My_Ref Fremen Edition - Build thread and tutorial

You made the assertion that Rcore are more efficient.
Are you going to reply?
Hi Andrew,

My reply was that I have no experience with R-core, so I can't defend it. Also, no data. I was only referencing the fact that Dario had been successful with an R-core smaller than the recommended 160 VA toroid for the My_Ref.

I don't have enough knowledge to speak to how you could design for efficiency in a transformer. I have obviously chosen the wrong word, "efficiency", in describing how the choice of an R-core might be different from a toroid. I gladly withdraw the R-core efficiency comment.

Jac
 
Last edited:
radical new topic

I have been experimenting with passive line level xover (PLLXO) in FE to replace active opamp xover. This may be a real cool application to incorporate with or add on to any FE.

First issue: does changing input impedance (R13) do anything other than change the load on the source and the corner of C13? I changed mine to 10k and it doesn't seem to have any other effect, but I changed so many other things that I can't really tell.

Second issue: can I use a variable resistor in addition to R13 to make the corner frequency of the PLLXO variable? Possibly put a 10k in series with a 100k pot so the load on the source would never get too great? If this will work, can someone please offer a guide for hooking it up? 😕 I am not very sharp when it comes to pots!

Dario, there is a lot of empty space on the FE board. Here is a possible use for some of it. Add pads for a small pot, with a jumper trace to be cut in order to use it, to make the MyRef FE into a ready-to-use biamp with built-in first order variable filter.

Other suggestion: add pads for balanced input/single ended conversion. One opamp, few resistors = balanced inputs.

This kind of stuff fully exploits SMT and a large PCB. Does any of this work for anybody else?

Peace,
Tom E
 
Hi Tom,

Interesting project and what a acronym, PLLXO.

Your description is a great starting point, but I could use a little more clarity on what you have done in order better understand the issues. Is it possible for you to make some hand drawn sketches of your input section, before and after?
 
Very straightforward, and not yet any changes to the circuit, but only values: so far, I have replaced 100k R13 with 10k R13. That changes the corner frequency of C13 by a factor of 10. After also changing C13, I now have a fixed first order high pass filter on the input at, say, 500hz. Now I would like to be able to vary the input impedance (R13) so that I can adjust the corner frequency of C13. If I leave the 10k R13 in place, but add a 100k pot in series with it, the input impedance can be varied from 10k to 110k. Can that be done?

Or a 10k - 100k pot could be put in parallel with a 100k R13 and the impedance varied that way. That might be a more practical addition to the current layout. Builders could decide whether or not to use it by populating it or leaving it open.

I've never heard of an amp with a variable input impedance. With all the interest in biamping speakers, I expect that a first order passive input filter could be a useful tool. With a somewhat more complicated circuit, one could install a second order passive filter, which would be even more useful. But the first order is so simple, it begs for implementation. The input cap C13 already forms a high pass, so why not make use of it for other purposes, and make it versatile, as well.

Peace,
Tom E
 
OK.

If anyone else has data to show one, or other, then please post.

Hi Andrew, I already talked about it in the past.

I think the correct word is regulation.

The transformers were:

Noratel toroid 225VA 2x25V
Selectronic (Shilcar) R-Core 120VA 2x24V

Unloaded both sets to measured voltages around 26-27V, when loaded the Noratel voltage goes down under 24V while the R-Core sets around 24V.

I'm recalling values by memory, maybe I'm remembering not perfectly exact values but the trend was that one.

Listening to both the R-Core clearly sounded louder and bolder (the latter is probably simply an effect of the loudness curves)

At the end it's simply this, the higher the voltage under load, the louder it sounds. Quite logical, isn't it?

Hi finally finished My My_ref built into a Sherwood 5:1 amplifier chassis, took every thing out and now just has 2 My Ref's and a 300va toroid in, best sounding amp i have owned plenty of power even at 8 Ohms. :]

🙂

As already asked by Jac:

Pics, please! 😉

Dario, there is a lot of empty space on the FE board. Here is a possible use for some of it. Add pads for a small pot, with a jumper trace to be cut in order to use it, to make the MyRef FE into a ready-to-use biamp with built-in first order variable filter.

Hi Tom,

just in these weeks I've struggled with 7th run boards to better route return currents so to reduce further the amp distortion.

Trust me or not... space (lack of) is the biggest constraint...

When designing a PCB accounting for optimal parts placement, routing and return current paths it's all but trivial.

Other suggestion: add pads for balanced input/single ended conversion. One opamp, few resistors = balanced inputs.

I've already thought about it and maybe a future revision will include that but the hard part is not the opamp and related resistors/capacitors but routing the PS lines...

Probably a cleverer approach would be to simulate a fully differential opamp with two LM318, where the second one feeds the howland current pump branch that actually goes to ground.
 
Last edited:
Probably a cleverer approach would be to simulate a fully differential opamp with two LM318, where the second one feeds the howland current pump branch that actually goes to ground.

Dario,

Thanks for explaining the R-core. I could only remember that you were OK with a smaller transformer.

As for a balanced approach, take a look at the Twisted Pear Sympatico. Russ shares both the active components used and a schematic, so you can get an idea of how he achieved a balanced amp.

As you know, Russ White started building amps with the original MyRef. The balanced Sympatico appears to have evolved from a MyRef-like architecture and still has a current pump, etc. I'm not saying that the Sympatico is the right approach, just an interesting one to study when thinking about a MyRef balanced amp.

Jac
 
The input cap C13 already forms a high pass, so why not make use of it for other purposes, and make it versatile, as well.

Peace,
Tom E

Hi Tom,

Thank you for that. I'm clear now.

I have no expertise in amp design, so I can't address what you are suggesting from knowledge. You proposal does raise an interesting question. If you used R13/C13 for a high pass filter, what would be the affect on the amp at frequencies above the corner frequency. I think it comes down to noise performance based on R13 and the output impedance of the preceding signal source. If the output impedance of the "pre-amp" was high and R13 was low, there might be a degradation.

So what value of R13 would be an issue? Seems like 20k or higher should be no risk for most "pre". Is 10k OK? 5k? I vaguely remember something about a 10:1 impedance ratio as being desirable.

Another question for you. If you used a PLLXO for the high frequency drivers, that is, a high pass filter, what do you do for the bass speakers? It seems like you still need R13/C13 to be "standard" so that the amp/speaker are protected from DC.

Jac
 
Dario,
Yes, PS always complicates layout. Maybe there isn't as much room as I think I see. Have you considered eliminating the local LM318 caps that no one seems to use? That would free up some space. My suggestions are somewhat outside the box, I know, but I think a few features could be incorporated to make the amp more versatile, although they are certainly not requirements for most systems. Of course, you've done a great job. 😀 This is my wishful thinking. If I had the guts, I'd design the FE/TE version!

However, a lot of people are using smaller main speakers with a subwoofer these days. Why tax these little chips with reproducing the lowest frequencies when the speakers can't pass them (and in fact could be harmful by increasing speaker IM distortion)? If users could better integrate their mains with subs, that might be useful and sound better. 20 - 20k is a standard bandwidth for power amps, but why? I think it's a marketing holdover from tube to SS transition 50 years ago, when everyone had huge, full range main speakers. Sometimes it's needed, sometimes not, especially these days, and might sound better with less. Just some thoughts.

Jac,
Russ's Sympatico uses the circuit that Dario wrote about: differential amp using both sides of balanced input and two amps, one for plus leg, other for minus leg. This is what doubles the power output, as well. I am referring only to balanced input/SE conversion, similar to what your little boards do, not a fully balanced amplifier.

Of course, with variable input impedance, you would need to be aware of too low an impedance causing problems. Yes, 10:1 is the standard minimum. That's why a 10k resistor could be put in series with a pot, so the input imp would never fall lower than that. There might be some increased noise with lower input impedance, but I haven't noticed any at 10k.

Any value cap at the input will block DC, and any cap will cause some phase shift. It's only a matter of how much at what frequency. That's something that must be reckoned with using any type of filter. For a first order filter such as this, it's 90 degrees shift. That can cause FR anomalies, but making the impedance variable (I'm not sure it can even be done), and therefore the corner frequency, could help tune the filter to be the least intrusive. For low frequencies, most subs have a built-in xover, but it's only a low pass for their bandpass, not a high pass for the mains. For my bass bins, I'm still using an active low pass xover.

Peace,
Tom E
 
Originally Posted by AndrewT View Post
OK.

If anyone else has data to show one, or other, then please post.
Hi Andrew, I already talked about it in the past.

I think the correct word is regulation........
The question came up about efficiency. OK it turned out that you meant regulation.

But having introduced efficiency, I asked if any Member had any evidence to show whether EI, or C-core, or Toroid, or R-core have any differences in efficiency.

I proposed that since the Rcore has an excess length of core that is not "inside" a working coil, then it's efficiency could be slightly lower than a core that is completely enclosed inside a working coil. But that's just a guess based on iron losses about which I don't know too much.

Any evidence? Anyone?
 
Perhaps this of general interest, so I'll post my impressions of a couple more C13 trials. While my application of these caps is for passive crossover duty as well as AC coupling, the high/mid frequency aspects will be useful for full range as well.

Had a pair of Rel-Cap Exotica TFT's laying around, so in they went. First I gave them an extensive break-in, because teflon's are notorious for requiring 100 hours minimum to begin to sound decent. Initially, they were painfully bright and etched, almost couldn't listen to them. After 30 or so hours they began to smooth out. Now, after at least 100 hours, they have become very open and detailed with just a hint of sharpness on top. The midrange is still a bit lean, but what's there is very accurate and layered with nice depth. These deserve more time, so I'll leave them in, and it's fun to hear almost too much detail.

Also tried a Jensen polystyrene/aluminum that is on sale at PC. Ridiculously cheap for what it is, and impressively large for the small value. It just fits on the board, and only 0.034uf! Jensen no longer makes them, and now I know why. After a dozen or so hours, the highs are still disappointingly shrill and closed in, sounding like that awful SS white noise type of hissing instead of high frequencies of real instruments. On a lesser amplifier, these would be too awful to bear, but I'll give it a bit more time. Midrange is okay but nothing special, and there seem to be little or no spatial cues. Unless something changes significantly in the next few hours, I would advise to not bother trying them.

I learned that the input impedance can easily be varied with a simple trimmer resistor, so I have incorporated a 50k pot in series with a fixed 10K, both in parallel to R13. If the trimmer should fail for some reason, either open or short, there will still be input resistance. Putting it in parallel with R13 allows for finer adjustments to the passive xover frequency. Still playing with that. I can state that it is far better than the active xover with opamp.

Peace,
Tom E
 
C13 Evaluation Update

Since Tom is reporting some of his experience with C13 caps, I thought I would update my experience with the Cornell Dubilier 942c.

All of my initial evaluation were using the FE amp with my main speakers and a 1 uF cap as a coupling or DC blocking cap (C13). In that system, I liked the 942c and thought it was a good value.

Based on those results and enjoying the dynamic sound of the 942c, I tried them in another project. For the tweeter crossover (10 uF total) of some speakers I am building for a family member, I added a 0.68 uF 942c to two 4.7 uF 940c caps. In addition, I am updating at BrianGT gainclone amp for this project. I tried several coupling caps with that amp and the speaker/crossover above.

My result is similar to some of my past experience, that is, you can have too much of a good thing. The combination of the 942c in both the crossover and the amp coupling cap had the desired benefit of neutral frequency response and a dynamic sound, but over longer listening periods, the tweeter sounded a little harsh and tiring. It could be a difference between tweeter behavior between my main speakers and this project. But my experience is that, as in cooking, you need a mix of flavors to get the best result.

So what does that mean for you FE builders? If your system is on the bright or very detailed side, the 942c might not be the cap for you. If your system is neutral to warm and you want to add some dynamics to your sound, then it is an moderately priced cap that might be worth a try.

Jac
 
Hello,
i am wondering about the possibilites to feed a mono board using a single socondary transformer, using this configuration:

An externally hosted image should be here but it was not working when we last tested it.


Connecting one secondary winding to AC1 + AC2 and the other to NAC1 + NAC2.
Could someone please explain me why that wouldn't work well?
According to schematics, there would be a common virtual ground referring to NAC1 and NAC2, so it would originate from the sum of the currents of 2 windings from a single secondary, whose AC values have been fltered by diodes.
Wouldn't that be OK? And why?
If i should use a transformer with 2 secondaries, what would be the difference? Current would come from a single 230VAC primary as well, isn't it?
So, if i understand correctly, the situation should be the following:

An externally hosted image should be here but it was not working when we last tested it.


So connecting A to AC1, B to AC2, C to NAC1 and D to NAC2 wouldn't it be the same than connecting A1 to AC1, B1 to NAC1, C1 to AC2 and D1 to NAC2 (as suggested in the build tutorial)?
This way i would preserve that AC1 and AC2 are connected to the same primary winding and NAC1 and NAC2 to the other primary winding, just the same than using a dual secondary transformer.
Am i wrong?
 
Last edited:
Sorry, i made a bit of confusion...
This is the correct image:

An externally hosted image should be here but it was not working when we last tested it.


So the problem would be that using a single secondary transformer both AC1/AC2 and NAC1/NAC2 would share the same winding wires, thus making impossible a virtual ground?
Why is virtual ground impossible when both NAC1 and NAC2 share the same winding? Wouldn't common ground rail's current be blocked by diodes?
 
A centre tapped transformer has three wires and a dual secondary has four. There is no virtual ground.

The diodes create a short.
 

Attachments

  • aImmagine.jpg
    aImmagine.jpg
    42.3 KB · Views: 399