[Porksoda]

So I was bored the other day when I was working on my dynalo and decided to hook it up to a small speaker I had laying around. It could actually push a fair ammount of sound, and I was quite impressed with the sound quality. Mind you, we're talking no more than a watt per channel here. I was thinking, "Hey, why not take a headphone amp design and tweak it to make a compact desktop-style speaker system?". So, why not? I figure you could get some good quality 3-4" full range drivers, make some compact cabinets for them, and BAM! Desktop bliss

Has anybody else tried making this kind of adaptation, and what was your experience with it? Anybody else have thoughts on what it would take to do this? My intention is to take existing printed circuit boards, (ppa, pimeta, dynalo, m3, etc.) and make something low-power that will not require heatsinking (no more than bolting parts to the case itself) or any other expensive or time-consuming modification.


Thoughts?

 

[tkam]

The M3 can be tweaked pretty easily for use as a small power amp.

 

[Porksoda]

Quote:

Originally Posted by tkam The M3 can be tweaked pretty easily for use as a small power amp.

I was just looking at the datasheets and I was thinking the exact same thing! Personally, I've never heard the M3, and I have a new-found affinity for discretes, so I was hoping to do something along the lines of a dynalo or a dynamid. Anybody know of a higher current alternative to the transistors in the dynalo that could make it happen, or is that just a lost cause?

 

[tkam]

Well there's always the dynahi

 

[Porksoda]

Dynahi is a bit much work

My other thought had been, perhaps if you bridged two dynalos you could decrease the power supply voltages to get a little more current out of them. I dunno somehow I don't think its gonna happen. I have a spare dynalo board here and it is staring into my SOUL daring me to build it

 

[amb]

The little TO-92 output transistors on the dynalo do not have enough current capacity to drive speakers effectively. It will take a bit more than simple substitution of higher rated transistors, though. The operating points are also not optimal for such an application. The Vas stage does not flow enough current either.

 

[Porksoda]

Quote:

Originally Posted by amb The little TO-92 output transistors on the dynalo do not have enough current capacity to drive speakers effectively. It will take a bit more than simple substitution of higher rated transistors, though. The operating points are also not optimal for such an application. The Vas stage does not flow enough current either.

Yeah, I figured some adjustment of quiescent current would be necessary. I was hoping against hope that somewhere out there I could find a good TO-92 audio quality transistor with more current output, even if at a lower voltage. I was thinking if I could bridge two dynalos then I could afford to drop the supply voltage in exchange for more current. But from my limited researching of TO-92 datasheets it looks like 150 mA is the max.

The M3 looks promising, though, and not too expensive. My only beef is that I wanted to cram it into the largest hammond enclosure, to match my dynalo, but it looks like even if I machined in ventilation, you need to use the 1.5" heatsinks to fit in the case, which is not quite enough area to get a decent ammount of wattage. I guess I could always fiddle with it, and just let it run a little warm. I mean as long as I put in ventilation, how hot could it REALLY get pushing 10 watts or so, with reasonably low quiescent current?

 

[amb]

Quote:

Originally Posted by Porksoda Yeah, I figured some adjustment of quiescent current would be necessary. I was hoping against hope that somewhere out there I could find a good TO-92 audio quality transistor with more current output, even if at a lower voltage. I was thinking if I could bridge two dynalos then I could afford to drop the supply voltage in exchange for more current. But from my limited researching of TO-92 datasheets it looks like 150 mA is the max.

Even if you find TO-92 devices capable of sufficient current, you'd still have to deal with heat dissipation. TO-92 is not conducive to efficient heat-sinking.

Quote:

The M3 looks promising, though, and not too expensive. My only beef is that I wanted to cram it into the largest hammond enclosure, to match my dynalo, but it looks like even if I machined in ventilation, you need to use the 1.5" heatsinks to fit in the case, which is not quite enough area to get a decent ammount of wattage. I guess I could always fiddle with it, and just let it run a little warm. I mean as long as I put in ventilation, how hot could it REALLY get pushing 10 watts or so, with reasonably low quiescent current?

See morsel's idea about using a piece of aluminum bar to couple the top of the heatsinks to the top lid of the Hammond case, which then becomes a portion of the heat-radiating structure. You would actually use the shorter 1" heatsinks for this solution.

 

[Porksoda]

The idea behind bridging was to decrease the voltage and thus decrease the power dissipation on the TO-92's.



I'm feeling lazy. Do you recall approximately where in that thread the idea is mentioned?

 

[amb]

Quote:

Originally Posted by Porksoda The idea behind bridging was to decrease the voltage and thus decrease the power dissipation on the TO-92's.

Power is voltage * current. While decreasing the supply voltage decreases the static quiescent dissipation, when driving speaker loads, the bridged configuration causes each amp to see half the load impedance so the current doubles.

Quote:

I'm feeling lazy. Do you recall approximately where in that thread the idea is mentioned?

Page 13 of the thread.

 

[Porksoda]

Quote:

Originally Posted by amb Power is voltage * current. While decreasing the supply voltage decreases the static quiescent dissipation, when driving speaker loads, the bridged configuration causes each amp to see half the load impedance so the current doubles. Page 13 of the thread.

Yes, but that means you have less than double the dissipation in the amplifier (double current but decreased voltage), but twice the current output, so four times the speaker output power. That means the amp dissipates at MOST half the power for a given output power produced. Ergo, at least twice the maximum power for dissipation-limited amplifiers.

EDIT: Thanks for the link!

 

[Garbz]

You could wait a little, I hear gilmore's next headphone amp will be 700watts

 

[Porksoda]

[size=xx-large]WHAT? SPEAK UP![/size]

 

[amb]

Quote:

Originally Posted by Porksoda Yes, but that means you have less than double the dissipation in the amplifier (double current but decreased voltage), but twice the current output, so four times the speaker output power. That means the amp dissipates at MOST half the power for a given output power produced. Ergo, at least twice the maximum power for dissipation-limited amplifiers.

You're being overly optimistic. You'll get four times the output power only if there is no reduction of supply voltage nor any current limiting. The fact that the dynalo's output impedance is in the order of a few ohms (which causes voltage loss and current limiting to the speaker load) and you won't see anything close to that level of linear scaling at speaker impedances.

Let's just assume you're aiming for 2Wrms output into 8 ohms. A simple calculation tells you that the output stage needs to be able to deliver need 0.5Arms or 1A peak. The dynalo is not anywhere near rail-to-rail (in non-bridged configuration it could swing just a bit over 4Vrms into 33 ohms before clipping with +/-15V rails, much less into 8 ohms or 4 ohms), so to maintain the 2W output goal you cannot really decrease the rail voltage much even though bridging would increase the swing. If you could find TO-92 transistors that could handle 250mA+ of current, the power dissipation on them would be over 2.5W each even if you decreased the rail voltage to +/-10V. Not gonna happen on those devices!

Even if you substitute some beefier output transistors and heatsink them, assuming that the their Hfe is 100, the base drive current would need to be 10mA peak which far exceeds what the Vas stage flows so it won't work either. You'd have to basically re-engineer the whole amplifier...

 

[Porksoda]

Quote:

Originally Posted by amb You're being overly optimistic. You'll get four times the output power only if there is no reduction of supply voltage nor any current limiting. The fact that the dynalo's output impedance is in the order of a few ohms (which causes voltage loss and current limiting to the speaker load) and you won't see anything close to that level of linear scaling at speaker impedances. Let's just assume you're aiming for 2Wrms output into 8 ohms. A simple calculation tells you that the output stage needs to be able to deliver need 0.5Arms or 1A peak. The dynalo is not anywhere near rail-to-rail (in non-bridged configuration it could swing just a bit over 4Vrms into 33 ohms before clipping with +/-15V rails, much less into 8 ohms or 4 ohms), so to maintain the 2W output goal you cannot really decrease the rail voltage much even though bridging would increase the swing. If you could find TO-92 transistors that could handle 250mA+ of current, the power dissipation on them would be over 2.5W each even if you decreased the rail voltage to +/-10V. Not gonna happen on those devices! Even if you substitute some beefier output transistors and heatsink them, assuming that the their Hfe is 100, the base drive current would need to be 10mA peak which far exceeds what the Vas stage flows so it won't work either. You'd have to basically re-engineer the whole amplifier...

The comment on four times the output power was merely to demonstrate that output power scales 4x with bridging, but dissipation only scales 2x. So, for a given setup with energy dissipation as a limiting factor, bridging should double its power output capability all else things being equal. So, if the dynalo were only capable of 1/4 Watt, bridging should bring it to 1/2 watt without making the transistors any hotter. I also made the secondary arguement that since the effective impedence is halved, you see less voltage swing so lower rail voltage is possible, meaning you could increase the power further for the same output wattage, within the limitations of the new rail voltages.

I realize that integral to that statement is the assumption that there is a solution to the problem of current limitations. The basic Dynalo transistors are capable of dissipating almost half a watt, which would probably suffice for this application. My question is if there are any with the current handling capability aswell, such that a tradeoff between rail voltage and output current could be made.

With regard to dissipation, when you are pushing one amp of current into 8 ohms, the output of the amp would be at 4 volts for a bridged amp. Suppose you knocked the output resistors to 10 ohms and decreased the quiescent current a bit to improve power handling. (250mA per transistor)*10ohms = 2.5V + 4V. The emitter of the output transistors would be at 6.5V, so if your rail voltage were, say, 10 volts, their peak dissipation would be a little under a watt. RMS, that is more like half a watt, which is not totally unreasonable as a maximum load condition with the right transistors.

As for the preamp current capability, that can always be adjusted with selection of the right resistors. Just drop the impedence of all the resistors in the gain stages and POOF! more current and approximately the same gain.

All in all, though, I am starting to agree that the dynalo, as it is, is not reasonably easily adapted to low-power speakers. I'd probably only get 1 to 1.5 Watts per channel out of it, which is less than I would ideally like for small desktop speakers. The M3 using the case as part of the heatsink looks promising though...