Maybe I missed it, but I do not remember ever coming across a discussion about going overboard on capacitance in a power supply and causing problems in the transformer because of excessive inrush current on turn on. Are there generalized guidelines, rules of thumb, etc., for determining how much capacitance a given transformer can handle based on amperage rating, VA rating, etc.?
Um... for FW rectification, 2000uF per amp of current draw is a general rule of thumb. Nets 1V of ripple at load, IIRC. Tube amps typically have much less capacitance per current because their voltage is much higher - 1V of ripple is 5% of a 20V SS amp's supply, but .25% of a 400V tube amp supply. No one's going to notice that, especially if it's PP and/or pentode output!
If you analyze how a power supply works, you'll see that power comes in large charging pulses at the peaks of the AC waveform. If C is small and ripple is large, the voltage where it starts charging is low. It stops charging just after the peak. You can clearly see that a larger C will make for shorter, much higher current pulses. This causes I^2R losses in the transformer, diodes and capacitor(s). Doubling the C will halve the ripple, as well as the time length of the charging pulses. As a result, peak current also doubles. You can calculate the RMS power dissipation and see that losses increase very quickly!
And that only covers static conditions - to say nothing of the first cycle of inrush (that you're after)! Suffice it to say, it ain't pretty...
Tim
If you analyze how a power supply works, you'll see that power comes in large charging pulses at the peaks of the AC waveform. If C is small and ripple is large, the voltage where it starts charging is low. It stops charging just after the peak. You can clearly see that a larger C will make for shorter, much higher current pulses. This causes I^2R losses in the transformer, diodes and capacitor(s). Doubling the C will halve the ripple, as well as the time length of the charging pulses. As a result, peak current also doubles. You can calculate the RMS power dissipation and see that losses increase very quickly!
And that only covers static conditions - to say nothing of the first cycle of inrush (that you're after)! Suffice it to say, it ain't pretty...
Tim
Sch3mat1c said:
Hmmm, now that sounds way too conservative to me. I believe I have seen many designs which use much more capacitance than that, even on small transformers for preamps. Even all the Aleph's use many times that, although that's not a good example because they use an inrush thermistor.
Hi Tim;
>>>...Um... for FW rectification, 2000uF per amp of current draw is a general rule of thumb...<<<
Dumb question over here - how would you count that on a split supply? In other words, if I'm using a 25VCT 2A PS trafo for a split supply off a bridge, would you add the capacitances in each leg (so the 2200uF per leg I'm using on my TDA2030A would pretty much conform to your rule) or would you treat the capacitance as "in series" (i.e. 1100uF from V+ to V-)?
Thanks in advance and all the best,
Morse
>>>...Um... for FW rectification, 2000uF per amp of current draw is a general rule of thumb...<<<
Dumb question over here - how would you count that on a split supply? In other words, if I'm using a 25VCT 2A PS trafo for a split supply off a bridge, would you add the capacitances in each leg (so the 2200uF per leg I'm using on my TDA2030A would pretty much conform to your rule) or would you treat the capacitance as "in series" (i.e. 1100uF from V+ to V-)?
Thanks in advance and all the best,
Morse
View each half (relative to ground) as a single supply. One amp flowing out the positive and into the negative is the same as one amp flowing from positive to ground plus one amp flowing from ground to negative.
The rule of thumb goes on current... that there's 2 in series doesn't change that. So you're probably talking about the caps effectively being in series, cutting their value in half... well it's a rule of thumb 😛
Tim
The rule of thumb goes on current... that there's 2 in series doesn't change that. So you're probably talking about the caps effectively being in series, cutting their value in half... well it's a rule of thumb 😛
Tim
Thanks and questions....
Thanks Tim;
If I'm reading you right, my 2A split power supply with 2200uF per rail should be about right. As you'd noted in your post, PS for valves have far lower capacitance (imagine trying to put 1000uF at 600V into an amp of reasonable size!) so I felt I was really going in for overkill when I used a pair of 2200uF caps (had 'em in my parts bins, so it was a case of "use what you have") in my TDA2030A. Imagine my surprise when I started reading up on solid state designs and discovered that it wasn't uncommon to see genuinely HUGE reservoir caps! 🙂
Now I just wish I could find a good source of cheap trafos....$10 each for 25VCT@2A isn't bad, but I'd like to find some with a bit more guts than that for multichannel efforts (say 10A current draw) on the cheap. Any ideas on cheap sources of iron?
Locally, our remaining electronics shop has an NOS Triad EI core 36VCT@4A for $40. Is that kind of price any good? If so, maybe I'll buy it for an LM4766 amp I'm planning.
Thanks for the info and thanks in advance for any suggestions on cheap iron.
All the best,
Morse
Thanks Tim;
If I'm reading you right, my 2A split power supply with 2200uF per rail should be about right. As you'd noted in your post, PS for valves have far lower capacitance (imagine trying to put 1000uF at 600V into an amp of reasonable size!) so I felt I was really going in for overkill when I used a pair of 2200uF caps (had 'em in my parts bins, so it was a case of "use what you have") in my TDA2030A. Imagine my surprise when I started reading up on solid state designs and discovered that it wasn't uncommon to see genuinely HUGE reservoir caps! 🙂
Now I just wish I could find a good source of cheap trafos....$10 each for 25VCT@2A isn't bad, but I'd like to find some with a bit more guts than that for multichannel efforts (say 10A current draw) on the cheap. Any ideas on cheap sources of iron?
Locally, our remaining electronics shop has an NOS Triad EI core 36VCT@4A for $40. Is that kind of price any good? If so, maybe I'll buy it for an LM4766 amp I'm planning.
Thanks for the info and thanks in advance for any suggestions on cheap iron.
All the best,
Morse
Re: Thanks and questions....
According to the guildlines above, if that power supply is for 2amps on the +ve and 2amps on the -ve, you should have 4400uF per rail for a total of 8800uf for the power supply.Morse said:
Hi AudioFreak;
Hmmmm, carefully re-reading Schematic's post makes me think you're right. Oh well - this PS may be a bit of a lightweight by ss standards, but it seems to work okay (biggest remaining problem with the amp is that I've got to replace a couple of bits of unshielded cable with the shielded stuff - and that's a task since it's buried in there....).
Thanks for the correction and all the best,
Morse
Hmmmm, carefully re-reading Schematic's post makes me think you're right. Oh well - this PS may be a bit of a lightweight by ss standards, but it seems to work okay (biggest remaining problem with the amp is that I've got to replace a couple of bits of unshielded cable with the shielded stuff - and that's a task since it's buried in there....).
Thanks for the correction and all the best,
Morse
- Status
- Not open for further replies.