my new NA323 phono stage hums: please help!

[jpk]

Here is my new phono stage, made of left overs, ALWSR boards and NA323 cards, and it makes 50Hz hum:

I checked these threads about wiring (especially grounding):

• Pictures of my new phono pre-amp
• NAC 72 (loss of signal on left channel)
• Modified NAIM NA323 boards - try it

My signal ground goes from the input to the output socket, from there directly to the NA323s, from the NA323s to the star-GND point between the PSU and the ALWSRs where all other GNDs are connected to. PE is connected to chassis (it's an old Mission PSU case). If I disconnect PE from the chassis the hum becomes incredible loud, so I keep it connected. I checked with my scope: I don't see any hum on the power in/out lines at the ALWSRs, only at the output of the NA323 boards I see hum. If I connect PE/chassis to signal-GND the hum improves a lot, but still has about 10mV at the outputs of the NA323s. I don't have a cartridge, so I tried to load the NA323s with 100k in series with the input and 220R across the input: no difference. What can I do?

 

[bugbear]

(topic drift - what's the nice little PSU pcb?)

 

[Pete MB&D]

I had to move the transformer in mine to an external case to get rid of a hum, and move it about on the rack.
I would connect each 0v to the star with a separate wire, and try moving the transformers out from the case a temporary lash up will be fine.
It’s a bit of a black art earthing imho.
That mains socket could do with an insulating boot.

Pete

 

[jpk]

Thanks I will try! The PCBs are cut-aways from other project's PCBs I draw in KiCAD. Actually the PSU has a gyrator on it, and the ALWSRs have VBEs fittet, so kind of redundant - should I take out one of them?

EDIT: the phono stage is dual mono with totally separate channels. Only their GNDs are tied together at the star earth point. I found out that if I disconnect one of the blue transformers the hum reduces dramatically. Only if both transformers are connected I see this at the ouput of the NA323s:

 

[PigletsDad]

If it was dual mono, disconnecting one transformer would lose power to one channel. If both channels run, you must have some cross connections. Check the wiring.

 

[jpk]

Yes, with disconnecting one transformer of course one channel goes off. Today I removed one transformer and rewired the project using one PSU for both channels, but the result wasn't any better. Only moving the remaining transformer around cured the hum, but there is not enough space in the case. So I need to follow the advice given in #3 and put the transformer into a separate case. Should I keep the PCB with the rectifier and smoothing caps in the same case with the ALWSRs and the phono stage or is it better to get everything related to AC out of that case?

 

[Pete MB&D]

Yes, with disconnecting one transformer of course one channel goes off. Today I removed one transformer and rewired the project using one PSU for both channels, but the result wasn't any better. Only moving the remaining transformer around cured the hum, but there is not enough space in the case. So I need to follow the advice given in #3 and put the transfomrer into a separate case. Should I keep the PCB with the rectifier and smoothing caps in the same case with the ALWSRs and the phono stage or is it better to get everything related to AC out of that case?

Just the transformers it won't mind if they are far away from the rest of the psu, you will need enough cores on your cable to do the two supplys and an earth connection.

Pete

 

[Avon]

You need to keep the transformer and rectifier / smoother board in the same case. Feed the raw DC output into the second case, containing the ALWSR regulators and phono cards. The shorter the distance between the transformer and rectifiers / smoothers, the better. Think carefully about the 0V arrangement too.

 

[Craig B]

Nice recycle of an old Mission Cyrus PSX case.

 

[Pete MB&D]

You need to keep the transformer and rectifier / smoother board in the same case. Feed the raw DC output into the second case, containing the ALWSR regulators and phono cards. The shorter the distance between the transformer and rectifiers / smoothers, the better. Think carefully about the 0V arrangement too.

If it was a power amp then yes shortest distance from the tansformers/rectifiers/smoothers but a low power phono amp then some resistance between the transformers and rectifiers can give a quieter supply.

Pete

 

[sktn77a]

I'd play with the grounds and see what happens. Grounds can be very unpredictable.

 

[john.luckins]

I can see that the sensitive end of the ALWSR's, i.e. the voltage references and their RC filters are within a couple of cm's of one of the transformer feeds and its respective rectifier. See if moving the ALWSR's further away from the lower transformer helps. Also , if you are using full wave rectification and a common 0v return path on the rectifier board for both rails, you might get success by swapping over one of the AC feeds (in or out) to just one of the transformers. These are both long shots but worth a try IMHO before offboarding the transformers.

I would also recommend twisting together the +ve and 0v outputs of the ALWSR's right up to the NA323's.

John

 

[jpk]

I took out the transformers from the case and modded the PSU board into a filter board, it now has CRCLCRC into a gyrator, from there into the ALWSRs. I kept the diode on the PSU board so that it's protected against reverse voltage. I played around with the ground lines and decided for signal-GND coming in from the output socket, from there to the NA323s, from there to the star-GND point where all other GNDs are connected except DC input at the "dirty" end of the PSU board.

I can see that the sensitive end of the ALWSR's, i.e. the voltage references and their RC filters are within a couple of cm's of one of the transformer feeds and its respective rectifier. See if moving the ALWSR's further away from the lower transformer helps.

Many thanks John for your insight! Hum only disappeared when I moved the transformer really far away. Also I tried various other transformer types, but the blue ones were best. Some off the transformers gave horrible big hum, I never expected such differences!

Also , if you are using full wave rectification and a common 0v return path on the rectifier board for both rails, you might get success by swapping over one of the AC feeds (in or out) to just one of the transformers.

This changed hum considerabely, but did not remove it completely. My above scope shot shows the best configuration I could achieve.

I would also recommend twisting together the +ve and 0v outputs of the ALWSR's right up to the NA323's.

The 0v lines go to the star earth point, so they can be twisted only for a short part. I also checked Naim PCB layouts to see how they do it, and the +ve tracks go completely separate because the 0v tracks go to star points.

Also I noticed that hum improves if the case is connected to earth, but I don't understand where is the best point to connect chassis to signal-GND. Please have a look at attached picture and let me know if I should ground the chassis at A, B, C or D:

If I see chassis-GND as part of the shielding I would think that where shield comes in is the best connection, that would be B. If I consider chassis-GND as a more "dirty" kind of GND I would guess to connect it to C or D. Or to A if I see it as just another GND "track" that ends nowhere. These are my amatuer thoughts, hopefully not too silly... Please enlighten me how I should think of chassis-GND!

 

[john.luckins]

I've just spotted something that looks fishy to me. If you take a look at the ALWSR manual, page 26, you will see figures 7 and 8 are the connection diagrams for local and remote sense. I'm pretty sure you are using local sense in which case the two pins below your "out -" sticker are linked together as per figure 7. If that is the case then you should be using this linked "out-" pin to feed 0v directly to your NA323 boards. What your voltage rails are seeing is the voltages developed from the return to 0v current of the regulators passing through the impedance of the black leads that feed the regulators. This voltage is effectively added to the regulators output if you take the 0v feed from the cap board. If you connect the regulators as per figure 7 and pass the 0 volt feed through the regulator and that linked "out-" pin you will remove this unwanted voltage.

John

 

[jpk]

I had it wired like that at the very beginning, but when I saw the wirings of the other projects (see the links in my first post) I removed the out-GND wires from the ALWSRs and wired them to the star-GND.

Now everything is twisted as tight as I could. The phono cards are secured with a piece of bamboo chopstick. GND goes through the ALWSRs again. The chassis is connected to the clean end of the PSU board where the star GND point is (marked "A" in my previous picture). I added a heat sink to the ALWSRs. The pass transistors on the NA323 boards at the +ve input are linked out.

I tried again with one blue transformer, but severe hum was there, so I will put the transformer in a spearate case. With the transformer far away there was no hum at all even with everything floating. Here is a pic:

Thanks all for your help!

 

[jpk]

Nice recycle of an old Mission Cyrus PSX case.

I got two of them for their big and nice H+F transformers that I wanted to use in a future power amp project. The other case houses my headline clone:

I replaced the 470R resistors on the phono boards with 1W parts as suggested here and fitted a DC input socket to the rear of the case:

Project completed! Thanks for all your input