transformer power for gainclone amp

Frarun

Member
2013-03-07 1:42 pm
Hi,
I have doubts regarding transformer power for Gainclone amp. I made a calculations based on
the datasheet for LM3875 page 20, but instead of assuming a output power and R load
assume the constant 2x19 AC voltage from transformer, and calculate a power required for
particular R load. This is what I get:

1. 30W / 8ohm -> 2x19V 105W
2. 40W / 6ohm -> 2x19V 140W
3. 60W / 4ohm -> 2x19V 210W

What if I use a 2x19V _140W_ transformer and connect 4ohm load? Will it broke a
transformer because the current will not be sufficient?

Regards.
 
The usual rule that works for virtually all power amplifiers is to use a transformer with a VA rating that is 1times to 2times the total maximum output power of the amplifier/s.

For 30W of output power use 60VA to 30VA.
for 60W use 120VA to 60VA.

However, due to the high regulation of very small transformers I recommend no smaller than 160VA.
That would support upto 160W of total power output.
This could be two 80W channels or three 50W channels or four 40W channels.
 

Frarun

Member
2013-03-07 1:42 pm
I should add, that in my calculations I totally skipped the 15% mentioned on page 21 of LM3875 datasheet.... and yes. My units was wrong :) Should be 'VA', not 'W'...

In deed this 150VA is my limit because of size of transformer (about 95mm). My problem is I need to choose a correct / optimal voltage for this transformer maximizing output power (or I mess something here?). Is there a simple rule to do that? Should I maximize this for specific load impedance?
 
The Voltage requirement is very different from the VA requirement.

The output from the amplifier is determined by the current and voltage delivered to the load, usually a reactive load.
The maximum voltage determines the maximum power that can be delivered to a resistive load.
That maximum output voltage is very dependant on the PSU voltage and in turn on the transformer voltage.


Look at the datasheet to find the maximum power you require.
Use that Pmax to determine the Vpk available to the load.
From Vpk and the losses through the amplifier determine the PSU voltage when the amplifier is delivering that maximum power.
Adjust the PSU voltage to the unloaded condition.
Determine the Vac of the transformer for that unloaded PSU voltage.
 

Frarun

Member
2013-03-07 1:42 pm
Thanks for your response. Looks like simple calculation.

I plan to use a 150VA transformer (more popular in my country than 160VA). This fits my requirement for about 95mm diameter for transformer (toroidal). As you said I could get something like 50% of that power. It will be 2x40W output power. The speaker impedance will be 6ohm...
Now the voltages.
Upk = sqrt (2x40x6ohm) = 21.9V

... losses through the amplifier ...
I am not sure what is that. Is this makes a heat? I will add here a +5V described in datasheet as 'dropout voltage' (still don't know if is the same):
Upk + 5V = 26.9V
and add 15% voltage (transformer regulation) to have an unloaded voltage:
U = 26.9V + 26.9V*15% = 30.9V DC

It gives a +-30,9 DC / 1.41 = +-22V AC transformer
Is that correct?
 

Frarun

Member
2013-03-07 1:42 pm
Do you mean 'Output Power vs Supply Voltage' chart?

For about 31V supply voltage It points to 55W for 6ohm. Is that mean I can get all below this 55W and only limit is my PSU?

For RL =4ohm, there is a power drop for 31V supply voltage on the chart :/ I will not be able to use such resistive load?
 

Frarun

Member
2013-03-07 1:42 pm
The 27V is below a standby 31V (15% margin). Hope this is a safe assumption.

What is really interesting, the impedance of speakers may vary from freq. So even If I connect nominal 6ohm speakers it may drop to like 4ohm(?). I fill this affects the PSU, and could generate some bad conditions for the design.

I saw lots of gainclones where the very big transformers are used (like 250VA). I was not sure it is really required. This is probably for some better peak output power...

Anyway. Thank for any comments/info.
 
.........I saw lots of gainclones where the very big transformers are used (like 250VA). I was not sure it is really required. This is probably for some better peak output power....
No.
A bigger transformer is good for continuous output power.

A small transformer shows up badly for continuous output power, but conversely does rather well when measuring peak or transient maximum power.

Similarly for smoothing capacitance. Big capacitance does well for continuous maximum power. Small capacitance does well for transient maximum power.

The 1times to 2times total maximum power is a very good guide. It only models poorly when very small transformers are adopted, i.e. <=100VA
 
Power supply design spreadsheet

I created an Excel spreadsheet that can be used to do power supply design calculations. It's well suited to answer your questions about how large or small of a transformer to use. Also, it tells you how large the capacitors need to be for a certain level of ripple on the rails. The caps smooth out the pulsed DC coming out of the rectifier but can't eliminate the pulsing totally, and this is called ripple. Usually you want to have only a couple of volts of ripple, because the amplifier can only drive the load close to the lowest voltage of the DC+ripple so more ripple means less output power capability.

Here's the link:
Power Supply Dimensioning Spreadsheet by charlielaub

Feel free to ask questions if you get stuck.

-Charlie
 

Frarun

Member
2013-03-07 1:42 pm
Thanks. xls sheet explains so mach details... I will have a much more fun with this PSU design than I though! :)

I focused on the same parameters: 6ohms/2chann/Vod=5V/50hz
Now I play with VA and Vac (unloaded)

150VA / 22V gives me -> 37W/ch (0dB crest factor) and
NOTE: Power output is limited by the rail voltage (±27.1VDC at onset of clipping)

150VA / 24V gives me -> 41W/ch (0dB crest factor) and
NOTE: Power output is limited by the transformer VA rating

is it better to limit the power via transformer power, or the voltage? I expected the answer is a voltage.

Wait a moment....... It was set a 1V ripple which requires a 44mF campacitors... and this is per rail! I planned to put 2*6800uF per rail. This gives a 3V ripple, which limits an output power to 31W for 22Vac and 38W for 24Vac.

This two volts on AC side makes a big difference.
 
The 27V is below a standby 31V (15% margin). Hope this is a safe assumption.

What is really interesting, the impedance of speakers may vary from freq. So even If I connect nominal 6ohm speakers it may drop to like 4ohm(?). I fill this affects the PSU, and could generate some bad conditions for the design.

I saw lots of gainclones where the very big transformers are used (like 250VA). I was not sure it is really required. This is probably for some better peak output power...

Anyway. Thank for any comments/info.

no matter how big your power traffo is, there will always be a voltage drop....nothing to lose sleep about, this is the nature of the beast....;)
 
Thanks. xls sheet explains so mach details... I will have a much more fun with this PSU design than I though! :)

I focused on the same parameters: 6ohms/2chann/Vod=5V/50hz
Now I play with VA and Vac (unloaded)

150VA / 22V gives me -> 37W/ch (0dB crest factor) and
NOTE: Power output is limited by the rail voltage (±27.1VDC at onset of clipping)

150VA / 24V gives me -> 41W/ch (0dB crest factor) and
NOTE: Power output is limited by the transformer VA rating

is it better to limit the power via transformer power, or the voltage? I expected the answer is a voltage.

Wait a moment....... It was set a 1V ripple which requires a 44mF campacitors... and this is per rail! I planned to put 2*6800uF per rail. This gives a 3V ripple, which limits an output power to 31W for 22Vac and 38W for 24Vac.

This two volts on AC side makes a big difference.

If you will be reproducing music with the chip amps, it's a good idea to enter a crest factor that describes the peak to average power ratio (that's what the crest factor is). A conservative value would probably be 10dB, and for music that is not compressed and often recorded at lower levels (classical?) 20dB can be used. If you listen to overcompressed pop music all day long at top volume, use 6dB, but you don't need to set this value to 0dB. The value you enter for the crest factor is used to calculate the average power that has to be dissipated by the transformer considering several factors. So I would guess that, for the case in which you find the power rating is limited by the transformer VA rating, if you used a 10dB crest factor the transformer would no longer be limiting.

Anyway, glad you like the spreadsheet. I use it all the time to get estimates of what I need to put together for a power supply, or to figure out now much power my chip amp will be capable of when using a transformer that I have in hand.

-Charlie
 
no matter how big your power traffo is, there will always be a voltage drop....nothing to lose sleep about, this is the nature of the beast....;)
and that voltage drop depends on the discharge of the smoothing capacitance. That smoothing capacitance is where the amplifier gets the current that is demanded by the speaker.
If the demand is a fast transient, the current is supplied by the decoupling not the transformer.

The transformer's job is to recharge the capacitance during the short time when the mains AC waveform is such that a pulse of charge can flow.
 

Frarun

Member
2013-03-07 1:42 pm
I have a brand new 150VA transformer for my GC... I increased my initial capacitance from 2x6800uF to 2x10000uF (per rail). It was hard to find some low profile (40mm height) capacitors, so finally I have a JAMICON. I don't really know if they are good, but for start I am happy :)

This gives me such values from excell sheet:
tran.: 150VA 2x24VAC, load res.: 6ohm, Vod: 5V, 50Hz
2.3V ripple voltage (-> about 20000uF per rail)

Calculated power:
40W [creast factor 0dB] (limited by the transformer VA)
41W [creast factor 10dB] (imited by the rail voltage)

Regarding capacitance. After some more reading about GC I see there ware a versions with only 1000uF per rail (with 200VA transformer), so this is all subjective thing ...


BTW. Do you guys recommend the soft-start machine for my small 150VA transformer and 4x10mF PSU?