Frank, most DIYers are put off the instant you utter the word "processor".
It appears we've been looking for solutions in similar places.
I dug up data from TI's site on chip power amps. I am currently mulling over TI's TAS 5630B, the well known LM 3886 and LM 4780. At first glance, they seem to be up to it all the way, my only quibble being their relatively low voltage slew rates. I honestly cannot say 20 V/uS is too low, but I would have liked to have seen something like 40 V/uS; on the other hand, this number corresponds well to their power levels, and IS above the bare minimum of 0.5 V/uS per every peak Volt of output with a reasonable margin.
I am particularly interested in a "double trouble" version, i.e. using them in both parallel and bridged verson. Bridging should theoretically produce four times the power, but in pratice this is more like three times the power, as there are always losses. Nevertheless, having well over 100 Wrms per channel is handy, even if you never actually need it all, but you do need to parallel them to restore good current delivery.
Another option would be Nat Semi's driver chips, LME 49811 bipolar and 48833 MOSFET. This approach has a very long tradition with Nat Semi, elders among us may remember their 381-xxx chip from the late 70ies, which did exactly the same, and did it well so long as you didn't ask for too much voltage gain and stuck to 26 dB maximum.
Much data to absorb.
Using a single National chip amp worked well for me, with a really solid power supply to back it up. The biggest "flaw" was that when driven really hard, the chip's internal protection mechanism cut in - from a soprano's long held note, of all things! So, going the 4 way sounds good to me; just make sure you use plenty of good heatsinking and beef up the supply - at the prototype stage, when the heatsink wasn't doing its job 100% my unit's chip would overheat at the drop of a hat ...
Frank
I have just been doing a small job and have used a 4017 decade counter ( CMOS from 1968 ! ) . It is described in one makers notes as an electronic Mobius loop . It got me asking myself would not a feedback amplifier be a Mobius loop if so ? the 4017 is a cascade of flip-flops with a final NOT gate that inverts the output to start the counting again . Mobius is a 2D object in a 3 D word ( or even 1D in concept ) . I read recently that space-time might be simlar and might explain Quantum phsiscs rather well . Putting that to one side as it is of no immediate importance could there be something to criticize in feedback amps if so due to how the loop works if Mobius ?
Don't you worry, Frank, I do have a solid stock of very efficient heat sinks (0.46) and I do plan on using them generously. I am in the habit of being an optimist with experience, i.e. a pessimist regarding cooling problems. You should see my PC, it's full of fans blowing each and every way, but it sure keeps the CPU cool enough. Check on it routinely every day.
I am not sure what you meant by "solid power supply". It's somewhat subjective, I find. To me, this means at least two 10,000 uF caps in parallel for every supply line used, usually with the same type, same make 4,700 uF cap thrown in to help out with the speed (a quirk Sony has been using forever, and while it may not help, it sure can't hurt). Backed up by a say 500 VA toroid per channel - haven't done any calculations, so this is an off hand figure. Point is, I like 'em big.
Off the cuff, enough is when the VA rating of the toroid matches what I would expect in Watts into 2 Ohms.
The true dilemma here is whether to use an all-in-one chip, or the one which does almost everything, but expects to drive output transistors. That way, one could scale the output stage to what one wants - even Wayne could.
Given that I already have a sizeable stock of power devices, I do have to think about it. Discrete devices do have the advantage of using the heat sink much more efficiently that the all-in-one devuices do, even if I'd have to use four per channel, thus also spreading them around the heat sink(s).
I am not sure what you meant by "solid power supply". It's somewhat subjective, I find. To me, this means at least two 10,000 uF caps in parallel for every supply line used, usually with the same type, same make 4,700 uF cap thrown in to help out with the speed (a quirk Sony has been using forever, and while it may not help, it sure can't hurt). Backed up by a say 500 VA toroid per channel - haven't done any calculations, so this is an off hand figure. Point is, I like 'em big.
Off the cuff, enough is when the VA rating of the toroid matches what I would expect in Watts into 2 Ohms.
The true dilemma here is whether to use an all-in-one chip, or the one which does almost everything, but expects to drive output transistors. That way, one could scale the output stage to what one wants - even Wayne could.
Given that I already have a sizeable stock of power devices, I do have to think about it. Discrete devices do have the advantage of using the heat sink much more efficiently that the all-in-one devuices do, even if I'd have to use four per channel, thus also spreading them around the heat sink(s).
Where feedback can go wrong is if it's treated as an afterthought, a by-the-way. Either don't use FB at all, or if you do use it go 100% in the effort. The way to think of it is that the FB is the most important thing to get right in an amp that does overtly use it. In high open loop gain opamps the behaviour in an ideal unit is completely set by the parts in the FB loop, the actual internal behaviours, idiosyncrasies of the opamp are completely swamped by the action of the FB mechanism.
Frank
I probably should have used the word "stiff",
. To me, the power supply is everything to making the amp do dynamics well, so how I went about it was using a 300VA per side, extensive smoothing by caps going into regulation circuitry, so that heavy bass didn't cause the rails to sag, and then very extensive, extremely local high frequency decoupling on the chip amp itself. Completely OTT for a 60W amp, perhaps - but it worked. I have an old Perreaux 2150B monster, and the chip amp easily knocked that fellow off the perch.A side benefit was tremendous energy reserves, I could pull the power cord out of the wall while playing at normal volume, and still get clean music for up to 5 minutes later ...
Frank
Last edited:
Is recoverable, like much gear back then the power supplies were not sorted out as well as would be done now. I went through the unit with a strong broom and it's now very decent. Put it this way -- it saw off most gear back then in raw form, and currently would keep back most of what I heard at the recent hifi show. I had a good listen to the Technical Brain monoblocks at the show, and it would be in the same ball park ...I had a 2150B sometime ago , nice case , great boat anchor .......
Frank
True, which is why I am not discarding them so easily.
I will not go below that figure, after all, I qualify for the title "gumpy old fart".
I've discarded the usual suspects (LM388x, TDA729x) for the better sounding climes of the old Philips car radio parts which come bridged as standard. Power output is more limited into higher impedance loads due to the restricted voltage (18V) so stacking might be the best way forward for me to increase the power
Flights of fancy are almost always doable ... I once seriously conceived a 2.4kW amp, done totally using chip amps; current drive of at least 50amps. Of course, this way you may as well go discrete, by the time you work out the hardware that's needed ... but still, a fun thing to do -- blow up someone's pro speakers using clean power from chip amps, just to prove it can be done ...
Frank
Frank
Surely the above comment is in some way related to the volume used. I mean, very high quality bass is available from others as well, but at higher power levels, this may (and most often does) change in favor of larger transformers.
However, for your "typical" room listening levels, there shouldn't be a problem caused by power supplies. I find that most amps start to have problems once you've come to about 1/3 of their nominal voltage output. In normal rooms, even a nominally 50 Wrms amp will start rattling window panes by then. Unless you have an uncommonly large room by Euro standards, in which case you purchased the wrong amp in the first place.
I think Thorsten is right when he says it all depends on whether the amp was designed for show (looks), a price level or a sound quality level. If you design it for sound quality, it should be all right under most conditions. Obviously, once you decide to design for sound quality, chances are it will have "worse" specs than if it's designed for a price level. But if you look deep down, it will invariably be better designed than the run-off-the-mill products from mass manufacturers.
There you go.
It wasn't hard to guess, actually, my own HK integrated amps have been consistently praised for their bass and mid drive energy and enthusiasm in the UK, German and Italian mags.
Truly, their grip on the bass drivers is amazing, BUT, as you might suspect, even if they hold out longer than others, eventually their power supplies come out short in comparison with the likes of Levinson. Though not as dismally short as the Levinsons and such like come in comparison with them in terms of prices.
In my view, there are two steps to take in such occasions:
1. Separate the VAS and output stage supply voltages, use regulated for the VAS. This will not aid actual power output much, if at all, but it will keep the sonic balance way longer than single supply units, and
2. Beef up the whole power supply, not just size up the capacitors. That's a good move only if you are upgrading an existing unit, but if you're doing it from scratch, go bigger right from the start, i.e. the transformer capacity.
For example, the mechanical and physical construction of my HK 680 integrated amps inhinbits bigger transformers, simply no space. Unfortunately, the space allowed for the 8,200 uF caps makes it impossible to use anything bigger than 10,000 uF caps, so I'll have to live with just a +22% upgrade. I would have preferred 12,000, or even 15,000 uF caps, but there you go, no space.
Not that I actually NEED more than it was designed with, but hey, if I'm going to do it, I might as well add as much as I can.
The subjective impression of good bass actually depends on the amp working correctly at the top end as well as the bottom end. Very few systems do this well, only the top of the line Bryston at the recent audio show demonstrated proper control at high impact levels. Most only achieve a type of wallowing effect, to my ears.
I'm reminded of when I auditioned active studio monitors: tried the top of the line Mackie, supposedly capable of better than 120dB, with a Sydney Opera House organ piece. What a mess it made of it ... and sound engineers are supposed to use these devices to assess "accuracy" ...?
Frank
I'm reminded of when I auditioned active studio monitors: tried the top of the line Mackie, supposedly capable of better than 120dB, with a Sydney Opera House organ piece. What a mess it made of it ... and sound engineers are supposed to use these devices to assess "accuracy" ...?
Frank
Last edited:
- Status
- Not open for further replies.