with today's caps much more improved than yesterdays'
the reverse can be made, small chokes and large caps,
i do it all the time.....470ufd/450v are numerous and cheap...
this link might help you....powersupplies
The SP2-XT speakers I own produce 101 db @ 1m 1W. Their maximum SPL (sound pressure level) is 126 db @ 600 W. I usually listen at 1 v or about 1/8 watt for soft passages, so 30 v is almost 50 db louder.
I've only measured about 24 v on my speaker terminals, but that was loud parts of a Rihanna CD measured with a VOM. The peak voltage could go quite a bit higher without an analog meter pointer showing because the peak is shorter in time than the needle will indicate.
Here's a math problem for you MagicBus: you mentioned 3300 uf was okay for a 120 W stereo amp because of the ripple percentage equation. Actually the rail voltage is 70 regulated down from 80 open circuit by transistors. So the ripple will be nil, the noise from the zener diode stack on the base of the regulator transistors (five parallel darlingtons) being more important.
What is important is how much peak current I can get out for how long into the speakers. 200 W would be 14 A for both speakers, or 7 A each side. Both sides draw out of the same capacitor. If the transformer didn't replenish amps in the 200 W peak, how much would the capacitor sag in voltage from 70 v? For how long? If the peak on speaker is 30 v, and rail is 70 v, then probably gain starvation might occur at 45 v rail. So how long is a peak current of 14 amps sustainable from a 3300 uf cap as it drops from 70 v to 45 v? Is that long enough to hear by humans, not mice? Remember one farad will discharge one volt from one ampere one second.
There is some quibble as to whether a NTE60 transistor will sustain that current, and on the plus side, the transformer contributes 6.75 amps average from a 1000 uf cap behind the regulator transistors all the time during the peak. MJ15003 with one datapoint soa match that might be equivalent to NTE 60 shows soa of 7 amps at 35 Vce, so it might just be doable with one output pair.
Cannon shot is assumed to be plus impulse square wave followed by logrithmic decay until you can hear the orchestra again. How long is the squared off part?
I've only measured about 24 v on my speaker terminals, but that was loud parts of a Rihanna CD measured with a VOM. The peak voltage could go quite a bit higher without an analog meter pointer showing because the peak is shorter in time than the needle will indicate.
Here's a math problem for you MagicBus: you mentioned 3300 uf was okay for a 120 W stereo amp because of the ripple percentage equation. Actually the rail voltage is 70 regulated down from 80 open circuit by transistors. So the ripple will be nil, the noise from the zener diode stack on the base of the regulator transistors (five parallel darlingtons) being more important.
What is important is how much peak current I can get out for how long into the speakers. 200 W would be 14 A for both speakers, or 7 A each side. Both sides draw out of the same capacitor. If the transformer didn't replenish amps in the 200 W peak, how much would the capacitor sag in voltage from 70 v? For how long? If the peak on speaker is 30 v, and rail is 70 v, then probably gain starvation might occur at 45 v rail. So how long is a peak current of 14 amps sustainable from a 3300 uf cap as it drops from 70 v to 45 v? Is that long enough to hear by humans, not mice? Remember one farad will discharge one volt from one ampere one second.
There is some quibble as to whether a NTE60 transistor will sustain that current, and on the plus side, the transformer contributes 6.75 amps average from a 1000 uf cap behind the regulator transistors all the time during the peak. MJ15003 with one datapoint soa match that might be equivalent to NTE 60 shows soa of 7 amps at 35 Vce, so it might just be doable with one output pair.
Cannon shot is assumed to be plus impulse square wave followed by logrithmic decay until you can hear the orchestra again. How long is the squared off part?
Last edited:
> in your scientific community.
This is not a "scientific" community. It is practical people using simple electric rules to build working toys.
> To my understanding signal envelope is the effect of the 200Hz on the 2kHz regardless the function of the psu
It does not have to be 200Hz (in fact it seems to be 220Hz).
The point is that they threw a MAXIMUM (heavy clipping) signal through the amplifier, here for about 0.1 Seconds 100 mS, the "envelope". The DC power rail "sagged". Now shown, but known from thought-experiment or actual amp abuse, the test frequency does not matter (within wide limits).
Is sag bad? (Good?) Can sag be avoided? What do we do to avoid it? Will it take money which would be better spent elsewhere?
All of which is NOT a new problem. You WILL be fine building stuff using the same general lump-sizes as other toys of similar scale.
> I have the feeling that engineering can be straightforward!
It IS! In many practical cases. Mostly about picking materials and proportions which suit the actual need.
There was a popular saying "WWJD", "What would Jesus do?". I can't pretend to know that. But I can guess "What would Joseph do?", Joe the "dad" of Jesus. He was a carpenter (or similar). Anyway, Mary says the Blessed Child is getting into the wise-men's incense, please put up a shelf! Does Joe put up a shelf 4 feet long but toothpick thin? No, saggy. Does he put up 4 feet wide and 2 feet thick? No, that's a waste of good wood and maybe too heavy for the wall. Did he look up in a building code? Did he simulate it in a software program? Uh, no.
A good carpenter (then or now) has some rules of thumb and some sense of proportion, at least for the average materials and loads in his jobs. A board 2 or 3 cubits wide, with a light load (jars of incense, not clay bricks) can be about one thumb thick. The units have changed, but this is still how we do a kitchen or book shelf.
Where did Joseph learn all this? He PLAGIARIZED! Or rather, he watched others. We can assume when old Joe was a wee lad, his father rented him out to a carpenter. Little Joe hauled logs and watered-wine, learned to chop straight, and incidentally learned what his boss/teacher used for beams, door frames, and shelves. (And what choices didn't work well...) After many years and many projects he knew enough that the other carpenters in town endorsed him as a master carpenter. Learning by looking at what others do.
> The idea of specifying PT according to speakers/room is new to me
That's again Engineering, and not a fussy rocket-science type.
The ideal goal is to have more power than you ever need. An infinitely powerful amplifier.
But infinitely powerful is another way to say "infinite cost". Also with great power comes great risk. Also electric bills.
We put power in the speaker so sound hits listener and lingers in the room until it is absorbed by the cat and carpet. How much power is that?
Speaker efficiency is poor. Listeners expectations have varied widely.
But if you can quantify all that (or make good WWJD guesses), then it is easy to get a workable system.
For most of the 20th century, a "home system" was a 4"-8" speaker on a 1 Watt amp. Millions of kitchen radios. Quite a few record-players and tape machines. Deluxe was a 5 or 15 Watt amp with a bigger speaker. "Hi-Fi" was often a 2-way speaker with 20 Watts. Change of amplifier economics and the need to fit two speakers for stereo led to 30W and 50W amps. Transistors made 200 Watt amps possible. 3,200W amps exist for larger rooms (drinking places).
Same as "how big an engine for a truck?" Well, an urban flower delivery or huge loads of ore from a mine? Could be 20HP, could be 800HP. You could move the ore faster with a 3,200HP engine, but the cost sinks the business. Get a general idea of "load" (flowers, ore, or room), estimate what that needs. Unlike a moon-rocket, it usually is not too critical or exact.
This is not a "scientific" community. It is practical people using simple electric rules to build working toys.
> To my understanding signal envelope is the effect of the 200Hz on the 2kHz regardless the function of the psu
It does not have to be 200Hz (in fact it seems to be 220Hz).
The point is that they threw a MAXIMUM (heavy clipping) signal through the amplifier, here for about 0.1 Seconds 100 mS, the "envelope". The DC power rail "sagged". Now shown, but known from thought-experiment or actual amp abuse, the test frequency does not matter (within wide limits).
Is sag bad? (Good?) Can sag be avoided? What do we do to avoid it? Will it take money which would be better spent elsewhere?
All of which is NOT a new problem. You WILL be fine building stuff using the same general lump-sizes as other toys of similar scale.
> I have the feeling that engineering can be straightforward!
It IS! In many practical cases. Mostly about picking materials and proportions which suit the actual need.
There was a popular saying "WWJD", "What would Jesus do?". I can't pretend to know that. But I can guess "What would Joseph do?", Joe the "dad" of Jesus. He was a carpenter (or similar). Anyway, Mary says the Blessed Child is getting into the wise-men's incense, please put up a shelf! Does Joe put up a shelf 4 feet long but toothpick thin? No, saggy. Does he put up 4 feet wide and 2 feet thick? No, that's a waste of good wood and maybe too heavy for the wall. Did he look up in a building code? Did he simulate it in a software program? Uh, no.
A good carpenter (then or now) has some rules of thumb and some sense of proportion, at least for the average materials and loads in his jobs. A board 2 or 3 cubits wide, with a light load (jars of incense, not clay bricks) can be about one thumb thick. The units have changed, but this is still how we do a kitchen or book shelf.
Where did Joseph learn all this? He PLAGIARIZED! Or rather, he watched others. We can assume when old Joe was a wee lad, his father rented him out to a carpenter. Little Joe hauled logs and watered-wine, learned to chop straight, and incidentally learned what his boss/teacher used for beams, door frames, and shelves. (And what choices didn't work well...) After many years and many projects he knew enough that the other carpenters in town endorsed him as a master carpenter. Learning by looking at what others do.
> The idea of specifying PT according to speakers/room is new to me
That's again Engineering, and not a fussy rocket-science type.
The ideal goal is to have more power than you ever need. An infinitely powerful amplifier.
But infinitely powerful is another way to say "infinite cost". Also with great power comes great risk. Also electric bills.
We put power in the speaker so sound hits listener and lingers in the room until it is absorbed by the cat and carpet. How much power is that?
Speaker efficiency is poor. Listeners expectations have varied widely.
But if you can quantify all that (or make good WWJD guesses), then it is easy to get a workable system.
For most of the 20th century, a "home system" was a 4"-8" speaker on a 1 Watt amp. Millions of kitchen radios. Quite a few record-players and tape machines. Deluxe was a 5 or 15 Watt amp with a bigger speaker. "Hi-Fi" was often a 2-way speaker with 20 Watts. Change of amplifier economics and the need to fit two speakers for stereo led to 30W and 50W amps. Transistors made 200 Watt amps possible. 3,200W amps exist for larger rooms (drinking places).
Same as "how big an engine for a truck?" Well, an urban flower delivery or huge loads of ore from a mine? Could be 20HP, could be 800HP. You could move the ore faster with a 3,200HP engine, but the cost sinks the business. Get a general idea of "load" (flowers, ore, or room), estimate what that needs. Unlike a moon-rocket, it usually is not too critical or exact.
> How do you calculate this? How much Vpk is needed for 15W into 5r0? Does Vpk means from side to side in a symmetrical psu?
Do you know, understand, how we turn raw boring DC into interesting audio?
Without any deep understanding of transistors, you should be able to work-out how "big" a Sine you can get out of a given DC.
Why we use "Sine" in audio is perhaps a different matter, but go with tradition.
Do you know, understand, how we turn raw boring DC into interesting audio?
Without any deep understanding of transistors, you should be able to work-out how "big" a Sine you can get out of a given DC.
Why we use "Sine" in audio is perhaps a different matter, but go with tradition.
> "crunch guitar". ... driving the tubes up in to the red plate zone where the tops were chopped off
The plate power *reduces* when the amplifier is pushed to clipping.
For ideal devices, device dissipation goes to zero.
As a practical case: I had SE 6550 idling at 39 Watts dissipation. This was an older 6550 which creaked as it heated or cooled. When I PUSHED it to hard-clipped square-ish wave output, I could hear the plate creak. The dummy load read about 24 Watts of square-wave, confirmed by the paint burning on the 20W resistor. The DC input hardly changed. So the tube dissipation *dropped* from 39W to 15W, and it creaked on cool-down.
That was A SE, but a "B" amp will also tend to go to zero dissipation when pushed to hard clipping. Class B amps fed Sines dissipate the most near 1/4-1/2 maxumum sine level (this is where they may red-plate); dissipation drops at higher level.
If you are melting OVER-driven tubes in 6 hours, look to your screens. When a tet/Pentode bottoms, G2 current rises. None of the Audio Power Pentodes were designed for "non-audio" (grossly distorted) conditions.
The plate power *reduces* when the amplifier is pushed to clipping.
For ideal devices, device dissipation goes to zero.
As a practical case: I had SE 6550 idling at 39 Watts dissipation. This was an older 6550 which creaked as it heated or cooled. When I PUSHED it to hard-clipped square-ish wave output, I could hear the plate creak. The dummy load read about 24 Watts of square-wave, confirmed by the paint burning on the 20W resistor. The DC input hardly changed. So the tube dissipation *dropped* from 39W to 15W, and it creaked on cool-down.
That was A SE, but a "B" amp will also tend to go to zero dissipation when pushed to hard clipping. Class B amps fed Sines dissipate the most near 1/4-1/2 maxumum sine level (this is where they may red-plate); dissipation drops at higher level.
If you are melting OVER-driven tubes in 6 hours, look to your screens. When a tet/Pentode bottoms, G2 current rises. None of the Audio Power Pentodes were designed for "non-audio" (grossly distorted) conditions.
If you use the proper size caps, the voltage sag of the rail supply, Vcc, is not fast at all. Same problem I put to you in post 42, calculate the worst case voltage sag of a worst case music transient coming out of an actual amp. I calculated 6 ms square wave for the cannon shot while watching TV last night, which to first approximation is piano note E3, which confirms my opinion that the cannon shot sounds rather like a bass note.
The fast rise you describe is "turn on thump" and many designs clamp the input or disconnect the output of the gain components to keep this thump from coming out the speaker.
The fast rise you describe is "turn on thump" and many designs clamp the input or disconnect the output of the gain components to keep this thump from coming out the speaker.
Last edited:
There is something about that I' m trying to clarify. It supposed that the "turn on thump" will be there on every transient hit and if sag is slow then inevitably the recharge will be also slow.
If we disregard the different way a psu sees transistors/pentodes and triodes as a load and also keep regulated supply out of the equation then we look for a final capacitor that could discharge fast on transient demand, sustain voltage for the soft passage and recharge fast provided it is supported by suitable transformer. How fast is fast for music? We can calculate the power for the cannons and for the choir. How long it will take after the bang to listen again to the choir? Rhetorical questions of course! I find your suggestions very interesting. I have to work it out on my amps (tubes). Neighbors will suffer tonight!
For hi-fi (but not necessarily guitar amps) any sag is bad; fast sag is worse. So if deliberately introduced it is a clear sign of poor design, caused by confusion and misunderstanding. If it happens accidentally then it may be a sign of merely ignorant design. You need to be aware that there are a few people who deliberately design poor PSUs for SET amplifiers, using big L and small C, and regard the resultant gain pumping as a sign of 'good dynamics'.MagicBus said:
slowish but not usually equal.
Look at at PSU rail voltage and examine the slopes on the rail, the ripple.
The recharging is much faster than the slower draw down. Typically the discharge time is 4 times as long as the charge time, 8ms:2ms for 50Hz mains.
Yes, I' m aware of such deliberately poor designs. That's why I try to understand. "Fast sag is worse" I think this is clear enough. Thanks DF96
Assuming large capacitors are used? I' ll check this vs small caps. It will be interesting.
Actually, I studied refrigerator turn off transients pretty thorougly at the refrigerator factory where I worked. We overload tested every compressor , which when it turned off could produce a 1300 v transient on the AC line, which would repeat every 15 seconds (the production rate) on my 200 mhz scope.There is something about that I' m trying to clarify. It supposed that the "turn on thump" will be there on every transient hit and if sag is slow then inevitably the recharge will be also slow.
Fortunately, E-frame transformers produce some supression of these power line transients on the output because of the characteristics of the iron. It is also reduced by the turns ration of the transformer. So on my 120 vac in 80 vac out ST120 transformer, a 1300 V refrigerator shut off spike would come out as 833 volts. Possibly less due to the iron. I don't have a scope at home to study it. Dynaco put a .01 uf 1000 v ceramic cap at the output of the bridge rectifier to suppress these transients, and it is effective - I don't hear refrigerator shutoff pops. I did hear a lightning strike induce a pop the other night with the thunder about 3 seconds away. So it was about 1/2 mile away strike.
I don't know if toroid transformers supress transient spikes this well, I don't own any that aren't in switcher supplies. Even cheap PCAT switcher supplies have a MOS transient supressor before the main filter caps to keep power line transients from shorting out the 200 v rated caps. Plus a .47 uf X rated cap.
Transients in the music are not supressed in hifi amps, only by the Ft of the transistors or inductance of coupler caps. Vacuum tubes are very fast, Ft is not even described. But many tube amps had wound paper coupler caps, which were pretty inductive since there was only one input and output pad (conductor) for all the windings. Wound Film caps are a little less inductive because the ends have a sprayed on metal connector to all the windings.
Guitar & PA amps have a high frequency suppressor on the music many times, to keep from blowing up the tweeter. Peavey calls it "DDT" and it measures the high frequency component of the music and turns down the volume if there is too much of it.
Last edited:
In 1966, the output transistors were pretty slow. I never heard the amp with the originals, they were burnt up when I got it, but people who simulated it said that the harmonic distortion from 10 khz to 20 khz was a lot better with the next generation of higher Ft output transistors, TIP3055. The 1966 amp did not have a ceramic cap parallel to the electrolytic cap after the regulator.
So now the amp is producing drum hits, cannon shots, piano hammer strikes, very well. Lots of high frequency content. I don't think I have a ceramic cap parallel to the main filter cap yet - all the 1 mf ceramics I have in stock are 50 v limited and the main supply is 70. So a 3300 uf 100 v CDE electrolytic cap supports hifi music pretty well, IMHO.
On the ST120 my original driver board with 1966 5 digit RCA driver transistors sounds a bit better on tinkly bells, than the replacement AX6 board I built for the left with 3 mhz Ft TIP41/42C driver transistors. So I think I'm hearing a little high freq limit from the slowish transistors on the AX6 side. Both sides share the rail capacitor, so that is not a limiter. I've bought faster driver transistors but I don't want to shut the system down to put them in yet until I finish some replacement amp. Meanwhile the television sound has turned to **** probably due to an 8 year old e-cap, and a friend's house is under supervision of the city til we finish work, so other projects get priority. I'm drying out now from working over there in the rain.
So now the amp is producing drum hits, cannon shots, piano hammer strikes, very well. Lots of high frequency content. I don't think I have a ceramic cap parallel to the main filter cap yet - all the 1 mf ceramics I have in stock are 50 v limited and the main supply is 70. So a 3300 uf 100 v CDE electrolytic cap supports hifi music pretty well, IMHO.
On the ST120 my original driver board with 1966 5 digit RCA driver transistors sounds a bit better on tinkly bells, than the replacement AX6 board I built for the left with 3 mhz Ft TIP41/42C driver transistors. So I think I'm hearing a little high freq limit from the slowish transistors on the AX6 side. Both sides share the rail capacitor, so that is not a limiter. I've bought faster driver transistors but I don't want to shut the system down to put them in yet until I finish some replacement amp. Meanwhile the television sound has turned to **** probably due to an 8 year old e-cap, and a friend's house is under supervision of the city til we finish work, so other projects get priority. I'm drying out now from working over there in the rain.
Last edited:
i am not a big fan of simulations, i do actual builds and listen....most of the time i liked what i heard....
in my psu builds, i follow PRR's advise, build it like a "strong bridge..."
and with today's costs for rectifiers and caps, how can i not?
also, surplus very good quality cores can be had here for almost a steal from metal recyclers.....
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.