D
Deleted member 148505
Latest layout is at post #15.
What I am trying to accomplish here is to have a very tiny version of My Ref Rev C using through hole parts. Current size is only 2 x 2.65 inches!
I didn't provide onboard power supply capacitors because I am thinking of powering these mini's with an SMPS
I already etched some boards here, hopefully I can finish it this weekend.
What I am trying to accomplish here is to have a very tiny version of My Ref Rev C using through hole parts. Current size is only 2 x 2.65 inches!
I didn't provide onboard power supply capacitors because I am thinking of powering these mini's with an SMPS
I already etched some boards here, hopefully I can finish it this weekend.
D
Deleted member 148505
I hate it when I started soldering and found out that I have incomplete components.
Still not yet finished
Still not yet finished
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I don't understand why AndrewT suggest to change rail-rail decoupling to rail-ground decoupling?
Original design by Mauro only use rail-rail decoupling.
And if I remember correctly, some members report that rail-ground decoupling reduce this amp's performance.
Sorry for my bad English.
Original design by Mauro only use rail-rail decoupling.
And if I remember correctly, some members report that rail-ground decoupling reduce this amp's performance.
Sorry for my bad English.
Last edited:
This sounds a bit impossible/crazy to me.
Small caps (decoupling) are nearly the same as the bigger caps, like the 220uF one. If those doesn't reduce the power, then smaller ones wont either. Those just have a plus effect of reducing HF noises.
One has to look at where the current flows.
If the output current flows from the output pin to the load and from the load to the Main Audio Ground and from the Main Audio Ground to the Power Ground and finally from the Power Ground to the Power Rail, then the decoupling MUST be from Power Rail to Power Ground, by the shortest route possible AT the chip PINs.
If the amplifier were a true balanced topology where the load is between two outputs and the current flows from rail to rail then decoupling would be rail to rail.
If some builder found worse performance from Their version of rail to ground decoupling then they did it wrong !!!!
There are very many builders that get the decoupling wrong. The majority of our Members do not understand why decoupling is needed, nor how it works and thus can never locate it correctly.
If the output current flows from the output pin to the load and from the load to the Main Audio Ground and from the Main Audio Ground to the Power Ground and finally from the Power Ground to the Power Rail, then the decoupling MUST be from Power Rail to Power Ground, by the shortest route possible AT the chip PINs.
If the amplifier were a true balanced topology where the load is between two outputs and the current flows from rail to rail then decoupling would be rail to rail.
If some builder found worse performance from Their version of rail to ground decoupling then they did it wrong !!!!
There are very many builders that get the decoupling wrong. The majority of our Members do not understand why decoupling is needed, nor how it works and thus can never locate it correctly.
It has to do with the feed-forward topology and unique compensation of the MyRef. There's also HF AC bypass path through rail-to-rail bypass cap C4, through both 220uF caps in parallel, to ground, through the load, R3 and back to the LM3886 output. This only works when the ESR of C4 is much lower the the shunted ESR of C1 || C2, thus causing the sonics to be dominated by the character of the small cap C4. When there's a rail-to-ground bypass shunting the 220uF caps, the sonics will be contributed by a combination of C4 + the shunted C of both those caps. Those caps will then have to be atleast as good as C4, or audible sonics will degrade perceptibly - and you can easily verify it by adding and removing the caps bypassing the 220uF electrolytics.
*Never* attempt to bypass +Vs and -Vs of the LM3886 to the lifted small-signal ground - you'll only end up contaminating the small-signal ground with HF hash from the LM3886, which the topology has gone to great lengths to avoid.
OTOH, a larger and/or higher quality rail-to-rail cap C4 improves sonics - I've gone as high as 0.47uF metalized polypropylene, and others have used even larger ones. 0.1 or 0.22 MKP10 are good compromises between size and sonics. MKP2 or PHE426 are also fine.
It's also worth experimenting with a film/foil polyester or polycarbonate at C4 - I'm going to experiment with the Russian K73-16 at that location eventually.
This explanation breaks the rules of where to locate HF and MF decoupling.
Maybe the separate HF and MF decoupling of BOTH chips is being compromised by omitting at least one of the eight required rail to ground decoupling capacitors.
D
Deleted member 148505
if one uses a source that already has a DC blocking capacitor on the output then one does not need a second DC blocker. In this situation one can short out the second blocker with a jumper.Here's the latest layout.
Nothing has changed except for the spacing and alignment of the components.. I also added a jumper for input capacitor.
So if anyone will use a preamp that has no DC output, they can bypass the capacitor by shorting the jumper.
This does NOT apply to a Source that presents a zero or near zero output offset that has resistance to audio ground.
That would change the the output offset of the amplifier if the DC operating conditions were changed by shorting out the only DC blocker.
Last edited:
D
Deleted member 148505
Mini My-Ref sound?
Could anyone comment on the sound this mini-my produces? I'm guessing it can't be as good as the real thing, but a quick comparison between the two and a "stock" gainclone maybe?
Buying all the parts and soldering SMD would be a real effort for My-Ref, so I'm thinking in building this version first. There is always something to learn in the process.
Could anyone comment on the sound this mini-my produces? I'm guessing it can't be as good as the real thing, but a quick comparison between the two and a "stock" gainclone maybe?
Buying all the parts and soldering SMD would be a real effort for My-Ref, so I'm thinking in building this version first. There is always something to learn in the process.
I think this mini version is an option if you already have existing power supply and speaker protection boards. If you don't have the power supply and speaker protect, the full my_ref would be a better option. By SMD, I assume that you have the FE version in mind. As and alternative, you can look for X-caliber version of Rudi and build the board yourself with through-hole components if SMD is a bit daunting for you. I have, and I'm very happy with the result.
But I admit, I would like to try the FE someday.
But I admit, I would like to try the FE someday.
Well I would use the PSU from my GC, I don't see why I should build a new one, the only difference is the voltage, so only a new trafo is needed.
Velleman speaker protection is always available and cheap, so no problems there.
Where would you position this version soundwise, compared to RevC, FE and X-Calibre? I know it's not a popular question, I'm just trying to put things in perspective - ie is it worth experimenting with cheaper solutions or go for the best one, since my GC is really working fine right now
Velleman speaker protection is always available and cheap, so no problems there.
Where would you position this version soundwise, compared to RevC, FE and X-Calibre? I know it's not a popular question, I'm just trying to put things in perspective - ie is it worth experimenting with cheaper solutions or go for the best one, since my GC is really working fine right now
D
Deleted member 148505
The original My_Ref is a through hole design.
It's my version (FE - Fremen Edition) that have SMD parts.
I can't say for sure, since the LM3886 decoupling is different from My_Ref original one... but I expect that jlester version should sound similar to a standard Rev C.
The X-Calibre should sound a bit better than the original RevC thanks to the double diode bridge and paralleled smoothing caps.
But the Fremen Edition sound (way) better than Rev C thanks to an optimized PCB and differerent grounding, LM318 regulator, conpensation, and use of SMD parts, etc.
I would rather have the PSU on a separate board anyway, which makes it easier to build a better one, with regulators or without, and provide a better grounding.
But why not build the FE like this, with optimizations, compensation and SMD parts, but keep the PSU (and even speaker protection) on separate boards?
But why not build the FE like this, with optimizations, compensation and SMD parts, but keep the PSU (and even speaker protection) on separate boards?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.