LPSU choices, new contender.....

[The Computer Audiophile]

Breaking news: @activist38 is @valveboy and @the_doc735  

 

who new 😁

 

 

[BigAlMc]

25 minutes ago, The Computer Audiophile said:

Breaking news: @activist38 is @valveboy and @the_doc735  

 

who new 😁

 

 

 

Sigh....

 

His rants, right after I defended his and the manufacturers right to a little more courtesy had me thinking it was defo him.

 

Too many "I'm new" comments mixed in with replies where he seemed to know who he was replying to. That and shall we say his somewhat unique style 🙄

 

Screw it. I stand by the civility and benefit of the doubt argument. Even if in this case the benefit of the doubt is resulting in a third conviction!

 

[I'd say a third conviction and incarceration but he's proven determined and adept at breaking out of pokey and coming back! 😉]

 

 

[4est]

24 minutes ago, The Computer Audiophile said:

Breaking news: @activist38 is @valveboy and @the_doc735  

 

who new 😁

 

 

WTF, really... I feel silly for defending him. Ugh.

[Lebouwsky]

39 minutes ago, BigAlMc said:

Screw it. I stand by the civility and benefit of the doubt argument. Even if in this case the benefit of the doubt is resulting in a third conviction!

 

[Jud]

2 hours ago, marce said:

Instead of spitting your dummy out of the pram, why not answer some of the points, it would gain more respect for your products...

 

 

This entire thread was worth it just to read the expression "spitting your dummy out of the pram."  

[jabbr]

18 hours ago, psjug said:

I don't get the argument over parts packages.  Just use what works for you

More important to look at ESR/ESL particularly if we are talking about bypass caps and high speed digital — so the application is important — basically SMD beats through hole in “lead” inductance, also it’s harder and harder to find old style jfets but the little critters are cheap and a roll will last a lifetime.

 

Some of the new packages have much lower inductance.

[Ralf11]

4 hours ago, The Computer Audiophile said:

Breaking news: @activist38 is @valveboy and @the_doc735  

 

who new 😁

 

 

 

Sandy nailed that right away - kudos

[marce]

13 hours ago, jabbr said:

More important to look at ESR/ESL particularly if we are talking about bypass caps and high speed digital — so the application is important — basically SMD beats through hole in “lead” inductance, also it’s harder and harder to find old style jfets but the little critters are cheap and a roll will last a lifetime.

 

Some of the new packages have much lower inductance.

Not just high speed digital, there is a lot of high speed analogue and microwave is creeping down onto (almost) standard PCB's (still specialist construction with one or both outer layers for the CPWG style routing. Power delivery is critical in all electronics and what goes on at a board level is far more important to the circuit operation than whether the main PSU is linear or switched. Most digital based designs will have several local PSU's, higher current will be SMPS, lower current and sensitive will be local low noise LDO's, some circuitry will have a filtered local supply from the main supply. Critical crystals (I am talking microwave, phase array) will have the PSU on the opposite side of the board as close as you can get it, effectively the power going up through the board stack, minimal inductance.

There are endless books and information on power delivery systems. We do on average at least on course or seminar every two years on the stuff, just to try and keep up...

When digital circuits switch, the instantaneous power does not come from the main PSU it comes from the local capacitance to the device, on die capacitance, any lead capacitance, PCB power plane capacitance, local decoupling capacitor capacitance, reservoir capacitors, power supply. All those are listed in order of importance and also the order of inductance, lowest first...

[4est]

3 hours ago, marce said:

Not just high speed digital, there is a lot of high speed analogue and microwave is creeping down onto (almost) standard PCB's (still specialist construction with one or both outer layers for the CPWG style routing. Power delivery is critical in all electronics and what goes on at a board level is far more important to the circuit operation than whether the main PSU is linear or switched. Most digital based designs will have several local PSU's, higher current will be SMPS, lower current and sensitive will be local low noise LDO's, some circuitry will have a filtered local supply from the main supply. Critical crystals (I am talking microwave, phase array) will have the PSU on the opposite side of the board as close as you can get it, effectively the power going up through the board stack, minimal inductance.

There are endless books and information on power delivery systems. We do on average at least on course or seminar every two years on the stuff, just to try and keep up...

When digital circuits switch, the instantaneous power does not come from the main PSU it comes from the local capacitance to the device, on die capacitance, any lead capacitance, PCB power plane capacitance, local decoupling capacitor capacitance, reservoir capacitors, power supply. All those are listed in order of importance and also the order of inductance, lowest first...

Although I do not disagree with these statement, I do not see how this is apropos through the meter long cable from this external power supply to the load. This seems especially true considering that not only are we fighting the lead network IRL, but that there is likely some sort of regulation between this power supply and the load as per your example. Perhaps I am daft, but it would seem to me that an external power supply cannot react at these frequencies. I am sure you know way more than I, and I'd like to understand. Perhaps you are speaking in generalization?

[jabbr]

3 hours ago, 4est said:

Although I do not disagree with these statement, I do not see how this is apropos through the meter long cable from this external power supply to the load. This seems especially true considering that not only are we fighting the lead network IRL, but that there is likely some sort of regulation between this power supply and the load as per your example. Perhaps I am daft, but it would seem to me that an external power supply cannot react at these frequencies.

 

Perhaps that’s the point. An external supply is only so important, often doesn’t need to be tightly regulated (or regulated at all) and the design that we are most interested in is the quality of the transformer, as well as solid capacitance and properly filtered diodes so that Hf spikes aren’t introduced. 

 

There seems to be an an obsession with external supplies, but an external supply can’t make up for a poorly designed board, nor a poorly designed local power distribution network. 

[4est]

10 minutes ago, jabbr said:

 

Perhaps that’s the point. An external supply is only so important, often doesn’t need to be tightly regulated (or regulated at all) and the design that we are most interested in is the quality of the transformer, as well as solid capacitance and properly filtered diodes so that Hf spikes aren’t introduced. 

 

There seems to be an an obsession with external supplies, but an external supply can’t make up for a poorly designed board, nor a poorly designed local power distribution network. 

OK then, just checking to be sure. I know you and he are beyond me in this stuff.

[jabbr]

10 minutes ago, 4est said:

OK then, just checking to be sure. I know you and he are beyond me in this stuff.

No you are right on. When I build my own supplies, I spend my $$ on good transformers, and then old fashioned filtering caps, or cap multipliers and good snubbers on the diodes. ... then a cable ... on the board board being supplied is where you place the LT3045s otherwise SMD versions of the Salas shunt etc... ie on board second stage regulation

 

I don’t always want to build my own because it’s work ... so I look for cost effective external supplies ... the main issue is that they are often too whimpy ... sorry low powered ... in the wattage dept. Some basic specs are expected. 

[4est]

TY again for confirming even more. One can begin to wonder and question themselves amid the hubbub. Iv'e been quietly promoting split bobbin xfmers (and choke input filters) here for many years. I don't have the knowledge or equipment here to properly test what I assumed to be true. What I understood is extrapolated off of RF amps.

[Ralf11]

ok, re good snubbers on the diodes....  what makes it 'good'?

 

is this a noise issue in the snubber diode itself?  or a regulation, or breakdown issue?

[Superdad]

3 hours ago, 4est said:

I've been quietly promoting split bobbin xfmers (and choke input filters) here for many years. 

 

I'm a fan of split bobbin trans too, but good luck finding then in anything larger than 48VA.  And they often are mechanically noisy (lam buzz) when you pull real current through then.  Unlike R-cores, which have near the performance, lower flux field radiation, are available in larger sizes--and are DEAD SILENT!

[Superdad]

29 minutes ago, Ralf11 said:

ok, re good snubbers on the diodes....  what makes it 'good'?

 

is this a noise issue in the snubber diode itself?  or a regulation, or breakdown issue?

 

No, I think Jonathan is just talking about well chosen values for an RC network to place across the trans secondaries (or diode bridge--same thing) to kill transformer ringing. Special parts not needed. 

[jaaptina]

18 hours ago, jabbr said:

 

Perhaps that’s the point. An external supply is only so important, often doesn’t need to be tightly regulated (or regulated at all) and the design that we are most interested in is the quality of the transformer, as well as solid capacitance and properly filtered diodes so that Hf spikes aren’t introduced. 

 

There seems to be an an obsession with external supplies, but an external supply can’t make up for a poorly designed board, nor a poorly designed local power distribution network. 

Not having the knowledge but thinking from this point. Is there a point in having a power supply, (long) cable, MPAudio SD-HPULN LT3045 dc-dc converter, very short cable, dac/streamer configuration? 

 

[marce]

22 hours ago, 4est said:

Although I do not disagree with these statement, I do not see how this is apropos through the meter long cable from this external power supply to the load. This seems especially true considering that not only are we fighting the lead network IRL, but that there is likely some sort of regulation between this power supply and the load as per your example. Perhaps I am daft, but it would seem to me that an external power supply cannot react at these frequencies. I am sure you know way more than I, and I'd like to understand. Perhaps you are speaking in generalization?

The final power supply is to slow to react to the instantaneous current requirements of digital switching, hence the local capacitance and reservoir caps as I mention.

[marce]

17 minutes ago, jaaptina said:

Not having the knowledge but thinking from this point. Is there a point in having a power supply, (long) cable, MPAudio SD-HPULN LT3045 dc-dc converter, very short cable, dac/streamer configuration? 

 

Shorter the better, inductance slows current delivery down, on PCB's for longish supply runs (not power planes) such as 24V to an section of circuitry the supplies are run as broadside coupled traces on multiple layers, so you have power, return, power, return, it gives a low inductance relatively high capacitance feed similar to bus bar routing for high power.

[jabbr]

21 minutes ago, jaaptina said:

Not having the knowledge but thinking from this point. Is there a point in having a power supply, (long) cable, MPAudio SD-HPULN LT3045 dc-dc converter, very short cable, dac/streamer configuration? 

 

 

It depends on what is being supplied, but if tighter regulation is helpful then yes. This is called a “pre regulator” and the design pattern might be to step down a supply and allow LDO which reduces the heat generated in the final board. In other cases on board LT3045s might be cascaded to generate the full complement of needed voltages. On board is best.

[marce]

11 hours ago, Superdad said:

 

I'm a fan of split bobbin trans too, but good luck finding then in anything larger than 48VA.  And they often are mechanically noisy (lam buzz) when you pull real current through then.  Unlike R-cores, which have near the performance, lower flux field radiation, are available in larger sizes--and are DEAD SILENT!

Worked on  a 16kW 230V SMPS, that had input and output differential and common mode chokes and a litz's wound transformer, the magnetic's guy was up there in the world of RF engineers... half mad scientist/half hippy.

 

[marce]

2 minutes ago, jabbr said:

 

It depends on what is being supplied, but if tighter regulation is helpful then yes. This is called a “pre regulator” and the design pattern might be to step down a supply and allow LDO which reduces the heat generated in the final board. In other cases on board LT3045s might be cascaded to generate the full complement of needed voltages. On board is best.

The general rule is main supply, which often depends on the final location for the design, then main onboard supplies, usually SMPS to keep heat generation down to provide the required main voltages, usually 3V3, 5V, +/-6-12V for analogue, then local LDO's for the device local supplies such as 2.5V, 1.8V, some FPGA's having up to 5 separate supplies, core, I/O, PLL's etc.

[4est]

17 hours ago, Superdad said:

 

I'm a fan of split bobbin trans too, but good luck finding then in anything larger than 48VA.  And they often are mechanically noisy (lam buzz) when you pull real current through then.  Unlike R-cores, which have near the performance, lower flux field radiation, are available in larger sizes--and are DEAD SILENT!

Oh, certainly. I was including them in the split bobbin category in my mind.

[Ralf11]

batteries, people

 

batteries

[mansr]

20 minutes ago, Ralf11 said:

batteries, people

 

batteries

Batteries need regulators too.