Arcam and the TDA1541
- Thread starter Fatmarley
- Start date
cubastreet
Espresso Fiend
I'm doin a bit of fiddling with the psu on my philips cd650 which has a tda1541. I took the psu transformer out of my dead marantz cd75 which has the same dac board and built an offboard psu in a case with the two transformers. I used the two main windings from one transformer (didn't measure them, but it the ones that end up supplying the +-15) and all the smaller windings from the other. This made a big difference by itself, but I'm too busy with work now to do any further work on it. I'll add better rectification at a later stage.
martin clark
pinko bodger
Well first off, the complement of all the signal current out of the TDA1541 also appears superimposed on the +5v pin, so decoupling here is critical... and a superreg would be an ideal supply to keep impedance low here. The supply on this pin is pretty much just pure DC, plus the signal components. It is the most critical in terms of direct effect on sound quality.
The -15v supply is the only pure analogue supply and should be quiet. It does not appear particularly sensitive to supply impedance, but low HF noise helps; a capacitance multiplier as a filter on the raw supply, followed by an LM337 does well here.
The-5v supply is the least critical, again a well-sorted LM317 is fine.
However - the internal current reference in the chip is between the -5V supply and the -15v supply, and tying these supply pins together with a cap really helps. 10uF oscon (+ to the -5v pin, remember) is a great start. Tweak your regs so that 10.0v is held between these pins, +/- 0.1v.
The -15v supply is the only pure analogue supply and should be quiet. It does not appear particularly sensitive to supply impedance, but low HF noise helps; a capacitance multiplier as a filter on the raw supply, followed by an LM337 does well here.
The-5v supply is the least critical, again a well-sorted LM317 is fine.
However - the internal current reference in the chip is between the -5V supply and the -15v supply, and tying these supply pins together with a cap really helps. 10uF oscon (+ to the -5v pin, remember) is a great start. Tweak your regs so that 10.0v is held between these pins, +/- 0.1v.
Hi Martin,
Thats very helpful. I am in the process of building a cd player based around the CDPRO-2M transport feeding the I2S into TDA1541A double crown and am just contemplating the power arrangements. I had intended to use three super regs for the +5, -5 and -15V, each super reg to be fed by seperate transformer with FET based Vbe. From your post, is there any sonic value ingoing to the full super-reg route? Secondly what would you suggest for arranging the local decoupling around the DAC chip? It was concerning me that the super regs might be several inches from the DAC and that this might not be perfect.
Cheers,
Ian
Thats very helpful. I am in the process of building a cd player based around the CDPRO-2M transport feeding the I2S into TDA1541A double crown and am just contemplating the power arrangements. I had intended to use three super regs for the +5, -5 and -15V, each super reg to be fed by seperate transformer with FET based Vbe. From your post, is there any sonic value ingoing to the full super-reg route? Secondly what would you suggest for arranging the local decoupling around the DAC chip? It was concerning me that the super regs might be several inches from the DAC and that this might not be perfect.
Cheers,
Ian
ron
Tweaker
ron
Tweaker
martin clark
pinko bodger
Yup, thats about the size of it; and up to 47uF of oscon seems to work well on the +5v pin.
BUT
Oscons that size alone will not 'play nice' with LM317/337s (too little ESR), so a little fettling/experimentation may be required. Adding about 0R25 to 0R33* in series with the LM317/337 output critically-damps the output impedance into a 47-10uF Oscon(respectively) - and the 1541 doesn't seem to mind the extra series resistance at all, decoupling and low noise seems to be more important (to my ears, anyway)
If you go the superregs route this will be irrelevant of course, although you may need to experiment somewhat with final choice of small cap at the DAC for best effect. The low output noise of superregs makes them a very attractive proposition
Edit - and I wouldn't worry about 2 or 3 inches from regs to pins. Keep the decoupling caps as tight as you can though, and lightly twist each supply-and-return wire pair between reg and supplied pin.
* I like stacking 1R SMT resistors to taste; 3 = 0R33, 4=0R25.
BUT
Oscons that size alone will not 'play nice' with LM317/337s (too little ESR), so a little fettling/experimentation may be required. Adding about 0R25 to 0R33* in series with the LM317/337 output critically-damps the output impedance into a 47-10uF Oscon(respectively) - and the 1541 doesn't seem to mind the extra series resistance at all, decoupling and low noise seems to be more important (to my ears, anyway)
If you go the superregs route this will be irrelevant of course, although you may need to experiment somewhat with final choice of small cap at the DAC for best effect. The low output noise of superregs makes them a very attractive proposition
Edit - and I wouldn't worry about 2 or 3 inches from regs to pins. Keep the decoupling caps as tight as you can though, and lightly twist each supply-and-return wire pair between reg and supplied pin.
* I like stacking 1R SMT resistors to taste; 3 = 0R33, 4=0R25.
martin clark
pinko bodger
Theoretically parallel operation gives better resolution / better S:N ratio because the linearity errors of each DAC get scaled by 1/SQRT(n), where n is the number of dacs in parallel - but only if these errors are truly decorrelated between dacs. You also get twice the current output, which would be useful for passive I/V using a resistor - except that the TDA1541 specifies a maximum 25mV output compliance (i. e. the resistor has to be a very small value, compromising S:N ratio).
I'm not convinced that the effort actually repays the effort; I can imagine a few other aspects becoming more difficult/introduce new sources of error; and the S/ ratio is already excellent, at -110dB.
However - if you had a pair of 1541s, one for each channel, and fed each complementary inputs, so that each dacs' outputs became channel +ve and channel -ve then all the signal current on the DAC supply pins would cancel, leaving only pure a DC draw and digital feedthrough noise separate from the output signal.
In otherwords, it would take the DACs PSU supplies out of the signal path (!); which might make for spectacular results...
I'm not convinced that the effort actually repays the effort; I can imagine a few other aspects becoming more difficult/introduce new sources of error; and the S/ ratio is already excellent, at -110dB.
However - if you had a pair of 1541s, one for each channel, and fed each complementary inputs, so that each dacs' outputs became channel +ve and channel -ve then all the signal current on the DAC supply pins would cancel, leaving only pure a DC draw and digital feedthrough noise separate from the output signal.
In otherwords, it would take the DACs PSU supplies out of the signal path (!); which might make for spectacular results...
ron
Tweaker
I've tried parallel TDA1541A's before and preferred single tbh, if your using passive I/V and a valve output stage it does give you more output but to me a single dac and discrete I/V and gain section sounded better
Which one do you have Dave? is it 6C45PI valves?
This is one I once was built
Which one do you have Dave? is it 6C45PI valves?
This is one I once was built