or...... we have wifes/kids that cant tell the diff from music and distortion and dont want to have to replace drivers every 6months....
This thread is educational and extremely interesting....
What would happen if you used an undersized (say 500VA) transformer to drive a 2ch amp like this into 8R speakers? Would the large cap banks give you the headroom to hit the 2x 250W rating without distorting? If the load was sustained for more than a few mS then I presume the rail voltages would collapse, but what would happen then? Would it simply reduce the output power or would it also generate significant distortion?
I guess another way of asking the question: is there any value in having an output stage that is oversized relative to the power supply?
What would happen if you used an undersized (say 500VA) transformer to drive a 2ch amp like this into 8R speakers? Would the large cap banks give you the headroom to hit the 2x 250W rating without distorting? If the load was sustained for more than a few mS then I presume the rail voltages would collapse, but what would happen then? Would it simply reduce the output power or would it also generate significant distortion?
I guess another way of asking the question: is there any value in having an output stage that is oversized relative to the power supply?
Ranchu32: I have always been told that one should always use twice the power for the power supply. For example, you have 3 500w amplifiers and you want to drive them off the same psu. That said, we have 1500w of amplifier power, therefore we need 3000w of psu power. I honestly think that this is because the psu sags and also, each topology has an efficiency. Class A is somewhere between 20 and 40%, class B is more like 70%, class D 85-90%. Class AB has the efficiency of class A while giving low power and it goes in a more convenient class B when the power reaches a certain threshold (although it doesn't just snap into class B but rather transient into class B).
Regarding your qs: Large cap banks would allow certain spikes be heard. They give the amp headroom but keep in mind, after you discharge them you have to charge them back! So it would be more like: there's a spike, the amp reproduces it with no distortion because of the large cap bank, but then the psu sags because it's charging the caps, the amp gets a maybe not so clean and lower voltage supply, therefore it distorts, maybe better said, it clips.
This is my 2 cents.
Regarding the thread, yes, it is educational and interesting especially if you do some research prior to and while reading it.
Kevin, i didn't know that you were working on the layout but i am certainly glad that you do. Watch out tho for resistors R47, R48, R49. They might need to be higher power, around 5w from my calculations. Also, about the coil... 22 gauge wire seems too thin for the power.
As for the filtering caps, depending on my budget i might etch a board that could house 12 caps and only populate half of it now and the rest later (4-5 months later). Oooor i might bite it and go straight for 12 caps... I repeat, depending on my budget. I was looking at the ts 6800uf 100v panasonic caps myself as i saw they have good performance for the value.
Also, which trans are more likely to be counterfeited? MLJ21193/4 or the MJL21195/6 ? Are they interchangeable ? Are there any drawback for using 95/96?
Regarding your qs: Large cap banks would allow certain spikes be heard. They give the amp headroom but keep in mind, after you discharge them you have to charge them back! So it would be more like: there's a spike, the amp reproduces it with no distortion because of the large cap bank, but then the psu sags because it's charging the caps, the amp gets a maybe not so clean and lower voltage supply, therefore it distorts, maybe better said, it clips.
This is my 2 cents.
Regarding the thread, yes, it is educational and interesting especially if you do some research prior to and while reading it.
Kevin, i didn't know that you were working on the layout but i am certainly glad that you do. Watch out tho for resistors R47, R48, R49. They might need to be higher power, around 5w from my calculations. Also, about the coil... 22 gauge wire seems too thin for the power.
As for the filtering caps, depending on my budget i might etch a board that could house 12 caps and only populate half of it now and the rest later (4-5 months later). Oooor i might bite it and go straight for 12 caps... I repeat, depending on my budget. I was looking at the ts 6800uf 100v panasonic caps myself as i saw they have good performance for the value.
Also, which trans are more likely to be counterfeited? MLJ21193/4 or the MJL21195/6 ? Are they interchangeable ? Are there any drawback for using 95/96?
Or are there any better sounding transistors?
Sorry for double post, i forgot to write this in the first one and just realized...
Sorry for double post, i forgot to write this in the first one and just realized...
Hi brlmat
Believe me when I say I've been doing my reasearch prior to and while reading this thread!
Using a transformer that's twice the amplifier power isn't some universal rule of thumb. If you look at most hi-fi power amplifiers the transformer is only about 1/4 to 1/3 of the combined rated power output. The multi-channel AV receivers use transformers that are far, far less than the combined outputs - typically only about 1/6 to 1/4.
I was always of the belief that this is OK because (a) quiescent current is usually very low; (b) the average power output when playing music are in the mA to single digit amp range, and; (c) the maximum power ratings are only ever approached for very short periods.
So the 50W, 100W, 250W or whatever output power provides sufficient dynamic range for the periods when the amplifier is asked to supply (say) 10X to 100X normal levels for very brief periods without clipping.
From what I understand, transformers are able to significantly exceed their rated maximum current draw for brief periods - and of course there is the filter bank that is better able to service these transient surges anyhow.
I don't understand why people here recommend double the transformer power.
Most people seem to use 4281/4302 outputs, rather than 21193/94 pairs. When you look at the datasheets the 4281/4302 devices have a better SOA and would be the better choice unless you can get the 93/94 devices cheaply. Personally I would only ever buy transistors from places like E14, Mouser, Digikey, etc., to be guaranteed of the genuine article.
http://www.mouser.com/ds/2/308/MJL4281A-D-113116.pdf
http://www.mouser.com/ds/2/308/MJL21193-D-95048.pdf
Believe me when I say I've been doing my reasearch prior to and while reading this thread!
Using a transformer that's twice the amplifier power isn't some universal rule of thumb. If you look at most hi-fi power amplifiers the transformer is only about 1/4 to 1/3 of the combined rated power output. The multi-channel AV receivers use transformers that are far, far less than the combined outputs - typically only about 1/6 to 1/4.
I was always of the belief that this is OK because (a) quiescent current is usually very low; (b) the average power output when playing music are in the mA to single digit amp range, and; (c) the maximum power ratings are only ever approached for very short periods.
So the 50W, 100W, 250W or whatever output power provides sufficient dynamic range for the periods when the amplifier is asked to supply (say) 10X to 100X normal levels for very brief periods without clipping.
From what I understand, transformers are able to significantly exceed their rated maximum current draw for brief periods - and of course there is the filter bank that is better able to service these transient surges anyhow.
I don't understand why people here recommend double the transformer power.
Most people seem to use 4281/4302 outputs, rather than 21193/94 pairs. When you look at the datasheets the 4281/4302 devices have a better SOA and would be the better choice unless you can get the 93/94 devices cheaply. Personally I would only ever buy transistors from places like E14, Mouser, Digikey, etc., to be guaranteed of the genuine article.
http://www.mouser.com/ds/2/308/MJL4281A-D-113116.pdf
http://www.mouser.com/ds/2/308/MJL21193-D-95048.pdf
Last edited:
IMHO use what makes you happy.
But, don't use AV recevers as an example!
I have found that there is no replacement for displacement!
That said, only you can decide where the point is that the sound dosent get any better.
The specs can show power supply weak links.
My good comercial amps have huge power supplys. And are good to low ohms at full bandwidth.
Kinda get what you pay for.
But, don't use AV recevers as an example!
I have found that there is no replacement for displacement!
That said, only you can decide where the point is that the sound dosent get any better.
The specs can show power supply weak links.
My good comercial amps have huge power supplys. And are good to low ohms at full bandwidth.
Kinda get what you pay for.
"I don't understand why people here recommend double the transformer power."
I don't understand why people "tune" little 4 banger pissed off hornet in a pop can cars!
But to each their own.
🙂
I don't understand why people "tune" little 4 banger pissed off hornet in a pop can cars!
But to each their own.
🙂
Hi sakellogg
I'm not suggesting we all use the small transformers found in AV receivers and low-end 2ch amplifiers. Why stop at "2x"? Why not "10x" if more is always better? Of course most people would scoff say that is silly; it's also impractical to have a 5kVA inside a stereo 250W amplifier.
There's always a point of diminishing returns and I'm suggesting that point is reached long before "2x" with typical class A/B designs. I read on this forum a post from a guy who designed amplifiers and the transformer specification was 70% of the combined output power on their high-end units.
I'm currently building a DX Blame MkIII with 20 output devices. I'm interested in this thread because the amps being discussed are similar. The DX can output 200W into 8R with very low distortion or 880W into 2R with low distortion figures also. My speakers are 8R nominal impedance. I guess most of the time they are drawing a few watts at most, but I like to have the headroom for the dynamic passages in the music. I also like the idea of driving 4R and 2R loads but would not need to drive them beyond (say) 200W.
Most of the guys who are building their MkIII's seem to be using 1kVA to 2kVA of transformers. But for normal listening levels in a living room it seems to be extreme overkill.
My current plan is to use a single 50V-0-50V / 500VA transformer. The rail voltages will be slightly higher than needed and specified. I've checked with the designer and this is OK provided suitable component voltage rating are observed. My assumption is that this way peak power outputs can be sustained for longer, once the rails start to sag.
The smaller transformer means I can build an enclosure is lighter, and importantly, more compact. I end up with better seperation between the transformer and the electronics for reduced EMI. The money saved will be spent on the filter bank, which, if my understanding is correct, has a bigger influence on how the power supply deals with transient power surges.
I'm approaching this next project as an engineering challenge, where every design choice comes with tradeoffs. Or I might be crazy to think about strapping a 500VA transformer to a 20 output amplifier might be a good idea 🙂
I'm not suggesting we all use the small transformers found in AV receivers and low-end 2ch amplifiers. Why stop at "2x"? Why not "10x" if more is always better? Of course most people would scoff say that is silly; it's also impractical to have a 5kVA inside a stereo 250W amplifier.
There's always a point of diminishing returns and I'm suggesting that point is reached long before "2x" with typical class A/B designs. I read on this forum a post from a guy who designed amplifiers and the transformer specification was 70% of the combined output power on their high-end units.
I'm currently building a DX Blame MkIII with 20 output devices. I'm interested in this thread because the amps being discussed are similar. The DX can output 200W into 8R with very low distortion or 880W into 2R with low distortion figures also. My speakers are 8R nominal impedance. I guess most of the time they are drawing a few watts at most, but I like to have the headroom for the dynamic passages in the music. I also like the idea of driving 4R and 2R loads but would not need to drive them beyond (say) 200W.
Most of the guys who are building their MkIII's seem to be using 1kVA to 2kVA of transformers. But for normal listening levels in a living room it seems to be extreme overkill.
My current plan is to use a single 50V-0-50V / 500VA transformer. The rail voltages will be slightly higher than needed and specified. I've checked with the designer and this is OK provided suitable component voltage rating are observed. My assumption is that this way peak power outputs can be sustained for longer, once the rails start to sag.
The smaller transformer means I can build an enclosure is lighter, and importantly, more compact. I end up with better seperation between the transformer and the electronics for reduced EMI. The money saved will be spent on the filter bank, which, if my understanding is correct, has a bigger influence on how the power supply deals with transient power surges.
I'm approaching this next project as an engineering challenge, where every design choice comes with tradeoffs. Or I might be crazy to think about strapping a 500VA transformer to a 20 output amplifier might be a good idea 🙂
Guys. Long time, no see. Think of it this way: you have a 30w class A amplifier. It's efficiency is 20%. This means that you have 30w of power to the speakers and 120w of heat. Now you add these together and you get 150w of needed power. This is why you need bigger transformers for a amp of a given power. Similar to this is class ab which has an efficiency of around 50 to 60%. This is where all this "twice the power" comes from. Also, one usually rounds thr number up just for safety and for less sag.it is true tho that a class ab amp is more like 65 % efficient at and near full power so it's safe to go 800w (even 750w but with big filtering caps) for a 500w amp.
I'd write a little more but with another occasion as I am extremely tired (it's 1.30am here in romania). I hope this post helps tho
I'd write a little more but with another occasion as I am extremely tired (it's 1.30am here in romania). I hope this post helps tho
brlmat, I understand what you're saying about efficiency. If you hooked a sine wave generator to the input and drove the speakers at full power, then yes, you would need a transformer that's 2x the amplifier rated power (give or take - assuming 50% efficiency). But music is not like this - the average power output will be something like 1/10 or 1/100 of the peaks.
The purpose of the filter cap bank is not only to smooth the rectified AC power but to provide sufficient power reserve for these short music peaks. A capacitor is going to respond better to transient loads than a transformer.
If you look at the SOA graph for a common output transistor, you see that the figures for 100ms don't look that much better than 1sec. I guess that explains why you need to parallel a bunch of devices to obtain high peak wattages even if the average music power is much lower.
The purpose of the filter cap bank is not only to smooth the rectified AC power but to provide sufficient power reserve for these short music peaks. A capacitor is going to respond better to transient loads than a transformer.
If you look at the SOA graph for a common output transistor, you see that the figures for 100ms don't look that much better than 1sec. I guess that explains why you need to parallel a bunch of devices to obtain high peak wattages even if the average music power is much lower.
Attachments
Yet, I would still go for a big transformer as more is not always too much. Maybe 800w or something similar.
I hear you...
I've managed to (nearly) convince myself that my logic is sound but then part of me is saying "just throw a pair of 500VA transformers in there and call it a day".
Would love to hear some thoughts from Struth or AndrewT.
I've managed to (nearly) convince myself that my logic is sound but then part of me is saying "just throw a pair of 500VA transformers in there and call it a day".
Would love to hear some thoughts from Struth or AndrewT.
Yeah, me too! I feel the same as you do...
PS: My whole amp is going to cost around 300 euros. It's a bit much, but what can i do?
PS: My whole amp is going to cost around 300 euros. It's a bit much, but what can i do?
Last edited:
Also, can someone recommend me some op-amps better sounding than tl074? It has to be 4 opamps in a package tho as i'll use them as opamps for the eq circuit and as buffers.
I almost shat myself when i saw how much the opa costs... lme would be cheaper, but still too much. I need 8 of them. The lme is 3.15 euros. Times 8 this goes up to 25 euros. This is a lot! TL074 retails for 0.512 euros. Times 8 it's 4.1 euros. That a lot more convenient. I'd go up to 10 maybe 12 euros for all 8 of them if that's possible.
Thank you!
Thank you!
Why do I hear a request for a repetition of my advice?
For the last few years I always recommend Transformer VA of 1times to 2times the total maximum output power.
Simple to remember and simple to apply. There is some arithmetic involved, but any primary school pupil should be able to cope.
For the last few years I always recommend Transformer VA of 1times to 2times the total maximum output power.
Simple to remember and simple to apply. There is some arithmetic involved, but any primary school pupil should be able to cope.
Hello Andrew
Based on your advice, a single 500VA transformer falls within that range and would therefore suit my application.
Regards
Based on your advice, a single 500VA transformer falls within that range and would therefore suit my application.
Regards
A 500 VA trafo will require higher unloaded rail voltage to get to the same power as using a 1KVA. Usually, the lower, stiffer rails sound better when driven heavily into clipping. Using the lower VA rating and higher rails requires more output transistors, as the SOA required is calculated from the unloaded rail voltage. Silicon is cheaper than iron, so most manufacturers choose the smaller trafo option. If you use a 500VA, you just might end up with lower power than you expect, unless you already planned for this. Calculating loaded DC voltage isn't necessarily straightforward - there is more to it than the 3% to 5% regulation rating of the trafo. That is for an AC load at 100% power factor and a rectifier/capacitor load is neither.
As far as what op-amp to use, there is nothing wrong with a TL074. Its weakness is in its output stage and that can be dealt with by biasing it a few mA into class A (hang resistor to negative rail), which you can do with any op amp. And usually, in a system where an equailzer is used to tune the sound to taste, getting .00000001% distortion isn't all that important anyway.
There are things I really like about the TL07x op amps and their LF-sries bretheren. Being nearly indestructible and immune to damage from input overload is one of them. Good thing for power amps that get subject to being turned up too loud.
As far as what op-amp to use, there is nothing wrong with a TL074. Its weakness is in its output stage and that can be dealt with by biasing it a few mA into class A (hang resistor to negative rail), which you can do with any op amp. And usually, in a system where an equailzer is used to tune the sound to taste, getting .00000001% distortion isn't all that important anyway.
There are things I really like about the TL07x op amps and their LF-sries bretheren. Being nearly indestructible and immune to damage from input overload is one of them. Good thing for power amps that get subject to being turned up too loud.
- Status
- Not open for further replies.