A major problm with all class A/B amps is the powersupply-transformer:
1) The output voltage is fixed and that means the output-power is fixed *** well. So if you load the amp with 8Ohm's the transformer will probably be too "big" and if the amp is loaded with 2 ohm's the transformer is probably to small.
2) The regulation is very bad and you will need big capacitors to fix' this..
3)If the transformer is overloaded there is no way to "turn it down" as it cools down, -it will have to be switched off.
My solution is:
Take a normal transformer with a output voltage thats to small for the amplifier but with a decent power-rating. After the transformer you put a switchmode boost converter and convert the voltage to the right level. The conversion could be done with 90% efficiency and with a pretty simple cicuit.
The advantage would be that the power-supply is now fully regulated (+/- 3% or so) and there is i possibility to control the output voltage.
The output voltage could be controlled with a standard SMPS-chip and a PIC-controller. PIC's with 8 A/D-channels are very cheap and could be used to monitor all the relevant things in the amp: temperature, current from the supply, output voltage, Peak-Mean-ratio of the signal and so on. On the basis of all the measurements the PIC could "decide" the correct output voltage from the boost-regulator, the fan-speed and so on. Furthermore the PIC could control a LCD that could display all the measurements and give the user the possibility to alter the different setting.
If you include a VCA (controlled from the PIC) before the power amp the whole amp would be extremely adaptive and fail-safe.
I know this approach probably would be most usefull with big PA amp's but I think it could be used with Hifi as well.
So, what do you think of my idea? - Has it been done before, have you tried it and why should it not work?
(It would of course be "easier" to make a SMPS that works directly of the mains, but as I understand the rules we are not allowed to discuss it here. - I would also ending up killing my selfe so maybe that's ok.... )
TroelsM
Denmark.
(as usual you will have to excuse my poor english)
1) The output voltage is fixed and that means the output-power is fixed *** well. So if you load the amp with 8Ohm's the transformer will probably be too "big" and if the amp is loaded with 2 ohm's the transformer is probably to small.
2) The regulation is very bad and you will need big capacitors to fix' this..
3)If the transformer is overloaded there is no way to "turn it down" as it cools down, -it will have to be switched off.
My solution is:
Take a normal transformer with a output voltage thats to small for the amplifier but with a decent power-rating. After the transformer you put a switchmode boost converter and convert the voltage to the right level. The conversion could be done with 90% efficiency and with a pretty simple cicuit.
The advantage would be that the power-supply is now fully regulated (+/- 3% or so) and there is i possibility to control the output voltage.
The output voltage could be controlled with a standard SMPS-chip and a PIC-controller. PIC's with 8 A/D-channels are very cheap and could be used to monitor all the relevant things in the amp: temperature, current from the supply, output voltage, Peak-Mean-ratio of the signal and so on. On the basis of all the measurements the PIC could "decide" the correct output voltage from the boost-regulator, the fan-speed and so on. Furthermore the PIC could control a LCD that could display all the measurements and give the user the possibility to alter the different setting.
If you include a VCA (controlled from the PIC) before the power amp the whole amp would be extremely adaptive and fail-safe.
I know this approach probably would be most usefull with big PA amp's but I think it could be used with Hifi as well.
So, what do you think of my idea? - Has it been done before, have you tried it and why should it not work?
(It would of course be "easier" to make a SMPS that works directly of the mains, but as I understand the rules we are not allowed to discuss it here. - I would also ending up killing my selfe so maybe that's ok.... )
TroelsM
Denmark.
(as usual you will have to excuse my poor english)
I would not alter the voltage when ever there was a shift in the input-signal, but only when the transformer or the amp was getting to hot.
That way the rail-voltage could change very slow and you would not get nasty switching spikes on the rails.
The amp could work with +/-80V when power was applied and only when it was loaded very hard it could turn down to say +/- 50V.
TroelsM
That way the rail-voltage could change very slow and you would not get nasty switching spikes on the rails.
The amp could work with +/-80V when power was applied and only when it was loaded very hard it could turn down to say +/- 50V.
TroelsM
I have not drawn a schematic at this time but it would probably all be very "standard"
Try googling: "Boost converter SPMS schematic" or similar.
The PIC-part is pretty easy and very flexible so people could make it just as they like.
If it seems to be a "do-able" solution I will work out a plan one of theese days.
TroelsM
Try googling: "Boost converter SPMS schematic" or similar.
The PIC-part is pretty easy and very flexible so people could make it just as they like.
If it seems to be a "do-able" solution I will work out a plan one of theese days.
TroelsM
I think the point is to allow a single PSU to be suitable fro a wide range of load impedances. For instance a high voltage, low current PSU might be great for an 8 Ohm load, but it would be overloaded with a 2 Ohm load. If it could be turned down so it's a low voltage, high current PSU then it would work great with a 2 Ohm load.joensd said:
One option would be to buy a ready-made off-line SMPS, that gets you around the safety issues and lets you start to built the system right away
.
These generally allow setting the output voltage to say 110% - 30% of nominal. You can control that with a PIC, reacting on the *average* load of the amp, a quick reaction is therefor not needed.
One drawback could be that SMPS's are generally not constant power but have a max current that is constant with output voltage. So you would say buy one for 60V 5 amps, and then vary it between 40V and 60V, but even at 40V you can only get 5 amps as well..
It is not as bad as it looks, in fact an often overlooked advantage of SMPS's is that they beat almost any 'normal' transformer-input suply for current capacity.
A transformer is effectively disconnected from the supply for 80-90% of the time (except when the diodes conduct at the top of the mains sine wave) so for that 80-90% of the time the capacitors have to supply ALL the current to the load. With a SMPS the 5 amps (in this example) is available continuously 100% of the time, and now the caps only have to supply the occasional peak.
Jan Didden
. These generally allow setting the output voltage to say 110% - 30% of nominal. You can control that with a PIC, reacting on the *average* load of the amp, a quick reaction is therefor not needed.
One drawback could be that SMPS's are generally not constant power but have a max current that is constant with output voltage. So you would say buy one for 60V 5 amps, and then vary it between 40V and 60V, but even at 40V you can only get 5 amps as well.
. It is not as bad as it looks, in fact an often overlooked advantage of SMPS's is that they beat almost any 'normal' transformer-input suply for current capacity.
A transformer is effectively disconnected from the supply for 80-90% of the time (except when the diodes conduct at the top of the mains sine wave) so for that 80-90% of the time the capacitors have to supply ALL the current to the load. With a SMPS the 5 amps (in this example) is available continuously 100% of the time, and now the caps only have to supply the occasional peak.
Jan Didden
A few more thoughts about the idea:
1) As far as I know a boost regulator will be able to deliver a constant power. That means a high voltage/"low" current or low voltage/"High" current.
2)The speaker impedance is not the only thing that has an influence on the "optimal" rail voltage. The peak-mean-ratio of the signal are probably even more important. In most modern "pop-music" the mean power is only 1/5 of the peak-power. Therefore the power drawn from the supply is much lower than the peak-power of the amplifier.
This means that the powersupply and the heatsink in a 250W-amp could be rated for only 50W.
3)The regulated rails should result in a "better" powersupply with a possible better sound...?
4)If the SMPS, PIC and VCA (lot of abbreviations...) means that all the "expensive" power-parts could be a lot smaller because the designer dos'nt have to oversize them.
5)in principle the PIC could calculate the exact power delivered to the speaker and protect it from over-heating.
I would like to hear from anyone who has build a boost-converter in the 300W-class.
TroelsM
1) As far as I know a boost regulator will be able to deliver a constant power. That means a high voltage/"low" current or low voltage/"High" current.
2)The speaker impedance is not the only thing that has an influence on the "optimal" rail voltage. The peak-mean-ratio of the signal are probably even more important. In most modern "pop-music" the mean power is only 1/5 of the peak-power. Therefore the power drawn from the supply is much lower than the peak-power of the amplifier.
This means that the powersupply and the heatsink in a 250W-amp could be rated for only 50W.
3)The regulated rails should result in a "better" powersupply with a possible better sound...?
4)If the SMPS, PIC and VCA (lot of abbreviations...) means that all the "expensive" power-parts could be a lot smaller because the designer dos'nt have to oversize them.
5)in principle the PIC could calculate the exact power delivered to the speaker and protect it from over-heating.
I would like to hear from anyone who has build a boost-converter in the 300W-class.
TroelsM
One major drawback IMHO is the fact that you will need some sort of current limiting circuitry in this SMPS to protect the semiconductors. This means that the SMPS either have to be a lot bigger watt vice that the amp it supplies or have a slow current limiting circuit. Both cases call for quite big semiconductors so for bigger amps I prefer a good old torrid with nearly infinite short term current any day.
There is also the issue of components, normally I would prefer N channel devises over P channel types due to smaller conduction losses. Can this idea be implemented without the use of both?
Also I feel that a SMPS in this application sort of misses its purpose if it is sitting behind a conventional supply. If a SPMS is designed to run directly from the mains however, I believe it can be very useful for medium power, power amps in e.g. an active system.
I seem to remember that CAD audio makes some SMPS based power amplifiers; maybe Karsten will be helpful and explain how to implement this in a DIY module that would be safe to use and build?
\Jens
There is also the issue of components, normally I would prefer N channel devises over P channel types due to smaller conduction losses. Can this idea be implemented without the use of both?
Also I feel that a SMPS in this application sort of misses its purpose if it is sitting behind a conventional supply. If a SPMS is designed to run directly from the mains however, I believe it can be very useful for medium power, power amps in e.g. an active system.
I seem to remember that CAD audio makes some SMPS based power amplifiers; maybe Karsten will be helpful and explain how to implement this in a DIY module that would be safe to use and build?
\Jens
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.