My_Ref Fremen Edition RC - Build thread

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
This method is simple and foolproof. If you hear a difference after the parts pass signal for a significant length of time (hours don't count--days at least), then orientation might be a factor.

Tom...

this method it's not so simple, IMHO.

In addition using mono amps to compare doesn't work, at least for me. Spatial informations are a not small part of the differences when reversing parts and phase accuracy and interaction betweeen channels can also influence other parameters like timbre, bass tightness, etc.

My tests and evaluations took me months, I'm sure of my (personal and subjective) results.

Other builders, using their personal evaluation methods, obtained similar subjective results.

If your're skeptic about orientation effect and the method you suggested satisfy your goals you should try it and find for yourself if orientation has merit for you.

You can try it with your method using a single cap, C7 will do, as any other FKP2 on the RevC.

If you want to test with resistors, since you're using PRPs, simply reverse the input resistor.

Since I value you as a competent listener I will read with interest and pleasure your findings about it, I'm less interested in speculations. ;)
 
Jac,

You have two monoblocks? That's two amps. Use one with optimal part orientation, the other opposite.

I've never tried them because I get a little spooked buying parts on eBay from Russia.

Peace,
Tom E

Tom,

I agree with Dario about the stereo effects being important, but I'm a sport and I'm in the right place in the project to experiment. So here is what I am going to do. At the moment, I have the Russian PIO + Amtrans bypass as DC blocking input caps and on fastons. They are in the orientation that I preferred before burn in and they are sounding pretty sweet. To honor Andrew's request, I'll let them play this way until about Dec 10. At that point, I'll evaluate orientation, keeping the combined cap as a unit, swapping both channels at the same time in an A-B-A comparison. If there is no difference, then I'm done. If there is a difference, then I'll try your test (each monoblock in a different orientation). I will follow your procedure and evaluate the monoblocks, both orientations, immediately and after one week. Finally, after comparing them to the styrene caps again, I'll try pulling the fastons out. Does that sound fair?

In the meantime, if you have a Paypal account, you are pretty protected when buying from Russia and Eastern Europe with Paypal buyer protection. I've done 2 orders from valtek in Bucharest and one from uralspirit with no problems. The only thing is to expect 3 weeks plus before the package arrives in the mail. In fact, I just put in an order for some FT-1 teflon bypass caps based on the recommendation of Jon L on Head-Fi.

Jac
 
Last edited:
Hi Guys. Missed yesterdays conversations as I lost internet connectivity late Saturday afternoon and just got back online. It's amazing how much can be accomplished around the homestead when one is rejected by Mr. Google. Who Knew :scratch2:

I missed the the opportunity to do some of the tests suggested and I should have thought about them on my own. I did complete the socket removal but with all the mess this weekend, haven't done a whole lot of listening. Hopefully my "Audio Memory" hasn't faded completely and I'll still be able to offer some valid information later today.

Fortunately, I retained the C13 fastons because I don't want to trim and solder pot the True Coppers till after the final board/design is built and tested. I'll take Dario and Tom's lead as to whether that's a wise move. As Tom suggests, when I got the TCs I cleaned the individual strands of the Litz leads mechanically as best I could, burned them off with a small butane torch and then soldered with lots of flux and the same torch.

I will take some time and extra care to do both the solid solder and Tom's pigtail alternative tests for the C13 faston removal using the Audyn Plus caps. Probably won't get to that till tomorrow.

A couple interesting findings from the past few days:

1. I mistakenly reversed the orientation of C-32 when mounting without the sockets. There was a notifiable increase in bass that disappeared once I corrected the error (the writing on the Amtrans cap oriented to the label on the PCB ). Had I not trashed the sockets, (a real pain to clean the vias), I would have done more testing - but due to the more than adequate low end from the TCs, I left them in the suggested orientation. That's not something that pleased me as it just further complicates the understanding of the interplay between components. I have complete confidence in stating changes in orientation can be easily heard, but the dynamics and reliability/repeatability of such choices continues to gain complexity.

2. For the interconnect hum I get on the RC and not on the beta - there was further confirmation that it is ground related. A quick recap: Any source can be connected to the beta build without a problem. Only the Mini 2496 DAC and the battery powered Walkman - with direct connection to the RC is clean. What I found yesterday is adding the self-contained Alps pot rig i made, (with all grounds and metal case acting as one big "Star" ground), introduces the hum to the RC build. The AC powered sources all have acceptable DC readings. The alps rig does not introduce hum when used with the beta build. Still a mystery but I suspect/hope it will magically disappear on the next build.

3. I should have bought the de-soldering tool Dario recommended a year ago.:cuss:

BONUS: If you really want to jump off into the deep end of the Phantom/Real wire and metallurgy universe - Check This - A little birdie gave me the link. :rolleyes: Too simple and tempting to not try ???
 
Last edited:
No Sockets Transition.

This change was more significant than I would have expected. The very first impression was "muffled". I actually did that 'shake my head" thing that does some kind of ear tube equalization. Keeping in mind and trying to resist the numerous claims of "hearing what one assumes", the change in sound was definite.

I you go to a good stereo shop to audition speakers the lower priced units might really impress/grab you. As you move up the scale the improved quality seems to come from the removal of artifacts in the sound and not a perceptible increase in loudness or impact. That's what I heard with the sockets removed.

It wasn't all that simple though. I got the impression of a more homogenized and stable presentation but a bit of the shimmer and forwardness had gone away. The side to side stage felt slightly more solid and defined - not a lot but hearable. The upper end bells, cymbals, fret noise, flutes, snares and such were clear and distinct. A little more immediacy and blend was perceived in the electric/acoustic bass - kick drum area.

My test mix has two keyboard tracks that have some static at dynamic peaks that I suspected could have been caused by poor connections at the sockets. It remained with the soldered components so is clearly in the recording.

What was very noticeable was a new graininess in vocals and solo instruments. It was pronounced enough to make me suspect an incorrect orientation at C-13. Instead of rotating the TCs I let it ride, and the majority of that element smoothed out somewhere between 45 minutes to an hour. In my mind that confirms Tom and Andrews claims dealing with significant impact on sound quality from very small changes, in very small pieces of various metals in the signal path. It seems there is always an internal adjustment between/among board components - Ex: on the level of the problems CPU designers found when the traces got so small the electron flow was influenced. Don't know all the science, but I'm sure electron flow is just as important for audio equipment. What I heard from the change in sound quality in that first hour appears to support Andrew's claim that initial "forming" or "burn-in" are the most significant at introduction to the circuit.

I'll still use the IsoTek process and listen for changes over time, but I am now fully aware of how the use of sockets is truly an interim technique.

This morning I did do the C-13 twirl dance for good measure. Still the same - the orientation in the photo has less bass and impact but produces an overall smoothness that will most likely pay off with less ear fatigue.

Today its Tom's "Pigtail Day" :D
 

Attachments

  • Nos4.jpg
    Nos4.jpg
    185.1 KB · Views: 236
  • Nos1.JPG
    Nos1.JPG
    152 KB · Views: 223
  • Nos2.JPG
    Nos2.JPG
    135.8 KB · Views: 208
  • Nos3.JPG
    Nos3.JPG
    148 KB · Views: 203
As you move up the scale the improved quality seems to come from the removal of artifacts in the sound and not a perceptible increase in loudness or impact. That's what I heard with the sockets removed.

What you heard is, in part, brass signature removal.

Sockets are good for testing parts but the amp sound better without... ;)

What was very noticeable was a new graininess in vocals and solo instruments. It was pronounced enough to make me suspect an incorrect orientation at C-13. Instead of rotating the TCs I let it ride, and the majority of that element smoothed out somewhere between 45 minutes to an hour.

It's the time usually taken by solder joints to settle.

Another reason against tests done with soldering/desoldering work in small intervals.
 
cheers Bob,

pretty impressed with your results!! Please let us know what is the rest of you system for testing, I wonder if I could find any difference by removing dozen of sockets on every pcb...

Btw, seems to me you have some sophisticated copper/aluminum (titanium :]] ) heatsinks, does it improve something as well?
I was planning to install two thin metal plates (0.5-1mm) between pcbs and toroids as they are pretty close now. Initially I thought about decent aluminum plate, then after some googling found that it should be copper or iron at least (will do but it's no use for low frequency noise from transformers) and Mu-metal at most (Nickel, iron and some other stuff here). So I decided to stay with copper, because generally I do not hear any noise already as it's now. Any suggestions in you copper/aluminum combination?
 
dtes,

The sockets were always intended as a temporary addition for component selection,testing and tuning. It is expected that all final builds will none for optimal operation. My plan is to have only the board power connections not soldered directly to the PCB eventually.

I honestly did not expect such an overall sweetening of the amp by just removing the sockets. I ran the Isotek process a few times last night and the system is simply stunning. Just finished listening to a Yo Yo Ma / John Williams orchestral collaboration. All the nuances of the full range of the cello were perfect, and when some dude hit what sounds/feels like a 20 foot diameter bass drum - my entire house vibrated. :D :D

My test system is very simple:

1. Jriver Media Center on an Asus AMD six core server with 5 GB USB external drive.
2. 16 foot USB, BNC S/PDIF and optical cables.
3. Mini2496 DAC (have Wielang DAC5 build also).
4. FE monoblock amps.
5. DIY Cat6 speaker cables.
6. Sunflower /open back mids speakers.

The only problem here is the need to pass the music through the Windows 7 audio drivers for volume control. If I can resolve my interconnect hum problem, I would prefer to use straight WASAPI Event Style and a bare pot for a cleaner feed. I always test with uncompressed FLAC files.

The intrinsic nature of the metals used has also renewed my interest in investigating brand, type and length choices for all internal wiring for the FEs installation in a chassis. It is well known as "The Wire Wars" and can lead to some ridiculously long and useless dialog. I don't intend to dilute this thread, but will make several PM/email requests to veteran builders for advice.

For cooling - I bought a bunch of those heat sinks on ebay several years ago. They are for Intel Xeon based servers. Similar pieces can still be found with a search. Mine were something like 6 for $30. I have not detected any hum from close mounting of the toroids in the BLAT build. I used 1/4 inch aluminum because I have a lot of it in my stash. I was prepared to buy mu-metal if necessary but discovered it wasn't. Were there a problem my next choice would have been 16 ga. cold rolled steel. I also took a clue from looking at the JC-2 build on the Parasound site. They use R-Core transformers (which I hope to own someday) but the thick aluminum appealed to me. If you still get just a warm touch on your compact chassis there probably is sufficient heat transfer. I still plan to do some performance/temperature readings on the liquid cooled build - time permitting.

Related - - I am trying to set up RMAA and/or other diagnostic software before I remove the fastons from the caps at C-13. I'm not shooting for anything as sophisticated as Jacs attempts, but hope to be able to present something more than my usual verbiage. Shooting for the weekend.

Major new project a la madisonears in development ;)
 

Attachments

  • jc2_inside1_big.jpg
    jc2_inside1_big.jpg
    113.4 KB · Views: 169
  • DSC00847.JPG
    DSC00847.JPG
    96.4 KB · Views: 110
  • DSC00848.JPG
    DSC00848.JPG
    116.1 KB · Views: 89
Last edited:
Bob's Hum

1. Jriver Media Center on an Asus AMD six core server with 5 GB USB external drive.
2. 16 foot USB, BNC S/PDIF and optical cables.
3. Mini2496 DAC (have Wielang DAC5 build also).
4. FE monoblock amps.
5. DIY Cat6 speaker cables.
6. Sunflower /open back mids speakers.

;)

Bob,

Just had a thought about your hum. Not sure why you would have it with one build of FE and not another. Anyway, if I understand it correctly, your server is 16 feet or so from your DAC and FEs. I am assuming that you have to plug the server and the DAC/FE into different wall power sockets. The Neutrals/Grounds in the wall are some distance apart and have finite resistance between them. Since the USB has a ground conductor, the USB + wall ground form a rather large ground loop.

There is no easy fix for this.

Having everything plugged into the same place might help. You could try running a power extension cord to plug the server into the same place as your audio. Unfortunately, the extension cord will also have resistance.

Check out this link for a discussion of this issue. Don't worry, it is not too crazy technical.

Ground Loops

I have a similar situation but send my signal by wifi using a wifi router plugged into the computer and a Logitech Squeezebox Touch at my audio system. By the way, this was a very nice improvement in sound quality over my old CD player, but I ended up using a separate DAC and the digital out from the Squeezebox. By using wifi, I am able to plug all of my audio system into a single power strip and minimize my ground loop.

Jac
 
Jac, you're reading my mind waves. I really appreciate you considering a fix, but the things you mention were checked early in the struggle. I have had a home A/V studio (I produce a few local TV commercials) with multiple computers and a rack of audio boxes for several years. I drove myself crazy back then and understand the loop complications. The switch to digital connections has greatly reduced those glitches.

Sunday I set up the complete system on a table and used short (6 to 12") interconnects. That didn't help. As I write this I'm finishing a little rig that might help the diagnosis. It's a bare SMD pot with solid grounds on one end and loose grounds on the other. There is also a jumper available to check if the L/R grounds need to be common.

Now I have to move everything back to the table and do some poking. Hopefully I'll have something good to report in the morning.

Thanks again....
 

Attachments

  • Hummer.JPG
    Hummer.JPG
    127.2 KB · Views: 314
I would say both, refer to this thread for project goals.

tuyen,

I would say that Dario is being modest. I can't find the exact post, but bcmbob has built both, loved the ultimate MyRef C, and found the Fremen Edition to be a clear improvement. Bob even put his FE Beta amp up against a $5000 tube amp owned by a friend. He said it didn't match the tube amp everywhere, but he was very pleased with the FE, even without considering costs. Both of those posts are in the Beta build thread, if you can find them.

Jac
 
Let me respond by simply showing my path.

Started with the ChipAmp BrianGT. I was amazed that it sounded as good or better than my Carver M-1.0t, Hafler 120 and a $1700 Denon AVR-2807. There was not as much gain as those three, but the price/performance ratio was outstanding.

Next was the Twisted Pear MyRef V1.2. The gain was about the same but it had a more relaxed sound and was a little less forward. At this point (at the risk of offending anyone's music choices) I got the impression the BGTs (and a set of BPA-150s I built) might be great for pop and party music, but the MyRef would do a better job with classical, small chamber and jazz combo offerings. Both amps are great and I would have been happy with either.

Linuxguru introduced the V1.3 which I viewed as a refinement of the TP design. The stock kit was very close to the previous version but seemed a bit more solid and detailed. The addition of his LF0-1 module removed almost all of the upper register graininess/edginess so often reported of the LM3886 chip itself. This is also when I moved from a 120VA transformer to 200VA. That brought the gain (bare pot and/or LDR attenuator) close to a point where no preamp was necessary. Again, I could have stopped right there and been satisfied for years.

When I completed the beta version of Dario's Fremen Edition I had to restrain from stating it was a quantum step beyond anything I had owned. I eventually did post that evaluation. That was the stock BOM with the old 120VAs, but the clarity, stage, sweetness and definition blew me away. Won't go into great detail but better main caps, higher grade resistors and small caps along with two new 300VA toroids created a truly impressive build. This is where I took the FEs to a local audio dealer and a friend with a high dollar tube amp for comparison. The results were very satisfying considering I had ~ $625 invested in the pair and the others cost thousands.

The RC version continued to refine the sound and smooth away that last bit of LM3886 edge on the top. Further testing and selection of components in the signal path continues and a real "Cloud 9" power amp is developing - IMO. With high quality sources and good speakers, the FEs will convince you the performers are in the room, be they a small folk group, large orchestra (with pipe organ) in a cathedral or any style and venue you fancy.

The only thing I haven't heard and most likely will try (for comparison) this spring is the Pass F5 house heater. I suspect someone like madisonears might even talk me out of that.

Any of the amps in this post are super projects for the DIYer. It's just a matter of budget and how much "Hi" you want in your "Hi-Fi" ;)

P.S. Linuxguru is developing a dual chip design that looks very promising and will probably fit well in the top choice LM3886 power amp group.
 
Last edited:
Cool! Thanks guys :)

The FE sound like it may still be worth the effort to try later on, even though I have the myrefC ultimate bom + lf-01 coming!

I'm looking for basically the ultimate sounding low powered (10-20watt is enough) ss amp possible. diy by design so as to save $$$. This My_Ref FE seems like one to put on top of the list. I have been using a modded TA2020 chip amp.
 
Hummmmm?

Bob,

You may already be doing a useful experiment for all of us who have yet to "box" our FEs. If I remember correctly, you were having a bit of a hum from your FE RC but none from the FE Beta. I also recall that you have 300VA shielded torroidal transformers for one and conventional, non-shielded torroids for another.

I was curious if, either boxed or boarded, transformer shielding made any difference? Have you already run the experiment of using both the RC and the Beta with the same transformer and box setup? Or both transformers with the RC? If so, is there any correlation to hum and transformer or board?

I note that Dario uses the R-Core transformers which are likely to be better for radiation.

Has anyone else built their RC into a box and been clean for hum? If so, can you detail your ground layout? Actually, the amp that Dario built for his brother may be a clue. Does that one have conventional torroids and no hum?

As for me, I am still no hum and on a board, but I am working on the box, rereading Douglas Self, and thinking about ground loops.

Jac
 
I note that Dario uses the R-Core transformers which are likely to be better for radiation.

Yes they radiate less but I don't think Bob's problem is related to the type of transformer.

Actually, the amp that Dario built for his brother may be a clue. Does that one have conventional torroids and no hum?

My brother's amp uses the same R-Cores as mine, both works with no hum, attached to AV receivers, DCB1 and simple pot.

Jac, can you measure voltage between PGND and safety ground?
 
Jac, can you measure voltage between PGND and safety ground?

Dario,

Just for clarification, I do not have any hum and I have the FEs mounted on a wooden board. That means, no connection between safety ground and PGND. But it was still interesting to measure, so...... with the amps on and connected to the DAC + bare pot, I measured 0.4 mV DC, then switched to the AC channel and measured about 10.4 mV AC. This was the same for both channels and stayed the same with music playing or not.

At this point, I am just researching and trying to understand proper grounding techniques. If my math is correct, 10.4 mV is only about 20 dB below a typical input signal, so connecting PGND to safety earth might be expected to generate a hum.

If we don't connect the FE amp to safety ground, but the chassis is properly grounded, then safety should be served as long as the box stays closed. I'm not sure this is true with Class 1 transformers. I need a better understanding.

It looks like the FE and other MyRefs refer to ground through the shield of the input signal via R11, so the DAC or other source provides the reference to ground. That assumes isolated single end connectors at the chassis.

If you follow Andrew's suggestion, then the input shield connects to chassis ground and, if you have a high quality star ground near that point for the safety ground, then you should minimize the hum risk and reference earth within the amp. This approach is recommended by Douglas Self. In neither case would PGND be connected, although it might be interesting to see what happens if you ran a wire from PGND to the star ground.

As I say above, I am just trying to better understand what is happening so that I can build a boxed version that stays quiet for hum.

Jac
 
Jac,

Dario Is trying to help with some unexpected voltages on my builds - possibly leakage or some defective parts. This is what I'm getting:

Neg probe to PGND
Pos probe to safety/earth Gnd.

RC Build
Left AC= 45.9V-----Right AC= 27V
Left DC= +200mV --Right = +158mV (these readings keep going up - stoped at 300)

Beta Build
Left AC= 88V ------Right AC = 88V
Left DC= -58mV ----Right DC = -47mV

These are both on boards. Both play music perfectly if correct source is used. I don't think it is a no-metal-chassis problem.

?????????
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.