I have the Arcam rDac, which has a 6V 600mA power requirement.
I have been running the DAC with the supplied SMPS wall wart, as well as a second wall wart. Both power supplies give off a reasonably loud high frequency noise, and I'd like to build a dedicated Linear Power Supply. Similar to something like this but within a reasonable price range.
I've got a bit of electronics experience, having built/tweaked elekit amplifiers, but would like to get input/advice on the circuit I've planned for the power supply.
Are there any further improvements that could be made to the design (such as a low-noise regulator), which would be worthwhile?
Appreciating any and all feedback!
I have been running the DAC with the supplied SMPS wall wart, as well as a second wall wart. Both power supplies give off a reasonably loud high frequency noise, and I'd like to build a dedicated Linear Power Supply. Similar to something like this but within a reasonable price range.
I've got a bit of electronics experience, having built/tweaked elekit amplifiers, but would like to get input/advice on the circuit I've planned for the power supply.
Are there any further improvements that could be made to the design (such as a low-noise regulator), which would be worthwhile?
Appreciating any and all feedback!
For easy implementation, I'd suggest a three-terminal regulator from the 7806 family, eg. ba178_ba78_series_tn-e.pdf
How about a single of these?
https://www.head-case.org/forums/topic/12269-goldenreference-low-voltage-power-supply/
https://www.head-case.org/forums/topic/12269-goldenreference-low-voltage-power-supply/
hello, arcam rdac already has classic linear regulators inside (5 ; 3 V etc.).
I built a normal external linear regulator for mine using LD linear regulator, works fine (I additionally use ferrite filters).
I draw your attention that Rdac is very sensitive to the supply voltage, it should be exactly 6V ( or better 6.1) under load, if it is 5.9 - it will work unstable, there will be no sound, even though all the indicators are on.
I built a normal external linear regulator for mine using LD linear regulator, works fine (I additionally use ferrite filters).
I draw your attention that Rdac is very sensitive to the supply voltage, it should be exactly 6V ( or better 6.1) under load, if it is 5.9 - it will work unstable, there will be no sound, even though all the indicators are on.
Attachments
Hi,
It is a standard LD1086 datasheet circuit (matched resistors so that the output is @ 6v). At first I used a fixed 6V regulator (L7806CV) but it had too much voltage drop and the power supply was not stable with my transformer (6VAC). That's why I switched to a low-drop regulator. Everything else is standard - a rectifier, some 4700uf capacitor, I added an extra choke on a ferrite core after capacitors (it seems from an old ATX power supply unit) - to reduce possible RF noise.
It is a standard LD1086 datasheet circuit (matched resistors so that the output is @ 6v). At first I used a fixed 6V regulator (L7806CV) but it had too much voltage drop and the power supply was not stable with my transformer (6VAC). That's why I switched to a low-drop regulator. Everything else is standard - a rectifier, some 4700uf capacitor, I added an extra choke on a ferrite core after capacitors (it seems from an old ATX power supply unit) - to reduce possible RF noise.
It is. Kevin retired from the university, hence broken links, but links to a google docs repository are available. That said, using it for voltages below 12V isn't straightforward. It is a very good very low noise supply. Look at the last two pages of that thread for more info on using it for lower voltages.
Absolutely. I would also suggest a three terminal regulator. Why make it harder than it needs to be.
So called low noise supplies are all well and good in the right application (not that a 7806 isn't low noise) but you have another factor to consider here and that is that if this is a wall wart type supply (so external to the equipment) then it connected by a wire and that wire with its own resistance absolutely destroys the low output impedance and low noise you might get from a dedicated low noise supply. In other words the result at the end of the wire is totally different to what you see at the regulator output.
So why not make it easy and use a well designed circuit for a 7806 🙂
If that is an issue then the 7806 (or use a 7805) can be easily tweaked to give any higher voltage. Using a transistor to lift the ground pin is the best option imo but diodes (and or Zeners) can be used and even just a resistor for the 'economy version'. In fact the latest data sheets seem to show only the resistive version.
LM340 and 7805
The suggested LM340 might be best for use case but I suggest another from the TI 7800 spec. It has a bypass transistor to divide the heat that the TO220 case can not dissipate...
Note that TI does not need 7806 because you can lift the 7805 ground pin.
Attachments
Should be 6.2 to 6.4 v.
2.2k resistor on zener 220 uf cap across zener
14 to 18 vdc as input voltage
2.2k resistor on zener 220 uf cap across zener
14 to 18 vdc as input voltage
For lower noise than 7806 you could use an LM317. This works:
Set R1 = 121 Ω; R2 = 464 Ω and you get 6.05 V on the output. C3 should be an electrolytic capacitor as the LM317 is not ceramic stable. I'd make C3 = C2 = 10 µF.
The figure is from the TI LM317 data sheet: https://www.ti.com/lit/ds/symlink/lm317.pdf
Tom
Set R1 = 121 Ω; R2 = 464 Ω and you get 6.05 V on the output. C3 should be an electrolytic capacitor as the LM317 is not ceramic stable. I'd make C3 = C2 = 10 µF.
The figure is from the TI LM317 data sheet: https://www.ti.com/lit/ds/symlink/lm317.pdf
Tom
If you're going to use the LM317 (which is a great choice for the purpose and it's easy to precisely tweak for the voltage you want), it's a good idea to include HF bypassing capacitors from the IC pins to ground. I uaually use small ceramic capacitors of 10n to 100n. These will help to prevent RF breakthrough into the IC (like a local AM station or even the tiny spark created by switching your room lights). The power supply should ideally be housed in an earthed metal box. I would also make R1 an 82Ω resistor in series with a 15-turn 100Ω cermet preset, to allow really precise adjustment of the output voltage.
NJM7806 datasheet: https://www.mouser.com/datasheet/2/294/NJM7800_E-1917431.pdf
LM317: https://www.ti.com/lit/ds/symlink/lm317.pdf?ts=1722485609747&ref_url=https%3A%2F%2Fwww.ti.com%2Fproduct%2FLM317%3Fbm-verify%3DAAQAAAAJ_____36_Lmicni9uffn-HZ-QOTIZN1Nx6f03SpZ8Ckk6OMy8PmFJbiz4Ik5hbl4CfDsMUNWNEIr7IdlmdDfC7PmZ7brxj6jFOIkECjiZTx8j97I4Kk9dwynrojoAQUMSCdDdlsrgXhBXhOVDRKCKoUgdlJOHzRtQ2ORlYphnpm_EFbiatRtbQcfTST_MfCHtl_sbVQel4Lmv6X2Zy7g3SEmfNFhY1i9uDmTzd9ID3ftp_WI98X6Kj6nm0nVzchF8sT-HjzAPHP_0aQ1_LgC4FHjgRd7VAz5tVXcPYVmEHzebmQwzWwf_lmefiQ
Looks to me like noise is spec'ed over different bandwidths and other different criteria.
May we know what comparison criteria you are using?
Other than that, 78xx series has an old reputation for sounding "better" than LM317, whatever that means and for whatever manufacturing processes were being used back then.
I guarantee you that the regulators sold today are not made on the same process as they were in, say, 1976 when the LM317 first saw the light of day. So you'd have to compare two modern versions of the LM7806 and LM317 with the resistors set for 6 V out.
I probably used the LM317's old reputation for being low noise rather than doing the math. Here's the math:
The National Semiconductor version of the LM317 (LM317-N) is spec'ed to have 0.003%Vo of RMS output noise. That works out to 180 µV RMS for 6 V output.
It looks like the LM78xx family has been pruned substantially. I was able to dig up an L7806 from ST Micro. It's spec'd to 10µV/Vo, which I interpret as 0.001% per Vo -> 0.001%Vo -> 60 µV RMS for 6 V output.
Both are spec'ed for 10 Hz - 100 kHz measurement bandwidth.
So it turns out the L7806 is lower noise. How 'bout that. Either way, I doubt it matters in this case. The Arcam DAC being powered likely doesn't use the 6 V directly. Rather, it derives various internal rails from it. So the point is likely moot.
I've yet to hear any sound emitted by an LM317 or 7806. Maybe I've been using them wrong. 😉 And also, as you love to point out, how something sounds can't possibly be captured by a single number. Maybe there's more to regulator sound (assuming that's even a thing) than RMS output noise. Right...?
Tom
I probably used the LM317's old reputation for being low noise rather than doing the math. Here's the math:
The National Semiconductor version of the LM317 (LM317-N) is spec'ed to have 0.003%Vo of RMS output noise. That works out to 180 µV RMS for 6 V output.
It looks like the LM78xx family has been pruned substantially. I was able to dig up an L7806 from ST Micro. It's spec'd to 10µV/Vo, which I interpret as 0.001% per Vo -> 0.001%Vo -> 60 µV RMS for 6 V output.
Both are spec'ed for 10 Hz - 100 kHz measurement bandwidth.
So it turns out the L7806 is lower noise. How 'bout that. Either way, I doubt it matters in this case. The Arcam DAC being powered likely doesn't use the 6 V directly. Rather, it derives various internal rails from it. So the point is likely moot.
I've yet to hear any sound emitted by an LM317 or 7806. Maybe I've been using them wrong. 😉 And also, as you love to point out, how something sounds can't possibly be captured by a single number. Maybe there's more to regulator sound (assuming that's even a thing) than RMS output noise. Right...?
Tom
Thank you for checking on that Tom 🙂
Regarding regulator sound, to the extent audio circuits have less than ideal PSRR could be a voltage regulator has some influence on sound. Various parameters come to mind including output Z versus frequency, etc. SMPS noise, maybe. Also, in some cases it sure seems like regulator error amplifier characteristics (loop gain, distortion, etc.), may come through to affect the circuitry being powered. Some people are convinced such things can happen and maybe do happen more often than some other set of folks may expect. As with many things in audio, probably it depends.
Regarding regulator sound, to the extent audio circuits have less than ideal PSRR could be a voltage regulator has some influence on sound. Various parameters come to mind including output Z versus frequency, etc. SMPS noise, maybe. Also, in some cases it sure seems like regulator error amplifier characteristics (loop gain, distortion, etc.), may come through to affect the circuitry being powered. Some people are convinced such things can happen and maybe do happen more often than some other set of folks may expect. As with many things in audio, probably it depends.
Last edited:
Perhaps "sounds" which are not part of the music input could be exposed by testing with REW. Or bypass the regulators with batteries. Then have a different person do the same tests.
It can be subtle stuff in sound that can be affected.
Steady state sine wave spectral analysis doesn't always provide sufficient insight into dynamics and transients. Thus, REW may not tell the whole story very well.
Beyond that, typical audio spectral analysis does not show phase information of audio signals nor of noise (at least not in an intuitively useful way). On an FFT its possible for noise to look like low level white noise, yet sound like loud popping or frying noise. Or it could sound like subtle hiss. Such can all be different types of white noise. However, it may take some study of how DFTs and FFTs work to understand why they can all look the same.
In addition, ESS says there are a number of things about sigma delta dac sound that do not show up in typical spectral analysis (which ESS refers to as PSS measurement).
Don't want to get too off topic here; if anyone has any interest in a deeper understanding of what audio FFTs show well, what they can show quantitatively and or non-quantitatively, and what they don't show very well at all, then maybe a new thread should be opened for that.
Steady state sine wave spectral analysis doesn't always provide sufficient insight into dynamics and transients. Thus, REW may not tell the whole story very well.
Beyond that, typical audio spectral analysis does not show phase information of audio signals nor of noise (at least not in an intuitively useful way). On an FFT its possible for noise to look like low level white noise, yet sound like loud popping or frying noise. Or it could sound like subtle hiss. Such can all be different types of white noise. However, it may take some study of how DFTs and FFTs work to understand why they can all look the same.
In addition, ESS says there are a number of things about sigma delta dac sound that do not show up in typical spectral analysis (which ESS refers to as PSS measurement).
Don't want to get too off topic here; if anyone has any interest in a deeper understanding of what audio FFTs show well, what they can show quantitatively and or non-quantitatively, and what they don't show very well at all, then maybe a new thread should be opened for that.
People can be convinced of the strangest things. Humans are not rational critters. That's why it's important to differentiate anecdotal evidence from scientific evidence.
The output impedance of batteries can be pretty horrid, especially when the battery is almost drained of charge. So batteries may not provide better performance than regulators.
Tom