No, I dont think so. And I dont think it would be audible in most single ended systems either. Unless max voltage is excided of course
Quick question.
I have a couple of shunts (v1.3) that have been used in a DAC which mostly sits in a closet. These are built with the standard parts supplied by Tea-Bag. Out of curiosity, I decided to try one of the shunts for powering the Sparkos op-amps in my Miro DAC in place of the Miro PS2 power supply (LT1963A/LT3015 regulators). The op-amps are run at +/-15v and supposedly consume less than 50 mA. As far as I recall, the shunt is configured for 200 mA.
I was quite surprised by the effect of the shunt on the sound of the op-amps: loss of high and low frequency extension, and the mids are somewhat sweet, almost wooly, with a loss of detail/airiness. I tried swapping the green C2 Nichicon (ES?) with KZs (all 220 uF) which returned a little of the high end/treble. Now, I am wondering if a shunt regulator is not appropriate for the Sparkos op-amps, or maybe the bias needs increasing for low end control, or the C2s to be swapped and/or bypassed? Something else? It is also entirely possible the sound of this shunt regulator in this particular setup is not for me.
I have a couple of shunts (v1.3) that have been used in a DAC which mostly sits in a closet. These are built with the standard parts supplied by Tea-Bag. Out of curiosity, I decided to try one of the shunts for powering the Sparkos op-amps in my Miro DAC in place of the Miro PS2 power supply (LT1963A/LT3015 regulators). The op-amps are run at +/-15v and supposedly consume less than 50 mA. As far as I recall, the shunt is configured for 200 mA.
I was quite surprised by the effect of the shunt on the sound of the op-amps: loss of high and low frequency extension, and the mids are somewhat sweet, almost wooly, with a loss of detail/airiness. I tried swapping the green C2 Nichicon (ES?) with KZs (all 220 uF) which returned a little of the high end/treble. Now, I am wondering if a shunt regulator is not appropriate for the Sparkos op-amps, or maybe the bias needs increasing for low end control, or the C2s to be swapped and/or bypassed? Something else? It is also entirely possible the sound of this shunt regulator in this particular setup is not for me.
Bypass C2 with small MKP underneath first. You can also try both low and high mA configuration. 120mA & 300mA. One step at a time. See where it gets you.
The sound you described does not remind Ubib, maybe bad interaction with what's hanging further on the rail. See about output wiring to be short and thick too. Funny enough a guy recently PMed me on how the bass went awesomely extended on his Sparkos SS 2590 Bal pre when with a Ubib vs Sparkos PSU... go figure.
The sound you described does not remind Ubib, maybe bad interaction with what's hanging further on the rail. See about output wiring to be short and thick too. Funny enough a guy recently PMed me on how the bass went awesomely extended on his Sparkos SS 2590 Bal pre when with a Ubib vs Sparkos PSU... go figure.
Twist & Shield as The Beatles once sang (but did not write).
29 uV RMS
29 uV RMS
Quick follow up question. For the C2 bypass cap, would there be any advantage to using a large format SMD? If so, which type?
I just started on a DCG3 to be powered by two uBibs. Should I just include bypass caps on C2 from the get go?
Cheers, Soren
I just started on a DCG3 to be powered by two uBibs. Should I just include bypass caps on C2 from the get go?
Cheers, Soren
You dont have anything to check for oscillations with? Opamps with its high common mode rejection should not react like that normally.
What kind of on board stuff sits on the rails of the Miro DAC?
What kind of on board stuff sits on the rails of the Miro DAC?
Last edited:
5mm pitch through hole MKP or PPS SMD should be the most practical. 50V-63V DC rating is enough. Better estimate impact by trial in each different application.
Thanks Salas.
I first added 0.1uF Wima caps to bypass C2, but am not sure there was a significant difference. However, changing R1 from 5 to 2.5 ohm improved things, with the sound 'opening up', giving similar frequency extension to the previous power supply. The sound is still a little fuzzy/soft, less focused than with the Miro PS2. Maybe this is due to the different flavors of caps used. Anyway, it seems the shunt does something to the mids that is worth pursuing 🙂
Another question. With the higher constant current (and larger heat sinks), M2 runs around 65C, which seems fine/as expected. However, M1 runs relatively cool around 30C and it made me wonder if 15v AC is too low for an output of 15 DC?
Thanks again,
Soren
I first added 0.1uF Wima caps to bypass C2, but am not sure there was a significant difference. However, changing R1 from 5 to 2.5 ohm improved things, with the sound 'opening up', giving similar frequency extension to the previous power supply. The sound is still a little fuzzy/soft, less focused than with the Miro PS2. Maybe this is due to the different flavors of caps used. Anyway, it seems the shunt does something to the mids that is worth pursuing 🙂
Another question. With the higher constant current (and larger heat sinks), M2 runs around 65C, which seems fine/as expected. However, M1 runs relatively cool around 30C and it made me wonder if 15v AC is too low for an output of 15 DC?
Thanks again,
Soren
Yes Salas,thanks,
For power the SS2590 Sparkos in my Academy Audio Muse equlibred Pream,I tested with the UBIB and Sparkos power supply with Sparkos discret regulator,the UBIB with C2 470µF gives a full and dynamic bass,the Sparkos in comparison is meager in the bass,but on the other hand it offers a more open medium less boring than UBIB,I specify that it is a bipolar UBIB,this may be important,I preferred to keep the configuration with the UBIB,I'm still adjusting and testing.
Also try less C2 value, say 220uF, maybe the bass masks the mids a little psycho-acoustically? R9 value is tonally influential too. In combination with the spare current value.
No need to pull R9 if for less value, just piggyback another R to halve its value and listen. Bipolar you mean dual +/-symmetric, not that you substituted Mosfet with BJT, yes? (Historically SSLV1.1 BiB could also use BJT output transistor).
No need to pull R9 if for less value, just piggyback another R to halve its value and listen. Bipolar you mean dual +/-symmetric, not that you substituted Mosfet with BJT, yes? (Historically SSLV1.1 BiB could also use BJT output transistor).
"Also try less C2 value, say 220uF, maybe the bass masks the mids a little psycho-acoustically"
That's exactly what I told myself,I will try with 220uF,
The consumption is 250mA for positive and 220mA for the negative,I currently have 1.5 Ohms R1,470R R9,I want to try R1 1 Ohm, what value for R9 so?
I still have the Mosfet,
what are the sonic differences between Mosfet vs BJT?
That's exactly what I told myself,I will try with 220uF,
The consumption is 250mA for positive and 220mA for the negative,I currently have 1.5 Ohms R1,470R R9,I want to try R1 1 Ohm, what value for R9 so?
I still have the Mosfet,
what are the sonic differences between Mosfet vs BJT?
Don't put TO-220 BJT instead of M2 in the SSLV1.3 Ubib. It will lose much open loop gain i.e. it will lose much performance. Thus it will sound worse. In the SSLV1.1 it was about trading acceptable open loop gain loss for >flat Zo extension if with BJT output element. It could also give 3.3V rail. Due to BJT Vbe loss deprived less voltage living space for the element than Mosfet Vgs.
As it is with 1.5 Ohm R1 I suggest you try another 470R over R9 (parallel) and listen first. See what you like (470R or 235R).
If you will also try 1 Ohm R1 remove the second 470R over R9. Theoretically even better is R9 1k when you allow so much spare current (and heat) since using R1 1 Ohm (about 0.55-0.6A CCS) for 0,22-0.25A load. What R9 does in technical terms is influencing the gain margin from zero gain to -180deg oscillation point. Higher value makes the reg extend shorter but to also work less on the edge in the several MHz ultrasonic region in simpler words.
As it is with 1.5 Ohm R1 I suggest you try another 470R over R9 (parallel) and listen first. See what you like (470R or 235R).
If you will also try 1 Ohm R1 remove the second 470R over R9. Theoretically even better is R9 1k when you allow so much spare current (and heat) since using R1 1 Ohm (about 0.55-0.6A CCS) for 0,22-0.25A load. What R9 does in technical terms is influencing the gain margin from zero gain to -180deg oscillation point. Higher value makes the reg extend shorter but to also work less on the edge in the several MHz ultrasonic region in simpler words.