Re: Help
Hi again,
Having done some more searching I think I've found the correct 1085 device, namely, - the LT1085CT (1.25v to 28.5v adjustable), available from RS.
But now though I'm having problems sourcing the LM337 from National Semiconductor. It appears they have been taken over by Texas Instruments. So do I still need to source one from Nat.Semi. or is the T.I. device going to be suitable?
Apologies for what may seem dumb questions to some.
Thanks.
Hi again,
Having done some more searching I think I've found the correct 1085 device, namely, - the LT1085CT (1.25v to 28.5v adjustable), available from RS.
But now though I'm having problems sourcing the LM337 from National Semiconductor. It appears they have been taken over by Texas Instruments. So do I still need to source one from Nat.Semi. or is the T.I. device going to be suitable?
Apologies for what may seem dumb questions to some.
Thanks.
Nearly there!
Hi folks,
Thanks for you're help this far. I have one more question then I think I should be okay.
I'm working from the drawing in post #248. Those resistors marked *, should my load (ie phono pre) be connected when deciding what value to use? - otherwise the value would just be 15V/ 0.046A = 326ohms, wouldn't it?
Regards.
Hi folks,
Thanks for you're help this far. I have one more question then I think I should be okay.
I'm working from the drawing in post #248. Those resistors marked *, should my load (ie phono pre) be connected when deciding what value to use? - otherwise the value would just be 15V/ 0.046A = 326ohms, wouldn't it?
Regards.
Dear ,John Bau, I have blue ray oppo 095 and I am planning to modify the power suply for the analoge section (the main power suply) by substituting the LM317T with the better ones LT1085, trying to implement your improved shematic.
The only problem is that due to the fact that the pcb that holds the power suply is double siteda and somehow tiny , I cannot put in series after the regulator the 20mΩ resistor you suggest .
I wonder whether this is a big problem, that avoids the hole effort to improve the power supply of my player or the regulators can work fine even with this omission ,with lets say, the 80% of the benefit that someone would expected by this changes. Or there is any other way to overcome this problem. Thanks Andy. Chara
The only problem is that due to the fact that the pcb that holds the power suply is double siteda and somehow tiny , I cannot put in series after the regulator the 20mΩ resistor you suggest .
I wonder whether this is a big problem, that avoids the hole effort to improve the power supply of my player or the regulators can work fine even with this omission ,with lets say, the 80% of the benefit that someone would expected by this changes. Or there is any other way to overcome this problem. Thanks Andy. Chara
Meanwhile this advice is not wrong but I don't think that this statement corelates with jbaus' experiment very well. He already noticed that his actual load is more than 100mA. So, required resistors' purpose must be providing at least 45mA and equal/balanced load on each rail, i think.
Hi,
If I wasnt out in the world right now, I would put up my article called Power Up... an overview of power supply considerations --- I noted that the linearlity of the series reg IC was highly non-linear at low current draw.... preloading, as I called it, moved the operating point to a more linear place. But there is also the fact noted that the output is an emitter follower and more current draw (more class A) increased the BW of the IC and thus the 'speed' and lower Z vs freq etc.
Lately, the better use for them is in a push-pull config were the outputs (+/-) are tracking one another -- also called master-slave.
With pre-reg and push-pull class A operation in tracking configuration you get the best possible of series/shunt operation and benefits of tracking operation to minimize diff levels to the circuit being powered.
Lets design it now. [I have circuits but not with me.... but the great designers here can figure it out]
THx-RNMarsh
If I wasnt out in the world right now, I would put up my article called Power Up... an overview of power supply considerations --- I noted that the linearlity of the series reg IC was highly non-linear at low current draw.... preloading, as I called it, moved the operating point to a more linear place. But there is also the fact noted that the output is an emitter follower and more current draw (more class A) increased the BW of the IC and thus the 'speed' and lower Z vs freq etc.
Lately, the better use for them is in a push-pull config were the outputs (+/-) are tracking one another -- also called master-slave.
With pre-reg and push-pull class A operation in tracking configuration you get the best possible of series/shunt operation and benefits of tracking operation to minimize diff levels to the circuit being powered.
Lets design it now. [I have circuits but not with me.... but the great designers here can figure it out]
THx-RNMarsh
Last edited:
Readers of Gray and Meyer's analog design textbook will no doubt recall that Gm is linearly proportional to Ic, even in the (emitter follower) series pass transistor of a voltage regulator. Higher output current gives higher Gm in the output device, which reduces its Re and increases loop bandwidth.
Attachments
LT1085 & LM337 regulator
With all the consideration of power supplies based on other regs, I'm a bit curious why very few have picked up the result of jbau's work with the LT1085 & LM337 combo and carried it forward or done any discussion of builds, layouts, or PCB's. Note post 248.
The only real outside reference I can find are a couple mentions by terranigma in 2014.
With all the consideration of power supplies based on other regs, I'm a bit curious why very few have picked up the result of jbau's work with the LT1085 & LM337 combo and carried it forward or done any discussion of builds, layouts, or PCB's. Note post 248.
The only real outside reference I can find are a couple mentions by terranigma in 2014.
Can't speak for others but I myself have gradually shifted to the multiple-stages-in-cascade design approach. First stage performs MHz to GHz RFI filtering. Second stage performs 2xfmains ripple elimination. Now the third stage can be optimized for low Zout, low noise, and wide bandwidth, knowing that its input is quite clean and free of anomalies. This costs extra money, extra PCB area, extra VA in the transformer, and extra design time. I don't mind.
Most probably because there are much better regulators or circuits for super-regulations for audiophile purposes nowadays.
That, or else all the skeptics didn't even try or discouraged others
jbau's findings are still very cool though, and something you would want to project even with the newer regs.
Have you known any other super duper reg design that have similar measurements here? I would be really interested if there any. Maybe the "mystery" that attained by over-engineered designs correlates with the "audiophile" concept perfectly, but this thread's main subject is a bit different.
I don't think I saw other measurements for other Regs like jbau did.
What I do know though, is when I built my DIY Linear Regulated PSU for my DAC using the LM317, I was asking questions around about a Super-Reg configuration for the LM317 and got some really 'jerk'-level replies from someone whose best advice was to read the datasheet.
As if the datasheet shows 'audiophile applications...
And secondly, when I was experimenting with various post Regulator capacitors, as well as revising the grounding/return scheme as in star-grounding properly, there were huge differences in sound quality, just by listening.
Measurements would certainly show up the discrepancies.
I did simulate my circuit and iterations with LTSpice and correlated that with what I heard.
It's quite enlightening (especially what happens with Bode plots and aural effects after the conventional 'audio-range'), and I would tend to agree about: aiming for lowest impedance as possible, and jbau's matching impedance if you use two lines (for my DAC, I only needed a +5v one as it's usually powered by USB which is unbelievably noisy).
The non-linearities above the conventional audio-range (conventional here means that many will stop any measurement efforts right there) can make other components in the device or the chain go into resonance/oscillations causing damage to overall SQ.
So, most of the time, you are actually battling the LM317's various weaknesses to achieve OK results, so that's why I think other regs and super-regs (like the Salas which is well-known around here) would be more suited.
However, with all its foibles, if you use the LM317 and try to understand and optimise your PSU, you actually learn many, many things along the way.
After I finished my PSU, a good while after, I saw a conference about power integrity and signal integrity and it appears I was actually learning to do the Power Integrity thing on my own. Much fun.
Checking various circuit modifications and looking at the Bode plots/Impedance graphs in LTSpice is a very interesting thing to do.
By the way, that's also an excellent question.
I have found (but many know this already) that when you get to a certain very good level of performance sound-wise with regulated or super-regulated circuits, things like proper grounding and PCB layout become super important.
Just proper PCB layout is a whole subject in itself. I briefly researched this on my own this year while learning KiCAD.
And it was interesting to see that much of the Nemesis I have been battling for a while at a macro-level in my audiophile setup, I also have to battle at the PCB Level - EMI/RFI, more generally EMC. And some new things as well: thermal effects.
Really fun stuff.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.