Hi Jan,
The reason is simple, we're using two positive voltage regs (LM338) for both rails.
As you can see, we join the G and + of the two regs at the output to make the ground (the other extremes are + and -).
I can't see other way to do this other than with independent secondaries and independent bridges.
The trafo I used for this amp was scavenged from an old amp and was a CT.
I opened it and separated the wires, transformed it for independent secondaries.
It's always a thing that people can do easily.
The reason is simple, we're using two positive voltage regs (LM338) for both rails.
As you can see, we join the G and + of the two regs at the output to make the ground (the other extremes are + and -).
I can't see other way to do this other than with independent secondaries and independent bridges.
The trafo I used for this amp was scavenged from an old amp and was a CT.
I opened it and separated the wires, transformed it for independent secondaries.
It's always a thing that people can do easily.
Yes sure does!
Going from memory with the non-regulated supply the + and - supply rails sag when the amp is under load. At about 1W o/p you can observe the supply rails dropping on a'scope and also supply rail modulation, can't recal the exact figures and I 'aint going to pull it apart to check but it's very noticable.
With the regulated supply there is no disernable sag even at full o/p under sine wave test and at 1W o/p there is hardly any modulation of the rails at all. At bass frequencies this supply can o/p full power for quite some time, I had to limit measurment due to the poor heatsinks I have on the 338's, when un-regulated the lines sag a huge amount I *think* I measured a drop from 26v nominal to 18v but don't hold me to that, I did not take notes.
NealG said:
Sorry to disagree on this point. If your AC is stable and connections are all fine, the main reason why supply rails will sag under load is because the transformer is too small. Going to a higher VA transformer will solve the problem or quite diminish it.
Supply modulation or ripple is directly related to capacitance, as well as capacitors type. If grounding is properly done and wired, higher capacitance will improve ripple.
Supply droping at 1W is certainly an indication of a small transformer. A transformer should only sag at the amp's highest output under heavier loads.
NealG said:
At a price, of course. You will be always wasting quite a few volts to regulate from input to output. You will also have to compute differences your AC line might have at some time in the day or week and add them to those input volts.
NealG said:
That's quite a huge drop! My advice would be to change your transformer, even if using a regulator.
Regulating or not regulating is a question of budget and taste. A properly designed unregulated supply should show little difference with a regulated one. If the difference is huge then there will be something wrong in the unregulated one. Of course that "little difference" between properly designed supplies might be huge to some, which I consider more a matter of taste.
It's a question of taste because some people do not like the sound of regulators, particularly those using feedback to regulate.
But I do agree that regulators are a shortcut to some problems, like lowering impedance and improving ripple.
Carlos
Carlos, the problem here is that these chips have the best sound quality with 1000 to 1500uf capacitance, with an unregulated PSU.
I made extensive tests with capacitance.
This is a big compromise.
I didn't make measurements but I can tell you that my unregulated GC has a much bigger transformer (2x24v, 384 VA) compared to the normal trafo I used on my regulated GC, much smaller (scavenged from an old amp).
My speakers (Epos 11) are difficult to drive properly, they need power, so I have good conditions to compare.
If you heard the difference, even on a blind test, you would say when switching for the regulated GC:
Oh yes, this is the Krell.
Bass is much better, tight like there's no tomorrow, and that brings improvements in midband, treble, soundstage, everything, as always.
It's like putting a good clock on a "jittery" CDP.
I made extensive tests with capacitance.
This is a big compromise.
I didn't make measurements but I can tell you that my unregulated GC has a much bigger transformer (2x24v, 384 VA) compared to the normal trafo I used on my regulated GC, much smaller (scavenged from an old amp).
My speakers (Epos 11) are difficult to drive properly, they need power, so I have good conditions to compare.
If you heard the difference, even on a blind test, you would say when switching for the regulated GC:
Oh yes, this is the Krell.
Bass is much better, tight like there's no tomorrow, and that brings improvements in midband, treble, soundstage, everything, as always.
It's like putting a good clock on a "jittery" CDP.
Hi Carlos, don't be sorry to disagree!
Curiosity go the better of me and I have just gone back to un regulated to check. I have a 300VA mains TX, I was way out on the voltage sag, it's no where as bad as I remembered, I have just measured a 8% approx. drop at full o/p before clip at 1KHZ and approx. 12% drop at 100HZ, this is with just 2000uF per rail setup favoured by some.
At 100HZ I seen to have LF instability and will have to look into that one later! At 1KHZ I see the o/p wave form on the supply rail at approx 250mV and if I increase capacitance to 10,000uF / rail it drops to approx 50mV at full o/p. Time to go back to regulated and try again!
Curiosity go the better of me and I have just gone back to un regulated to check. I have a 300VA mains TX, I was way out on the voltage sag, it's no where as bad as I remembered, I have just measured a 8% approx. drop at full o/p before clip at 1KHZ and approx. 12% drop at 100HZ, this is with just 2000uF per rail setup favoured by some.
At 100HZ I seen to have LF instability and will have to look into that one later! At 1KHZ I see the o/p wave form on the supply rail at approx 250mV and if I increase capacitance to 10,000uF / rail it drops to approx 50mV at full o/p. Time to go back to regulated and try again!
tbla said:
i was referring to my krell fbp300......its not playing yet, it will be converted to 230V/50hz next friday
and then i will compare it to the triamped 12 x lt1083 regulated 6 x 300VA 12 x 10000Uf/63V cerafine bla bla bla supersystem
I think it is not easy to compare Tri-amp system with another system. That will not be a amp to amp compare. I had similar situation once but bi-amp and the closest I've done was to replace a set of the amps that drive the mid-high and compare, but ideally I wish I have the same number of amps each side to really compare them within a particular system. So how are you planning on with this compare? I'll be interested to see what other options are there?
Regards,
Chris
NealG said:
Now that seems about right. A larger transformer may probably sag even less, which seems to be the advantage of using them. My guess is maybe yours are not toroidal, as my experience with sagging transformers were usually EI types, smaller ones.
The lack of control some have told about at low frequencies, which I could myself listen on the tests I did, seem to be related to smaller capacitors. That's why I thought this arrangement using 1000uF capacitors was better for active bi-amping, powering mid/tweeter. Using larger caps on woofer side should provide the best of both worlds.
Regulators act like isolating filters, which in your case the bigger cap seems to be doing. Larger capacitors store the charge better, allowing smaller transformers work less.
But once I read that the best way to deal with Class B amp's supplies was to deal with them as if they were class A. Big transformers, big capacitors. I think that's about right.
Pity there's no way to modify the idle current setting in the 3875/3886, as that might bring further improvements.
Please pay attention that I never said that an unregulated amp should sound better or worst than a regulated one, or viceversa. They are just different.
Carlos
carlosfm said:You can treat them as op-amps, wich is what they are.
A resistor from V- to output.
But some people have tested that and it seams that in this case there's no advantage.
Then it's not the same. Using the "class A trick" on chips never sounded worst, as some reported here for the GC.
Carlos
Hi Carlos,
The transformer is a toroidal one!!
I am now back with regulated supply and large heatsinks on the 338's. First thing I noticed was large oscillations on both power rails with the negitive rail voltage dropping from 25V to 22V at full o/p 1KHZ sine wave. 
I had not measured under load before so did not notice this, it works out that the .1uF caps I used on the o/p of the 338's where causing it, removing them the oscillations went away and increasing to .47uF they also went away so there must be some interation causing a resonant circuit somewhere, I am now at .47uF on the 338 o/p and all is well.
This regulated supply can supply full power at 50HZ with no supply rail drop, for me with 25v rails it's about 35W and sounding great!
The transformer is a toroidal one!!
I am now back with regulated supply and large heatsinks on the 338's.
First thing I noticed was large oscillations on both power rails with the negitive rail voltage dropping from 25V to 22V at full o/p 1KHZ sine wave. I had not measured under load before so did not notice this, it works out that the .1uF caps I used on the o/p of the 338's where causing it, removing them the oscillations went away and increasing to .47uF they also went away so there must be some interation causing a resonant circuit somewhere, I am now at .47uF on the 338 o/p and all is well.
This regulated supply can supply full power at 50HZ with no supply rail drop, for me with 25v rails it's about 35W
and sounding great!
carlosfm said:It may be me, but I have doubts of the advantage of puting an unregulated supply with lots of ripple on the output of the chip.
Who says an unregulated supply has lots of ripple? It certainly will not if capacitance is enough. The minimum capacitance can be computed to have minimum ripple according to the current you will be demanding. There's a formula to get that number.
Carlos
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.