For 2 ohm operation into 85v rails, I'd go 8 pairs of the 5200/1943's. I would drive these with the same type i.e. 5200/1943.
This will optimize distortion for 4e operation and make 2e operation pretty practical. Using standard drivers will let you drive 8e safely but any reactive loads or lower impedances will strain the smaller drivers.
I have used this combinition with excellent results.
Also While using the 5200/1943 as drivers, I used a single (common) emitter resistor of 20 ohms bypassed with a 1 uF switch-off cap across this floating emitter resistor.
This will optimize distortion for 4e operation and make 2e operation pretty practical. Using standard drivers will let you drive 8e safely but any reactive loads or lower impedances will strain the smaller drivers.
I have used this combinition with excellent results.
Also While using the 5200/1943 as drivers, I used a single (common) emitter resistor of 20 ohms bypassed with a 1 uF switch-off cap across this floating emitter resistor.
Thanks
Now the following is final for my amp ;-
Power Transformer Sec 60-0-60v when pri is 220v ( this will be 65-0-65v when pri voltage increases to 240v at night ) I hope this is ok ?
Caps 40,000MFD 100v per rail , will a stereo version require more MFD ??
Output Transistors 8 pairs of the 5200/1943's , mounted on extralarge heatsink with cooling fan .
Drivers 5200/1943
as the output stage is a tripledarlington what should I use to drive the 5200/1943 drivers , will MJE 15034/35 be ok or should I look for a Toshiba device will a 250v 5w 0.5A device be enough ,
sajti & K-amps
I have an intution that somewhere there is a mistake , this amp will deliver say max 800w at 4E which is 56.5vrms and the output current will be 14.28 amps , peak current will be higher , how will each transistor dessipate 23w only. You have passed the same for 2E too.
Now the following is final for my amp ;-
Power Transformer Sec 60-0-60v when pri is 220v ( this will be 65-0-65v when pri voltage increases to 240v at night ) I hope this is ok ?
Caps 40,000MFD 100v per rail , will a stereo version require more MFD ??
Output Transistors 8 pairs of the 5200/1943's , mounted on extralarge heatsink with cooling fan .
Drivers 5200/1943
as the output stage is a tripledarlington what should I use to drive the 5200/1943 drivers , will MJE 15034/35 be ok or should I look for a Toshiba device will a 250v 5w 0.5A device be enough ,
sajti & K-amps
I have an intution that somewhere there is a mistake , this amp will deliver say max 800w at 4E which is 56.5vrms and the output current will be 14.28 amps , peak current will be higher , how will each transistor dessipate 23w only. You have passed the same for 2E too.
rajeev luthra said:
The peak output voltage is 80V, the peak output current is 20A. But the average current from the power supply is Ia=2/Pi*Ip=12.74A. This means 1082W. If the output is 800W, 282W is the dissipated power for 16 output devices.
Please note, the this is not the maximum dissipation, because maximum dissipation exist at 63% of the output voltage, which is approx 40% of maximum output power.
sajti
rajeev luthra said:
Crunched some numbers for you. These are approximates and yours could differ. Using 60-0-60 and 1 kvA tranny rated at 5% regulation, this is what you get:
Inputs
Number of output devices: 8 Pairs
Voltage rails (per rail): 83 volts
Emitter resistance (per device): 0.22 ohms
Bias voltage per Emmiter resistor: 9 mv
Idle bias per device 40.9 mA
Speaker ohms 8 ohms
Results
Idle bias per device: 0.041 Amps
Total Amplifier bias (per rail)** 0.33 Amps
Total Amplifier bias (both rails) 0.65 Amps
Total Dissipation (per channel at idle) 54.3 Watts
Dissipation per device pair at idle 13.6 watts
Class-A output: Peak 3.4 Watts peak
Class-A output: RMS 1.7 Watts RMS
Efficiency 3.15 %
Max Class-AB RMS (1 kVA Toroid p/CH 5% reg) 391.77 Watts RMS
And at 65-0-65 the numbers look like:
Inputs
Number of output devices: 8 Pairs
Voltage rails (per rail): 90 volts
Emitter resistance (per device): 0.22 ohms
Bias voltage per Emmiter resistor: 9 mv
Idle bias per device 40.9 mA
Speaker ohms 8 ohms
Results
Idle bias per device: 0.041 Amps
Total Amplifier bias (per rail)** 0.33 Amps
Total Amplifier bias (both rails) 0.65 Amps
Total Dissipation (per channel at idle) 58.9 Watts
Dissipation per device pair at idle 14.7 watts
Class-A output: Peak 3.4 Watts peak
Class-A output: RMS 1.7 Watts RMS
Efficiency 2.91 %
Max Class-AB RMS (1 kVA Toroid p/CH 5% reg) 460.64 Watts RMS
The wattage you will get will depend a lot on the size and regulation of the transformer. For a stereo amp, you are looking at close to 2kvA. Anything less and you will not get 800w at 4e. If using a smaller transformer, you will need to increase DC rails.
Thats why high end designers use huge trafo's because they can lower rails for a given output and make it warmer sounding compared to a hi-volt low current design. IMHO. 😉
Wow, these are numbers ! 60watt idle dissipation per channel ?
This make a heatsink of ~0.1k/w just to keep idle temp at ~30° ! (per channel)
I think this amp will become some 50kg monster !
Are you sure that you need that much power ?
This make a heatsink of ~0.1k/w just to keep idle temp at ~30° ! (per channel)
I think this amp will become some 50kg monster !
Are you sure that you need that much power ?
looking at the soa curve won't this be a little cramped by the 2nd breakdown locus, are you designing for +/- 45 degree complex load currents?
bridged output's lower supply voltage lets you run nearer the bjt device power dissapation limit curve
bridged output's lower supply voltage lets you run nearer the bjt device power dissapation limit curve
MikeB said:
My experience is, that not possible to build this amp without fan cooling. I already made some high power amplifier, and they work properly, with low noise, speed regulated fan.
sajti
K-amps / sajti
What will be the min & max base current of the bank of output devices and the base current of the driver transistor
Reg transformer rating the coments by Dan Fraser in the thread "What should be the transformer rating in respect to wattage of a power amplififier" are worth going through.
We also have a lot of storage in the filter caps again I will quote Mr Frasers text
((However, since energy storage goes up by linearly with capacitance but by the square of the voltage in the capacitor, the amount of energy I have stored is quite massive. As well, the parallel arrangement makes the effective series resistance of the capacitors much smaller, allowing for more efficient of dumping of power into the load.))
MikeB
(( Wow, these are numbers ! 60watt idle dissipation per channel ?
This make a heatsink of ~0.1k/w just to keep idle temp at ~30° ! (per channel)
I think this amp will become some 50kg monster !
Are you sure that you need that much power ? ))
I have already stated that I will use an extra large heatsink with cooling fan , also this is not my first 800w amp , the only differance is that the previous ones were bridged amps.
I have just completed a dual 15" sub for Home theatre and I was testing with a cartoon DTS dvd , I could feel vibrations all over my body and my hair stoodup when the lion roared , this is what 800w can do.
What will be the min & max base current of the bank of output devices and the base current of the driver transistor
Reg transformer rating the coments by Dan Fraser in the thread "What should be the transformer rating in respect to wattage of a power amplififier" are worth going through.
We also have a lot of storage in the filter caps again I will quote Mr Frasers text
((However, since energy storage goes up by linearly with capacitance but by the square of the voltage in the capacitor, the amount of energy I have stored is quite massive. As well, the parallel arrangement makes the effective series resistance of the capacitors much smaller, allowing for more efficient of dumping of power into the load.))
MikeB
(( Wow, these are numbers ! 60watt idle dissipation per channel ?
This make a heatsink of ~0.1k/w just to keep idle temp at ~30° ! (per channel)
I think this amp will become some 50kg monster !
Are you sure that you need that much power ? ))
I have already stated that I will use an extra large heatsink with cooling fan , also this is not my first 800w amp , the only differance is that the previous ones were bridged amps.
I have just completed a dual 15" sub for Home theatre and I was testing with a cartoon DTS dvd , I could feel vibrations all over my body and my hair stoodup when the lion roared , this is what 800w can do.
If the amp is intended for a subwoofer, wouldn't a ClassD be a better choice ?
You can easily use even more power without this kind of dissipation.
Mike
You can easily use even more power without this kind of dissipation.
Mike
output transistor voltage rating??
Please don't laugh at my amplifier newbee question>
Loudspeakers are my game but I've got this amp bug now!
I understand it to be true that an output device needs to be rated at in excess of twice the dc rail that it sits on. (assuming complimentary design) Please correct me if I am wrong.
Now, this being the case I am wondering why it needs to be rated at > the total +-voltage if it only resides on the top..0 to+ (or bottom..0 to -) In what case does it see the full rail voltages? I am not disagreeing with convention at all. I am just trying to understand the convention.
Also, anyone ever used the 2N5631/6031 pair. Are we talking about a good audio output here?
Thanks in advance,
Marc
Please don't laugh at my amplifier newbee question>
Loudspeakers are my game but I've got this amp bug now!
I understand it to be true that an output device needs to be rated at in excess of twice the dc rail that it sits on. (assuming complimentary design) Please correct me if I am wrong.
Now, this being the case I am wondering why it needs to be rated at > the total +-voltage if it only resides on the top..0 to+ (or bottom..0 to -) In what case does it see the full rail voltages? I am not disagreeing with convention at all. I am just trying to understand the convention.
Also, anyone ever used the 2N5631/6031 pair. Are we talking about a good audio output here?
Thanks in advance,
Marc
For example, if the amp puts out full positive voltage, the lower transistor
has its collector connected to - supply, the emitter to output. But
output is at nearly max +voltage, so it has to withstand the full
supplyvoltage at this moment. The other has nearly 0volt.
Mike
has its collector connected to - supply, the emitter to output. But
output is at nearly max +voltage, so it has to withstand the full
supplyvoltage at this moment. The other has nearly 0volt.
Mike
K-amps
I have noticed that in these High power amps idle current of 7to 8ma per device is enough , this also brings down the total dissipation in idle mode ,
does the 1 uF switch-off cap across this floating emitter resistor make much differance??.
Regarding
(( Thats why high end designers use huge trafo's because they can lower rails for a given output and make it warmer sounding compared to a hi-volt low current design. IMHO. ))
My concept is that when its a " hi-volt low current design " the power supply regulation is better than " hi-current low volt design " also the same MFD will have more storage capacity , hence the power output will be more , Yes the cost of high voltage devices and caps is much higher and this could be the reason why some commercial amp designers do not go for this topology .
jcx
If we tune the protection circuit properly I think we can get away with it. But we can increase the devices if required .
I have noticed that in these High power amps idle current of 7to 8ma per device is enough , this also brings down the total dissipation in idle mode ,
does the 1 uF switch-off cap across this floating emitter resistor make much differance??.
Regarding
(( Thats why high end designers use huge trafo's because they can lower rails for a given output and make it warmer sounding compared to a hi-volt low current design. IMHO. ))
My concept is that when its a " hi-volt low current design " the power supply regulation is better than " hi-current low volt design " also the same MFD will have more storage capacity , hence the power output will be more , Yes the cost of high voltage devices and caps is much higher and this could be the reason why some commercial amp designers do not go for this topology .
jcx
If we tune the protection circuit properly I think we can get away with it. But we can increase the devices if required .
Hello ,
jcx,s text has made me do some thinking and last night I did some reading , and decided that before we proceed it will be wise to look into the following :-
I did some calculations which are :-
I took the worst case which is when mains is 240v at night , here the secondary voltage of the transformer will be 65-0-65v and DC will be +/- 92v .
The peak output voltage will be around 87 v ,
The peak output power at 4E will be 1892 w ,
The peak load current at 4E will be 21.75 A ,
At 2E the figures will be double,
Other things to be kept in mind are ;-
The quoted power ratings of the output devices is normally at 25 digrees centigrade and the device operates at higher temp in the amp hence we have to reduce the internal power dissipation so that the device junctions are not damaged by overheating,
The internal dissipation of the device will be 40% of the of the max power output of the amp with pure sinewave inputs , but with music input , odd harmonics increase the internal dissipation until one reaches the worst possible case of internal dissipation of 50% of the rated power output , now add another 10% to allow adequate margin of safety by assuming a maximum internal dissipation of 60% or higher of the rated power output of the amp .
Please let me have your suggestions and views .
jcx,s text has made me do some thinking and last night I did some reading , and decided that before we proceed it will be wise to look into the following :-
I did some calculations which are :-
I took the worst case which is when mains is 240v at night , here the secondary voltage of the transformer will be 65-0-65v and DC will be +/- 92v .
The peak output voltage will be around 87 v ,
The peak output power at 4E will be 1892 w ,
The peak load current at 4E will be 21.75 A ,
At 2E the figures will be double,
Other things to be kept in mind are ;-
The quoted power ratings of the output devices is normally at 25 digrees centigrade and the device operates at higher temp in the amp hence we have to reduce the internal power dissipation so that the device junctions are not damaged by overheating,
The internal dissipation of the device will be 40% of the of the max power output of the amp with pure sinewave inputs , but with music input , odd harmonics increase the internal dissipation until one reaches the worst possible case of internal dissipation of 50% of the rated power output , now add another 10% to allow adequate margin of safety by assuming a maximum internal dissipation of 60% or higher of the rated power output of the amp .
Please let me have your suggestions and views .
What's E on about?
Hi Rajeev,
Just out of curiosity why are you using the letter E to denote ohms? Or do you not intend to mean ohms?
Hi Rajeev,
Just out of curiosity why are you using the letter E to denote ohms? Or do you not intend to mean ohms?
rajeev luthra said:
Somehow i hoped you have... 😀
But i haven't read these large threads about classd here...
Mike
Don't forget that E = mc^2 😀
For your info, here's a site I find really useful for keeping track of standard symbols included in the International System of Units:
SI units
It is unfortunate that Ohms are represented by the Greek letter Omega which is tricky to type. Fortunately "ohms" is quite short.
For your info, here's a site I find really useful for keeping track of standard symbols included in the International System of Units:
SI units
It is unfortunate that Ohms are represented by the Greek letter Omega which is tricky to type. Fortunately "ohms" is quite short.
Rajeev,
You are now almost done with the output stage design. Looks good, eight pairs should be enough for government work, small scale mild steel welding with moderate reliability should be possible........😀
One point though, you do need to address current sharing in the output devices.
How will you guarantee it? Incredibly precise matching of your eight devices on each rail? Or very large emitter resistors? How big?
Now, on to the voltage amp and driver stage!!
Will you use full complementary diff input stage with two voltage amplifiers facing one another from opposite rails?
Or will you use a single diff pair followed by a single-ended voltage amplifier?
If the former, are you able to control the offset properly? Remember, high voltages here, and we are juggling low impedance power supplies which eat voice coils for breakfast!!
If a SE VAS, will you use current source as a load, or a bootstrap?
What is the transistor choice for the VAS? What current will you run? Will the beta be high enough to easily drive from the input stage?
For eight output pairs, you will have an appreciable current. Let us imagine a 2R load; if you dump 84V into it, this is 42 amperes, spread over eight devices, this is 5.25A apiece. That's OK, but you must supply all necessary base current. If we assume beta of 50 at this moderate CE current, each output will require 105mA of drive, and so eight will need 840mA of base current just to drive the outputs, not counting any reactive kickback which might mandate even more (and threaten the SOAR of the outputs).
Driving this beast is beginning to look like a nightmare, and I strongly suggest that you drive all eight outputs on each rail with another like output device, viz a 2SC5200/2SA1943, easily capable of supplying this current at high beta, let's assume 50.
So, to drive the driver you will need a base current of 840/50 = 16.8mA, so for a reserve of 100% on the voltage amplifier, you will need to run 33mA through it, giving a dissipation of more than 3W at your overrun rail voltage of +/-92V. This narrows your choice of voltage amplifier, as the truly fast ones (remember this component is in common collector, so speed is important) with a decent dissipation margin are quite tricky to source.
You have a choice here; either run a high current voltage amplifier (which bulks up the input stage as well for obvious drive purposes) OR run a conventional low power input and VAS and follow the VAS with two complementary emitter followers to drive the drivers.......
You therefore have input stage => voltage amplifier => predriver emitter follower => driver stage => output stage. These five stages must support a global feedback loop to derive the very low output impedance required for an amplifier of this current capacity, and the difficulties of designing unconditional stability at these sorts of criteria now abruptly reveal themselves.
Let me just say that you will need a very tight layout with extremely low parasitics, and at least a 150MHz VAS and input stage, with maybe 50MHz devices as the triple emitter followers. These devices are all readily available, incidentally; I'd suggest the 2SC4793/2SA1837 as the pre-drivers, and a suitable VAS would be the Sanyo 2SC3782.
The lag compensation and phase lead dimensioning to achieve stability would take about two months of intensive work for real world loads, though you can accelerate this with PSpice and an IHF load for testing. But while this takes care of the engineering issues, until now no attention has been paid to the sonics.
If you want this thing to sound good you will need same batch output device matching (this reduces sample size to perhaps 100 for 8 exactly alike devices), and very careful voicing of all operating points and capacitors in the circuit for best sonics.
It's all doable, but unless you are very experienced, it could easily consume many months of full time, rigorous R&D.
Ready?
Cheers,
Hugh
You are now almost done with the output stage design. Looks good, eight pairs should be enough for government work, small scale mild steel welding with moderate reliability should be possible........😀
One point though, you do need to address current sharing in the output devices.
How will you guarantee it? Incredibly precise matching of your eight devices on each rail? Or very large emitter resistors? How big?
Now, on to the voltage amp and driver stage!!
Will you use full complementary diff input stage with two voltage amplifiers facing one another from opposite rails?
Or will you use a single diff pair followed by a single-ended voltage amplifier?
If the former, are you able to control the offset properly? Remember, high voltages here, and we are juggling low impedance power supplies which eat voice coils for breakfast!!
If a SE VAS, will you use current source as a load, or a bootstrap?
What is the transistor choice for the VAS? What current will you run? Will the beta be high enough to easily drive from the input stage?
For eight output pairs, you will have an appreciable current. Let us imagine a 2R load; if you dump 84V into it, this is 42 amperes, spread over eight devices, this is 5.25A apiece. That's OK, but you must supply all necessary base current. If we assume beta of 50 at this moderate CE current, each output will require 105mA of drive, and so eight will need 840mA of base current just to drive the outputs, not counting any reactive kickback which might mandate even more (and threaten the SOAR of the outputs).
Driving this beast is beginning to look like a nightmare, and I strongly suggest that you drive all eight outputs on each rail with another like output device, viz a 2SC5200/2SA1943, easily capable of supplying this current at high beta, let's assume 50.
So, to drive the driver you will need a base current of 840/50 = 16.8mA, so for a reserve of 100% on the voltage amplifier, you will need to run 33mA through it, giving a dissipation of more than 3W at your overrun rail voltage of +/-92V. This narrows your choice of voltage amplifier, as the truly fast ones (remember this component is in common collector, so speed is important) with a decent dissipation margin are quite tricky to source.
You have a choice here; either run a high current voltage amplifier (which bulks up the input stage as well for obvious drive purposes) OR run a conventional low power input and VAS and follow the VAS with two complementary emitter followers to drive the drivers.......
You therefore have input stage => voltage amplifier => predriver emitter follower => driver stage => output stage. These five stages must support a global feedback loop to derive the very low output impedance required for an amplifier of this current capacity, and the difficulties of designing unconditional stability at these sorts of criteria now abruptly reveal themselves.
Let me just say that you will need a very tight layout with extremely low parasitics, and at least a 150MHz VAS and input stage, with maybe 50MHz devices as the triple emitter followers. These devices are all readily available, incidentally; I'd suggest the 2SC4793/2SA1837 as the pre-drivers, and a suitable VAS would be the Sanyo 2SC3782.
The lag compensation and phase lead dimensioning to achieve stability would take about two months of intensive work for real world loads, though you can accelerate this with PSpice and an IHF load for testing. But while this takes care of the engineering issues, until now no attention has been paid to the sonics.
If you want this thing to sound good you will need same batch output device matching (this reduces sample size to perhaps 100 for 8 exactly alike devices), and very careful voicing of all operating points and capacitors in the circuit for best sonics.
It's all doable, but unless you are very experienced, it could easily consume many months of full time, rigorous R&D.
Ready?
Cheers,
Hugh
- Status
- Not open for further replies.