Sorry - I was off the forum for a while due to health issues and some other unrelated stuff, but I finally got around to doing an LM3886 monoblock version of the MiniRef. I started with my older MyRef Rev C v1.4 layout and tweaked the feedback loops to use the Walt Jung double non-inverting topology of the MiniRef (MyRef is double inverting, and has a unique compensation network that limits the usable opamps to less than a handful - LM318, TL071 and a few other slow and venerable opamps). I also transplanted the opamp rail shunt-regulator from the MiniRef V1.04.
The result is the VersaRef - it should be a direct drop-in replacement for the MyRef Rev C v1.3 and v1.4 boards, but with a large range of usable opamps for the outer loop - LT1360, LM6171, OPA627, LME49990 and maybe some discrete opamps, should work with minor tweaks. Here's the schematic and layout if you want to re-do it in Eagle or anything else. Alternatively, I'll try to get PCBs fabbed in due course from Gerbers after re-checking everything to make sure that there's no major screw-up.
The main problem now is that the LM3886TF is becoming unobtainium, but it should still be available from TI online. It won't be cheap-cheap
with shipping, but it's still exceptional value for what it offers. There isn't any comparable Class-AB chipamp with similar power-handling capability and durability on the market.
This will probably be my last LM3886 design - I may switch to discretes driven by an LME49810 or similar, for future projects. It's not as convenient, for sure.
The result is the VersaRef - it should be a direct drop-in replacement for the MyRef Rev C v1.3 and v1.4 boards, but with a large range of usable opamps for the outer loop - LT1360, LM6171, OPA627, LME49990 and maybe some discrete opamps, should work with minor tweaks. Here's the schematic and layout if you want to re-do it in Eagle or anything else. Alternatively, I'll try to get PCBs fabbed in due course from Gerbers after re-checking everything to make sure that there's no major screw-up.
The main problem now is that the LM3886TF is becoming unobtainium, but it should still be available from TI online. It won't be cheap-cheap
with shipping, but it's still exceptional value for what it offers. There isn't any comparable Class-AB chipamp with similar power-handling capability and durability on the market.
This will probably be my last LM3886 design - I may switch to discretes driven by an LME49810 or similar, for future projects. It's not as convenient, for sure.
Attachments
Hi Linuxguru, welcome back.
Yours is a venerable design, i know.
What is the benefit of the jung type feedback tho compared to a single cap of say 470pf value?
Also i see that youve omitted the 3k3 1nf stabilizer between the input pins of 3886 and is the design still rock stable without it?
Ive been using a design almost identical to yours for the past year. (Actually still kind of yours because you described it here years back which i followed)
I did add the 3k3+1nf string just mentioned and the regulators are cascoded lm317 337.
I along with several people whos had chance to take it home love the sound. I think the super high psrr offered by the input stage already makes it a special amp.
I too have found -+12v to be subjectively the best range for the opamp.
My next build ill be using opa551 and run the whole board at -+28v, omitting the regulators. I anticipate it will sound more 'pure' because of the regulators do have noise and phase trade offs
Yours is a venerable design, i know.
What is the benefit of the jung type feedback tho compared to a single cap of say 470pf value?
Also i see that youve omitted the 3k3 1nf stabilizer between the input pins of 3886 and is the design still rock stable without it?
Ive been using a design almost identical to yours for the past year. (Actually still kind of yours because you described it here years back which i followed)
I did add the 3k3+1nf string just mentioned and the regulators are cascoded lm317 337.
I along with several people whos had chance to take it home love the sound. I think the super high psrr offered by the input stage already makes it a special amp.
I too have found -+12v to be subjectively the best range for the opamp.
My next build ill be using opa551 and run the whole board at -+28v, omitting the regulators. I anticipate it will sound more 'pure' because of the regulators do have noise and phase trade offs
Single cap at C10? I think 470 pF is a bit excessive, but 47 pF might be right - will have to build and listen to the sonics.
Thanks - In the hurry to do a quick mash-up, I forgot the inter-stage Zobel of 3k3 + 1nF. I'll add something there - maybe just plain 220 pF as in the NatSemi AN, or 1k + 3n3. Maybe also a small pF value parallel to R6 to ground on the chipamp inverting input.
OPA551 is news to me - great high-voltage rail specs. Should be workable for the MiniRef/VersaRef, but you probably want to filter the opamp rails anyway, with or without a shunt regulator. Maybe just a series RC or LC network on each rail.
Thanks - In the hurry to do a quick mash-up, I forgot the inter-stage Zobel of 3k3 + 1nF. I'll add something there - maybe just plain 220 pF as in the NatSemi AN, or 1k + 3n3. Maybe also a small pF value parallel to R6 to ground on the chipamp inverting input.
OPA551 is news to me - great high-voltage rail specs. Should be workable for the MiniRef/VersaRef, but you probably want to filter the opamp rails anyway, with or without a shunt regulator. Maybe just a series RC or LC network on each rail.
Linuxguru could you perhaps sim what i did with the interstage zobel?
3k3+ 1nf was too bright sounding (hint of instability?) So i changed the R to 2k. Treble became more calm. Then i tried 270ohm+ 330pf. This was even better subjectively. Better behaved treble, everything more transparent.
As i understand it the value of the cap corrolates to the range of freq it attenuates and the value of the resistor how much of it?
Id like to see if 470ohm+ 330pf is stable in the sim (the values i will be trying next) and what the lowest values i can get away with is.
220pf by itself was unstable, btw.
And here is my build. Kept to your schematic to the tee. Except for the czcrc filter (cap-zener-crc) for the opa551. Triple current pump. Mono mono chasiss
Sounds amazing. Completely hum free.
3k3+ 1nf was too bright sounding (hint of instability?) So i changed the R to 2k. Treble became more calm. Then i tried 270ohm+ 330pf. This was even better subjectively. Better behaved treble, everything more transparent.
As i understand it the value of the cap corrolates to the range of freq it attenuates and the value of the resistor how much of it?
Id like to see if 470ohm+ 330pf is stable in the sim (the values i will be trying next) and what the lowest values i can get away with is.
220pf by itself was unstable, btw.
And here is my build. Kept to your schematic to the tee. Except for the czcrc filter (cap-zener-crc) for the opa551. Triple current pump. Mono mono chasiss
Sounds amazing. Completely hum free.
Attachments
Hi Linuxguru,
I am interested in this new board, I do have a couple of LM3886TF and LM318 in stock.
So maybe when you are done with Testing and there is GB please do let me know, A pair would sufficient for me.
Currently using the stereo board of Ref A, ( plan to update to Ref C).
Also, have a build of Circlophone in the background - planning to a Swappable system such that I can use the same heatsink and either of the amp.
You have also built one right - circlophone by elvee?
This thread has been quite for sometime.
Has anyone built the circuit that LinuxGuru has generously posted for the community in post #203 ?
I was going to design a board without the power supply part of the circuit as I already have a few of those left over from precious chipamp projects. It also allows me to power it with SMPS without the those big caps there. Also i have speaker protection boards so I would leave that part of the circuit out too.
I would be using original LM1875 and would believe that coming from LinuxGuru the circuit would be good to go with just minor adjustments depending on the opamp.
Any thoughts from someone who has made this circuit or from the Guru himself would be great.
I have made a couple of LM1875 amps over the last 12 months paying careful attention to component layout, ground planes and star grounding (go to do something when you cant get LM3886s) and have been pleasantly surprised by the end result.
Now time to try a more evolved design.
Has anyone built the circuit that LinuxGuru has generously posted for the community in post #203 ?
I was going to design a board without the power supply part of the circuit as I already have a few of those left over from precious chipamp projects. It also allows me to power it with SMPS without the those big caps there. Also i have speaker protection boards so I would leave that part of the circuit out too.
I would be using original LM1875 and would believe that coming from LinuxGuru the circuit would be good to go with just minor adjustments depending on the opamp.
Any thoughts from someone who has made this circuit or from the Guru himself would be great.
I have made a couple of LM1875 amps over the last 12 months paying careful attention to component layout, ground planes and star grounding (go to do something when you cant get LM3886s) and have been pleasantly surprised by the end result.
Now time to try a more evolved design.