Double-amping with two amplifiers -- will it work?

February 4, 2008 6:44 AM

Double-amping fun with Class-T amps: Can it be done?

I'm continuing my experiments with the Sonic Impact Class-T amps, as detailed last week.

They arrived but one was faulty, so had to be returned. I'm expecting the replacement this week.

It turns out that they don't go very loud on my speakers, and if the dial goes past half-way, distortion creeps in, especially on the high notes.

My CD player has no less than three line outputs, so what I'm thinking of doing is what I call (perhaps incorrectly) double-amping -- one line out to one amp, another to the second amp, and connecting both in parallel to the speakers. NOT bi-amping. Both high and low inputs on the speakers get 2x amplifier voltages.

Is this going to damage anything? I'm thinking more about the output from one amp going into the second amp if it isn't turned on, for example.

posted by humblepigeon to sports, hobbies, & recreation (16 comments total)

I'm not 100% sure I understand what you're suggesting - but if you're suggesting hooking up two amps to the same exact inputs, then that's a bad idea. Don't do it.
posted by kickingtheground at 7:36 AM on February 4, 2008

I don't imagine what you propose could be a good idea. You need to get some more efficient speakers.
posted by zemblamatic at 7:44 AM on February 4, 2008

> one line out to one amp, another to the second amp, and connecting both in parallel to the speakers. NOT bi-amping. Both high and low inputs on the speakers get 2x amplifier voltages

Sorry, doesn't work that way. Theoretically, this would double the available current (which wouldn't necessarily result in "louder", depending on your speakers), but I suspect that most power-amps would be unhappy about it. Don't do it.

You might want to try bridging two amps into one speaker, which WILL double the voltage across the speaker. The hassle with this is that you will have to make an inverter to flip the phase of the signal going into one of the amps (assuming both amps are unbalanced in)
posted by Artful Codger at 8:21 AM on February 4, 2008

If you connect things this way, each amplifier's output will see the other's output in parallel with the speaker. This is bad.

From reading your earlier post, it seems that your speakers are nominally 6 ohms and 86dB/W efficient. That's probably a tough load for a fairly low-powered amp. I would try running one amp into each speaker.
posted by the duck by the oboe at 8:39 AM on February 4, 2008 [1 favorite]

I would try running one amp into each speaker.

But I think I'm right in saying that, if you disconnect one speaker from an amp (in standard configuration), it won't then make the other channel twice as loud.

Will it have to work this way:

line output 1 left -- input left of amp 1
line output 2 left -- input right of amp 1
Amp 1 -- left speaker (left and right outputs)

line output 2 right -- input right of amp 1
line output 2 right -- input right of amp 2
Amp 2 - right speak (both left and right outputs)
posted by humblepigeon at 9:00 AM on February 4, 2008

Sorry, that last posting should be:

CD player line output 1 left -- input left of amp 1
CD player line output 2 left -- input right of amp 1
Amp 1 -- left speaker (left and right outputs)

CD player line output 2 right -- input right of amp 2
CD player line output 2 right -- input right of amp 2
Amp 2 -- right speak (both left and right outputs)

Effectively one amplifier per channel.
posted by humblepigeon at 9:06 AM on February 4, 2008

uh, what?
posted by Artful Codger at 9:11 AM on February 4, 2008

I am going to disagree with the group here. I think it might work, and at the price of the T-amp if it doesn't who cares, even if both are destroyed (which I doubt will happen). Employing parallel output sections is a well known technique used in single ended triode tube amp designs to boost their flea power. Voltage is not the issue here, power is, and thus since the speaker is a relatively constant load the issue is amperage. Each op-amp in the T-amp has just so much ability to deliver current, so by doubling up here you can deliver more. I am not sure how great it will sound, but you won't know until you try. One potential issue is that you are not just paralleling the output sections but also the pre-amp sections. How are you going to get them to work nicely in parallel? If one amp is trying to output a different voltage than the other will one pull the other one down, will that affect the sound, perhaps even stress the circuit? Please let us know if you do try. Also try bi-amping and compare the two schemes. Which works better? Bi-amping is the traditional approach here by the way.
posted by caddis at 9:24 AM on February 4, 2008

Is it speaker distortion or amp distortion?

My understanding of the digital power amp technology is that it can't be overdriven because it uses a PWM mechanism for volume control. Unless the output filter is lame and can't handle what the amp is giving it.

So is the CD player putting out the same signal the T-Amp is expecting? Maybe the incoming signal isn't clean and you are simply amplifying noise.
posted by gjc at 10:36 AM on February 4, 2008

You're of course welcome to disagree, caddis, but your suggestion is based on a faulty premise.

Yes you can double up output devices in an amp (eg triodes) but remember that tube amps almost always have an output transformar, which would be selected to correctly match that 2-triode output to the speaker load. Discrete transistor amps often have paralleled output devices, again because they are designed thus.

The biggest error is in thinking that more current would solve the OP's problem. For a given load, the only way to push more power into the load is to raise the supplied voltage. The OP reported distortion beyond a given level, which is to me an indication that the output was running to the rail and clipping; and parallelling the amp outputs won't increase the available voltage.

I'm all for experimenting, but there's no point risking experiments that you're pretty sure won't be successful.
posted by Artful Codger at 10:44 AM on February 4, 2008

I don't think this will work.

The normal solution in this case (multiple amps driving one speaker) would be a bridge tied load, as mentioned above. Not all amps are capable of this and it's also worth noting that bridging divides the effective impedance seen by each amp in half, which may be a problem even if it works topologically.

Here are the problems that I see:

-nearly all amp topologies use a negative feedback loop to reduce error in the output stage. I think "class T" is a fancy marketing version of a class D (PWM) amp which means that this one does as well. This means that in your parallel configuration, you will have two separate feedback channels. Let's say that amp A drives a little hot compared to amp B. Amp B sees a higher than expected output voltage and correspondingly drops its (already low) voltage. Amp A does the opposite. Hello, runaway output. Goodbye, output transistors.

-properly parallel output will reduce the amount of current that each output transitor pushes/pulls. It is very likely that this amp is designed to not overload the transistors given its rail voltage and typical load. Therefore, it's very unlikely that output current is a limiting factor, and it's almost certainly not the source of the distortion. Finally, as other people have noted above, it will not make the speakers louder. Reducing the output impedence (i.e. by paralleling two speakers, but then you're driving two speakers with two amps...) or raising the rail voltage (surgery required) would be two ways to increase the volume if you knew you were paralleled.

my conclusion - bridge the amps or bi-amp if you want 4 amps / 2 speakers. or, just get 2 more speakers.
posted by aquafiend at 12:47 PM on February 4, 2008

Tripath TA2020 data sheet

Looking at the outputs of the chip that is the core of the Sonic Impact amp, and reading the operational specs, I don't think it is straightforward. It's an odd bird... differential digital outputs with spread spectrum high frequency (super audio) switching stuff in the ouput. There are some questions I'd have on the manner in which these components appear on the right and left outputs that aren't answered in the data sheet. So it's a gamble, but probably not a big one.

Were I doing it, , I'd stick two small valued resistors in series with each output... 1/4 ohm or so on both the high and low sides... (you'd need 4 per amp). That would at least provide some isolation between the channels.

Also, while you are playing, be aware that these chips have a 7-amp per channel current limit in them which will shut down the chip if you over-current the outputs. You just have to cycle power to get them to come back on, so I don't think you are going to killl the amps. It may not work, but I think you're safe experimenting.

Above 10 watts output (per channel) the THD goes to hell, too.

Be sure to post results of your mad scientist efforts!
posted by FauxScot at 12:51 PM on February 4, 2008

Almost all amps, T-Amps included, are current limited, not voltage limited. The class D amp, like the T-Amp, does have a hard voltage limit, but that would only be the limiting factor into a very easy load, and your six ohm speaker is not an easy load. If the rail is 12 volts, and not accounting for losses or other effects, you are looking at 24 watts, but the T-Amp only provides something like six watts at low distortion. Bridging (applying left and right channels to one speaker, turning the amp into a mono amp at twice the power) is a good way to go except that the T-Amp is already internally bridged. That leaves paralleling or bi-amping. Paralleling two amps is fraught with difficulties, but it is done. A quick googling of "parallel amplifiers" will show some of the difficulties. Bi-amping is the more traditional solution and will work, but the limiting factor is still the woofer and you only get one amp on the woofer, while the tweeter can sing with ease. Of course either way you will need a separate pre-amp or some other way to adjust the volume. Anyway, if your speakers really are 86 dB/W even this will be rather underpowered. What the T-Amp really thrives on are high efficiency speakers.
posted by caddis at 1:40 PM on February 4, 2008

Let me take this opportunity to evangelize on behalf of this receiver that I own. It uses a similar "class D" amplifier, and it sounds tremendous. And check out those specifications. This thing rocks, and that sonic t-amp should similarly rock when you get it working. And if any readers are considering moving to this realm, consider this a vote in favor.
posted by gjc at 7:02 PM on February 4, 2008

I'd suggest a kit from 41Hz Audio, but they aren't for beginners.
posted by Chuckles at 12:37 AM on February 5, 2008

FauxScot, it's a TA2024 chip, not 2020. Not sure if that makes any difference.

Anyhow, I've now got no less than three of these amps. No mean feat bearing in mind I live in UK and quite literally bought the last one available in the country (the retailer told me as much). The other two were imported from ThinkGeek.

As requested, here's what I did, and what I think.

In a bi-amp configuration they sound pretty smooth, and plenty loud enough for my modest-sized lounge. I'm no expert on audio quality but they really win out on picking out instruments in the mix. Instruments that melt in the background with my standard amp can suddenly be clearly heard. They're just fun to listen to.

One amp takes care of the tweeter, one the low-end. The beauty of this configuration is that I can diminish/enhance the brightness without any signal processing. But the main benefit for me is that I can make everything a notch louder without THD stepping in. The THD steps in at around 6 amps on my speakers, I believe, and if the low-end amp isn't driving the tweeter too, that means it can go a little louder before THD becomes a pest. I'm not sure how much of a load the tweeter normally is in terms of output wattage, however.

They're running off the 12v supplies that came with them. The amps are the 2nd gen models, not the originals, and I read somewhere that the max DC input they'll take is 13v, before shutting down for safety (the original model could apparently take up to 13.8v, which boosted the volume). But 13v power supplies are hard to get for a decent price (at least in the 2-3A range), so I'll stick with the ones I got.

If anybody else is thinking about doing this, bear in mind there's a weird flaw with these amps. They buzz when you touch them, or even hover your hand over them. All three do it! It's something to do with the thick ribbon cable that connects to the headphone socket on the side — I've opened them up and when you touch the cable, the buzz goes crazy. It's nowhere near as bad if you use 8 x AA batteries, but it's still there.
posted by humblepigeon at 4:26 AM on February 6, 2008