What power transformer to buy

[Alan0354]

Hi

My amp is going to be about 70Wrms per channel, 2 channels amp. I need to find a power transformer. My +B=400V. From loadline calculation, peak current is 600mA per channel that is 1.2A peak for both. This will give RMS current of 1.2 X 0.707 = 850mA just for the power tubes at maximum power output on both side.

I don't need 5V as I don't use tube rectifier. The 6.3V has to be able to provide 8A. It wound be nice to have like a 30V or so tap to get the -ve bias voltage for grid 1.

I know I need secondary voltage to be 400V X 0.707 = 283V or 285V with rectifiers. My questions are:

1) do I get 285-0-285 CT or just 285 non center tap and use bridge rectifier. Seems like 285-0-285 has the advantage of ripple at 120Hz rather than 60Hz, that will ease the requirement of the filter caps.

2) We all know we don't run maximum power all the time. I don't think we need 850mA max current. The question is what is the standard calculation given the maximum is 0.85A? 500mA? or 400mA?

I want to get some suggestion of what brand I can look at. This is not OPT, so I go as cheap as possible. Chinese transformer is OK with me. I am born cheap. I don't want to spend a penny more than I have to on the power transformer.

Thanks

 

[thorpej]

AnTek is excellent. This one has the juice you require: http://www.antekinc.com/as-4t320-400va-320v-transformer/

 

[whoaru99]

I'm using AnTeks in my forthcoming KT88 PSE monos. One 2T350 (200va - 350V) per amp. About 412V net plate is expected considering the roughly 47V cathode bias, drop across OPT, etc.

 

[whoaru99]

1) do I get 285-0-285 CT or just 285 non center tap and use bridge rectifier. Seems like 285-0-285 has the advantage of ripple at 120Hz rather than 60Hz, that will ease the requirement of the filter caps.

Non-CT with FWB gives ripple at 120Hz too.

 

[Alan0354]

Non-CT with FWB gives ripple at 120Hz too.

You're right. So I don't need CT.

 

[Alan0354]

BTW, I am not planning to use cathode bias. I want the bias to be more predictable not charging the cathode up upon large signal. I plan to use grid bias.

 

[whoaru99]

BTW, I am not planning to use cathode bias. I want the bias to be more predictable not charging the cathode up upon large signal. I plan to use grid bias.

I understand, but you do still need to consider drop across any filter chokes and the OPT. With fairly low resistances it won't be too much but it will be.

Also, for grid bias supply a transformer like 12.6V:120 run backwards from the filament winding is one way to get 50V, give or take. It's really small current so the transformer can be is small. I used one taken from a small wall wart.

Also, with a toroid it's easy to add another winding right over the top as long as the volts per turns ratio doesn't make the hand winding too much pain in the butt.

Here is example I did to add two additional windings to get 16V CT. Using the 12.6V CT original of the transformer plus the additional windings in series gives the ~16V CT I wanted.

 

[Alan0354]

I am not planning to use a filter inductor. This is for the two power tube only, I am going to use regulated supply ( tap off from this) to power the screen grid and the input stages. The OPT primary is only about 100ohm, so the drop at 50mA idle is only 5V.

I found hammond 370MX 550V CT but it's $175.

It would be nice if I can take out some turns of the Antek transformer. So close.

 

[whoaru99]

Yeah, that's why the recommendations for AnTek. The price will be roughly 1/3 for the transformer but then a little more for some sort of cover.

 

[Alan0354]

I am working on the load line. 320V AC translates to +B=450VDC after rectification. I drew the loadline, I have to use 5K primary. I loss a little of the class A region. But looks like it's OK. I do gain a lot of power. Now the amp is 84W for 2 pairs of 6L6GC, this translates to about 67W in real world!!!! And 12.8W of class A region in the real world. Heat produce is 72W per channel due to the bias current. Total heat produce is 144W( bias) + 45.5W ( heater) = 189W. It's going to be hot, but not the end of the world.

The Antek might be a good one, thanks both you guys. Here is the load line with 5K primary per pair at 450V.



Nice thing is it has dual 320V secondary completely separated, I can totally separate the two channels by using separate bridge rectifier for each secondary to power the two channels separately.

 

[Alan0354]

AnTek is excellent. This one has the juice you require: http://www.antekinc.com/as-4t320-400va-320v-transformer/

Hi Jason

How have you been? I am down the wire in choosing components for my amp. I think I decided on 2 pair of 6L6GC, 2.5K primary OPT, I am planning of using the 100W 2.5K Edcor of the prototype amp on wood board and go to either Lundahl or Sowter in the real amp. Now the Antek sound pretty nice, So I have a lot of components picked out.

Alan

 

[maxhifi]

Hi Alan,

You can use either a bridge, or two diodes.

I see you calculated voltage will be 1.414xVsec, for a bridge rectifier. Usually this ends up being a little less, especially if you don't have a huge filter capacitor (and wirh push pull, you don't really need it)

I strongly suggest to simulate the power supply, using the correct values of resistance for the power transformer primary and secondary, and figure out the voltage droo across the output transformer primary too. It will also let you figure out the correct capacitors for the target value of ripple, and not go overboard. Less chance of disappointment this way. The simulation will also tell you exactly how much current you need.

This will be more accurate than just using 1.414 times Vsec, which never really turns out right for me.

Antek does look like one of the best deals. Another way to go if you're cheap is use a lower voltage transformer and a voltage doubler supply.

The load line looks really good, I checked it and the calculation. Another good looking line is 450V B+ and 400V,G2 with a 5600 ohm load. Makes a bit more power, but less class A region.

-Max

 

[Alan0354]

Hi Alan,

You can use either a bridge, or two diodes.

I see you calculated voltage will be 1.414xVsec, for a bridge rectifier. Usually this ends up being a little less, especially if you don't have a huge filter capacitor (and wirh push pull, you don't really need it)

I strongly suggest to simulate the power supply, using the correct values of resistance for the power transformer primary and secondary, and figure out the voltage droo across the output transformer primary too. It will also let you figure out the correct capacitors for the target value of ripple, and not go overboard. Less chance of disappointment this way. The simulation will also tell you exactly how much current you need.

This will be more accurate than just using 1.414 times Vsec, which never really turns out right for me.

Antek does look like one of the best deals. Another way to go if you're cheap is use a lower voltage transformer and a voltage doubler supply.

The load line looks really good, I checked it and the calculation. Another good looking line is 450V B+ and 400V,G2 with a 5600 ohm load. Makes a bit more power, but less class A region.

-Max

I think it should be close enough. I use two Antek and they are like 2V higher than calculation for 40V. So it's only 5% difference. It's really not that sensitive. The speaker impedance is all over the place, the load line is only giving the starting point. 5% variation is like 5% variation of the primary impedance or the variation of the tubes. I think it's going to be fine.

I am going to use excess filter to lower the ripply. I believe the more caps the better. Too bad this is going to be 450V, I need to use at least 500V caps.

I am planning to use regulated screen and for input stage, I think that is more important.

 

[Alan0354]

..........

 

[maxhifi]

OK here is an example, I used Antek AS-4T320 - 400VA 320V, one per channel. I used 400mA as the maximum load (at g1=0, I=280mA, 280*0.7071*2 sets of tubes= approx. 400mA),

I calculated 18 ohms winding resistance from Antek datasheet. You can see that the output voltage isn't close to 1.414*Vsec RMS, but reality is this voltage is almost ideal for your amplifier.



Under no load (except bias current), the voltage rises, as below:

 

[Alan0354]

I have another idea: http://www.antekinc.com/as-3t275-300va-275v-transformer/
I use two, one for each channel. This will give 275V X 1.414 = 390V.

Or two of this 200VA 300V: http://www.antekinc.com/as-3t275-300va-275v-transformer/

This gives 425V .

 

[PakProtector]

That both tubes are conducting simultaneously is not the only requirement for Class A operation.

You can run the standard, bean-counter power trans and filter, and it will still pass signal. There is no reason to do this for DIY builds unless you choose to bean counter yourself. Powerful amps are heavy.

There is also the option to use a single L-C filter, in which case the output voltage is about .9x the rms input. There is no peaky current delivery to charge the caps, and the peaky current delivery does not infect the cathode AC. You can then put a large capacitor( or many in || ) and have a B+ that could not care less what the AB bias is requiring of it...as long as your power Iron can keep up.
cheers,
Douglas

 

[kjello]

I have another idea: http://www.antekinc.com/as-3t275-300va-275v-transformer/
I use two, one for each channel. This will give 275V X 1.414 = 390V.

Or two of this 200VA 300V: http://www.antekinc.com/as-3t275-300va-275v-transformer/

This gives 425V .

My Nissan Leaf has a 40 kWh 400 Volt battery

 

[PakProtector]

My Nissan Leaf has a 40 kWh 400 Volt battery

I can do 40kW-hr with 18650's... and have. Batteries are not to be despised.
cheers,
Douglas

 

[Alan0354]

I think the Antek 400VA 320V-320V is a good choice. I am going to put the rectifier circuit and the voltage regulator circuit for screen and input stage on pcb, one pcb per channel. I am going to put the provision of having 3 5W 3.9V zener diode in series with the output of the rectifier before the big filter cap. Depends on the voltage of the transformer, I want to adjust the working +B=425V by putting zener if needed, I don't have to have all 3, I can put jumper if I get 425V.

With 425V, I can go back to 2.2K primary for 2 pairs of 6L6GC and have the load line cut the Vg1=0 curve clearly on the right side of the knee to guaranty 3rd harmonic is low. I get about 30W real output power per pair or 60W total per channel. I get the first 17.6W of real class A power.