. Caps should still be resized and diodes added.
I fail to see the purpose of resistors R2 and R3 which add 1R resistance to the power supply rails after the LMs present a virtually nil resistance (i.e. LM's output voltage is virtually invariant with load current within operational limits).
The 1 ohm is needed for the amp circuit.
The LM317 can handle higher capacitive loads on its output, the LM337 not so much.
Given the values of the used caps in the Magni it would have been advisable to add the diodes. I suppose Schiit didn't, they probably found the LM's were fine used this way and did not (have to) use the reverse diodes.
 
Last edited:

ElJaimito

Member
Joined
Mar 28, 2019
Messages
36
Likes
38
Having used these devices for too many years, I never had issues with large output capacitors (1000-4700uF). If this seems a lot, imagine the devices are connected to and powering a small mixing console, with 470 uF local bypass on each rail on 24 channels, a typical application for them in the 80's. However I would never use an output capacitor of any size without the reverse bypass diode across the device. The diode prevents device failure due to reverse voltage if the input is shorted or the input rails go down quicker than the outputs on power-off. And I would never in any circumstances use a tantalum capacitor anywhere ever across a power rail if I wanted to keep my hair... (wait!) They are the most unreliable components ever made IMHO. 317/337 were convenient back in the day, but they are seriously ancient tech. Surprised they are still used. The 337 is not very good, a negative rail built with a 317 is more robust IME. Oh, and these devices are not LDO, need 3V across them to regulate...
 
Last edited:

Ken Tajalli

Major Contributor
Forum Donor
Joined
Sep 8, 2021
Messages
1,995
Likes
1,773
Location
London UK
Magni 3+ uses a 15VAC transformer. It uses a halfwave rectifier (I see only two diodes on the board) to create +/- voltages.
Output voltages after the regulators are marked at +/- 17VDC.
My mind has gone a blank, but is there enough headroom for the regulators to work?

magni 3+.jpg
 

sam_adams

Addicted to Fun and Learning
Joined
Dec 24, 2019
Messages
976
Likes
2,368
replacing C3 and C4 with 10 microfarad tantalum caps.

You'll be sorry if you do this. Better to use a high quality electrolytic in the 22 µF to 35 µF range. Use more bulk capacitors, 10 µF, distributed around the board.

However I would never use an output capacitor of any size without the reverse bypass diode across the device.

Seconded.
 
Top Bottom