Just trying to understand the concepts behind putting the first cap in this configuration and in partucular why we can only use a small value with tube rectifiers and much larger values with SS recitifiers.
eg.
with my 6X5 I use only a 0.47uF
but in a SS we can use maybe a 47uF or more.
What effect does using a 0.47uf/1uF first cap in the SS set up have instead of 47uF?
eg.
with my 6X5 I use only a 0.47uF
but in a SS we can use maybe a 47uF or more.
What effect does using a 0.47uf/1uF first cap in the SS set up have instead of 47uF?
Brit01 said:
Hi,
Are you talking of the first cap after the rectifier (tube or ss)? It seems your values are very low. Common values are 50-100uF for tubes and 10,000-20,000uF for ss.
Anyway, there are a few issues. First of all, the cap holds the energy to supply the amp during the interval that the secondary transformer voltage is lower than the cap voltage and the rectifier doesn't conduct. You may be surprised that that is some 80-90% of the time. So during that time, the cap is the only energy source. Since the energy is a function of voltage and capacity, it follows that lower voltage (ss) tends to require higher capacity, especially since ss amps are usually more powerfull (require more current) than tube.
So, if your cap is too small, it will discharge rapidly between top-ups and that increases the ripple.
OTOH, too large a cap can really stress the rectifier: when the diodes again start to conduct when the secondary transformer voltage rises above the cap voltage, a huge cap looks like a bottomless pit and takes a LOT of current to top up. In ss it is not uncommon to see current spikes into the cap of 10A and more. SS rectifiers can take some abuse, but for tubes the cap value should not be too large.
Does this help?
jd
I'm referring to the first cap after the bridge before any resistors.
In my 6x5 psu I've got
0.47uf
1k
100uf
1k
200uf
But maybe in a SS I could use:
47uf
1k
470uf
1k
680uf
Now what would be the result of using 1uf or 0.47uf as the first cap in a SS bridge? No point at all?? I can use a high quality PIO or something as in my 6x5 psu.
In my 6x5 psu I've got
0.47uf
1k
100uf
1k
200uf
But maybe in a SS I could use:
47uf
1k
470uf
1k
680uf
Now what would be the result of using 1uf or 0.47uf as the first cap in a SS bridge? No point at all?? I can use a high quality PIO or something as in my 6x5 psu.
It never hurts to follow the original manufacturers' spec for cap input filter - some are very low.
A quick search at
http://tdsl.duncanamps.com/tubesearch.php
will reveal that the 6X5 calls for max of 4mF
http://www.pmillett.com/tubedata/HB-3/Receiving_Tubes_Part_2/6X5_6X5-GT_G.PDF
A quick search at
http://tdsl.duncanamps.com/tubesearch.php
will reveal that the 6X5 calls for max of 4mF
http://www.pmillett.com/tubedata/HB-3/Receiving_Tubes_Part_2/6X5_6X5-GT_G.PDF
Brit01 said:
One of the specs on a rectifier tube is the max size of the first cap in the filter. The reason for the limit on cap size is "in rush current". When the AC mains power is flipped on the un-charged cap looks to the rectifier just like a dead short and a very large current goes through the tube. The only limit is the resistance in the transformer. The tube rectifiers simply can not handle this kind of current for any length of time, or they melt. SS Diodes can handle a longer in-rush period but they have specs for this too. The in-rush current is limited by what the transformer can supply but the in-rush period is determined by the size of the capacitor. The rectifier is between the two. All three must be "in balance" or you get either smoke or you have paid to much for one of the parts.
If you want a larger input cap but are limited by the in-rush energy limit of the rectifier there is still a way. There are in-rush limiters. These look like 200 ohm resisters when powered off but. Put one in series with the reccitifier as the device heats up the resistance drops. The 200R limits current in the first few seconds. After that they act like 2R. Every switch mode computer power supply uses these and they cost about $2 each. The correct term to Google is Current Limiting Thermistor.
You see different designs over the years because 50 years ago capacitors where very expensive and chokes where cheaper. Today we have cheap caps and chokes cost more. Today with SS diodes we cn have larger input caps so we can lower the values of the others. It's all a cost trade.
A lot of what you see in amp designs is an engineer's attempt to be cost effective.
Brit01 said:
Yes, you can use low values with a solid state bridge. But the question is; what are you trying to accomplish? What is your transformer delivering? What voltage output do you want? What ripple level do you want? How much dampning do you want?
All these things you can predict if you model the supply in PSUD. It's free from the Duncan Amps site. It's easy. It's accurate.
Sheldon
Thks Shedlon.
Yes I'm a regular user of this. It's great software.
I was trying to use a 1uF cap here just because of financial restraints. Got a load of WIMAS 1uF and a box of 470uf and 680 uf caps.
the 47uF caps are expensive here. But I'm getting 4 new ones in a few weeks from someone travelling.
But in the meantime I simulated using a 1uF as the first cap and everything looked fine. Very low ripple with the big caps after.
275V transformer secondary.
Trying to get the max possible out of this to run 150-200mA.
Yes, definitely look at the ripple current through the rectifier in PSUD. PSUD will warn you if you hit the absolute maximum current rating of the tube, but not the steady-state peak current. You need to make sure that you don't hit this. The result of exceeding the latter is usually shortened tube life and nothing spectacular like tube arcing.
So if you look at the spec of a rectifier tube, they often do the math for you and tell you what the maximum cap size is for a capacitor-input filter (where the first letter is a C, like CLC). For a choke-input filter (LCLC, etc), there is no practical limit since the first choke eats most of the ripple current assuming it is sized appropriately.
So if you look at the spec of a rectifier tube, they often do the math for you and tell you what the maximum cap size is for a capacitor-input filter (where the first letter is a C, like CLC). For a choke-input filter (LCLC, etc), there is no practical limit since the first choke eats most of the ripple current assuming it is sized appropriately.
What is really going on here ...
If you have an LC filter and your L is large, you have what is called a choke input filter. The output voltage of that filter will be lower than the if you have direct capacitive loading of a rectifier (i.e C is the first thing you see). Note that the loading of the transformer will be lower and you can typically extract significantly more energy out of it than with C input even though the voltage is lower (Cu losses are proportional to the square of the current, so pulsing the current is not a good idea and one of the reasons why people like to use ridiculously large transformers for audio). Ripple with L input rectification will also tend to be more or less sinusoidal if L is large enough. Think through why this may be so - hint the inductor lengthens the charge pulses so the energy can flow over a longer period of time.
If the input C is very low, all you do pretty much is take som spiking off the output of the rectifier which is caused by the input choke and these voltages can be nasty. The quality and voltage rating of that capacitor must be high - it can literally physically sing like a loudspeaker. You are still choke loaded.
As the capacitor value increases you get into a capacitive input device where you pulse energy into the capacitor at twice line frequency (if using a bridge). The quality and rating of the capacitance is no longer as radical as before. If your numbers are "hot" all around, you will get an output voltage which is the peak of the output voltage of the transformer. The current load will severely peak and you will send junk back to the grid which will probably get picked up by your other audio equipment. Note that if you have chokes in a C input stack, you still have capacitive loading, so even thought the choke may do a lot of good, it does not do as great of a job alone as before, and the requirements on choke current capacity and thermal dissipation rating are relaxed. Again consider why this is so: There is only very short time where the transformer voltage is higher than the output voltage - this is the only time that energy can flow from transformer and thus the energy density of that time (current) will be very high. For tube rectification, this is ameliorated somewhat by high impedance of the rectifier tubes themselves at the expense of more output ripple voltage.
So to make it clear that there is no free lunch:
The choke of a choke loaded system lengthens the charge pulses and puts less current strain on transformer and rectifier at the expense of lower output voltage and much higher voltage stress on the rectifiers. Ripple approaches sinusoidal. Utilization of transformer can be higher and a smaller model selected.
A capacitive loaded system puts very high current stress on transformer and rectifier but low voltage stress on rectifier. Ripple (both input current and output voltage) is nowhere near sinusoidal. Adding chokes after a capacitively loaded system does not make your system choke loaded - or more precisely it makes it choke loaded at that point where the operating conditions are quite different thus the end result will be quite different. Any classic textbook will intimately cover a capacitively loaded rectification system.
It is somewhat surprising to me that "nobody" seem to understand these very basic ideas.
PSUD is your friend for these sort of things. Start with the simplest possible circuit and build from there.
Petter
If you have an LC filter and your L is large, you have what is called a choke input filter. The output voltage of that filter will be lower than the if you have direct capacitive loading of a rectifier (i.e C is the first thing you see). Note that the loading of the transformer will be lower and you can typically extract significantly more energy out of it than with C input even though the voltage is lower (Cu losses are proportional to the square of the current, so pulsing the current is not a good idea and one of the reasons why people like to use ridiculously large transformers for audio). Ripple with L input rectification will also tend to be more or less sinusoidal if L is large enough. Think through why this may be so - hint the inductor lengthens the charge pulses so the energy can flow over a longer period of time.
If the input C is very low, all you do pretty much is take som spiking off the output of the rectifier which is caused by the input choke and these voltages can be nasty. The quality and voltage rating of that capacitor must be high - it can literally physically sing like a loudspeaker. You are still choke loaded.
As the capacitor value increases you get into a capacitive input device where you pulse energy into the capacitor at twice line frequency (if using a bridge). The quality and rating of the capacitance is no longer as radical as before. If your numbers are "hot" all around, you will get an output voltage which is the peak of the output voltage of the transformer. The current load will severely peak and you will send junk back to the grid which will probably get picked up by your other audio equipment. Note that if you have chokes in a C input stack, you still have capacitive loading, so even thought the choke may do a lot of good, it does not do as great of a job alone as before, and the requirements on choke current capacity and thermal dissipation rating are relaxed. Again consider why this is so: There is only very short time where the transformer voltage is higher than the output voltage - this is the only time that energy can flow from transformer and thus the energy density of that time (current) will be very high. For tube rectification, this is ameliorated somewhat by high impedance of the rectifier tubes themselves at the expense of more output ripple voltage.
So to make it clear that there is no free lunch:
The choke of a choke loaded system lengthens the charge pulses and puts less current strain on transformer and rectifier at the expense of lower output voltage and much higher voltage stress on the rectifiers. Ripple approaches sinusoidal. Utilization of transformer can be higher and a smaller model selected.
A capacitive loaded system puts very high current stress on transformer and rectifier but low voltage stress on rectifier. Ripple (both input current and output voltage) is nowhere near sinusoidal. Adding chokes after a capacitively loaded system does not make your system choke loaded - or more precisely it makes it choke loaded at that point where the operating conditions are quite different thus the end result will be quite different. Any classic textbook will intimately cover a capacitively loaded rectification system.
It is somewhat surprising to me that "nobody" seem to understand these very basic ideas.
PSUD is your friend for these sort of things. Start with the simplest possible circuit and build from there.
Petter
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.