My_Ref Fremen Edition - Beta build/Fine tuning

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
In this thread we will discuss about the beta build of the My_Ref Fremen Edition and fine tuning of it.

I've attached the actual beta PCBs picture and schematic.
 

Attachments

  • IMG146 1024x768.jpg
    IMG146 1024x768.jpg
    233.3 KB · Views: 5,377
  • My_Ref_FE - Schematic.pdf
    46.5 KB · Views: 1,709
Last edited:
Member
Joined 2005
Paid Member
Some pictures of the build process.

I think the SMD soldering was quite easy and with good results.

The secret is to use thin solder and thacky flux ;)
 

Attachments

  • IMG159 1024x768.jpg
    IMG159 1024x768.jpg
    273.9 KB · Views: 1,194
  • IMG152 1024x768.jpg
    IMG152 1024x768.jpg
    268.7 KB · Views: 4,090
  • IMG150 1024x768.jpg
    IMG150 1024x768.jpg
    282.8 KB · Views: 4,286
  • IMG149 1024x768.jpg
    IMG149 1024x768.jpg
    311.6 KB · Views: 4,539
  • IMG147 1024x768.jpg
    IMG147 1024x768.jpg
    329.3 KB · Views: 4,990
Up and running!

I've finished the build and it works!

So, apart the poor manifacturing, the PCB design don't have any major problem.

It's a bit early to judge how it sounds but it seems promising :cool:

It seems to sound as good as the modded TP boards, in fact better, there is more detail and depth.

All components are brand new so I expect things to improve further with a minumum of burn-in.

Just one (incredible) note:

In pictures you can see the 390R Takman REX orientation.

In this config it sound strangely thin, I've reversed that single resistor (matching the orientation I've on TP boards, red band on ground) and the full sound I'm accustomed to got back...
 

Attachments

  • IMG171 1024x768.jpg
    IMG171 1024x768.jpg
    261.4 KB · Views: 1,891
  • IMG172.jpg
    IMG172.jpg
    545.8 KB · Views: 1,646
Member
Joined 2005
Paid Member
Thanks

The amp now sound gorgeous, just a problem with one channel.

It has a strange electric noise when volume is very high, like there is arching between tracks.

In fact the corresponding PCB is the worst among the batch, pads and copper are not well aligned.

I've checked visually all the others and they seems better.
 
Beta PCBs ready for shipping

Beta PCBs are ready for shipping, tomorrow mornig, that it means in 10 hours, packages will be shipped.

I can't wait beta testers opinions and suggestions! :)

In the meanwhile I was puzzled by some strange artifacts I was hearing...they disappeared when finally I've connected PGND of each module to the star ground (which is connected to safety earth).
 
Member
Joined 2005
Paid Member
Packages shipped

Packages shipped.

You'll find, apart PCBs, Elnas, zeners and partial kits (for whom asked them) also some thin solder (Sn/Pb/Ag) for SMD soldering.

I've highlighted with white paint the manifacturer 'fixes' on PCBs, take maximum care on that sections, consider them as without the solder mask and after soldering check with a multimeter for shorts.

The ground plane connections are more difficult to solder, thermals width too large (it will be fixed in the final boards). Use a higher soldering temperature on those.

Don't forget to short LM3886's ground to PGND or 0V using a solder bubble on the solder jumper (see attachment).

Please let me know when packages arrive.
 

Attachments

  • IMG152.jpg
    IMG152.jpg
    98.3 KB · Views: 1,707
Last edited:
In the meanwhile I was puzzled by some strange artifacts I was hearing...they disappeared when finally I've connected PGND of each module to the star ground (which is connected to safety earth).
So, before that, the "GND" potential was floating, with respect to Protective Earth? If so, any noise current between your multiple, seperate (?) transformers was forced to mainly flow through through the 1R's, then through the signal cable shields to the point when these combine at your source device (CD?) for then to possibly interact with the source's own GND system.

EDIT: I see something even more suspicios, you used full bridge on each rail. So any leakage current gets chopped, it only flows when the diodes are conducting.
 
Last edited:
So, before that, the "GND" potential was floating, with respect to Protective Earth?

Exactly

If so, any noise current between your multiple, seperate (?) transformers was forced to mainly flow through through the 1R's, then through the signal cable shields to the point when these combine at your source device (CD?) for then to possibly interact with the source's own GND system.

You can see in the attached picture how things are connected.

Neither the DCB1 or the SACD player are grounded to safety earth.

EDIT: I see something even more suspicios, you used full bridge on each rail. So any leakage current gets chopped, it only flows when the diodes are conducting.

It sounds like a bad thing... Am I right?

But it's like all Gainclone's PS around here...
 

Attachments

  • IMG174.jpg
    IMG174.jpg
    141.2 KB · Views: 1,396
Double bridges are not problematic per se, but it can become a problem when the chopping action finds a way into the signal path. The point is that chopped hum current (assuming 50/100Hz hum as the dominant noise) is worse than unchopped when it makes itself apparent in the signal, via a voltage drop accros the 1R resistors. For every mV drop you'll get 30mV at the output.
 
I was also wondering about the grounding. Simple circuit simulation shows not connecting to protective earth ground is a problem. But if you put the power supply into consideration, the speaker current flows through the power ground. Although I do not have problems because of the power I use, but I have not looked at the linear power used here. If there is a problem with not connecting ground to earth, then it would be a problem for areas that do not have earth ground. We did not require earth ground here until the last 10 years or so.
 
The point is that chopped hum current (assuming 50/100Hz hum as the dominant noise) is worse than unchopped when it makes itself apparent in the signal, via a voltage drop accros the 1R resistors.

I doubt that the eventual ground noise is 50/100Hz related beacuse it is not constant, it's not a noise, it seems more a signal correlated artifact, if it exists at all.

Yes, if it exists...

Since this is my first project I'm, maybe, a bit paranoid and I tend to relate any problem to the project.

In fact I've listened to my player via headphones and at least part of the problem seems also present.

In the next days I'll do more listening sessions and experiments to confirm it or not.

It seems that connecting PGND to the star ground only masked the problem since in this way it sounds fuller and restricted thus partially masking the artifacts but this evening I've heard them again.

BTW it seems to sounds better (airy and natural) if left floating (PGND not connected)

I was also wondering about the grounding. Simple circuit simulation shows not connecting to protective earth ground is a problem.
...
If there is a problem with not connecting ground to earth, then it would be a problem for areas that do not have earth ground.

Let's see what are further tests results... maybe it's only me...

BTW what problems you found through sims?
 
Hi frequency resonances. The higher, the larger. This may not be what happens in reality, but does indicate something needs to be dealt with.

Since our APT does not have a ground wire on normal mains, I was amazed that I did not run into much problems, but one channel seems to have some high frequency resonance after playing some time, and I have to shut it down and power back up after it cools off. The resonance seems to happen only in the channel with the original LM318 compensation.
 
Last edited:
Hi frequency resonances.

Interesting

The higher, the larger.

I suppose you mean higher the frequency, larger the resonance, right?

This may not be what happens in reality, but does indicate something needs to be dealt with.
...
one channel seems to have some high frequency resonance after playing some time, and I have to shut it down and power back up after it cools off. The resonance seems to happen only in the channel with the original LM318 compensation.

So it seems to happen also with the single diode bridge... if we're talking about the same problem.

With or without C12?
 
Yes, higher the frequency, larger the resonance. I did not notice artifacts, but it was noticed when I was switching channels comparing modification differences between channels. The funny thing is it can only be heard when I swap the channels. I operate without C12 all the time now, and also took away the high frequency filter. Since I just modified the input section, I will see if the problem persists or not.
 
Last edited:
Just one (incredible) note:

In pictures you can see the 390R Takman REX orientation.

In this config it sound strangely thin, I've reversed that single resistor (matching the orientation I've on TP boards, red band on ground) and the full sound I'm accustomed to got back...

OK, not to flame on your parade, but are you implying that the orientation of a resistor makes your amp go from "strangely thin" to "full sound"?

If so I have some cables and crystals that will really wow ya.
 
OK, not to flame on your parade, but are you implying that the orientation of a resistor makes your amp go from "strangely thin" to "full sound"?

Should be read as thinner than usual / full as usual, not a big difference but one I can hear.

Please don't start the n-th war on resistor orientation/cables/capacitors/etc., simply ignore my comments on it if you can't hear/don't trust such things.
 
OK, not to flame on your parade, but are you implying that the orientation of a resistor makes your amp go from "strangely thin" to "full sound"?

That's a typical symptom of either a dry (non-ohmic) joint on one of the resistor leads, or a flaky resistor with a cracked/loose/non-ohmic end-contact. Reversing it probably fixed the dry joint or made the end-contact sit snugly, thus solving the problem - probably not related to the resistor orientation itself, though that's also possible (most resistors are slightly non-ohmic - when measured with a high-precision DMM, they'll differ slightly in value when measured in each direction. I typically see a few ohms difference on "good" 22k metal films, but that could be 10s of ohms on carbon films, and even more when they're flaky).

Note that this is a relatively low-value (390R) resistor, where rectification artefacts due to non-ohmic/dry joints could well be audible. It could also account for some of the hum/crackling artefacts reported - in my experience, the MyRef is sensitive to dry joints at several locations: the Howland resistor networks, the GNFB resistor network and the lift resistor R11 come to mind. Actually, every solder joint in and around the small-signal section/LM318 should be re-checked and re-flowed if it looks suspect.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.