If I put my notes here, I might be able to find them again later!
Conduction Angle, Or Why You Need a Bigger Power Transformer Than You Think You Need
Posted 29th March 2012 at 04:35 PM by rjm
I tell people: "Buy a nice, heavy power transformer. It will sound better."
They are skeptical, because the circuit only draws a couple of watts, and less than 100 mA current.
The image below shows how the power transformer, and rectifier diodes, actually work much harder than you would estimate from looking at the output power.
It shows a zener regulated supply with a load drawing 100 mA at 20 V. That's 2 W.
As a result of the capacitor input filter directly after the diodes, however, the diodes and transformer do not conduct current all the time, but instead for just a couple of milliseconds twice every cycle of the AC wave. They have to supply all the output current in just that short space of time. As you can see in the simulation, the diodes are pushing peak currents well in excess of 1A or 10x the output current. This is a typical "normal" power supply with a initial ripple ratio of a modest 1/40, things only get worse in an "audiophile" power supply, with higher input capacitance, lower capacitor impedances, and ultrafast diodes.
As a rule of thumb, consider that the power transformer is running pulse-wise at 10x the DC output current. I suggest rating the transformer accordingly : i.e. getting VA for the peak A. That's a bit of an overestimate, but a decent guide nonetheless. Those sharp pulses can do all kinds of mischief with circuit inductances, too, and have lots of high frequency energy that can couple over into the audio circuit if improperly shielded.
It's my empirical observation that large, high quality transformers do a better job of damping out these spikes than small and cheap ones. Or perhaps more accurately stated: less likely to do odd and nasty things with them.
Your experiences may vary. However, it's worth keeping in mind.
They are skeptical, because the circuit only draws a couple of watts, and less than 100 mA current.
The image below shows how the power transformer, and rectifier diodes, actually work much harder than you would estimate from looking at the output power.
It shows a zener regulated supply with a load drawing 100 mA at 20 V. That's 2 W.
As a result of the capacitor input filter directly after the diodes, however, the diodes and transformer do not conduct current all the time, but instead for just a couple of milliseconds twice every cycle of the AC wave. They have to supply all the output current in just that short space of time. As you can see in the simulation, the diodes are pushing peak currents well in excess of 1A or 10x the output current. This is a typical "normal" power supply with a initial ripple ratio of a modest 1/40, things only get worse in an "audiophile" power supply, with higher input capacitance, lower capacitor impedances, and ultrafast diodes.
As a rule of thumb, consider that the power transformer is running pulse-wise at 10x the DC output current. I suggest rating the transformer accordingly : i.e. getting VA for the peak A. That's a bit of an overestimate, but a decent guide nonetheless. Those sharp pulses can do all kinds of mischief with circuit inductances, too, and have lots of high frequency energy that can couple over into the audio circuit if improperly shielded.
It's my empirical observation that large, high quality transformers do a better job of damping out these spikes than small and cheap ones. Or perhaps more accurately stated: less likely to do odd and nasty things with them.
Your experiences may vary. However, it's worth keeping in mind.
Total Comments 11
Comments
-
Nice read. Makes a lot of sense, although I would love to learn a bit more about what the current spikes can cause.
Anyway, your simulation might be a bit optimistic too. You are forgetting about the big startup inrush current because of the empty filter caps, which on a very rough simulation on my pc started/peaked at about 12A and eventually balanced at about 1A spikes (with a similar but simpler circuit - just a resistor in parallel to the filter capacitors).
Reducing the filter cap size (to something like 100uF) caused the current spikes to be reduced to almost half, but the voltage ripple was quite bigger.
I guess it is a give-take situation...
The thing is that we can afford more current (bigger transformer/bridge) but bigger ripple is always a big problem, although smaller caps might be nice.
So QFT... do not cheap-out on transformer and rectifier specs...Posted 29th March 2012 at 06:11 PM by dimkasta
-
Hi,
Your telling some of us things we already know, and assuming
you know more than the people who actually design transformers.
They are not stupid, are well aware of the issues, and design
mains transformers for real use and give them real ratings.
rgds, sreten.Posted 29th March 2012 at 10:51 PM by sreten
-
Richard never said anything about designers not knowing what they do. They create specific stuff with specific properties.
From there, it s our job to understand and decide, try, reject or accept what works best for our projects.
And I completely agree that at least on preamp/phono level, enough VA to cover our mA consumptions is just that... Just enough... Bigger transformers do give a better sound. Richard just tried to give a logical explaination for those that are sceptical.
If you have something to contradict this, or provide an alternative explaination, please do
Posted 29th March 2012 at 11:45 PM by dimkasta
-
Sreten, the VA rating is quite precisely defined as the maximum AC current that can be drawn before the voltage falls below VAC_rated(1-regulation/100). This is a very valuable parameter, but does not tell you how big a transformer you might want to use in silicon-diode-rectified, capacitance-input power supply circuit for high-end audio. There is only a "rule of thumb" for the scaling factor s=VA/(output watts).
My point was that "s" should take conductance angle into account. Anyone who designs power supplies knows that. Not everyone who builds DIY audio does.Posted 30th March 2012 at 01:16 AM by rjm
-
Another way of explaining it is to consider the current that flows through the transformer, diodes and filter caps without ever passing through the load.Posted 30th March 2012 at 03:50 PM by PChi
-
True. In other words we have to take into consideration the power consumption of the filter/regulator as well...Quote:Originally Posted by PChi
Another way of explaining it is to consider the current that flows through the transformer, diodes and filter caps without ever passing through the load.
In the simulation if you check the current passing through the filter cap, it is almost the same as the that passing through the rectifiers...
But even then, a 12VA transformer should have been enough to provide for the 1A spikes.
And of course that consumption is way below the power abilities of a 160VA toroid @12V for example (13A to cover for 1A spikes).
And still upgrading from a 48VA to a 120VA and then to a 160VA all made a big difference to the sound of the phono preamp.
Perhaps there are other issues to consider too. Maybe bigger transformers are built with higher standards or precision to reduce risks and flactuations that should be less evident on smaller currents.Posted 30th March 2012 at 06:19 PM by dimkasta
-
Yes, but the voltage drop when attempting to do so will be much greater.Quote:Originally Posted by dimkasta
But even then, a 12VA transformer should have been enough to provide for the 1A spikes.
The 12 VA (say 12 VAC) transformer has a regulation of ~20%, so at 1 A the output drops to about 10 V. A 2 V drop.
A 120 VA transformer would typically have a regulation of about 5%, and that's for 10 A! So the voltage while making the 1 A current peak (1/10th VA rating) only drops on the order of 5%/10=0.5%, 11.95 V. A drop of just 50 mV. Rough numbers.
Big difference!
Now, how much difference this has on the actual sound is something we can debate all day, but the numbers suggest that it could be an influence.Posted 31st March 2012 at 01:23 AM by rjm
Updated 31st March 2012 at 01:29 AM by rjm -
i agree with the author rjm, 100%...sometimes when someone echoes one's thoughts.....i get that feeling that i am not alone......
i build all my traffos for all of my projects and them some, but i am not into it commercially, just to support a hobby i build some for friends....
at least in this aspect i am free from commercial choices.....i'd say i am "transformer independent" if there is ever one....
i asked the question in some forums before, the answer i got was "get the biggest transformer that will fit in your box", from then on i never looked back......Posted 31st March 2012 at 10:43 AM by TonyTecson
-
actually custom is the only way to go - standard parts are wound for minimum cost to the manufacturer in Copper, Iron that can optimistically be argued to meet the spec
you should almost always want to run at lower core B than the standard parts - so you neen more turns, more Copper - not just a bigger VA rated "standard" part
the pri no load current is almost always set deep in nonlinear core bias - even 10 % higher line can more than double the current in Lm
for 120 V line we can use the 240 pri connection of "universal" tansformers to cut core B, reduce mechanical lamination buzz/hum - at least for headphone or pre-amp sized loadsPosted 31st March 2012 at 09:44 PM by jcx
Updated 31st March 2012 at 10:00 PM by jcx -
hi,
Just my 2 cents but It boils down to this ,In a small signal phono,preamp,ect. you NEED enought power to cover the load and another half for safety for transits,sangs ect.Engineers make the circuit work just enought to get it out the door,On most equipment,Cost = profit,Amps are a little different,IF you don't have a efficient power supply your amp doesn't push loads of bass out because the power supply lags,sags,and plays catch up all the time except at low volume levels,That's why we build out equipment because we want the best we can afford,right? I use a current clamp to monitor whats going on at the speakers using a test tone input, and power supplies,should be a pretty steady reading,unless you have a small supply then it's all over the scale,especially bass notes,Have you ever noticed a LED flickering on a power supply when the volume is up,whats that tell you? turn the volume down and it stops,lol.NEED more BEEF,lol.
I have built 4 F5 boards now I am using 2 1.5 amp @18v sec on a nelson PSU board,I test them on 200 watt 15" monitors,
They run fine ,current is level ,I am running bias at .6 for now,
I just got my F5c boards,waiting on parts,going to run them at +- 50 to start,some errors on bom R115 and R116 belive to be extra,or some one tell me where they go? C101 and 102 will have to be bigger for your power supply,mines 50+- so 220uf@63vdc,
Anyone got theirs running HOW do they sound?
Got any heat sink gurus out there,what size for F5 c aluminum?
Thanks for your help!!!!!!!Posted 23rd June 2012 at 02:48 AM by noSmoking
Updated 23rd June 2012 at 02:50 AM by noSmoking (spell check) -
Interesting! I leaned to bigger is better based on what I have absorbed over the years. Good to hear sound [bad pun intended] reason why.Posted 5th February 2013 at 01:11 AM by KMossman
