"And then there is the marketing angle - I'm sure you guys are aware that often designers lay out their PCBs and parts with the beautiful internal pictures that Stereophile will print in mind ;-)
Jan"
TRUE!
I remember such a circuit where we found that the right channel had some audible hum after using 200 dollars worth of regulators
At least it looked really good for the press !
Jan"
TRUE!
I remember such a circuit where we found that the right channel had some audible hum after using 200 dollars worth of regulators
At least it looked really good for the press !
If people woldn't be prone to snobism engineers wouldn't be leaning to fake performance and you know that too, don't you?
We live in an era where we wouldn't need to reinvent anything to build ourselves the best possible audio setup just by using expired patents...yet lots of manufacturers pretend for at least 15 years to bring something better than before and there's no hard evidence for that.
We live in an era where we wouldn't need to reinvent anything to build ourselves the best possible audio setup just by using expired patents...yet lots of manufacturers pretend for at least 15 years to bring something better than before and there's no hard evidence for that.
The 7815 and 7915 have been good enough for high-end studio mixing consoles for decades now, at least when they put regulators on the individual channels strips. Not all of them do- many just rely on the regulation from the main power supply (which probably uses a uA723).
Now... I can understand the dislike for the 78/7915. Their manufacturing tolerance isn't great, and the noise isn't as good as it could be. That is all solved using the LM317 and LM337.
Designing a power supply entirely from discrete components (especially without any op-amps) is still worthwhile as a design exercise (and not as trivial as some might think), but for most things there's really no point. Sometimes a good old geriatric uA723 is a good choice too, especially for big linear power supplies.
If you need to regulate high voltages, the LR8 is a very cool device that can work really well. You can even go above 450 volts by floating it. If it doesn't suit your needs, then you may choose to use op amps and transistors to accomplish what you need.
Keep in mind that designing a good regulated power supply using discrete parts is not as easy as one might think (and I don't mean wacking a single zener and pass transistor at it), and it's very easy to build a 2 MHz signal generator instead of a power supply.
Now... I can understand the dislike for the 78/7915. Their manufacturing tolerance isn't great, and the noise isn't as good as it could be. That is all solved using the LM317 and LM337.
Designing a power supply entirely from discrete components (especially without any op-amps) is still worthwhile as a design exercise (and not as trivial as some might think), but for most things there's really no point. Sometimes a good old geriatric uA723 is a good choice too, especially for big linear power supplies.
If you need to regulate high voltages, the LR8 is a very cool device that can work really well. You can even go above 450 volts by floating it. If it doesn't suit your needs, then you may choose to use op amps and transistors to accomplish what you need.
Keep in mind that designing a good regulated power supply using discrete parts is not as easy as one might think (and I don't mean wacking a single zener and pass transistor at it), and it's very easy to build a 2 MHz signal generator instead of a power supply.
It's a great part, but it is still over 50 years old. For an electronic component, that qualifies as geriatric.
Shindo Labo is using IC integrated CCS all the time in their tube amps and with good effect.
Its always know how to do it to sound good. Its never a single feature or implementation of an idea that makes a whole unit sound good or bad. Its the summary of wrongs or rights that makes the difference.
Its always know how to do it to sound good. Its never a single feature or implementation of an idea that makes a whole unit sound good or bad. Its the summary of wrongs or rights that makes the difference.
Last edited:
86 uV RMS typical from 100 Hz to 10 kHz at 5 V output voltage when you don't filter it, according to https://www.ti.com/lit/ds/symlink/lm723.pdf , so that's almost 860 nV/sqrt(Hz). An 85A2 glow discharge voltage reference tube has less noise at a 17 times higher voltage.
LM317/337 is not the solution. In fact, it does not make much of a difference
which one of the crowd you use. Noisewise, they are all in a 10 dB band.
LT3042, 3045, 3094? for neg. is a completely different league.
< VoltageRegulators | Output noise of some voltage regulators.… | Flickr >
Zeners at <= 3V3 are wonderful. The onset of avalanching already at < 5V
destroys their noise behaviour. The TC compensation at 6V8 relies on the
opposite TC of the Zener and the avalange effect. Those diodes are
temp-stable, but they are noise generators. Some examples are in the
pictures to the right/left.
The 723 with it's 6V zener is not so wonderful. OK, it can stand it's ground
against the noisy bandgaps. I wonder what tricks LT did pull to compete
successfully with the Zeners in the LT6655.
Cheers, Gerhard
Last edited:
Spend another dollar and install a passive filter between the VREF generator (buried Zener, avalanche diode, zener diode, LM329, bandgap, Jung GLED431) and the error amplifier. Voila, noise is reduced to negligibility.
Spend another three dollars and install an ACTIVE filter before the passive filter. Voila, noise is now obliterated.
Spend another three dollars and install an ACTIVE filter before the passive filter. Voila, noise is now obliterated.
actually there are countries where people can't by anything like that with 4 dollars.There are lots of things that cost 4 dollars in the US and 6 dollars in some parts of Europe.In Romania i have to pay an aditional 20...25 dollars for the transport of that lt3042 ...I just checked mouser and digikey and i need to pay abut 30 dollars in total for that lt3042.Then it's smd...which makes for the fact that i need to pay an additional 20 dollars for the minimum pcb quantity i can get from china and i also need to spend some hours to prototype that pcb too...All in all i pay 50 dollars for an application of that special IC...
Last edited:
+1000
Because usually they do not know the PSRR of their circuit, they do not know the target of signal to noise ratio. Even they do not care if at transient signal, some PSU can oscillated.
Just stick to IC regulator, because they do not know how to design better than IC regulator or lazy to compare better discrete regulator with IC regulator
I've mentioned this before in some other power supply thread a few years back: It seems that "center tapped" power transformers have been replaced in recent decades by dual secondaries (with effectively the center tap cut and both connections brought out). Yes, you can connect the opposite phases of the two windings together to make the common connection in the "typical" supply with complementary voltage regulators for + and -, but there's a good option I don't see people taking advantage of.
With just using two bridge rectifiers, one for each secondary, you can make two identical, isolated, filter-and-regulator sections whose outputs can be put in series for symmetrical +/- DC outputs. One can start using the better performing (usually positive) of the two complementary regulator types, and add whatever super-regulator stuff you want. This does so much for either one-off personal use (simplicity, you can make the exact same circuit twice instead of two nearly-identical circuits where you have to keep track of different polarities) or commercial production (reducing the number of different types of components stocked), that it seems it easily pays for the four extra rectifiers.
With just using two bridge rectifiers, one for each secondary, you can make two identical, isolated, filter-and-regulator sections whose outputs can be put in series for symmetrical +/- DC outputs. One can start using the better performing (usually positive) of the two complementary regulator types, and add whatever super-regulator stuff you want. This does so much for either one-off personal use (simplicity, you can make the exact same circuit twice instead of two nearly-identical circuits where you have to keep track of different polarities) or commercial production (reducing the number of different types of components stocked), that it seems it easily pays for the four extra rectifiers.
It's not so different, as the LT3042 also gets its good noise performance from low-pass filtering between the reference and the rest. When using a filtered reference as a DAC reference, the noise below the cut-off frequency still causes sidebands around the desired signals, but you can make that cut-off frequency quite low if you want.
I don't care what DIYers do; I prefer following Rohde, Hill&Horrowitz
or Jim Williams. The few pearls like Andrea's won't go unnoticed.
I's outright dumb to throw out a part that is 40 dB better right from the start.
LT4042 is handicapped at low frequencies by the preamp and
FFT analyzer noise.
Attachments
Last edited: