We are started with the design of the 360W SMPS, actually it's the same design as the 180W but double the power. Probably we will go for two transformers running in parellel.
Later this year we will start with the more power full SMPS. I plan to have this availble as samples for the next Frankfurter Prolight & Sound fair April 2009.
Please note that we rate our SMPS for power at 20Hz! Whereby our competitor rates the products at 1kHz. Our SMPS will do the double power at 1kHz, but nobody is intersted for full power at 1khz.
So we have decided to rate the SMPS with realistic power levels at 20Hz..... 😀
Regards,
Jan-Peter
Later this year we will start with the more power full SMPS. I plan to have this availble as samples for the next Frankfurter Prolight & Sound fair April 2009.
Please note that we rate our SMPS for power at 20Hz! Whereby our competitor rates the products at 1kHz. Our SMPS will do the double power at 1kHz, but nobody is intersted for full power at 1khz.
So we have decided to rate the SMPS with realistic power levels at 20Hz..... 😀
Regards,
Jan-Peter
About defining output power:
A 2*400/600W (sin/music) amplifier can draw 35-40 A at peaks from a 70 V rail. This is 2800 W, and only from one puffer capacitor if you don't change the polarity of one amp. This amplifier needs approx. 900 W power for nominal sinus output at 1 kHz. I think one should really be careful about different power definitions, output capacitors, and output overload characteristics! DC spec. alone is almost useless.
But extra power at DC or 1 kHz is also unneccessary.
A 2*400/600W (sin/music) amplifier can draw 35-40 A at peaks from a 70 V rail. This is 2800 W, and only from one puffer capacitor if you don't change the polarity of one amp. This amplifier needs approx. 900 W power for nominal sinus output at 1 kHz. I think one should really be careful about different power definitions, output capacitors, and output overload characteristics! DC spec. alone is almost useless.
But extra power at DC or 1 kHz is also unneccessary.
Hi
So? Peak is peak, I don't care, for that you have caps, now if you can get say 600w from rail, your amp can do 600w peak too, but you still have energy stored in caps, so amp can do more then just 600w peak. If all you have is power converters (smps, Damp), little power is wasted in heat,... everything else goes to load, so I don't see why I would need 1kw smps to power 600w amp.
From what I've seen smps had less work to do then amp
So if you can explain why you are right and I am wrong.
So? Peak is peak, I don't care, for that you have caps, now if you can get say 600w from rail, your amp can do 600w peak too, but you still have energy stored in caps, so amp can do more then just 600w peak. If all you have is power converters (smps, Damp), little power is wasted in heat,... everything else goes to load, so I don't see why I would need 1kw smps to power 600w amp.
From what I've seen smps had less work to do then amp
So if you can explain why you are right and I am wrong.
OK, you have cap, but how much? For example if you want to power up a half cycle of 20 Hz boom with less then 15 % drop (~30 % drop in output power!), then you need approx. 40000 uF per rail. This is quite expensive, big, and not likely to exist in most PSU. You can't assume that caps automatically equalize power fluctuations, because in most cases they don't!
But for example if SMPS is capable of give 200% of rated power for a short period, or you equalize power consumption in other way, you can dramatically decrease puffer cap costs without any negative side-effect. But this needs intelligent, selective power spec.!
I agree, you don't need, if temporary power capacity exceeds significantly continuous power. But this have to be specified properly!
But for example if SMPS is capable of give 200% of rated power for a short period, or you equalize power consumption in other way, you can dramatically decrease puffer cap costs without any negative side-effect. But this needs intelligent, selective power spec.!
I agree, you don't need, if temporary power capacity exceeds significantly continuous power. But this have to be specified properly!
Hi
I see, no I don't have 40kuF per rail, just 6600uF, but I assume it is enough because of high freq. used in smps and because it is unregulated + I don't play music under 25Hz more 30 and up, coz 12"(30cm) speakers can't play good under ~30Hz, not with such small power
I see, no I don't have 40kuF per rail, just 6600uF, but I assume it is enough because of high freq. used in smps and because it is unregulated + I don't play music under 25Hz more 30 and up, coz 12"(30cm) speakers can't play good under ~30Hz, not with such small power
SMPS freq is not related to puffer cap requirements. The load is the relevant.
20 Hz needs 40 mF -> 6,6 mF is "enough" for 120 Hz if every condition is the same.
PS.: 4 ohm load, stereo, in-phase, half bridge amps.
Hi
But as I understand you this would be if there wouldn't be any supply present...but you do have it, so caps power only small portion of signal when there is max. I don't see any amp that have that much capacitance and do drive 20Hz signals and they don't have supply rated a lot more then amp puts out.
And a nother thing: Half period of 20Hz is 25ms, if I load supply for that time one rail, there will be more power consumed then if amp was connected and he would reproduce half sine of same freq., max voltage od sine is same as DC value.
Only way that they would consume same power is if sine would have higher max value.
Right?
But as I understand you this would be if there wouldn't be any supply present...but you do have it, so caps power only small portion of signal when there is max. I don't see any amp that have that much capacitance and do drive 20Hz signals and they don't have supply rated a lot more then amp puts out.
And a nother thing: Half period of 20Hz is 25ms, if I load supply for that time one rail, there will be more power consumed then if amp was connected and he would reproduce half sine of same freq., max voltage od sine is same as DC value.
Only way that they would consume same power is if sine would have higher max value.
Right?
Dont forget that input cap of SMPS damp the peak power (another ''power reserve''...). That's not like a Transfo power supply that you just have output cap. Good rule of thumb is to keep 2000uf per rail for each KW and 1000uf at input of switching PSU for each KW too, based on a ration of 2:1 on input output power supply with 50-100KHZ switching frequency. Without regulation, that's give 15% voltage variation from full load to no load and vice-versa. I use that rule of thumb with sucsess and QSC too....
Look at ST L6571 to have cheap solution to got nearly 1KW with PNP buffer for mosfet drive discharge...That's cheap, simple and work very well!
Not exactly. I assumed there is nominal current from SMPS, but a slightly overdriven amplifier (wich is a very usual situation). As I said, max. power is 2800 W, while nominal power is 900 W! Difference is 1900 W, not a small portion! (Of course not during the whole half cycle.) But please don't make an argue about the exact value of discharging current, unless you make the calculations! The point is: peak and average differs very much, and capacitor can't equalize this unconditionally, ergo: DC spec is not enough!
I've build some.
I didn't calculate 25 ms direct load. If I had, the result was 60000 uF.
Pafi said:
Hello Pafi
I understand you are still in prototype stage although near completion.
Once your tests are complete, are you going to share the schematics and other construction information of your SMPS? It sure looks interesting to me.
Thanks!
Pafi said:
uF/kW??????? Regardless of topology, load impedance and number of channels???????
Class d amplifier transfert power, not like analog amplifier that use plain voltage and transfert load current to power supply! If you have 200W into 4 ohms, that's means you nead at least 40V of power supply, involve 28V RMS at output and 7.15A at load. You will draw 5A on power supply...Not 7.15A. That's another storry for analogue amp...You will draw the 7.15A of the speaker + the idle current. That's means a minimum of 286Watts...Compared to 200W (+ loss..) for a class d....Even worst with higger power amplifier. Impedance on class d will not change a lot the current draw on power supply FOR SAME OUTPUT POWER at same PSU voltage. Think about that Mr Pafi..With class d amplifier on switching power supply, you can use my rule of thumb without probleme.. 🙂
Do you really think this is new for me? And do you think this has any relevance to my question?
Do you want to say that every ClassD amp works with the same supply voltage???
And you didn't tell anything about topology and number of channels!!!
OK, take a 2*50 W amp, 2 ohms load, half bridge, in-phase drive! So you say 200 uF/rail. Enough? Doubtfully. If you drive all channels with the same signal, supply pumping current will be 1,75 A peak, so 8.8 V/ms increase in opposite capacitor, wich is unacceptable! Rail voltage can increase many times of normal voltage easily!
Your rule of thumb is not a rule at all! It's just an example.
Hi Pafi!
Fredos wrote about his experience. He is skilled. He has an amplifier business. In my experience, the average current draw from the supply is not so much with Class-D output stage, because of the inductive load.
You can't share your schematics or you don't?
Fredos wrote about his experience. He is skilled. He has an amplifier business. In my experience, the average current draw from the supply is not so much with Class-D output stage, because of the inductive load.
You can't share your schematics or you don't?
Gyula!
Yes, he seems to be skilled in building and selling amplifiers. So? This topic is not about skills! Fredos generalized his experience. His experience comes from a very limited sample, about 0,001 % of all ClassD amps in the world, without real theoretical basement, and despite of this he interpreted as a rule, while it is clearly not.
Also this experience is useless without giving some neccessary background information!
If a phisics teacher said 2+2=5, then would you agree with him, because he is skilled?
Inductive or capacitive! And because of the working principle, and nature of music signal. This was absolutely not a question! Didn't you read what I wrote?
Average current is much smaller then peak, hence you have to specify both on a PSU. This is what I'm explaining continuously! Please think a little! If you take a shower, and you get 10 C degree water for 5 minutes, and then you get 60 C degree for 5 min, will you be pleased?
There is no difference.
Yes, he seems to be skilled in building and selling amplifiers. So? This topic is not about skills! Fredos generalized his experience. His experience comes from a very limited sample, about 0,001 % of all ClassD amps in the world, without real theoretical basement, and despite of this he interpreted as a rule, while it is clearly not.
Also this experience is useless without giving some neccessary background information!
If a phisics teacher said 2+2=5, then would you agree with him, because he is skilled?
Inductive or capacitive! And because of the working principle, and nature of music signal. This was absolutely not a question! Didn't you read what I wrote?
Average current is much smaller then peak, hence you have to specify both on a PSU. This is what I'm explaining continuously! Please think a little! If you take a shower, and you get 10 C degree water for 5 minutes, and then you get 60 C degree for 5 min, will you be pleased?
There is no difference.
Pafi said:
Pafi....If you do stupidity like to put in phase 2 half bridge 50W amplifier on a 100W power supply, I think we should stay there and did not talk more....You should use your brain and be smart! If you really whant to play with peak, use my rule to determine bus cap for peak. That will work. Use a 100W PSU with a 100W load and YES, the 200uF will be enought at frequency over 50Khz..Now if you plan to draw peak of 1000 watts on that 100 watts power supply (if he can give it!) use a minimum of 2000uF on output to reach it, and this depend of how long is your peak. Dont forget to design the input bulk cap in the same way. And before you ask me, YES, use a good grade Cap with high ripple voltage at working frequency.
And no I did'nt assume that all class d work at same voltage, but just realize that the variation of current draw from an amp running at 40V (that could give nearly 200W into 4 ohms) is not double that an amp running at 60V and produce 400W into 4 ohms....Do you see now that my rule is nearly the reality? This is not scientific at all, just a basic approximation to save time when prototyping.
Finaly, for sure multi chanel will have influence on that ( a bit), but it still a question of total power draw....Peak or RMS this one? 😉
And yes I design and manufacture class d amplifier for over 10 years, my business is a building of over 2000 squared foot, I have 10 employes, and I still like to help other in prototyping because I begin too in my basement when I was younger...
i agree that more caps will be better for high peak power.
and the peak power is quite important .
for example:
a 400w version calss-d amp , will need a very high current output to drive the speaker,
normal calculate: 60v to 4ohm is 17A(single side),
the max peak power is 60v x17a is 1024w
but in our experience, 17A is not enough for peak power,
when we design the MCD-400,
if use 18A, the protection will action when a big dump coming
so we change the design to 22A,
dont forget the load of the speakers is active , not just 4ohm, sometimes lower.
after that , we get very good dynamic.
we can calculate at 65v,22A ,is about 1500W peak.
we are designing a smps is peak power output >1.5kw,
rms 500w, peak power 1.5kw, this will give us a very good smps for audio.
and i don't agree that just using caps for the peak power.
need a diffrent thinking for audio smps, not just care about the rms power, put more care about the max output peak current.
so the pwm design will be different with the RF smps.
another: the most important things is the noise floor of the smps. and the unbalanced +/-rails
rg
fumac
and the peak power is quite important .
for example:
a 400w version calss-d amp , will need a very high current output to drive the speaker,
normal calculate: 60v to 4ohm is 17A(single side),
the max peak power is 60v x17a is 1024w
but in our experience, 17A is not enough for peak power,
when we design the MCD-400,
if use 18A, the protection will action when a big dump coming
so we change the design to 22A,
dont forget the load of the speakers is active , not just 4ohm, sometimes lower.
after that , we get very good dynamic.
we can calculate at 65v,22A ,is about 1500W peak.
we are designing a smps is peak power output >1.5kw,
rms 500w, peak power 1.5kw, this will give us a very good smps for audio.
and i don't agree that just using caps for the peak power.
need a diffrent thinking for audio smps, not just care about the rms power, put more care about the max output peak current.
so the pwm design will be different with the RF smps.
another: the most important things is the noise floor of the smps. and the unbalanced +/-rails
rg
fumac
- Status
- Not open for further replies.