advertisement


What IS ultra-low output impedance ?

N

nojobtoosimple

Active Member
Hi all,

I have been reading through threads about ALWSRs and TeddyRegs/STRs with a view to understanding the categorisation of output impedance levels for regulator solutions. Can anyone tell me the sort of values at which output impedance would be considered to be high, low, very-low or ultra-low.

I seems likely that output impedance is frequency dependent (many parameters seem to be so in the real world) in which case are there particular frequencies at which output impedance is described to enable comparison ?

I expect this sort of question is easier asked than answered but I'd be grateful for any information, however 'wooly' it may seem to a proper engineer.

Thanks, in anticipation, for any help.

Regards,
Jon
 
Take a look at the Belleson regulator comparison page there's a good list on there, but bear in mind absolute noise level, spot noise, bandwidth, step response, and a whole host of other factors affect how a reg works in any circuit.

Also consider how the measurements are done, the lt3045 looks great when dropping 3v from input to output, much less so when only dropping 200mV.
 
nojobtoosimple said:
Hi all,

I have been reading through threads about ALWSRs and TeddyRegs/STRs with a view to understanding the categorisation of output impedance levels for regulator solutions. Can anyone tell me the sort of values at which output impedance would be considered to be high, low, very-low or ultra-low.

I seems likely that output impedance is frequency dependent (many parameters seem to be so in the real world) in which case are there particular frequencies at which output impedance is described to enable comparison ?

I expect this sort of question is easier asked than answered but I'd be grateful for any information, however 'wooly' it may seem to a proper engineer.

Thanks, in anticipation, for any help.

Regards,
Jon
Click to expand...

It's all relative of course....

It's mainly (though not entirely) down to that pretty much most important thing in hi fi but which rarely gets mentioned, NFB. Hence, like an amplifiers output impedance and distortion, it is frequency dependent yes and for EXACTLY the same reason....and a voltage regulator pretty much IS a power amp but reconfigured to give a DC voltage out.

1Ohm would be shite and <a micro-Ohm (a millionth of an Ohm) is quite feasible at the other extreme...which would no doubt be less than an inch of welding cable!

There are often bigger fish to fry though.... The wire used to connect a good regulator to its load will have far higher impedance than the regulator itself, hence "sense" wiring on some regulators which moves the output of the regulator to the point of load... but only to THAT point. Things like stability, transient response and noise can be more important for specific applications and in fact stability is about the biggest issue with very high performance regs that are required to work over a wide bandwidth.
 
+1, stability is everything.

- At one end, the Teddy Reg is probably liked so much simply because its output is essentially resistive - which means it can be put to use anywhere, even implemented really badly - and nothing bad will happen; it simply cannot interact with what follows - owing to its intrinsic high, if load-dependent impedance. To be clear - I certainly don't care for that latter attribute for performance reasons; but it is incredibly forgiving, esp. for DIYers of any & all aptitudes.

- At the other, just the one tiny under-considered tiny-value-cap placement choice on my part on [a developed DIY, massive NFB Vreg with remote sensing and a few other tricks aimed at rather better than ALWSR] temporarily made my preamp into an outstanding 2.5MHz oscillator - a full 40v pk-pk. (It all survived ! And I solved the problem ... learning a few things on the way...)



There's a huge, comfy, space between these extremes where 'good enough' really, really ..IS.
 
martin clark said:
+1, stability is everything.

- At one end, the Teddy Reg is probably liked so much simply because its output is essentially resistive - which means it can be put to use anywhere, even implemented really badly - and nothing bad will happen; it simply cannot interact with what follows - owing to it high, if load-dependent impedance. I don't care for that latter attribute.

- At the other, just the one tiny under-considered tiny-value-cap placement choice on my part on [a developed DIY, massive NFB Vreg with remote sensing and a few other tricks aimed at better than ALWSR] temporarily made my preamp into an outstanding 2.5MHz oscillator - at 40v pk-pk. (It survived !)



There's a huge, comfy, space between these extremes where 'good enough' really, really ..IS.
Click to expand...

Fully agree. A similar reg I designed was stable after loads of faffing about but oscillated the next morning... it was winter and with the heating on the ESR of an electrolytic was just right but in the cold the next day it was high enough so it oscillated! I sorted it in the end.

Just designed a reg for an amp that's a WIP but has maybe the worst PSRR ever seen (prob why I've not seen the reasonably obvious topology used before but it has so many good points otherwise that I'm determined to crack it) and have deliberately sacrificed theoretically ultimate performance for (hopefully!) better stability and excellent PSRR from the reg itself.
 
The dark secret is that low output impedance can often be wasted effort! Low enough noise is always a good thing, although the target for that depends on what it is driving - a moving coil phono stage needs much lower noise levels than the input stage of a power amp. But impedance is more subtle.

Impedance tells you how much the output voltage changes per unit of current. If it is a pre-amp stage, the signal currents are small, so the effect on the voltage rail is small, and unless it is an exotic circuit (like Jez's mystery power amp) you probably have a reasonable amount of PSRR, so the effect of that small voltage disappears into the noise floor. For example, if you have a Naim style line stage putting out say 2V, with typical load and internal impedances, you might get 0.2mA of signal current. A very moderate regulator might have say 500mOhm impedance, giving you 100uV on the voltage rail. Unimproved Naim stages do much better than 40dB of PSRR, so that 100uV on the rail causes much less than 1uV on the output, over 120dB down.

There are only a couple of cases I can think of where regulator output impedance becomes critical, and all of these can be better fixed by using local regulation. If you run a high current stage, like a headphone amp, off a shared supply, you can get much bigger signal currents and so bigger signal voltages. If you have a very high gain stage, like an MC stage, the PSRR can easily be negative (the effect on the output can be bigger than the change on the rail).
 
Thank you so much everybody for your comments. As seems often to be the case I find myself massively unqualified to conclude anything substantive for future projects from the insights provided. Although my personal MO has always been to make an effort to understand things rather than just parrot the efforts of others or take a flier, in the area of electronics, and in my 60's, I think I'm going to have to accept that a general engineering bent and sciencey careers in Geology and IT do not even come close to giving me a grounding that might help me to understand anything much of this at all. So, from now on I'll be walking only well trodden paths and limiting myself to asking for practical advice when necessary. Hopefully the community will continue to be as brilliantly supportive as it has been thus far.
Best regards to all,
Jon
 
The lower the DC impedance, the more the regulator looks like an inductor. At some frequency you have an LC resonance. Some DC resistance really helps stability.
 
nojobtoosimple said:
Thank you so much everybody for your comments. As seems often to be the case I find myself massively unqualified to conclude anything substantive for future projects from the insights provided. Although my personal MO has always been to make an effort to understand things rather than just parrot the efforts of others or take a flier, in the area of electronics, and in my 60's, I think I'm going to have to accept that a general engineering bent and sciencey careers in Geology and IT do not even come close to giving me a grounding that might help me to understand anything much of this at all. So, from now on I'll be walking only well trodden paths and limiting myself to asking for practical advice when necessary. Hopefully the community will continue to be as brilliantly supportive as it has been thus far.
Best regards to all,
Jon
Click to expand...

hey always feel free to ask:) Some things are more intuitive than others;)
 
davidsrsb said:
The lower the DC impedance, the more the regulator looks like an inductor. At some frequency you have an LC resonance. Some DC resistance really helps stability.
Click to expand...

Yeah pretty much so:) Same with all NFB hence using a Thiel network to partially isolate a capacitive load from a power amp.... or not, in Naims case....:eek: Great for selling inductive speaker cable:rolleyes:
 
Designing a regulator that can handle transient loads at both mA and A levels at the frequencies needed for a Class AB output stage is hard and one reason regulated power amplifier output stages are so rare.
 
Here you go @Arkless Electronics I've got a name for your 'Monster' amp:
Charybdis (Greek Mythology): Described in Homer’s Odyssey as a sea monster that dwelt in the Strait of Messina, this many-toothed serpent emerged from under rocks and was so colossal it could writhe and turn in the deep and create enormous maelstroms on the surface above. :)
(Reference courtesy of SwimTrek)
[Edit: Oh dear, maybe Arkless Charybdis sounds too much like a medical condition...]
 
You must log in or register to reply here.


advertisement


Back
Top