OK, now I'm a little confused. Does the high heat generation come from the voltage output of the transformer, regardless of load/volume?
I don't want to drastically underpower my one-transformer GC, but on the other hand, I don't want to buy so much rail voltage that I generate excess heat, either.
Which would be better (for a two-channel, one-trannie system), given that I would only be listening at moderate volumes on 89db, 8-ohm speakers?
18/18/250VA
or
25/25/500VA?
I am using an LM3875TF, with a buttload of heatsink for the chips.
I don't want to drastically underpower my one-transformer GC, but on the other hand, I don't want to buy so much rail voltage that I generate excess heat, either.
Which would be better (for a two-channel, one-trannie system), given that I would only be listening at moderate volumes on 89db, 8-ohm speakers?
18/18/250VA
or
25/25/500VA?
I am using an LM3875TF, with a buttload of heatsink for the chips.
Hi,
Those are two separate issues.
If your cicrcuit needs a 18/18/250VA xformer to function you may not have to burn off xtra voltage but the xformer may get hot.
So if we keep the secondary voltages at 18/18 but we up the VA rating, the xformer will become more effecient and load/line regulation will improve and the entire amp won't get as hot either.
Hope it helps,
Those are two separate issues.
If your cicrcuit needs a 18/18/250VA xformer to function you may not have to burn off xtra voltage but the xformer may get hot.
So if we keep the secondary voltages at 18/18 but we up the VA rating, the xformer will become more effecient and load/line regulation will improve and the entire amp won't get as hot either.
Hope it helps,
Hmm. The choices I had were the closest two, sticking with parts express' offerings. I am sure the 22/22/400VA would do it. I was just wondering of the 18/18/250 would give the needed rail voltage.
I thought I heard someone say that they were running 3875's at that voltage with no problems. This may be a dumb question, but shouldn't the VA req's decrease proportionately with the lower secondaries' voltage?
I.e., if I am not supplying as much voltage, my amp isn't going to draw as much current (and thus power), right?
If I am not going to demand window rattling out of the amp, will the 18/18/250 drive efficient speakers to modest volume levels in a 3875 GC?
I thought I heard someone say that they were running 3875's at that voltage with no problems. This may be a dumb question, but shouldn't the VA req's decrease proportionately with the lower secondaries' voltage?
I.e., if I am not supplying as much voltage, my amp isn't going to draw as much current (and thus power), right?
If I am not going to demand window rattling out of the amp, will the 18/18/250 drive efficient speakers to modest volume levels in a 3875 GC?
18-18 rails should be ok as long as you don't need very much power... just look at the LM3875 datasheet, there's a graph that shows power vs. supply rail for 4 ohms and 8 ohms. 250VA is more than enough power to satisfy that graph @18V supply.
you are sort of right that VA can go down with secondary voltage, because ultimately the chip is current-limited, not voltage-limited. using a 500VA 18-18 transformer probably won't get you any more power than a 250VA, tho there may be some sonic benefits to the former.
p.s. oops, i think i actually meant that at low supply voltages into average loads, the chip is VOLTAGE limited, not current-limited... so VA can go down with secondary voltage. this wouldn't be the case if you are using low impedance speakers however, then the chip would be current-limited, and you may indeed still be able to use the extra power at lower supply voltages, i think.....
you are sort of right that VA can go down with secondary voltage, because ultimately the chip is current-limited, not voltage-limited. using a 500VA 18-18 transformer probably won't get you any more power than a 250VA, tho there may be some sonic benefits to the former.
p.s. oops, i think i actually meant that at low supply voltages into average loads, the chip is VOLTAGE limited, not current-limited... so VA can go down with secondary voltage. this wouldn't be the case if you are using low impedance speakers however, then the chip would be current-limited, and you may indeed still be able to use the extra power at lower supply voltages, i think.....
Yes, you could settle for lower VA rating att lower voltages, as the required current will not increase at lower operating voltages (rather decrease, if anything).
Grahamn & analog:
The regulation depends not only on the wire resistance, but also the construction (E-core, toroid or whichever). Ultimately, they have different properties, which reflects on their performance.
Toroids have very low loss (less heat at idle) while E or C cores often have better coupling, which lowers regulation. This is because the output impedance (related to the regulation) is partly depending on the input impedance (primary side), multiplied by a coupling factor and (if I remember correctly, the in/out ratio squared).
Something like that, anyway... it's been too long for details...
I hope this makes at least SOME sense, since english isn't native to mee, and it's a tricky subject.
Jennice
Grahamn & analog:
The regulation depends not only on the wire resistance, but also the construction (E-core, toroid or whichever). Ultimately, they have different properties, which reflects on their performance.
Toroids have very low loss (less heat at idle) while E or C cores often have better coupling, which lowers regulation. This is because the output impedance (related to the regulation) is partly depending on the input impedance (primary side), multiplied by a coupling factor and (if I remember correctly, the in/out ratio squared).
Something like that, anyway... it's been too long for details...
I hope this makes at least SOME sense, since english isn't native to mee, and it's a tricky subject.
Jennice
Peter - Thanks for the graph. I am assuming that is rail voltage, not seconday voltage? You mentioned that the supply voltage is going to be a little higher than the secondary voltage for a couple of reasons.
I am assuming the 18/18/250 would get me into at least the low-to-mid 20's on a rail voltage?
If so, that would give me plenty of power into 8 ohms, according to your graph.
GnD
I am assuming the 18/18/250 would get me into at least the low-to-mid 20's on a rail voltage?
If so, that would give me plenty of power into 8 ohms, according to your graph.
GnD
ok. some basics.
with capacitor filtration and no load, your DC rail voltage will be the peak-to-peak voltage of the transformer secondary, i.e. RMS * sqrt(2), or about 1.4. so, 18V secondary gets you about 25.4VDC on the rails after filtration. however there are two factors that will make this drop in practice. one is that with load, the capacitors can no longer maintain peak voltage throughout the cycle, so as load increases this will drop back towards the RMS voltage. second is that you must figure in the voltage drop across the rectification diodes. for the HEXFRED types most recommend, this is somewhere between 1V to 2V. let's say 1.5V for now. another complication is the regulation factor of the xformer, which increases the secondary voltage at no load vs. the voltage @rated load. let's say 6%.
so, under no load, your DC rails will be ((18*1.06)-1.5)*1.4 = 24.6VDC. under a heavy load, this drops toward (18-1.5) = 16.5VDC. in actual use with the gainclone circuit, i'd be surprised if it gets quite this low; this is pure conjecture, but with your transformer i'd say the rail voltage will hover around 20V. try using this as your operating point for max power calculations.
with capacitor filtration and no load, your DC rail voltage will be the peak-to-peak voltage of the transformer secondary, i.e. RMS * sqrt(2), or about 1.4. so, 18V secondary gets you about 25.4VDC on the rails after filtration. however there are two factors that will make this drop in practice. one is that with load, the capacitors can no longer maintain peak voltage throughout the cycle, so as load increases this will drop back towards the RMS voltage. second is that you must figure in the voltage drop across the rectification diodes. for the HEXFRED types most recommend, this is somewhere between 1V to 2V. let's say 1.5V for now. another complication is the regulation factor of the xformer, which increases the secondary voltage at no load vs. the voltage @rated load. let's say 6%.
so, under no load, your DC rails will be ((18*1.06)-1.5)*1.4 = 24.6VDC. under a heavy load, this drops toward (18-1.5) = 16.5VDC. in actual use with the gainclone circuit, i'd be surprised if it gets quite this low; this is pure conjecture, but with your transformer i'd say the rail voltage will hover around 20V. try using this as your operating point for max power calculations.
GrahamnDodder
Use the charts that Peter posted and compare them to your expected speaker load. In my case, I have low impedence drivers so I don't bother with any transformer and run it straight from batteries. For those still curious about transformers you should check this out. The guy who started it never concluded it but I think that he was on the right track.
Use the charts that Peter posted and compare them to your expected speaker load. In my case, I have low impedence drivers so I don't bother with any transformer and run it straight from batteries. For those still curious about transformers you should check this out. The guy who started it never concluded it but I think that he was on the right track.
Dorkus
Your numbers are basically right but if we are talking about the LM family of inverted gainclones, the load is nearly insignificant once you have chosen the optimum rails as per the chart posted earlier by Peter. The reason is that the chips run full class A internally at a miniscule current and behind a hefty -120dB ripple rejection wall which work well to absolve the problems you are aiming to address with your suggestions. From what I have read, energy delivery is not this amps weakness but RF hash and ground plane phase issues are. The fact is that the chip seems to work better with a cap that is always partially drained. Contrary to earlier comments, 24V rails will work sufficiently well with 2 80VA transformers and would probably work even better if you put them in series in order to reduce the flux density. Lower rail voltage and lower impedence loads will require bigger transformers but 400VA is about as big as you should go IMHO. This is what I've been able to understand from my readings on this and other forums but I leave it to the more experienced members to correct me.
Your numbers are basically right but if we are talking about the LM family of inverted gainclones, the load is nearly insignificant once you have chosen the optimum rails as per the chart posted earlier by Peter. The reason is that the chips run full class A internally at a miniscule current and behind a hefty -120dB ripple rejection wall which work well to absolve the problems you are aiming to address with your suggestions. From what I have read, energy delivery is not this amps weakness but RF hash and ground plane phase issues are. The fact is that the chip seems to work better with a cap that is always partially drained. Contrary to earlier comments, 24V rails will work sufficiently well with 2 80VA transformers and would probably work even better if you put them in series in order to reduce the flux density. Lower rail voltage and lower impedence loads will require bigger transformers but 400VA is about as big as you should go IMHO. This is what I've been able to understand from my readings on this and other forums but I leave it to the more experienced members to correct me.
Another noob
Ok, According to Peter's graph. I will need around 26-28 VDC supply for a 35W 8Ohm Speaker. After that I decided to find a Transformer which Primary around 115 VAC, Secondary around 19-20 VAC. Am I right? When I connect the supply to the bridge, the voltage rail will be around 19-20VAC * 1.414 = 26.866-28.28VDC, right?
Second question, Is the Toroidal you guys talking about (330VA
25/25) Had dual Primary and Secondary? If I decided to buy a 550VA 25/25, and I am in US. Secondary will be like [120/(500/2)]*25= 12VAC, right? Then 12*1.414 withe bridge = 16.968.
Third question, don't we have to calculate the current too, then determine how many Watt will the PS have? P=VI? How can I determine the Current no?
Forth question, If I supply 240VAC to a 115VAC Transfromer, will the transformer burn?
Ok, According to Peter's graph. I will need around 26-28 VDC supply for a 35W 8Ohm Speaker. After that I decided to find a Transformer which Primary around 115 VAC, Secondary around 19-20 VAC. Am I right? When I connect the supply to the bridge, the voltage rail will be around 19-20VAC * 1.414 = 26.866-28.28VDC, right?
Second question, Is the Toroidal you guys talking about (330VA
25/25) Had dual Primary and Secondary? If I decided to buy a 550VA 25/25, and I am in US. Secondary will be like [120/(500/2)]*25= 12VAC, right? Then 12*1.414 withe bridge = 16.968.
Third question, don't we have to calculate the current too, then determine how many Watt will the PS have? P=VI? How can I determine the Current no?
Forth question, If I supply 240VAC to a 115VAC Transfromer, will the transformer burn?
Hi,
What do you think? Geeezzz. of course it will.
When picking a powerxfomer you need to work backwards, meaning that you need to consider what the cicruit consumes statically and dynamically.
Once you know that, you can add some margin to the VA rating, add some more if load/line regulation is a bonus to you and all else is up to quality of that xformer.
By quality I don't mean wide bandwidth, we leave that to OPTs. In fact most of the time limited BW xformers are preferable unless you want to filter out RF grudge afterwards...
You'd want that xformer to be quiet, reasonably cool ( hardly over room temp) and never run out of steam with the load hooked up...
So, how hard can it be ?
A good amp always starts with a good PSU and a good transformer should definetely be part of that.
Cheers,
What do you think? Geeezzz. of course it will.
When picking a powerxfomer you need to work backwards, meaning that you need to consider what the cicruit consumes statically and dynamically.
Once you know that, you can add some margin to the VA rating, add some more if load/line regulation is a bonus to you and all else is up to quality of that xformer.
By quality I don't mean wide bandwidth, we leave that to OPTs. In fact most of the time limited BW xformers are preferable unless you want to filter out RF grudge afterwards...
You'd want that xformer to be quiet, reasonably cool ( hardly over room temp) and never run out of steam with the load hooked up...
So, how hard can it be ?
A good amp always starts with a good PSU and a good transformer should definetely be part of that.
Cheers,
S.C., i don't quite get the math, what is [120/(500/2)]*25? did you mean [120/220]*25 if you were to use the entire 220V primary for 120V?
actually, this brings up an interesting point. i have noticed a lot of larger xformers come only with higher-voltage secondaries, e.g. the 500VA Avel at part expess comes only as low as 25-0-25. however, notice that they have 115 + 115 primaries, which would normally be wired in parallel for US operation (series for 220V). you can wire these in series and run them at 115VAC, and end up halving the secondaries... so 25-25 becomes 12.5-12.5 obviously. not only is this convenient for obtaining lower voltage, but i believe Phred indicated performance is improved this way as well (less distortion, less loss, more linear operation). then you could use the 625VA 45-45 part to obtain 22.5-22.5, right around what Peter's recommends. of course 625VA seems a little excessive, but hey it can't hurt right? as long as size isn't an issue i see no reason not to go with it, it's only $18 more than the 330VA part anyway.
this is all, of course, for us 120VAC peeps here in the states... apologies for any provincialism.
actually, this brings up an interesting point. i have noticed a lot of larger xformers come only with higher-voltage secondaries, e.g. the 500VA Avel at part expess comes only as low as 25-0-25. however, notice that they have 115 + 115 primaries, which would normally be wired in parallel for US operation (series for 220V). you can wire these in series and run them at 115VAC, and end up halving the secondaries... so 25-25 becomes 12.5-12.5 obviously. not only is this convenient for obtaining lower voltage, but i believe Phred indicated performance is improved this way as well (less distortion, less loss, more linear operation). then you could use the 625VA 45-45 part to obtain 22.5-22.5, right around what Peter's recommends. of course 625VA seems a little excessive, but hey it can't hurt right? as long as size isn't an issue i see no reason not to go with it, it's only $18 more than the 330VA part anyway.
this is all, of course, for us 120VAC peeps here in the states... apologies for any provincialism.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.