I was reading the past few days about different supply designs. I'm experimenting with mic preamp designs, and later will do a small 5-6 channel mixer. Looking for +-12 or +-15v.
I would appreciate it if someone could set me straight about which sounds best and a little bit as to the whys.
As I understand it the best approach would be center tap mains transformer --> secondary ~3 volts over supply voltage ---> preregulation ---> filtering/smoothing/more regulation
?
Something like the Belleson superpower, superteddyreg, or the newer lt regulators like lt30xx fixed outputs seem like the best options.
The audiophile people seem to avoid any switching converters like the plague, but then I've read if it's done very carefully you can get decent or good results.
However I am trying to keep things ultra-small and portable and much prefer using batteries if at all possible, and I'm unclear on the some potential sonic disadvantages of batteries and pitfalls of using rail splitters or virtual grounds. Ultimately I'd like everything on one small-as-possible board: power supply and preamps, but this might not be possible.
Keeping things small it looks like possibilities are
-external battery-->switching step up converter--> a lot of smoothing/regulation
-center tap wall wart-->diy belleson/superteddyreg type thing
-larger battery pack with rail splitter/divider
also these larger usb rechargeable battery packs sold for tablets that put out 5,9,12 and other voltages, are these noisy / switch converter outputs unsuitable for audio?
I'm getting better at keeping the preamp footprints very small - leaving more room for power...trying to keep everything altoid-size, but this might not be possible going for good bipolar power.
I would appreciate it if someone could set me straight about which sounds best and a little bit as to the whys.
As I understand it the best approach would be center tap mains transformer --> secondary ~3 volts over supply voltage ---> preregulation ---> filtering/smoothing/more regulation
?
Something like the Belleson superpower, superteddyreg, or the newer lt regulators like lt30xx fixed outputs seem like the best options.
The audiophile people seem to avoid any switching converters like the plague, but then I've read if it's done very carefully you can get decent or good results.
However I am trying to keep things ultra-small and portable and much prefer using batteries if at all possible, and I'm unclear on the some potential sonic disadvantages of batteries and pitfalls of using rail splitters or virtual grounds. Ultimately I'd like everything on one small-as-possible board: power supply and preamps, but this might not be possible.
Keeping things small it looks like possibilities are
-external battery-->switching step up converter--> a lot of smoothing/regulation
-center tap wall wart-->diy belleson/superteddyreg type thing
-larger battery pack with rail splitter/divider
also these larger usb rechargeable battery packs sold for tablets that put out 5,9,12 and other voltages, are these noisy / switch converter outputs unsuitable for audio?
I'm getting better at keeping the preamp footprints very small - leaving more room for power...trying to keep everything altoid-size, but this might not be possible going for good bipolar power.
The issue with opamps is relatively poor PSRR. This can be addressed in a number of ways, however running from batteries narrows down the options. I've been using arrays of ICL7660 to do step-up converters as I figure inductors are more likely to cause interference in a small envelope than caps. I also prefer local shunt regulators (TL431 and transistor assisted variant of same). Where opamp loading is relatively heavy (such as you'll have in the first stage of a mic pre) then use a current source to bias outputs into classA, perhaps consider using balanced paths to equalize the load current variations.
<edit> In the past couple of days I've acquired two of those power bricks you referred to (5V USB output, 1A rated) and intend to have a look at their output noise levels.
<edit> In the past couple of days I've acquired two of those power bricks you referred to (5V USB output, 1A rated) and intend to have a look at their output noise levels.
Last edited:
for battery power you don't need regulation.
If you use two 9Vdc batteries in series you end up with a dual polarity +-9Vdc supply.
Many chips will work from that. Some will operate down to +-2.5Vdc.
You could use a single 9Vdc battery, either in a single ended topology, or as a +-4.5Vdc with a virtual zero volts ground.
If you want to work with a lower battery voltage, then you will need some kind of DC to DC step up converter.
This puts you into switching, i.e. smps, although Elvee recently posted a non switching converter.
Most smps can be regulated to suit your chip's voltage range.
Don't use a LED to indicate good voltage. It needlessly wastes battery power.
Only use the LED as a bad voltage indicator.
Look at ultra low quiescent current chips.
Add in a DC input plug for when mains power is available.
If you use two 9Vdc batteries in series you end up with a dual polarity +-9Vdc supply.
Many chips will work from that. Some will operate down to +-2.5Vdc.
You could use a single 9Vdc battery, either in a single ended topology, or as a +-4.5Vdc with a virtual zero volts ground.
If you want to work with a lower battery voltage, then you will need some kind of DC to DC step up converter.
This puts you into switching, i.e. smps, although Elvee recently posted a non switching converter.
Most smps can be regulated to suit your chip's voltage range.
Don't use a LED to indicate good voltage. It needlessly wastes battery power.
Only use the LED as a bad voltage indicator.
Look at ultra low quiescent current chips.
Add in a DC input plug for when mains power is available.
<Looking for +-12 or +-15v.>
I'm working on a bunch of different designs that I didn't mention that require the above. Also I'm sticking with certain op amps that sound best to me and are recommended by other trusted ears - usually higher Iq. For battery power I'm not interested in economy of power but smaller size in relation to power transformers and more complex regulating schemes.
Another reason for trying to get a good power solution is to further design experiments and help to rule out bad power as a variable.
I'm working on a bunch of different designs that I didn't mention that require the above. Also I'm sticking with certain op amps that sound best to me and are recommended by other trusted ears - usually higher Iq. For battery power I'm not interested in economy of power but smaller size in relation to power transformers and more complex regulating schemes.
Another reason for trying to get a good power solution is to further design experiments and help to rule out bad power as a variable.
Last edited:
Thanks guys,
Andrew, that supply looks good, but a bit large. If I could rework the layout and shrink then maybe, but...
The sjostrom jsr02 x 2 looks about perfect. two and room for a little pregregulation board will fit in an altoids. So an a/c wallwart into a simple preregulator feeding jsr02's ???
Andrew, that supply looks good, but a bit large. If I could rework the layout and shrink then maybe, but...
The sjostrom jsr02 x 2 looks about perfect. two and room for a little pregregulation board will fit in an altoids. So an a/c wallwart into a simple preregulator feeding jsr02's ???
Which datasheets are you looking at?! Many, many modern (and inexpensive) audio opamps, including several low power ones, have quite high PSRR. To go in with that assertion largely invalidates your solution to a (largely) nonexistent problem.
E.g., figure 4: http://www.ti.com/lit/ds/symlink/opa1642.pdf (between 50-60 dB @ 20 kHz on the negative rail), even oldies like the OPA2132 are 50 dB at 20 kHz. Something like the LM4562 is even lower, but you'd probably be looking at fet-input opamps for a mic preamp.
If you're trying for the last N'th of performance for something like an instrument preamp, yes, regulation makes sense, but to suggest opamp PSRR is "poor" is, well, a very "poor" word choice and far more likely to send someone down the wrong path than be helpful.
OP: go look at Scott Wurcer's, syn08, and Samuel Groner for some very high quality preamp circuits (that maintain high psrr to boot, especially at lower gains). Joachim Gerhard (sp?) is probably another individual worth doing a search on. I'm sure there are others (not meaning to slight anyone!), but those were the guys that came to the top of my head right away.
E.g. http://www.diyaudio.com/forums/solid-state/165961-simple-60db-discrete-low-noise-amplifier-lna.html
http://www.diyaudio.com/forums/soli...1000-low-noise-measurement-amp-ikoflexer.html
They'd need different compensation for lower gains, but might get you off to a good start. The architectures given above are for G=1000 at extremely low noise (but may not load your cartridge correctly).
I agree that the datasheets show in many cases impressive figures for PSRR. But those plots are often the result of simulation rather than test and the conditions under which they are run don't mirror the situations under which people use opamps. So take the plots with a pinch of salt.
One of the reasons that they're not fully indicative of practice is they're normally taken without a load on the output (LM4562 is a notable exception to this). Another difference is they're taken with no signal applied - when I ran a simulation myself I found this makes a fairly large difference in practice due to the output stage being lightly biassed classAB in most cases. So if you measure PSRR with no signal then both output transistors are contributing but in practice one or other of those transistors is cut off most of the time.
I share my experience when I say its 'poor' - getting good sound from opamps means addressing this issue. Opamps tend to have a somewhat 'poor' reputation amongst audiophiles and my hunch is this PSRR issue is a significant contribution to that. Fortunately its not that difficult to fix up when the cause is understood.
Discrete opamps I'm very interested in - note they're often run at much higher biasses in the OPS than integrated ones, this helps a fair bit with PSRR in practice I think.
That, if anything, has me using more of them on a regular basis. Did you take Mooly's cascaded opamp test recently? He loaded the heck out of them and did nothing special psu-wise. Most of us could, if anything, barely tell a difference (and that was mostly noting the TL072 in the mix, which is output current starved). And that was with 5 cascaded.
http://www.diyaudio.com/forums/ever...-trying-understand-what-we-think-we-hear.html
Samuel Groner's work on opamps lays out few compliments towards any of the discrete opamps he tried.
Methinks the lady doth protest too much
No I didn't try Mooly's cascaded opamp test - could not see the point. I have studied Samuel Groner's work on opamps - the discrete ones he tested I'd also not recommend. Weren't they Burson's ? Burson has been running something of a hate campaign on opamps for their own marketing purposes but recently they included one in a design (inside a digital volume control). It was fun to read their rationalization of that.
To which lady does the honourable gentleman refer?
To which lady does the honourable gentleman refer?
Help I was hijacked
What do you guys think of this
Discrete Voltage Regulators - Sparkos Labs.com
for a good low profile solution
What do you guys think of this
Discrete Voltage Regulators - Sparkos Labs.com
for a good low profile solution
All data available points to very good performence in these, so if you can afford them and have an idea of how to get rid of Pdiss, by all means.
Don't skimp too much on input (smoothing) capacitance though, at least if you plan on using a mains xmfr that isn't vastly oversized. All the PSRR in the world is for naught if you drop out of regulation. Consider CRC or CLC filtering if you're not much into big caps.
Don't skimp too much on input (smoothing) capacitance though, at least if you plan on using a mains xmfr that isn't vastly oversized. All the PSRR in the world is for naught if you drop out of regulation. Consider CRC or CLC filtering if you're not much into big caps.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.