> info about signal envelope!
I just passed a plot in Neumann. See attached.
From Guitar Amplifier Overdrive Neumann, Irving, ISBN 132959665X
The run starts with a 2KHz tone about 6dB below clipping. The B+ is steady with some ripple. (Here "A+" is the first B+ filter node, not the filament supply.)
Then a MUCH larger 200Hz tone is added. The power stage is about 14dB past clipping. Because this is an AB stage, and the drive is past Sine to Square, average B+ current increases. The B+ sags. You also see more ripple. The overdriven 200Hz tone starts from the un-sagged B+ but drops slightly as B+ sags. The peaks of the clipping shift with B+ ripple (making IM at 200Hz +/-120Hz).
The blue line is an approximation of the "envelope" of the output wave (some bent because of overload effects).
When the abuse stops, the B+ un-sags and ripple goes back to before.
Incidentally, while OVER-driven, coupling networks charged, bias shifted. When abuse stops, the bias recovers. Notable is no gross wonkiness after a burst of OVER-load. (Some amplifiers "faint" due to grid-blocking.)
This is !NOT! how we use hi-fi amps!! We do clip, but very rarely. Our average is 20dB down from clipping, not the -6dB used in this test. We usually aim for lower ripple than this. (But due to push-pull action, and reduced ripple at idle, amps like this {even with exceptionally efficient speakers!} can have low hum in a quiet club or worship service.)
When I modded big tube amps for "good" sound (reinforcement of clean acts), I did test with HUGE overdrive to be sure they would not go "acccckkT!" if hit a bit hard. I did not obsess about how much virgin llama-oil was in the filter caps, or strive for "no sag". IIRC, a 300W amp which idled at 600V B+ would sag to 550V in OVER-drive, but working live-sound to the point I could hear some clips it only dipped to 590V.
The amp shown is a Fender 5F6a. You can find the plan(s) everywhere. This is perhaps a minimal starting point for a 40+W tube amp with 1950s economics. Many fine hi-fi amps used the same 40uFd first cap right to the OT. Most filter the Screens with R-C not L-C (because they are not meant to play at-max). It is ugly on 'scope but it mostly cancels in the OT. 10+dB NFB almost hides the residual. In today's economics, MUCH bigger caps are low-cost (and Si rectifiers can stand larger peaks than 5U4) so hundreds of uFd might be used in a Hi-Fi 2x6L6 amp.
The referenced book is a dense read, and not suggested for beginners. It also focuses on *distortion*, so by definition has little for the Hi-Fi fan. But if you want to know "why" we do a power supply a certain way, this specific example shows what happens at an extreme.
I just passed a plot in Neumann. See attached.
From Guitar Amplifier Overdrive Neumann, Irving, ISBN 132959665X
The run starts with a 2KHz tone about 6dB below clipping. The B+ is steady with some ripple. (Here "A+" is the first B+ filter node, not the filament supply.)
Then a MUCH larger 200Hz tone is added. The power stage is about 14dB past clipping. Because this is an AB stage, and the drive is past Sine to Square, average B+ current increases. The B+ sags. You also see more ripple. The overdriven 200Hz tone starts from the un-sagged B+ but drops slightly as B+ sags. The peaks of the clipping shift with B+ ripple (making IM at 200Hz +/-120Hz).
The blue line is an approximation of the "envelope" of the output wave (some bent because of overload effects).
When the abuse stops, the B+ un-sags and ripple goes back to before.
Incidentally, while OVER-driven, coupling networks charged, bias shifted. When abuse stops, the bias recovers. Notable is no gross wonkiness after a burst of OVER-load. (Some amplifiers "faint" due to grid-blocking.)
This is !NOT! how we use hi-fi amps!! We do clip, but very rarely. Our average is 20dB down from clipping, not the -6dB used in this test. We usually aim for lower ripple than this. (But due to push-pull action, and reduced ripple at idle, amps like this {even with exceptionally efficient speakers!} can have low hum in a quiet club or worship service.)
When I modded big tube amps for "good" sound (reinforcement of clean acts), I did test with HUGE overdrive to be sure they would not go "acccckkT!" if hit a bit hard. I did not obsess about how much virgin llama-oil was in the filter caps, or strive for "no sag". IIRC, a 300W amp which idled at 600V B+ would sag to 550V in OVER-drive, but working live-sound to the point I could hear some clips it only dipped to 590V.
The amp shown is a Fender 5F6a. You can find the plan(s) everywhere. This is perhaps a minimal starting point for a 40+W tube amp with 1950s economics. Many fine hi-fi amps used the same 40uFd first cap right to the OT. Most filter the Screens with R-C not L-C (because they are not meant to play at-max). It is ugly on 'scope but it mostly cancels in the OT. 10+dB NFB almost hides the residual. In today's economics, MUCH bigger caps are low-cost (and Si rectifiers can stand larger peaks than 5U4) so hundreds of uFd might be used in a Hi-Fi 2x6L6 amp.
The referenced book is a dense read, and not suggested for beginners. It also focuses on *distortion*, so by definition has little for the Hi-Fi fan. But if you want to know "why" we do a power supply a certain way, this specific example shows what happens at an extreme.
Attachments
Last edited:
That is in a house, by yourself. I'm using about 1/8 watt base level (1 v ) in my music room.
But playing to 200 seats on December 19 when everybody is wearing sweaters or wool , you need a lot more.
In a noisy bar or dance hall, a whole lot more.
And outdoors from 50' away, 3/4 W is just not loud enough.
Playing a crowd of 400 in an open field, if the Harley riders club decides to ride by (as they do by the city park frequently, & no mufflers enforced here) you need more like 110 db base level to be heard. Hundreds of watts just base level.
Last edited:
And was that indoors?
At the city park, nearest listeners are 30' from the stage. The back row is 150' away, the corners 400' away. It would be more effective to put speakers everywhere & use delay lines, but that stuff would be stolen 3 hours after the concert if it wasn't taken down immediately. Putting the sound system on the stage near the loading dock seems to be appropriate. Two big sets of speakers on the stage seem adequate, even when a 20000 hp barge string is accelerating out of the lock on the river behind the stage. Bands you can really hear from the back row use two 8 driver 15" speaker enclosures, IMHO. Plus tops. That is I suspect more than 140 W.
Well, admitely much more info than expected!
I understand that there isn't a ready recipe for designing a psu, but I have the feeling that engineering can be straightforward! The idea of specifying PT according to speakers/room is new to me and very interesting. Add capacitance for tranient peaks headroom, avoid clipping - PRR, great info 
- use what you have at hand, spend wisely, no room for snake oil! That said, you are invited to share your subjective approach to the task WRT hifi mostly but not only.
To start with, I present the method I used to design PSUs. Since there is no bench equipment and I' m not familiar with LTSpice and the similar, I play with Ohm's law and PSUD. I was reading a paper published on the internet about how to use PSUD to optimize a power supply;http://www.dhtrob.com/overige/pdf/dhtrob_psu.pdf
It refers to fast recovery time and damping factor and it ends up with huge chokes and tiny caps carrefully damped to avoid overshooting. The target is a sharp and clean roll off - or rising - which improves sound. Perhaps with a compromise to low frequency extension? Any ideas?
I understand that there isn't a ready recipe for designing a psu, but I have the feeling that engineering can be straightforward! The idea of specifying PT according to speakers/room is new to me and very interesting. Add capacitance for tranient peaks headroom, avoid clipping - PRR, great info - use what you have at hand, spend wisely, no room for snake oil! That said, you are invited to share your subjective approach to the task WRT hifi mostly but not only.To start with, I present the method I used to design PSUs. Since there is no bench equipment and I' m not familiar with LTSpice and the similar, I play with Ohm's law and PSUD. I was reading a paper published on the internet about how to use PSUD to optimize a power supply;http://www.dhtrob.com/overige/pdf/dhtrob_psu.pdf
It refers to fast recovery time and damping factor and it ends up with huge chokes and tiny caps carrefully damped to avoid overshooting. The target is a sharp and clean roll off - or rising - which improves sound. Perhaps with a compromise to low frequency extension? Any ideas?
this is a good way to specify a power amplifier. But we rarely see it.
Manufacturers have adopted POWER ratings instead.
20Vpk is 25W into 8r0
30Vpk is 56W into 8r0
40Vpk is 100W into 8r0
50Vpk is 156W into 8r0
60Vpk is 225W into 8r0
If those were converted to dBv:
the 20Vpk equals +23dBv and the 60Vpk equals +32.5dBv (a range of just 9.5dBv)
Last edited:
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?
Yes, you are confused. Don't worry, we were all confused by PSUs at some time in the past. I think part of your trouble is that you are going about your learning in the wrong way. You have discovered some words, and you are asking us what they mean - but in some cases you don't have enough background knowledge to understand our answers.MagicBus said:
Better to start back at the beginning. Design and build a safe PSU which will provide a particular voltage and current at a particular ripple. Play with it. Find out what it does well and what it does badly. At this stage don't worry about envelopes or 'sound' or 'fast PSUs'. You need to realise that some of what you have read is nonsense, but you don't yet know which is which. Some people never understand this!
Ohm's Law. P=IV. Or P = V^2/R. Remembering (or learning?) that Vpk = 1.414Vrms.
I think that by asking this question you have confirmed that you are not yet in a position to understand the finer points of PSU design.
Rearrange the formula to give W= V squared/R
therefore,
V = square root of W * R
Sqr root of (15*5) is 8.6 volts. That is 8.6 volts rms. I will let you calculate the peak voltage from that
This will help:
Measuring Vibration - Page 6
https://en.wikipedia.org/wiki/Root_mean_square
One thing I admit is my lack of studies on electronics! But I'm used in advanced reading, terminology, maths and physics. This doesn't make me a self educated Electronic Engineer neither is my intention to make an intrusion as such in your scientific community. I enjoy being a first year student in diyaudio.com university! I chose these forums deliberately to avoid all the nonsense floating on the internet. I come here with enough diy backround, many soldering hours, I' ve done the simple circuits, I can design my one and be sure they won't explode. At this point I felt that I have to respect electronics same way I do for science in generall. It was not that easy because I mess with electronics since chilhood. It is people like you that forced me to change my opinion.
That said I think I perfectly understood Indianajo's and PRR's posts. I can't tell if I could go more advanced than this but it was pure brain food for me and I'm grateful! The rest of the posts also appreciated, needless to say. Andrew's calculations obviously involved Ohm's law but I missed RMS voltage. I thought it was about DC power rails. I' m still learning.
Please feel free to throw in theory. I' ll take as much as I can, the rest won't be wasted I think... And practical tricks if any. We agreed that psu design is subjective!
Is this relevant to the formula for calculating final cap in static conditions - found on the net; C=2*I/f*V
Indeed, I read about it in M. Blencowe's book. Besides it presents in a different form - C=Idc/2fVpp which I assume is the correct - it refers to reservoir cap and how to determine the desired ripple. In the article I was reading previously it refers to the final cap wrt providing stable voltage to the load. Does this has any meaning?
Here is the deep water... To my understanding signal envelope is the effect of the 200Hz on the 2kHz regardless the function of the psu. This compined signal will modulate the psu and be modulated by the psu even if it doesn't reach to clipping. The result will be IM at the frequency of the trancient +/- 2*mains frequency. Is there any particular reason that we should mind only for the low frequency trancients or I miss something? Of course, I may missed the whole thing!
While PRR's post about guitar amp design is a little over my head, some of it is quite understandable. People find squarish sine waves with rounded corners very pleasant to hear. This is called "crunch guitar". This was originally achieved by driving the guitar signal into a 6L6 push pull tube circuit, then driving the tubes up in to the red plate zone where the tops were chopped off by the power limit, but the fact they were tubes caused the edges to be rounded. By contrast, sine waves chopped off flat by transistor output stages, are very unpleasant. lots of odd harmonic (I think) fourier components, which people hate.
Driving 6L6's in the red plate zone costs a lot, about $70 for a 6 hour tube life, so engineers spent a lot of hours trying to achieve the sound, without the cost. Plus tubes are subject to shorts from the chock of carrying around and setting up, running the cost up more.
Hence this special guitar amp power supply. PS is required to deliver a lot of high freq even harmonics, more than a sine wave or flute or violin signal. Because square top waves heat up the transformer more than sine waves, high volume guitar amp transformers have less "copper loss' ie internal resistance, than the transformer sold for equivalent hifi market wattage.
I'm a physicist too, and I learned ohms law, kirchoff's laws, complex arithmetic models of inductors and capacitors, how to calculate linear approximations of current flow in complex circuits, blah blah, in 2nd year physics. By contrast with most of the rich guys I went to school with, I was repairing car radios and stereo's my parents owned and wore out, and graduated to serious hifi when I bought worn out dynaco equipment my junior year. So I had a physics lab in Dad's garage. Better than electronics lab at school where the wire was all cut up into 1" pieces and half the components were blown and put back on the shelf.
The trouble with math models is they don't figure in cost. There is design/prototyping cost, and there is production cost. Some people have referred to wire versus iron design in transformer design. Actually, until you work for a megacorporation, you don't get to choose any of that. A prototype inductor will cost you $$$$ unless you can build it yourself from a kit supplier, or have the prospect of a sale of 5000 units to back up your request. So use what is in catalogs and stocked at distributors, or cheaper, what was in burnt up junk. I can buy a dozen blown up amps for the cost of one prototype inductor.
That PS design with a tiny cap and a huge choke strikes me as hilarious. That thing might work great for TV commercials, where the waveforms are fattened up to be as loud as possible and the volume is close to the max allowed all the time.
For music, which varies a lot, people use caps. Even though they wear out in 1 to 15 years depending on the sealant and water reserve size. For example, by contrast with crunch guitar with even harmonics and high power drive, piano requires huge power peaks at the start when the hammer strikes, then a much lower actual average power during the sustain portion of the note. So power supplies that supply classical music can use smaller transformers and heat sinks than guitar amps, but have huge headroom for peaks. My ST120 is set up for classical music now. The heat sinks will only sustain only about 10 W average, but with a rail voltage of 70 I can put out 200 W peaks into 8 ohms on occasion, like the cannon shot in 1812 overture.
Have fun.
Driving 6L6's in the red plate zone costs a lot, about $70 for a 6 hour tube life, so engineers spent a lot of hours trying to achieve the sound, without the cost. Plus tubes are subject to shorts from the chock of carrying around and setting up, running the cost up more.
Hence this special guitar amp power supply. PS is required to deliver a lot of high freq even harmonics, more than a sine wave or flute or violin signal. Because square top waves heat up the transformer more than sine waves, high volume guitar amp transformers have less "copper loss' ie internal resistance, than the transformer sold for equivalent hifi market wattage.
I'm a physicist too, and I learned ohms law, kirchoff's laws, complex arithmetic models of inductors and capacitors, how to calculate linear approximations of current flow in complex circuits, blah blah, in 2nd year physics. By contrast with most of the rich guys I went to school with, I was repairing car radios and stereo's my parents owned and wore out, and graduated to serious hifi when I bought worn out dynaco equipment my junior year. So I had a physics lab in Dad's garage. Better than electronics lab at school where the wire was all cut up into 1" pieces and half the components were blown and put back on the shelf.
The trouble with math models is they don't figure in cost. There is design/prototyping cost, and there is production cost. Some people have referred to wire versus iron design in transformer design. Actually, until you work for a megacorporation, you don't get to choose any of that. A prototype inductor will cost you $$$$ unless you can build it yourself from a kit supplier, or have the prospect of a sale of 5000 units to back up your request. So use what is in catalogs and stocked at distributors, or cheaper, what was in burnt up junk. I can buy a dozen blown up amps for the cost of one prototype inductor.
That PS design with a tiny cap and a huge choke strikes me as hilarious. That thing might work great for TV commercials, where the waveforms are fattened up to be as loud as possible and the volume is close to the max allowed all the time.
For music, which varies a lot, people use caps. Even though they wear out in 1 to 15 years depending on the sealant and water reserve size. For example, by contrast with crunch guitar with even harmonics and high power drive, piano requires huge power peaks at the start when the hammer strikes, then a much lower actual average power during the sustain portion of the note. So power supplies that supply classical music can use smaller transformers and heat sinks than guitar amps, but have huge headroom for peaks. My ST120 is set up for classical music now. The heat sinks will only sustain only about 10 W average, but with a rail voltage of 70 I can put out 200 W peaks into 8 ohms on occasion, like the cannon shot in 1812 overture.
Have fun.
Last edited:
More than fun, satisfaction of learning! Thanks Indianajo! The fact that I highlight parts of your posts doesn't mean that I don't enjoy them all. I' m just trying to focus my query.
Obviously IMD is desired in guitar amps but it happens to HiFi as well. Good thing to fight it (big word
).What is the SPL of the speakers you running with ST120? Do you feel it takes the whole headroom for the cannons?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.