"noise lower than on RBCD" maybe an ADD one
He didn't have a LT1028 but it was a "real" OPA604 good ps, stable, with gain, mm from the i/v resistor and dac output, and you could easily hear the hf noise between tracks, tried to filter it out, but also affected the hf detail and air.
Played around with it for ages trying to quieten it down to silence, and in the end gave up on I/V resistors. Went to active i/v and never looked back.
Cheers George
Ah the use of OPA604 explains it - its about 10X noisier than the LT1028 in that application so you'd get at best 75dB SNR. Around Dolby C noise levels from a cassette.
No, LT1028 would give a noise lower than the dither noise even on a DDD recording (which would typically be around -92dBfs)
No, LT1028 would give a noise lower than the dither noise even on a DDD recording (which would typically be around -92dBfs)
Strange the Cary 303/200 I had myself used the 604 with gain for the output buffer with a 627 as the I/V from a PCM10704, it was dead quite.
Cary 303/200 CD player Measurements | Stereophile.com
Cheers George
Cary 303/200 CD player Measurements | Stereophile.com
Cheers George
About 6 V peak-peak (i.e. 1k5 running +/- 2mA) is about 2.2 volt RMS. (dont ask me for the calculus
)So with 25 ohms I/V getting 0.2 Vpp we would need a 40 dB amplification (=ECC83) to get a "normal" output level. Probably 33 ohms and a 35 dB tube (6SL7, ECC88) would do too.
Not at all strange when you take into account the different signal levels - post an OPA627 I/V stage we'd expect the working level to be at least a volt or so, at least 40dB above the level on your 15ohm resistor. Which would make the OPA604's own noise totally insignificant.
So with all this, is what your saying that the 1541 with I/V resistor and a tube I/V will be quite? I would say it would be noisier than with the 604 given the same gain.
Cheers George
Of course, a good ECC83 can perfectly deliver a MM pick-up signal of 3-5 mV to a line level of 0,7 Volt (RMS) without noise and has done so for decades. (Telefunken used E83CC and the likes in microphones.)
Now our standards are higher - we demand a lower noise floor than 40 years back - but with all tricks we now have such as DC filaments with high impedances, mu-stages plus power supplies with shunts, we can again deliver silence like we demand.
Well, OK, a bit noisier than solid state due to its inherent thermal noise but still dynamic enough.
For instance but OT, my driver for a 300B is based on an ECC35, it can pump out almost 250 Vpp with ultra low distortion, no tube engineer dreamed about that 80 years ago.
I have no idea really, I don't do tube design at all so I'm completely clueless about the noise levels.
The noise level with the OPA604 depends on the value of the I/V resistor. Obviously 15R makes it way too noisy. I seem to recall that the limit for passive I/V was in the region of 100mV peak signal (I think Thorsten hinted at this) in which case the OPA604 could almost be quiet enough, but I'd prefer something like ISL28210 myself or, better still M5240P and run with a somewhat smaller I/V R.
SRPP Noise...
As far as hiss goes there was none. Hum was a different story. Not like the old guitar amps I have heard throughout my life though. Had to really amplify the output to hear any of that. Will have to see if shielding might help that.
As far as hiss goes there was none. Hum was a different story. Not like the old guitar amps I have heard throughout my life though. Had to really amplify the output to hear any of that. Will have to see if shielding might help that.
TDA1541
Halving the gain as suggested with twice the current likely would work. I actually thought of another experiment to try before finishing this parallel dac unrelated to that. An advantage to paralleling is you gain by averaging a 1/2 bit LSB so 16.5 bit accuracy. In passive approach we are limited to the compliance range of the 1541. So there is a limit to how big the passive resistor can be. Hence the reason for the low I/V values. Most people choose to high a value....
Halving the gain as suggested with twice the current likely would work. I actually thought of another experiment to try before finishing this parallel dac unrelated to that. An advantage to paralleling is you gain by averaging a 1/2 bit LSB so 16.5 bit accuracy. In passive approach we are limited to the compliance range of the 1541. So there is a limit to how big the passive resistor can be. Hence the reason for the low I/V values. Most people choose to high a value....
When I mentioned about noise with my dual tda1541a dac and passive i/v + tube buffer, it's not that the noise is as bad as you can hear it sizzling on the speakers but to my ears the minor details were masked and the sound became a bit sticky compared to good active i/v.
I used 34 ohm i/v resistor (2x 68 ohm in paralel) just slightly higher than 25ohm and two ecc88 (1950s very low noise tubes). Two dac chip with same date code 8910 holland made Tube buffered was assembled on two layer PCB and it was a widely used SRPP configuration.
I used 34 ohm i/v resistor (2x 68 ohm in paralel) just slightly higher than 25ohm and two ecc88 (1950s very low noise tubes). Two dac chip with same date code 8910 holland made Tube buffered was assembled on two layer PCB and it was a widely used SRPP configuration.
Have you linked pin16 of the dacs together?
This cured noise when I was experimenting with dual dacs.
(I used 500pf caps on both dacs).
I tried that very briefly but then decided to remove the link. It was about 10 years ago.
As I said the tone of the DAC was quite nice. If one listen to it alone you can say there is nothing wrong with the sound. At first I even thought it was quite detail but after long listening I determined that the sound was slightly foggy though I could still hear a lot of things happing behind the thin fog plus a bit of tubey. I have moved on to using ad844 for quite some time. The tube buffer and the surplus tda1541a are still lying around the room. I take time to optimize the ad844 stage and buffer.
The level of transparency I got from these ad844s is very high. They reveals the quality of wire that I use to connect from dac board to i/v board as well as the wire from ad844 to the output buffer. I had to tear out some Supra interconnect cable to get the wire to connect these boards.
Last edited:
Noise...
Hi quantran, On my build the noise is below the threshold of 16 bits. So it isn't masking anything. In some ways the SRPP is rivaling the PCM1704 especially in the mid range. Overall the 1704 build is better as it has the discrete op amps now. I should write up a review of that at some point.
Hi quantran, On my build the noise is below the threshold of 16 bits. So it isn't masking anything. In some ways the SRPP is rivaling the PCM1704 especially in the mid range. Overall the 1704 build is better as it has the discrete op amps now. I should write up a review of that at some point.
Linking pin 16?
Hi batteryman, My pin 16's and 17's each have there own 120 pF PPS capacitor for DEM. I have not linked them. So each dac has the same input data however they are independently converting the data to analog which is summed at the I/V resistor. So is that just pin 16's, or are you using the same capacitor for DEM on both?
Hi batteryman, My pin 16's and 17's each have there own 120 pF PPS capacitor for DEM. I have not linked them. So each dac has the same input data however they are independently converting the data to analog which is summed at the I/V resistor. So is that just pin 16's, or are you using the same capacitor for DEM on both?
I have 500pf on each dac and just the pin 16s linked.
My newly designed dual dac pcb is arriving from Eurocircuits tomorrow and I've designed it for the same dem configuration, although with a header for optional external reclocking. I plan of adjusting the dem clock to 176Khz and hope it synchronises by crosstalk to Fs.
This new one has only the dacs & low noise regulators on-board, and will use iancanada's I2S to PCM boards in differential / simultaneous mode.
I/V conversion will be on a separate board as well, although initially I am using only my IV transformers connected between dac outputs with 100R i/v resistors. (No worries about the 2ma offset when connected in true differential working.)
IanJ
Last edited:
Dual TDA project
Hi Ian, Thanks for sharing. I am using 705 Khz DEM which is supposed to lock. This works well with the 1541A S1 Crowns. I remember one of my 1541A's didn't like that and required 470 or 500 pF. Keep us posted on your progress. After I recover from my surgery I plan to look at iancanada's FIFO. Just to much going on for the next 2 weeks.
Hi Ian, Thanks for sharing. I am using 705 Khz DEM which is supposed to lock. This works well with the 1541A S1 Crowns. I remember one of my 1541A's didn't like that and required 470 or 500 pF. Keep us posted on your progress. After I recover from my surgery I plan to look at iancanada's FIFO. Just to much going on for the next 2 weeks.
Just got the OPA1641 & ISL28110 . It turns out I ordered the wrong part. These OPA1641 are in MSOP8 package! Too bad... I need to get the correct adapter before listening to them. However, the ISL28110 is excellent. The bass is clean and defined. I'd say it's better than LT1468 on my circuit (thanks, Albraxalito).