I might throw in, that for a period, about 10 years ago, I was half getting back into audio, and was using my wife's pretty decent Sony integrated amp, of the early 80's, as a fill-in - the Perreaux had a dead channel and I had to rev up motivation to fix it. Very typical of the period, with a million switches on the front 🙂 - which was most of its problems; if I did a round of the little tweaky things before each play it did pretty well for itself before going dead by the end of the album - the "quite boring" that davym mentions.
I never did an inside fiddle - those Japanese amps are a nightmare inside, lots of sub-circuits hooked together with connectors, a rat's nest of cabling - just too much work to tidy it up electrically, in the ways that count. Just look at the insides of highly rated units - super, super tidy, that alone gets rid of a lot of the problems ...
I never did an inside fiddle - those Japanese amps are a nightmare inside, lots of sub-circuits hooked together with connectors, a rat's nest of cabling - just too much work to tidy it up electrically, in the ways that count. Just look at the insides of highly rated units - super, super tidy, that alone gets rid of a lot of the problems ...
Last edited:
You must not've followed up the link I posted a few days back, Mr Blue Circle is the archbishop of cap banking, I'm merely a deacon in this hierarchy....
When it come to caps, why settle for less when you can have more (up to a point). It does seem to be a failing of many a commercial amp that they go and fit a miserable power supply. I blame the bean counters.
Yes, in previous decades that seems to have been a typical failing, the power supplies were far too flimsy - it amazed me at times how even the "monster" amps of the period, almost to a man, would fall into a heap once a bit of volume was called for ..
Last edited:
Wow we seem to have reached an agreement! Just need to find a source of free beer now and we're sorted.
A little tip ... when I listen to something, I'm never listening to the amp, or the speakers, or whatever - I'm listening to the music, the sound event that's taking place in front of me, that's the only thing that's meaningful to me. And I assess whether it sounds 'right' or not - that's the only criterion I use. Talk about midrange, treble, bass, frequency range emphasis, is meaningless gibberish to me - has nothing to do with what I consider important ...
My preliminary conclusion about amp PSUs is that lack of chokes is the most common failing. Very, very rare to find CLCLC type supplies. Adding more caps without including chokes would just overwhelm the trafo, hence the need for an 'oversized' one.
Interesting comment, Richard ... when I was doing some serious analysis of the waveforms delivered to the smoothing caps - this is many years ago, unfortunately - I considered inserting L - and from memory I dismissed this, because the size of the chokes would have been huge, to have had significant impact. What I used instead were very low resistances, the CRCRC type of thing - this did the job, with relatively minimal losses in the R's, all things considered ...
I'm not suggesting the traditional choke input type supply, that does indeed employ huge chokes. But lower value chokes do provide HF suppression, and series R in addition.
The series chokes will help until they saturate and add to the cacophony. I explored that in depth and a too small choke is pretty much a waste of time as a choke input filter. A smaller cap and then a choke where the current won't change a lot is different. Still big and way expensive compared to a cap.
A bit unfair ... some pics are of the unit only partially connected up, and it is using point to point wiring for a valve stage, which is always going to look a bit messy - I would like to see the internals of a fully finished unit, as purchased by a consumer before passing judgement ...
Here you go: SGW Blog: World First: Lampizator 'BIG SIX' DAC - "The best sound we can possibly make" Nothing like this: http://media.oregonlive.com/silicon-forest/photo/tek-innards-randyjpg-30653e8021787d4a.jpg
Actually the electrons don't care about visual neatness but servicing or getting consistency really does require attention to the details.
Actually the electrons don't care about visual neatness but servicing or getting consistency really does require attention to the details.
Stand back guys , Frank shifted gears , " bench test" and "analysis", wooha.... 🙂
Cant recall Seeing a production amplifier with CLC , maybe Accuphase , my understanding only high bias class -a benefits from the choke, where the inductor is large enough to handle the bias . Demian's response is inline with what most designers think about using inductors .
What say you RM ...?
Cant recall Seeing a production amplifier with CLC , maybe Accuphase , my understanding only high bias class -a benefits from the choke, where the inductor is large enough to handle the bias . Demian's response is inline with what most designers think about using inductors .
What say you RM ...?
I would suggest that "oversize" anything is a miss. Never mind if it's the transformer, or the caps.
In my view, the trick is to have a nicely designed PSU which is well balanced overall. Enough but not too many caps, too many will unduly load the rectifier diodes and evenetually the transformer, which may in some instances start buzzing.
Somebody mentioned CLCL supplies - what do you want L for? To filter out the line borne junk? But isn't that a bit too late down the line? That grid borne junk should have been filtered out BEFORE it ever got to the transformer. That way, you recover 5-10% of the transformer's efficiency, or if you like, you use 5-10% less power from the grid for the same effect. Less heat that way, too.
How much capacitance you need has a nice formula for working it out. Basically, experience shows that we need 1-2 Joules of energy for every 10W of dissipated power, depending on how difficult the load is. Work it out. If you want 100/200/400W of power into 8/4/2 Ohms, and if you are a pessimits, you need 80 Joules for those 400W/2 Ohms.
Joule = (V+ * V+) F /2
Never let me down yet.
The only danger you might encounter is that if you use milk bottle sized caps, you moght find them a bit too slow on some transients. I would always suggest putting caps in parallel; for example, if you want say 22,000 uF, put 10,000 // 10,000 // 2,200 uF rather than using just one single 22,000 uF cap. You get lower output impedance and, more ften than not, more available current than from a single cap.
The other problem with large capacitor banks is stray inductance. Get rid of it by placing in series a 1 Ohm/2W resistor and a say 220 nF cap to the ground as near to the power output stage as you can. The small cap value will depend on the capacitors, might be a it larger, or a bit smaller. One just has to do it in situ. A bit of a pain, but you just might be surprised at how this may open up the upper mids and treble ranges and make them come alive.
In my view, the trick is to have a nicely designed PSU which is well balanced overall. Enough but not too many caps, too many will unduly load the rectifier diodes and evenetually the transformer, which may in some instances start buzzing.
Somebody mentioned CLCL supplies - what do you want L for? To filter out the line borne junk? But isn't that a bit too late down the line? That grid borne junk should have been filtered out BEFORE it ever got to the transformer. That way, you recover 5-10% of the transformer's efficiency, or if you like, you use 5-10% less power from the grid for the same effect. Less heat that way, too.
How much capacitance you need has a nice formula for working it out. Basically, experience shows that we need 1-2 Joules of energy for every 10W of dissipated power, depending on how difficult the load is. Work it out. If you want 100/200/400W of power into 8/4/2 Ohms, and if you are a pessimits, you need 80 Joules for those 400W/2 Ohms.
Joule = (V+ * V+) F /2
Never let me down yet.
The only danger you might encounter is that if you use milk bottle sized caps, you moght find them a bit too slow on some transients. I would always suggest putting caps in parallel; for example, if you want say 22,000 uF, put 10,000 // 10,000 // 2,200 uF rather than using just one single 22,000 uF cap. You get lower output impedance and, more ften than not, more available current than from a single cap.
The other problem with large capacitor banks is stray inductance. Get rid of it by placing in series a 1 Ohm/2W resistor and a say 220 nF cap to the ground as near to the power output stage as you can. The small cap value will depend on the capacitors, might be a it larger, or a bit smaller. One just has to do it in situ. A bit of a pain, but you just might be surprised at how this may open up the upper mids and treble ranges and make them come alive.
D,
Only your big toe , back or wallet would get hurt from too much transformer, ohh yeah, the undersized output stage might let out the magic smoke ..
🙂
Only your big toe , back or wallet would get hurt from too much transformer, ohh yeah, the undersized output stage might let out the magic smoke ..
🙂
I think the reality of it is 98% of the time it isn't used . The 2 % matters and is a bonus . Bob Carver himself said he was surprised it could work .
I suspect with care they can sound fine . Lets say we have an old Quad 405 case . We might build an amp in class G times times as powerful in that case .
I recalculated the Bob Stewart's amp . Not knowing how the FET current source behaves I have to look to the series resistor and zener . 28V/4K7 = 6 mA . For the zener to bight it must be far lower CCS current . Lets say 3 mA could be used in trans-conductance as an unrealistic maximum .We then have the greedy 100 pF VAS cap . If so it could slew at a descent rate . My guess is half of what I used . Realistically 6 V/uS . Other than that an interesting amp that might surprise people .
Nope. As you say its far better off filtered out at mains voltage where the currents are lower. Rather its to attenuate the higher harmonics of the ripple and here an inductor does a better job than a resistor. But yeah a resistor of lowish value in series between caps is better than none.
- Status
- Not open for further replies.