Running a stereo amp in parallel instead of bridged

Status
Not open for further replies.
You would think that as a physicist some analytic prediction of the results of parallelling the amplifiers would be in order. A simple spreadsheet would suffice.

You will quickly find that the only part that matters is the power supply - at least to a first order approximation.

The schematic showed an interesting combination of 4700 // 3300uF (or 8000uf per rail).
Lets throw a couple more equations into the discussion:
C = Q/V
Differeniate WRT time.
C = dQ/dt / dV/dt

Now we all know dQ/dt is current.

C = I / dV/dt

You are in the USA so the mains is 60Hz, and a full bridge rectifier charges 120 times a second - to use some mangled english to describe how it works🙂 So "dt" = 1/120 sec.

You said the amplifier is 30W, so this is 1.9A RMS in 8 Ohms, and lets simplify this to 3A peak in an 8 Ohm load. This is entirely consistent witht he current limit circuit by the way. So lets pop 3Amps into the above equation.

8000uF = I / dV/dt
8000E-6 = 3/ dV/dt
dV/dt = 3/8000E-6
dV/dt = 375V/sec

between cycle peaks this is a drop of: 375/120 = 3.1V

If you successfully double the amplifiers in parallel without any loss, you will drop this to 1.6V. Or gaining about that much in output. We know we will be putting some expensive external resistors in to reduce the current that willl result from an output voltage mismatch. If you go 0.3 Ohms, this will be one of those precious volts!

Run a test and see.... You could simply pop some extra caps in the power supply too of course.
 
  • Like
Reactions: tonyEE
I suppose the amplifiers (individually) will have to be examined as well; the output stage may not have emitter/colletor or drain/source resistors, in which case I would (personally) not attempt to parallel them. Even with amps having them (resistors), I'd use an oscilloscope to make sure there are no oscillations.
 
I suppose the amplifiers (individually) will have to be examined as well; the output stage may not have emitter/colletor or drain/source resistors, in which case I would (personally) not attempt to parallel them. Even with amps having them (resistors), I'd use an oscilloscope to make sure there are no oscillations.

I got a feeling I'm on a Pyrrhic quest.

OTOH, I might get a scope and some fancy resistors out of the deal.
 
You would think that as a physicist some analytic prediction of the results of parallelling the amplifiers would be in order. A simple spreadsheet would suffice.

You will quickly find that the only part that matters is the power supply - at least to a first order approximation.

The schematic showed an interesting combination of 4700 // 3300uF (or 8000uf per rail).
Lets throw a couple more equations into the discussion:
C = Q/V
Differeniate WRT time.
C = dQ/dt / dV/dt

Now we all know dQ/dt is current.

C = I / dV/dt

You are in the USA so the mains is 60Hz, and a full bridge rectifier charges 120 times a second - to use some mangled english to describe how it works🙂 So "dt" = 1/120 sec.

You said the amplifier is 30W, so this is 1.9A RMS in 8 Ohms, and lets simplify this to 3A peak in an 8 Ohm load. This is entirely consistent witht he current limit circuit by the way. So lets pop 3Amps into the above equation.

8000uF = I / dV/dt
8000E-6 = 3/ dV/dt
dV/dt = 3/8000E-6
dV/dt = 375V/sec

between cycle peaks this is a drop of: 375/120 = 3.1V

If you successfully double the amplifiers in parallel without any loss, you will drop this to 1.6V. Or gaining about that much in output. We know we will be putting some expensive external resistors in to reduce the current that willl result from an output voltage mismatch. If you go 0.3 Ohms, this will be one of those precious volts!

Run a test and see.... You could simply pop some extra caps in the power supply too of course.

thanks.... I need to get parts.... need to get scope.... 2 channel, max 100Mhz. USB. Not overly expensive like the behemoths at work.

Ideas?

For function/signal generator I can use a PC with an external DAC.

Yeah, those expensive external resistors will prove to be the kabosh on the practicality of the endeavor, but it should be fun. Hopefully it won't be like the very first power supply I built... it had NO on/off switch because then, at the tender age of 22, I know exactly what I was doing. Hmmm...
 
This one is an overkill

View attachment 1100214

The cheaper one will do. Maybe get 4 x 8 ohms, so you can play with 16/8/4 ohms (stereo, both channels driven) combinations..?

If paralleling the amps, make sure they don't oscillate. You could add a bit of capacitive load... just as a simulation of speaker wiring capacitance.

I figure I'd run a 200 Hhz signal to start, put two 1 ohm in line resistors and one 8 ohm load and measure the current and voltage across the load. Not drive it over 1 watt to start with.

But oscillation, I figure I'd be looking for other signals other than what I'm driving?

I also have a number of midrange/mid-bass old speakers I pulled out from a stereo console.
 
You dont need fancy kit.

Put an (at a guess) 220R - 820R resistor in SERIES with your speaker and use it to monitor the output of your amplifier. The aim here is to let you hear the output attenuated by a lot of dB while the amplifier is running flat out. I am guessing 220-820R to get 30-40dB of attenuation. Call this the monitor output.

Use a DVM to measure the AC voltage at this point.

Then run the thing with a sinewave at say 400Hz and turn it up until you hear the distortion on the monitor. You will hear distortion at clipping, and if you are careful, as low as 1% THD is quite audible on a sinewave.

Parallel the amplifiers in your manner of choice.

Test again per above with no load other than the monitor. This will tell you:
  • Does the thing explode into flame?
  • Is it stable. The 400Hz will be clear. If it is not stable you will quickly hear a cruddy tone, and if that fails by the smell of smoke.
  • Remeasure the clipping voltage. Compare.
  • Has your mod actually degraded the no load output? It well could you know! A slight mismatch in gain will pull easily 1A peak from your amp if you use low value "matching" resistors.
Get a four off, 5 or 10W resistors and create a parallel series combination of a 20-40W resistor. If you can only get 20W worth, dont worry, just dont run it too long. Four 8R2, two series and two parallel will do the trick. This chould cost you less than 5 bucks. Trust me, investing megabucks in a dummy load for this is a waste of money, and you do need a dummy load. So do it as cheaply as you can.

Repeat the above.
  • For bonus points measure the power supply rail DC voltage during the tests and look for the changes predicted.
  • Again, all you need is that DVM and cheap dummy load and clipping monitor.

Invest further funds in good red wine or beer.
 
Last edited:
You dont need fancy kit.

Put an (at a guess) 220R - 820R resistor in SERIES with your speaker and use it to monitor the output of your amplifier. The aim here is to let you hear the output attenuated by a lot of dB while the amplifier is running flat out. I am guessing 220-820R to get 30-40dB of attenuation. Call this the monitor output.

Use a DVM to measure the AC voltage at this point.

Then run the thing with a sinewave at say 400Hz and turn it up until you hear the distortion on the monitor. You will hear distortion at clipping, and if you are careful, as low as 1% THD is quite audible on a sinewave.

Parallel the amplifiers in your manner of choice.

Test again per above with no load other than the monitor. This will tell you:
  • Does the thing explode into flame?
  • Is it stable. The 400Hz will be clear. If it is not stable you will quickly hear a cruddy tone, and if that fails by the smell of smoke.
  • Remeasure the clipping voltage. Compare.
  • Has your mod actually degraded the no load output? It well could you know! A slight mismatch in gain will pull easily 1A peak from your amp if you use low value "matching" resistors.
Get a four off, 5 or 10W resistors and create a parallel series combination of a 20-40W resistor. If you can only get 20W worth, dont worry, just dont run it too long. Four 8R2, two series and two parallel will do the trick. This chould cost you less than 5 bucks. Trust me, investing megabucks in a dummy load for this is a waste of money, and you do need a dummy load. So do it as cheaply as you can.

Repeat the above.
  • For bonus points measure the power supply rail DC voltage during the tests and look for the changes predicted.
  • Again, all you need is that DVM and cheap dummy load and clipping monitor.

Invest further funds in good red wine or beer.

Thanks.... I do need dummy loads to match the dummy running the test.

Funds... well, I just did the shopping, fridge in the garage has plenty of good IPA.... the kind that is 8% and 9%, plus I got beaucoup de wine, spirits and cigars. I might have to invest in a fire extinguisher and run this in the garage.... I'm very good at smoking electronics. Too good indeed. ;-P
 
The guy has been told few dozens times, that it would not work, but still insists.
Looks like a troll or student practicing how to handle the dispute. Stop feeding him.
Although... I might be mistaken.
Here it comes:
You wrote:

"Looks like a troll or student practicing how to handle the dispute. Stop feeding him.
Although... I might be mistaken.
Here it comes: "


You are posting on DIY Audio... and you have NO interest in the possibilities of using amplifiers in different ways?

By pure definition per your post... you are trolling. You are the one knowlingly making an unsolicited comment with the intent of triggering an angry response from me.

ignored. Adios.
 
Last edited:
By pure definition per your post... you are trolling.
By saying one can't take NO for an answer?
By the way you did not answer the relevant part 3 (three) times: just do it and judge by a result.
Anyhow, it's rather easy to force current sharing, even the dynamic one in switching circuit using coupled inductance that can take volt-second inequality. Another way may be applicable for an amp with rather high output impedance, so-called droop one. There's a point of degeneration resistors use in EF when connected in parallel, again well within outer feedback loop.
Moreover often than not such solution will not cure the cause. Best way would be still select parts with near equal parameters and make junction temperatures as equal as possible like in integrated circuit.
Again if you tie together two SISO systems you'd better be sure you know what a response would be. It's much easier to derive than divergence of a curl you've posted earlier.
 
Last edited:
  • Like
Reactions: ejp
Forget it. I ask for a couple of simple questions... people go off on a tangent and they can't let go of it. I mean, people really spend time and effort on such things.

I suppose I shouldn't bring up the time invariant inductive circuit resonance at this time..

I didn't expect the Spanish Inquisition here of all places... What is this... the forums over at ASR too slow nowadays?

Let me explain why "voltage has nothing to do" here... something so fundamental that it's right there yet you don't see it.

In general

Pmax(t) = ( Vmax )( I(t) )

And for peaks, again the classic problem taking into account a perfect circuit:

Imax = Pmax / Vmax

However, the circuit too can reach its current maximum due to heat, internal component power dissipation, design characteristics, low load impedances...

Meaning at it is possibly to reach Imax when Vmax has not been achieved: When you get into peaks. where the POWER than one channel can deliver is reached, you will see that I becomes Imax... meaning... no more power. If you try to push it harder, it just won't.

In this case, it is not the power supply holding back the delivery of power, but other parts of the circuit.

Under such circumstances, if you can deliver more current, as in having a second amplifier that can deliver current, then you can hold the voltage up to to Vmax and you will be delivering more power indeed.

Do you now understand my point?

Can you refute it?

Now, don't be nasty, just tell me, is my analysis wrong... and if so. WHY? Please don't just give me Ohm's Law because that is not in discussion in this thread, it was NEVER meant to be so, the whole point of my bringing this thread up was to discuss the output impedance of a parallel topology and its ability to deliver more current, and power, on peak transients, not on a steady state, RMS, condition.
Your analysis is wrong because you're making a wrong assumption. Speakers DRAW current from the amplifier due to their impedance, amplifiers do not PUSH current into the speakers. Somewhat simplified: all an amplifier does is offer the speaker voltage. As long as the impedance of the speaker stays the same, and the voltage of the amplifier doesn't sag, that speaker will never draw any more current due to Ohm's law. The only way to increase the current draw from the speakers is to increase the voltage. Your idea is to increase the AVAILABLE current, but that doesn't change the equation, it only prohibits the voltage from sagging. You could pretty much solve that problem by increasing the total capacitor value on the power supply of the amplifier and in extreme cases exchange the transformer with a type that supplies the same voltage but with a higher VA rating.

Someone already mentioned the good old lightbulb for comparison: no matter how much current is available to a 60W lightbulb (as much as the fuse for the circuit allows), that lightbulb will always draw the same amount of current due to two factors of the equation staying constant: the voltage and the impedance/resistance.
 
Those are OPamps. and they are talking of output drive capability. You do get that. But to make use of that capability, something have to change. Same goes for this case. You do increase the output drive capability....for lower impedance use, in a case where a singel amp ch would not be able to deliver the needed current for the voltage at that resistance. But that is not the case her.
 
  • Like
Reactions: JMFahey and ejp
That is when you are trying to drive lower impedances than a single amplifier was intended to handle. Amplifiers aren‘t perfect voltage sources - they have limits. One of two things can happen when you try to pull more current than they were designed for - they either limit early or blow up. Paralleling two lessens the possibility of that happening. That is all. If a single amplifier does not hit its limits when driving the desired load, then paralleling won’t add any power. If it does go into a limit condition (like one of those AD op amps would) then you do get the increase by paralleling.

What happens with most audio amplifiers is that they will happily drive a heavier load, right up until things go south. Overload protection in home audio equipment is pretty lax. Paralleling when driving a lower than 8 ohm load can prevent the magic smoke. You do need small ballast resistors in series with the outputs when paralleling anything. These will eat up any efficiency advantage you get from paralleling so don’t even expect the extra 5 watts or so you could theoretically get. You can just drive a 4 ohm sub driver off two paralleled “minimum 8 ohms” channels. The power would be the sum of both channels driving individual 8 ohm loads. You would get th SAME power if you used only one amplifier channel to do it - it just wouldn’t like it very much and may decide to punish you by making you buy new output and driver transistors.
 
  • Like
Reactions: ximikas and ejp
You have more than 10 x Parasound Z yet you want us to tell you what's the best thing to do with them and you're a big shot in physics...
Just bridge each of the amps , then you can parallel 5 Parasound Z on each channel...What's so f...difficult to do?
You don't need our advice cause you f...know it all! We're f...waiting you here to tell us that the result blew all your expectations, windows, neighbour's patience...
 
well alot of infighting and conjecture in this thread but after some thought and some quick research it seems an industry giant agrees with tonyEE;s thinking....
https://www.analog.com/en/technical-articles/paralleling-amplifiers-increases-output-drive.html
Sure thing. You can find this for other parts as well, but not for all parts, and, most likely for the parts on the same die.
Those f...ers are matched to mV if not uV levels. Newer ones often use CE output stage which is "a current source".
Connecting in parallel as done by Self, JFI, require some knowledge, experience and, God forbid, some testing.
Power things paralleled I've seen were BUF634, which again, were well matched diamond buffer IC.
BTW if a giant agrees with someone it does not make one right or otherwise.
 
Status
Not open for further replies.