I theory any number of identical amplifiers can be connected in parallel to drive the same load to achive more power than a single unit can. My question is, because in the real world identical amplifiers doesn't exist, wich is the general way to compensate this small diferences?. What I'm trying to do is to combine power amp ICs in bridge/parallel configuration to get more power and keep things relatively simple. Thanks!
Parallel Amps
Amp paralleling is all most amplifier do internally with special circuits
so all the output transistors work in harmony. Don't hook up two
Amplifiers in parallel . . .it would be bad. No simple way to increase
power except by careful planning and layout of circuits designed to
amplify and distribute the current.
Chris
Amp paralleling is all most amplifier do internally with special circuits
so all the output transistors work in harmony. Don't hook up two
Amplifiers in parallel . . .it would be bad. No simple way to increase
power except by careful planning and layout of circuits designed to
amplify and distribute the current.
Chris
I wanted to do this on a Forte Model 3 amp. Asked Nelson and he suggested paralell bridge the output stages. He said tie the input and output and watch out for DC offset.
I asked him if the gain of the 2 channels differed and caused issues... he didnt reply... but he did say to tie the two OP stages with a 0.1 resistor.
My approach was to feed both channel OP stages from one channel's drivers.
I asked him if the gain of the 2 channels differed and caused issues... he didnt reply... but he did say to tie the two OP stages with a 0.1 resistor.
My approach was to feed both channel OP stages from one channel's drivers.
Parallel Amps
Some sort of resistor or network should be employed to prevent
the output transistors from killing each other. . .I understood this
to be putting two different but similar amps in parallel. . . something
that will fry the output transistors of one amp or the other. . . interally
anything goes. . . as long as it is designed correctly
Chris
Some sort of resistor or network should be employed to prevent
the output transistors from killing each other. . .I understood this
to be putting two different but similar amps in parallel. . . something
that will fry the output transistors of one amp or the other. . . interally
anything goes. . . as long as it is designed correctly
Chris
With regards to parallel amps.
Beware of any difference between the amps, not just DC offset. There is noise and interference. I've killed amps instantly trying to do this. It is inelegant at best, in my opinion.
What do you hope to achieve? You'll get the same output power unless the amp was under much stress before in which case I feel you should try a better match for your speakers instead. If you are driving two speakers from one amp you should use them separately. The halved Zo is not worth this particular effort in my opinion.
Beware of any difference between the amps, not just DC offset. There is noise and interference. I've killed amps instantly trying to do this. It is inelegant at best, in my opinion.
What do you hope to achieve? You'll get the same output power unless the amp was under much stress before in which case I feel you should try a better match for your speakers instead. If you are driving two speakers from one amp you should use them separately. The halved Zo is not worth this particular effort in my opinion.
luisma1972 said:
K-amps said:
Kamps approach is paralleled output stage and completely different from paralleling chipamps to allow lower impedance loads to be driven.
Have a look at the datasheets from a selection of manufacturers. Some are very helpfull with guidance on paralleling and bridge parallel for more power.
ps there is another chipamp thread running on exactly this question.
It appears your search button is not working properly.
I reckon it has been discussed before: but whats better using one driver to drive 8 OP devices or 2 drivers running 4 each?
As far as sonic preference goes, I prefer non-parallelled approach, however from a pure current delivery standpoint would 1 or 2 drivers be better:
Keep in mind with more OP devices running on their high Hfe area, the need for beta droop on the driver is reduced...
As far as sonic preference goes, I prefer non-parallelled approach, however from a pure current delivery standpoint would 1 or 2 drivers be better:
Keep in mind with more OP devices running on their high Hfe area, the need for beta droop on the driver is reduced...
Hi,
I think it depends on the load and how that affects the output stage,
Some say that any paralleling reduces performance i.e parallel output devices.
The ultimate expression of this non-paralleling argument is the large group that only listen to single ended amplifying stages.
Others say that with careful design/selection the performance loss from paralleling can be reduced or even eliminated.
If the driver is becomming overloaded when a severe load is put on the output then paralleling the drivers would seem to be obvious. But the alternative is a bigger/more capable single pair of drivers.
I think one would have to try the two options, each in many guises before one can come to a conclusion.
Personally, I had planned to use 2sb649/d669 as twin drivers for my KSA50 Klone. Lower capacity drivers doubled up to drive reactive loads down to 4ohm. I now plan to do similar to the KSA100 Klone and use bascially the same output toplology but with modern faster less robust devices since I will NEVER consider loads less than 4ohm.
I think it depends on the load and how that affects the output stage,
Some say that any paralleling reduces performance i.e parallel output devices.
The ultimate expression of this non-paralleling argument is the large group that only listen to single ended amplifying stages.
Others say that with careful design/selection the performance loss from paralleling can be reduced or even eliminated.
If the driver is becomming overloaded when a severe load is put on the output then paralleling the drivers would seem to be obvious. But the alternative is a bigger/more capable single pair of drivers.
I think one would have to try the two options, each in many guises before one can come to a conclusion.
Personally, I had planned to use 2sb649/d669 as twin drivers for my KSA50 Klone. Lower capacity drivers doubled up to drive reactive loads down to 4ohm. I now plan to do similar to the KSA100 Klone and use bascially the same output toplology but with modern faster less robust devices since I will NEVER consider loads less than 4ohm.
By drivers we (Andrew and myself) mean the BJT's powering the OP stages. (As opposed to the thread starter)...
Issue is: according to self and sloan, beta droop due to excessive current demanded by the load mostly affects the driver not the OP stages (if they live) becaus ethe OP stage having a constant gain simply transfer this "load" on to the driver.
On the other hand if there are enough paralleled OP devices, they may have enough gain so that the load (hence beta droop) they transfer on to the drivers is less.
However, in the case of high biasing (like I plan) there is a constant load on the driver and this load increases with having to drive more OP devices.
As Andrew says, it is a best guess at this point because of the following:
More OP devices mean less load on drivers because of increased current gain of the OP stage. (Signal)
Inversely more OP devices need to be biased hence load on driver. (No signal)
On the other hand, with the 1302/3281's their hfe actually INCREASES remarkably as the temperature of the OP devices goes up.
This perhaps explains wny I like the sound with the amp warmed up .
Issue is: according to self and sloan, beta droop due to excessive current demanded by the load mostly affects the driver not the OP stages (if they live) becaus ethe OP stage having a constant gain simply transfer this "load" on to the driver.
On the other hand if there are enough paralleled OP devices, they may have enough gain so that the load (hence beta droop) they transfer on to the drivers is less.
However, in the case of high biasing (like I plan) there is a constant load on the driver and this load increases with having to drive more OP devices.
As Andrew says, it is a best guess at this point because of the following:
More OP devices mean less load on drivers because of increased current gain of the OP stage. (Signal)
Inversely more OP devices need to be biased hence load on driver. (No signal)
On the other hand, with the 1302/3281's their hfe actually INCREASES remarkably as the temperature of the OP devices goes up.
This perhaps explains wny I like the sound with the amp warmed up .
Re: Parallel Output Drivers or Parallel Output Devices
Is this a statement or a question?
I think there are a few amplifiers out there that dedicate a driver to each output device. Keeping in mind the cost of doing it this way, the designer must have thought there was some return.
Thought through from the load end. A pair of 250W 200degC To3 devices could easily take a 15W driver up to it's thermal and SOA limits.
I would suggest that consideration should be given to adding extra driver pairs when ever the paralleled output devices exceed 2pairs.
But in all cases the driver SOAR should be checked even when driving one or two pair of output devices as part of the design process.
It is possible with correct device and load selection for one driver pair to adequately drive MANY paralleled output devices, but the designer always checks this rather than leaving it to chance.
gni said:
Is this a statement or a question?
I think there are a few amplifiers out there that dedicate a driver to each output device. Keeping in mind the cost of doing it this way, the designer must have thought there was some return.
Thought through from the load end. A pair of 250W 200degC To3 devices could easily take a 15W driver up to it's thermal and SOA limits.
I would suggest that consideration should be given to adding extra driver pairs when ever the paralleled output devices exceed 2pairs.
But in all cases the driver SOAR should be checked even when driving one or two pair of output devices as part of the design process.
It is possible with correct device and load selection for one driver pair to adequately drive MANY paralleled output devices, but the designer always checks this rather than leaving it to chance.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.