Sorry - been away for a while. To clarify:
In Mauro's Rev A, C22 was a 100nF snubber cap that bypasses the AC rails of bridge rectifier on one channel, while C5 served the same function on another channel. If you're using a single trafo for both channels, it suffices to use just one of these. Also, the design that ended getting the most traction was the Twisted Pear Monoblock Rev C, so there again, the C5 designation was retained and C22 was dropped.
When I revised the Rev C ver-1.4.x monoblock layout ~5-6 years ago, I was unaware of the prior use of the C22 designation in Mauro's earlier designs, and I accidentally reused the C22 parts designation for a different cap in the Ver 1.4.x PCBs - it bypasses or substitutes C4 as described above, and it's optional. In practice, I have no recollection of having ever used it, so it's probably best to drop it in future iterations.
On a related note, a metallized polyester seems to be the sonically best option for C5 - the modest ESR of a metallized polyester seems to help the snubbing function. I found either Wima MKS3, MKS4 (PETP) or ERO KT1822 to work well at this location, though there are undoubtedly numerous alternative choices available that will work just as well.
I'm not sure that the R43 designation above is the same as what is used in the monoblock boards (Twisted Pear v1.2, and my v1.3 and 1.4 which are derivatives of the Twisted Pear v1.2). There can be some confusion in parts designations across Rev A and Rev C layouts, especially when talking about some early Rev A dual monoblock designs on a single PCB. To avoid confusion, I'm talking only about the v1.2/v1.3/1.4 single monoblocks below:
R43 is in the LM318 inner feedback and compensation loop, and the value is 22k. It should have no effect on the DC speaker-protection relay circuit. It's purely related to AC stability and compensation. 22M is definitely wrong, and it would be almost an open circuit with no effect on the AC feedback - apart from some stability and sonic changes, it would have had no effect on the DC speaker protection.
R14 is nominally 180R, reduced from the value on the Twisted Pear v1.2 board, to allow lower rail voltages and trafos compared to the original 24-0-24 VAC spec.
R21 is in the speaker-protection circuit, again reduced to 100k nominally, a bit lower than the 220k on the Twisted Pear board, to allow reliable operation at lower rail voltages.
Everything else in the speaker protection circuit is the same, and in practice I've seen no problems from the slightly altered operating points - it was mainly intended to allow reliable operation with 15-0-15 or 18-0-18 trafos, while still working fine even at 24-0-24 or higher.
If you find any issues at higher voltage rails, you can always revert to the original Twisted Pear values for R14 (470R?) and R21 (220k).
To repeat, all of this applies to the Rev C monoblocks in through-hole PCBs (Twsited Pear v1.2, and my v1.3 and 1.4). The Rev A and dual-monoblock 2-channel speaker protection circuits probably differ in some significant ways, including parts designations.
The Mains Bulb Tester (MBT) works as protection at first power ON.
It does NOT protect against abuse after power is established.
Use the MBT as a check for your wiring and connections at power ON.
After this has been proved, then power ON direct from the mains.
Last edited:
Andrew, It's a temporary safety measure against high load / current conditions till I fix the speaker protection. It's in series with the power strip because there's another 4 channels in building and several other transformers I try. It does also act as a good load indicator and did give me a few seconds to quickly turn the power off. I think I could have fried something otherwise.
Btw does the speaker protection protect from shorted outputs?
Btw does the speaker protection protect from shorted outputs?
Last edited:
The MBT does not protect against abuse after power is established.
The MBT works very well to protect against wiring faults during the power ON phase.
probably not.
The speaker isolation relay opens when the detector loses mains supply and/or when high level DC is present and/or when very high level very low frequencies are present at the output.
Shorting the output does all the damage before the detection circuit has even seen any anomaly.
Shorting the output pulls the output voltage to near audio ground. This leaves the detector with near enough no signal to measure.
The primary protection from MBT is safety of the in-house mains wiring.
Also the MBT acts as a soft-power-on. But it also acts as a high load indicator. If the transformer etc. is shorted, the bulb will be on full glow. I doubt it "protects" the circuit but it does give you a few seconds to get lucky and turn things off.
When the speakers got shorted it was on full glow and indicated to me that a powering off was required.
Like you said, the speaker protection doesn't protect from output shorting. In that case too much output current for too long a time would either have fried the LM3886 or the transformer itself.
I once connected a 12-0-12 DC supply in reverse to the OPA2134 crossovers. It glowed for a few seconds. It did fry all the 6 OPA314 s.
I don't know if you mean anything deeper or some other sort of behaviour of the MBT but I think I covered the long and short of it?
Also the MBT acts as a soft-power-on. But it also acts as a high load indicator. If the transformer etc. is shorted, the bulb will be on full glow. I doubt it "protects" the circuit but it does give you a few seconds to get lucky and turn things off.
When the speakers got shorted it was on full glow and indicated to me that a powering off was required.
Like you said, the speaker protection doesn't protect from output shorting. In that case too much output current for too long a time would either have fried the LM3886 or the transformer itself.
I once connected a 12-0-12 DC supply in reverse to the OPA2134 crossovers. It glowed for a few seconds. It did fry all the 6 OPA314 s.
I don't know if you mean anything deeper or some other sort of behaviour of the MBT but I think I covered the long and short of it?
Can anyone help me understand "feelses"? How does the amplifier interact / behave with the load?
After 11(!) years i have finally finished my amp...... Finally having a home studio i decided to give this build a final try, only to find that i had connected the powertransformer incorrectly....oh well....
I coupled the amps with the Darwin source selector and and Joshuatree volume control.
The amp pushes some diy speakers and i really like the sound: detailed and it pushes some low end.
Thanks Mauro ande the guys at Twisted Pear!
I am so sorry!
I want to bump this up. Recently I changed speakers. From 25 year old Thiel CS-3.5 to new Klipsch Heresy iii. The impedance is close, around 5+ohms for both as a minimum impedance. The efficiency difference is large, 87 dB for the CS-3.5 and 99 dB for the Heresy iii.
I wondered how my Evo single with the last mod would sound at several milli watt output. The linearity and tone are way better than expected. The graphs and measurements done by George are vindicated. These amps are spooky quiet and deliver dynamics at sub watt output. Used with Lowther’s many years ago. But the coloration from the drivers masked how well the amps worked. Plus maybe the 15 ohm impedance was a little issue.
With a more normal load and very low distortion speaker the amplifier is so transparent. The sound is easily the best I have heard.
I wondered how my Evo single with the last mod would sound at several milli watt output. The linearity and tone are way better than expected. The graphs and measurements done by George are vindicated. These amps are spooky quiet and deliver dynamics at sub watt output. Used with Lowther’s many years ago. But the coloration from the drivers masked how well the amps worked. Plus maybe the 15 ohm impedance was a little issue.
With a more normal load and very low distortion speaker the amplifier is so transparent. The sound is easily the best I have heard.
I will order one set of MyREF !
I will test them on my Earl Geddes (Gedlee) Abbey loudspeakers.
The Abbey's are a bit bright and depending on the music played can be irritating at times. So far Usher R1.5 Amp was the best suitable amp for the Abbey's, but started looking for something more smooth and with less power consumption.
The MOD-86's I build a few months ago are extremely transparent, and amazing sound staging on the Abbey's. But being a bit more bright than my Usher R1.5 Amp, this is to much of "good thing".
I build the Sympatico hoping to get a more smooth sound. And OMG .. Sympatico is extremely smooth. Exactly what the doctor ordered. No more overly brightness from the Abby's.
First impressions was some loss of details and sound staging, but listening for a few days now I think that is not the case. It's just different sound than MOD-86. Very natural sound. Nothing nasty stands out anymore, allowing to "just enjoy the music".
I fail to understand why the Sympatico sounds so much more smooth than the MOD-86 on the Abbey's. Any theories from you guys ? Earl Geddes has studied the audibility of distortions types and found amplifier "crossover distortion" most audible to the human ear. Is the Sympatico symmetrical design canceling out most of this ?
I am very interested to know what the MyRef FE will do in my system.
The Abbey's are a very special design using : 12" B&C driver 80-1000 Hz. Passive cross over to a B&C 1.75" compression driver on Earl Geddes Waveguide.
The Abbey's are a very easy load with 8ohm and 95dB @ 1 Watt @ 1meter.
No problem for any of these Chip amps.
The Abbey's are completed with multi subs (according Earl Geddes advice).
Resulting in even less load on the amps for the Abbey's.
Lets see which of these chip amp designs is most suited in combination with the Abbey's.
So far the Sympatico amp is best suited in my system. But would like to get some of the MOD-86 spectacular sound staging & detailing back.
I will test them on my Earl Geddes (Gedlee) Abbey loudspeakers.
The Abbey's are a bit bright and depending on the music played can be irritating at times. So far Usher R1.5 Amp was the best suitable amp for the Abbey's, but started looking for something more smooth and with less power consumption.
The MOD-86's I build a few months ago are extremely transparent, and amazing sound staging on the Abbey's. But being a bit more bright than my Usher R1.5 Amp, this is to much of "good thing".
I build the Sympatico hoping to get a more smooth sound. And OMG .. Sympatico is extremely smooth. Exactly what the doctor ordered. No more overly brightness from the Abby's.
First impressions was some loss of details and sound staging, but listening for a few days now I think that is not the case. It's just different sound than MOD-86. Very natural sound. Nothing nasty stands out anymore, allowing to "just enjoy the music".
I fail to understand why the Sympatico sounds so much more smooth than the MOD-86 on the Abbey's. Any theories from you guys ? Earl Geddes has studied the audibility of distortions types and found amplifier "crossover distortion" most audible to the human ear. Is the Sympatico symmetrical design canceling out most of this ?
I am very interested to know what the MyRef FE will do in my system.
The Abbey's are a very special design using : 12" B&C driver 80-1000 Hz. Passive cross over to a B&C 1.75" compression driver on Earl Geddes Waveguide.
The Abbey's are a very easy load with 8ohm and 95dB @ 1 Watt @ 1meter.
No problem for any of these Chip amps.
The Abbey's are completed with multi subs (according Earl Geddes advice).
Resulting in even less load on the amps for the Abbey's.
Lets see which of these chip amp designs is most suited in combination with the Abbey's.
So far the Sympatico amp is best suited in my system. But would like to get some of the MOD-86 spectacular sound staging & detailing back.
with such an easy load & rather high efficiency, I would try some good SET (single-ended triode) amp, too. A few good Watts are more than enough...
While a low-power, Class A design would likely be my personal choice for similar speakers, indeed a carefully built FE could be what you're looking for.
Beware: in spite of Mauro's careful and sophisticated circuit design, component selection is paramount. Depending on what you use to stuff the circuit you can get very different results.
Check the dedicated topics for the FE, starting from this one: "My_Ref Fremen Edition - Build thread and tutorial"
(also check here: "MyRef: famoso amplificatore © Mauro Penasa" - Italian language, use some translator)
Do not forget to consider George's modifications (different OpAmp in the first stage), which makes a big difference.
BTW: other possible options could be the classic JLH Class-A, Susan Parker's Zeus, ... or possibly this one: "My ZD-50 ultralow distortion chipamp".
Last edited:
I've been driving a lovely 4 channel myref_C biamped with ESP project 09 crossover. It's been over 3 happy years. But time and again I've had to use an eq. to boost the low-end < 100Hz. (I just use the android or PC app to do this).
I've always felt the bass missing. And I think it's not a fault of the myref_C but instead of the way I've coupled it with the speakers and the sound source etc.
Is this because of line-level mismatch? Given that the LM318n is running at +/-12Vdc, does it mean I can use an input signal that swings at a theoretical max of +/-12Vdc to hit the full blast?
PS: I was using an 18-0-18Vac xfr on the crossover which went kaput and now I'm left with a 9-0-9Vac (LM317 + LM 337 filter). And now I long for the bass all the more. Is it in my head or a reality?
I've always felt the bass missing. And I think it's not a fault of the myref_C but instead of the way I've coupled it with the speakers and the sound source etc.
Is this because of line-level mismatch? Given that the LM318n is running at +/-12Vdc, does it mean I can use an input signal that swings at a theoretical max of +/-12Vdc to hit the full blast?
PS: I was using an 18-0-18Vac xfr on the crossover which went kaput and now I'm left with a 9-0-9Vac (LM317 + LM 337 filter). And now I long for the bass all the more. Is it in my head or a reality?
What are your speakers?
The My_Ref has flat response on the low end, but the Evo version uses larger input coupling caps and a lower value output resistor. Both changes improve the low end.
The voltage rails for the LM318 are not the issue. And the gain should be around 27 dB, 1,5 volts will drive to full output.
It may be the ESP is not mating well. Try driving one full range to see if this helps the bass.
The My_Ref has flat response on the low end, but the Evo version uses larger input coupling caps and a lower value output resistor. Both changes improve the low end.
The voltage rails for the LM318 are not the issue. And the gain should be around 27 dB, 1,5 volts will drive to full output.
It may be the ESP is not mating well. Try driving one full range to see if this helps the bass.