Group Buy: Mauro Penasa MyRef Rev C kit

[Audiodoc]

Here are the premium boards almost finished. Two capacitors and a wire bridge is all that needs to be added after confirmation from linuxguru.

Need to know the audio difference between the rev c and rev e.

 

[gopib]

Nice work.I did the ordinary version of the kit. Still struggling with the enclosure though. You may want to double check the soldering, because of the double sided PCB.

 

[quad]

Very nicely done, Doc. Those red boards look awesome.

I took a look at the schematic - the 3.3k, 10k and 100k are
not in any critical locations, so the 5% should be fine.

There are two resistors, R1 and R4 (1K, 1W) where the silk
screen suggests to raise it by about 1cm. It would have
been good to follow that, as it allows for all round air flow to cool
them in case they get hot. (unless Linuxguru has advised otherwise).

The 47ohm/150pf - these are the compensation network,
which was tweaked by Mauro Penasa for his next gen "Evolution" amp.
Dario started a new thread about it on diyaudio.
Not sure if the change suggested by Linuxguru is to implement
this mod which is supposed to have much improved sonics.
Await your or linuxguru's update on this.

My_Ref Fremen Edition - need help on PCB evaluation - diyAudio

 

[linuxguru]

However 6 resistors were +-5% metal films (10k, 100k and 3.3K). I supposed all should have been +-1% or lower for a kit costing 4000 per monoblock.

LED is not provided.

Plus linuxguru has made some changes that I could no comprehend as they were labeled a crosses.

Hi audiodoc - I've PM'ed the Rev E changes to you. The crosses are omitted components (usually extra bypass caps in parallel with other caps, or an integrated bridge rectifier instead of discrete rectifiers).

The 5% resistors are actually mostly CFRs, which are audibly better in those locations - more airy, less analytic sounding, but YMMV. Dario (ClaveFremen) and I have both independently auditioned these and come to more or less the same conclusions - all shunt resistors in/near the signal path should be CFRs.

I actually prefer Allen-Bradley Carbon Composition at one 3.3k-resistor location (input series after the input coupling cap), but they tend to drift with time and temperature, so I safely left it as a CFR (next best). Go ahead and substitute it with a 1% MFR if you wish, but you'll find it harsher and more analytic than a CFR.

The best resistors there are either a Kiwame 2W CFR, or a Takman REX CFR - both unobtainium in India. BTW, the Watts CFR that I provided is pretty good, but I'll try to find a better one in due course, if available.

 

[linuxguru]

Very nicely done, Doc...

Seconded - audiodoc has a very steady hand and superb assembly skills. Probably the cleanest board I've seen, cleaner than the ones I've done myself.

However, please carefully do the following changes:
1. Interchange C4 and C5 (one is Wima MKP10 polypropylene, other is MKS-3 polyester) - the one near the LM3886TF should be polypropylene for optimum sonics.
2. Remove the 22pF C0G at C10 and install it at C34.

There are two resistors, R1 and R4 (1K, 1W) where the silk
screen suggests to raise it by about 1cm. It would have
been good to follow that, as it allows for all round air flow to cool
them in case they get hot. (unless Linuxguru has advised otherwise).

Yes, it helps to raise them a bit, especially if you use rails higher than the recommended +/- 32V. The resistors are amply over-rated at 2W, but they still get a bit toasty at high rails.

The 47ohm/150pf - these are the compensation network,
which was tweaked by Mauro Penasa for his next gen "Evolution" amp.
Dario started a new thread about it on diyaudio.
Not sure if the change suggested by Linuxguru is to implement
this mod which is supposed to have much improved sonics.
Await your or linuxguru's update on this.

My_Ref Fremen Edition - need help on PCB evaluation - diyAudio

Yup, Dario's is Rev FE (Fremen Edition), and mine is Rev E. The changes in Rev E (wrt to Rev C) are:

R42: short
C32: 330 pF
C10: 27 pF Silver Mica
C34: 22 pF C0G ceramic

The last two are not critical if you have it interchanged - the soundstage will get slightly recessed, but it will still be unconditionally stable either way.

 

[omishra]

I have two suggestions -
1. Why not to rotate PCB's to make heat sinks at backside of chassis and exchange positions of input and output jacks. This will make shorter input wires.

2. Also transformer can be moved forward near front side power switch. This will make input wires away from transformer.

 

[linuxguru]

Need to know the audio difference between the rev c and rev e.

Compared to Rev C, Rev E :

1. Has an immersive, forward and palpable soundstage;
2. Has explosive and percussive mid-bass dynamics;
3. Is slightly more edgy (less smooth) in the upper mids (but this can be eliminated by using the LF01 Class-A hybrid/discrete opamp instead of the LM318).

The tradeoff is worthwhile, IMHO. The upper-mid edginess actually helps some vocals and wind instruments by adding breath to aspiration sounds. Check it out with saxophone tracks, e.g. "Take Five" by Paul Desmond, etc.

Allow the Muses and the Black Gates at least 100 hours to break-in, during which the edginess will gradually decrease. With C21 installed, it may take even longer for C9 to break in fully - maybe 200-300 hours, but with very slight changes after 100 hours.

 

[Audiodoc]

Made the changes, except for the airmounting of the resistors, according to Rev E as guided by linuxguru. Almost everything went well except the desoldering part of the 22pf capacitor whose leg came apart. Had to do a small surgery to fix it and check with a LCR meter. Now everything seems fine. If any problem arises will have to trouble linuxguru with a spare.

Enclosures will arrive in the next week. The problem is that I have two toroids a 25-0-25 250VA and a 24-0-24 160VA but I can use the 160VA one only because the internal dimensions of the enclosure are 45 x 175 x 290 mm only.

Heat sink is also ready just need to cut it after the enclosure arrives.

Any suggestions.

 

[linuxguru]

... Almost everything went well except the desoldering part of the 22pf capacitor whose leg came apart. Had to do a small surgery to fix it and check with a LCR meter. Now everything seems fine. If any problem arises will have to trouble linuxguru with a spare.

Enclosures will arrive in the next week. The problem is that I have two toroids a 25-0-25 250VA and a 24-0-24 160VA but I can use the 160VA one only because the internal dimensions of the enclosure are 45 x 175 x 290 mm only.

Good work - everything looks fine. No problem about the 22 pF, I'll send you a spare if necessary. You can also use a 22pF Silver Mica without any probs - it should be available at the local electronics markets, or maybe in Delhi. If you switch to Silver Mica, change them on both channels to preserve imaging.

Trafo: 160 VA is adequate, and 24-0-24 is better for sonics than 25-0-25.

From the pic, all component orientations seem to be correct, but recheck them anyway before applying power. Check all electrolytics, including the big PSU caps. Also do a quick sanity check with a multimeter for shorts and opens on all the rails.

 

[jls001]

I finally got my toroidal today from Toroid Engineers. This thing is BIG, much bigger and heavier than I envisaged. 1600 bucks. Ordered this morning, got it at 4 pm. He mistakenly built me two!

Couldn't get the cabinet (from Dinrack) as they didn't have any 2U in stock. Going back next Sat to get a custom size.

Question time: the proposed cabinet is full aluminium, 2.5 mm thick. Can I use the cabinet's side panels as heat sink if I maintain proper contact with the amp chip and screw it down? In other words, will it provide sufficient cooling? Or do I need to use dedicated heat sinks?

The heat sink I bought is of the wrong form factor, and going back to market to buy an alternative will be another day wasted.

 

[linuxguru]

Question time: the proposed cabinet is full aluminium, 2.5 mm thick. Can I use the cabinet's side panels as heat sink if I maintain proper contact with the amp chip and screw it down? In other words, will it provide sufficient cooling? Or do I need to use dedicated heat sinks?

It will work as long as the contact is good. It only dissipates about 10-15W/channel at +/- 32V rails. You'll be ok as long as it is not run at max amplitude all the time, and even then, all that will happen is that the SPike protection may trigger, causing the LM3886 to clip from time to time.

 

[gopib]

Heat sink is also ready just need to cut it after the enclosure arrives.

Audiodoc,

The heatsink looks good. Where did you get it, if I may ask.

TIA

 

[quad]

If you are using the LM3886T package, make sure to use a
mica or similar insulator (and heat sink paste) as the metal tab
is at the negative power rail voltage.

 

[Audiodoc]

Got it from the local electronics market in Kolkata for 500 bucks.

 

[captrajesh]

It will work as long as the contact is good. It only dissipates about 10-15W/channel at +/- 32V rails. You'll be ok as long as it is not run at max amplitude all the time, and even then, all that will happen is that the SPike protection may trigger, causing the LM3886 to clip from time to time.

This would also depend on the ambient temperatures I guess.

 

[Audiodoc]

Most ready made amplifiers based on LM3886 from ebay use the 3 mm aluminum cabinet body as the heat sink however they use lower supply voltage rails. It is always better to use a dedicated heat sink because it gives you a leeway to play things louder for a prolonged period of time without letting the internal protection circuitry to get activated which significantly raises the distortion figures of the chip.

 

[gobble]

If you are using the LM3886T package, make sure to use a
mica or similar insulator (and heat sink paste) as the metal tab
is at the negative power rail voltage.

i dont getit. If we use an insulator how will the heat transfer/dissipation happen? What metal tab is at -ve power rail voltage and what will happen if we dont use mica?

Please enlighten.

Thanks
--G0bble

 

[Audiodoc]

There are two variants of LM3886. One is the LM3886T which has a open metal tab as you seen in transistors/ regulators like 7812. The open metal tab is connected to the negative rail so it will be -30-35V DC depending on your supply voltage. It would cause no harm as such but could easily short circuit components as soon it finds a ground. It will also short circuit if you are using a star ground as is obvious. It will also destroy your speakers if it accidentally touches the output terminals. By placing a mica sheet and non conducting washer with some heat sink compound the heat will be transmitted without connecting the back plane electrically.

Better buy the LM3886TF which is fully enclosed in heat conducting plastic and doesn't need the isolation.

TC

 

[josedandrew]

Hi guys,
Completely oblivious to this thread till now.
is the Ref C kits still available.
would like to buy a pair.
thanks.

 

[josephjohnt]

Hi guys,
Completely oblivious to this thread till now.
is the Ref C kits still available.
would like to buy a pair.
thanks.

Standard and premium kits with v1.4 PCB are usually available with Siva. I have PM'ed you his contact info.