This PSU I made is quite good, but its application to headphones is still unwise. With low sensitivity headphones, it’s fine, with higher sensitivity headphones, the noise floor is clearly audible.
Or maybe it’s the Connex SMPS, as my bench supply performs much better without additional filtering.
Or maybe it’s the Connex SMPS, as my bench supply performs much better without additional filtering.
Mark's filter is designed around a certain frequency range where the switcher/s typically used for a number of the projects here operate - if I recall its around 10kHz. You can also try adding some additional input filtering typical of most of Nelson's recent projects where a simple R-C-R-C-R-C is used (look at the ACP+ or the other Pass line-level projects).
The most conclusive way to figure things out is to put your setup on a scope and see where you are seeing the AC / noise. If you are unfamiliar with how to do this, check further up in this thread where Mark posted on this or DM me and I can help. The SMPS you are using might simply be using a switcher / setup that is producing AC/noise outside of this band and this is the easiest way to determine that.
--Tom
The most conclusive way to figure things out is to put your setup on a scope and see where you are seeing the AC / noise. If you are unfamiliar with how to do this, check further up in this thread where Mark posted on this or DM me and I can help. The SMPS you are using might simply be using a switcher / setup that is producing AC/noise outside of this band and this is the easiest way to determine that.
--Tom
Thank you, @thomasnadeau .
I think my PSU build might not be as neat as I thought it was.
I am referring to this picture:
https://www.diyaudio.com/community/...a-korg-nutube-etc.354213/page-57#post-7635408
When I measure the AC voltage between output ground of the filter and earth ground, I get about 0.25Vrms.
In other words, the amp's circuit ground is lifted by that much, which may be modulating that noise into the output.
That lift must be introduced by the Connex SMPS.
So, I need to revisit the wiring to understand if I did everything right.
I think my PSU build might not be as neat as I thought it was.
I am referring to this picture:
https://www.diyaudio.com/community/...a-korg-nutube-etc.354213/page-57#post-7635408
When I measure the AC voltage between output ground of the filter and earth ground, I get about 0.25Vrms.
In other words, the amp's circuit ground is lifted by that much, which may be modulating that noise into the output.
That lift must be introduced by the Connex SMPS.
So, I need to revisit the wiring to understand if I did everything right.
For measure, that looks pretty high. I tried a number of 24V meanwells and what I measured never exceeded .02V so it may be the Connex.
Were you able to determine where the frequency/ies were centered around? That will tell you if its outputting in the effective band on Mark's filter or not.
--Tom
Were you able to determine where the frequency/ies were centered around? That will tell you if its outputting in the effective band on Mark's filter or not.
--Tom
@thomasnadeau Didn’t put on the scope yet, but would be shocked if it were not 60 Hz, as it really sounds like it.
Still trying to ascertain how to correctly install the Connex. It has an earth tab on its input and I found out yesterday it’s connected to the bottom “cooling” plate, which itself has continuity to the chassis, which itself I connected to earth ground.
So, clearly, my setup is not quite right.
I wish they provided a schematic for those.
Still trying to ascertain how to correctly install the Connex. It has an earth tab on its input and I found out yesterday it’s connected to the bottom “cooling” plate, which itself has continuity to the chassis, which itself I connected to earth ground.
So, clearly, my setup is not quite right.
I wish they provided a schematic for those.
If its down at 60 you might have some ground loop / return path issues. What you describe might well be your earth-ground situation creating a loop. Try simply putting a CL60 between the ground pin on the IEC connector and your chassis star-ground point and see what you get. There are a number of very well done articles / threads here, but this is one I quickly found:
https://www.diyaudio.com/community/threads/grounding-arrangements-for-star-ground.258197/
I generally use one of Brosky's "house ground" kits which are very effective, inexpensive, very compact and work. From what I understand, it would also successfully pass grounding compliance testing:
http://glass-ware.store.turbify.net/housegnd.html
Basically you connect the earth ground/green wire from your inlet AC IEC/fuse to it, and then on the other side connect that to the chassis star ground point.
--Tom
https://www.diyaudio.com/community/threads/grounding-arrangements-for-star-ground.258197/
I generally use one of Brosky's "house ground" kits which are very effective, inexpensive, very compact and work. From what I understand, it would also successfully pass grounding compliance testing:
http://glass-ware.store.turbify.net/housegnd.html
Basically you connect the earth ground/green wire from your inlet AC IEC/fuse to it, and then on the other side connect that to the chassis star ground point.
--Tom
@thomasnadeau I suspect the idea with that Connex is that the only wire from IEC earth goes to the earth tab and the chassis is grounded via the cooling plate.
Earth tab > cooling plate,
Earth tab > “ground breaker” > SMPS circuit
Gonna have to verify this, of course.
Earth tab > cooling plate,
Earth tab > “ground breaker” > SMPS circuit
Gonna have to verify this, of course.
@thomasnadeau Validated it’s 60 Hz and its harmonics. The 120 Hz harmonic is actually at a greater level than the 60 Hz fundamental.
I changed the wiring of the PSU box, so there’s only one path to earth ground, but I didn’t add anything between IEC earth tab and SMPS earth tab.
I simplified the setup to investigate ground loops.
With the left board only connected (so, +, - and common ground connected to that board), I still get the same hum in the left channel.
Further, if I short input to ground, silence.
Clearly, when only +, - and common ground are connected to the left channel amp board, and XLR inputs are left unconnected, there’s no possibility of ground loop.
I am stomped.
I changed the wiring of the PSU box, so there’s only one path to earth ground, but I didn’t add anything between IEC earth tab and SMPS earth tab.
I simplified the setup to investigate ground loops.
With the left board only connected (so, +, - and common ground connected to that board), I still get the same hum in the left channel.
Further, if I short input to ground, silence.
Clearly, when only +, - and common ground are connected to the left channel amp board, and XLR inputs are left unconnected, there’s no possibility of ground loop.
I am stomped.
In the post #1,138 where you showed a picture of your build, you are connecting the inlet IEC connector's ground pin (green wire) directly to the screw on the chassis, where you are also connecting from that point to the switcher. That is where you want to isolate the earth ground because you essentially are offering a second return path for current on the IEC connector. For sake of pinpointing where the loop is, try putting a small resistor (10 Ohms or less) or CL-60 between the green wire that goes to the switcher. The basic premise is that you want to insert "a little resistance" between the chassis return paths and earth ground so that the earth ground is only used as an exception due to its (slightly) higher impedance.
It looks like your connector has two wires inside and has a shell shield. First, the outer shield of the connector / cable should only be connected at the source end (the chassis with the switcher inside) otherwise you will have TWO current return paths from the destination device, and likely a ground loop. This is similar to why you isolate the earth ground as I said above. Next, test continuity between the pins on your output connectors and the chassis the star ground point - you should not have any continuity between these points. You should only see continuity between the output cable's outer shield and chassis.
If you don't know how to do this, put your DVM in "ohms" mode and most have an audible beep/tone option. Connect your negative lead to the star ground and then use the positive lead to probe and connect each of the pins on the IO connectors and shield. If you touch the chassis anywhere you should get a beep and show continuity with the shield and all other parts of the chassis including the heat sink of the switcher. You should not see continuity between the star ground point and the pins on the I/O connectors.
--Tom
It looks like your connector has two wires inside and has a shell shield. First, the outer shield of the connector / cable should only be connected at the source end (the chassis with the switcher inside) otherwise you will have TWO current return paths from the destination device, and likely a ground loop. This is similar to why you isolate the earth ground as I said above. Next, test continuity between the pins on your output connectors and the chassis the star ground point - you should not have any continuity between these points. You should only see continuity between the output cable's outer shield and chassis.
If you don't know how to do this, put your DVM in "ohms" mode and most have an audible beep/tone option. Connect your negative lead to the star ground and then use the positive lead to probe and connect each of the pins on the IO connectors and shield. If you touch the chassis anywhere you should get a beep and show continuity with the shield and all other parts of the chassis including the heat sink of the switcher. You should not see continuity between the star ground point and the pins on the I/O connectors.
--Tom
Thanks, Tom, but I don't think it's any of that.
Today, I simplified the setup to the extreme:
1) I removed the IEC PE connection in the PSU box. (FUTURE READERS, NEVER DO THIS, UNLESS IT'S FOR A QUICK TEST.)
2) I connected a single umbilical from the PSU box to a single amp board. So, only +/-/Common going to the amp board.
3) I shorted the input on the amp board.
4) I removed the headphones protector board that sat in between the amp board output and the headphones.
After all that, same hum level in that one channel.
And, it cannot be attributed to any ground loop or to signal input.
Other possibilities:
- The SMPS spews out "stuff" that somehow makes it through to the amp board output.
- The amp has too much gain. (17 dB)
- The headphones I use for testing are too sensitive. (94dB)
A combination of these factors.
I'll fetch my sigma22 and see if that change things.
Of course, I suspected the amp board, but the hum level is the same on both boards...so, unless I made the same mistake twice...
PS: I also did a test bypassing the filter board in the PSU box, which didn't change anything, as anticipated.
You might have some induced noise. Definitely compare with the Sigma to rule out the switcher. Try another switcher also for reference too. Looking at your pictures, I wonder if its the switcher's transformer. I just noticed from your picture that its an open frame design so that could be radiating next door to the filter board and so on. Also try twisting your wires together as well as adjusting their spacing near other cables / the transformer. In the end, it might just be the switcher which might have been designed to be put into its own shielded container or the applications it was designed for didn't worry too much about induced RF/etc...
--Tom
--Tom
Quiet as a mouse with sigma22.
I realize even though this SMPS is designed for audio, it is not designed for headphones amplifiers.
I am also willing to believe the enclosure I picked is too small for this PSU.
I don't really have another SMPS on hand at the moment, but I could take all the parts out of the small box and spread them around to see if it changes anything.
Thanks for your help with this, Tom!
I questioned everything, but I didn't question the SMPS unit itself enough, even though it was the elephant in that small PSU box.
Glad you were able to narrow things down. I love my Sigma22 BTW. 
Try one of these in the meantime. I've used a bunch of these and they're quiet enough (especially with Mark's filter) and for fifteen bucks you can't go wrong:
https://www.amazon.com/dp/B019GYOM5O?psc=1&ref=ppx_yo2ov_dt_b_product_details
--Tom
Try one of these in the meantime. I've used a bunch of these and they're quiet enough (especially with Mark's filter) and for fifteen bucks you can't go wrong:
https://www.amazon.com/dp/B019GYOM5O?psc=1&ref=ppx_yo2ov_dt_b_product_details
--Tom
This is a real thing in EE and pops up in all sorts of places. I am sure there is an EE class many of us have missed but should have taken.
If you've had the pleasure of building the B1K and getting it to work without microphonic feedback, you'll know this affect first hand. It took me quite a while to avoid micro-phonics (audible at least) until I realized that both the chassis vibrates just sitting there as well as has induced vibrations due to things like the feedback from the environment.
In another example here, Wayne talks about how he uses an HP vibration analyzer as part of his design engineering in the Whammy thread:
-Tom
If you've had the pleasure of building the B1K and getting it to work without microphonic feedback, you'll know this affect first hand. It took me quite a while to avoid micro-phonics (audible at least) until I realized that both the chassis vibrates just sitting there as well as has induced vibrations due to things like the feedback from the environment.
In another example here, Wayne talks about how he uses an HP vibration analyzer as part of his design engineering in the Whammy thread:
-Tom
Dear mr @Mark Johnson final report! My mentor mr @Brijac made to me paralel 3 filters. The result is fascinating. I use MeanWell LRS 350-48 and only after installing this filter I got a sound one step better. I tested with several types of music, The result is fascinating. I take this opportunity to inform @Chasmo57 that this solution works. About the technical details of how to do it in the most beautiful way, it is best to contact @Brijac because he is the author of this solution.
As Mr. Galileo Galilei said: It still moves!
As Mr. Galileo Galilei said: It still moves!
