|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
141.239.172.61
Most of the elements, caps, chokes, trans, resistors, are self evident but the output tube whether power or signal, how does one determine or represent that?
I have been using the "current tap" and using the expected or estimated current requirement for the tube after the filter section of the power supply. Any other tips is also appreciated. cheers.
Follow Ups:
What I look at most is once I have the proper voltage and ripple within what I set as targets, I use the stepped load option. I always set my load as a CCS, as that is what I usually use as a load, (that is of course optional). Then with the stepped load, I can see how fast my power supply can respond to variations of current demand. Remember CCSs aren't perfect and there is always some variation in demand, even more so with other loads, so chose your stepped demand carefully. To me, the response to variations of time/current are key.
I can use various values of caps, chokes, resistance, number of sections, as well as choke or cap input, to effect the speed of power delivery (while of course keeping voltage and ripple requirements within limits). I try and get the quickest response from my supply, while staying within the voltage/ripple criteria that I want.
twystd
"I try and get the quickest response from my supply"Why?
The quickest response means the quickest voltage change due to current draw change.
Wouldn't the slowest voltage change due to current draw change represent better voltage regulation?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 05/29/17
The amount of voltage sag remains the same, so regulation is the same. The power supply is just quicker to stabilize to changing demands of current.
twystd
"The amount of voltage sag remains the same, so regulation is the same."We might be talking about two different things but I don't think the above is true. I am talking about a supply that Dennis Fraker would use.
A supply filter built to take longer to "stabilize" after a current draw change will be less regulated if the current draw increase (or decrease) is sustained. The chokes will be bigger with more DCR and the caps would be bigger.
If the supply is operating a Class A circuit then the only thing that will cause the current to increase or decrease from the idle current is the music.
BTW the current will never increase (or decrease) instantaneously as it does with the "stepped response" feature in PSUD2. It will, instead, follow the music and it will take time for the current to increase and decrease.
The longest duration the current will increase (or decrease) due to the music is 12.5ms ( 1/4 of a 20Hz wave cycle) before it heads the other way.
If you look real close to a supply designed to stabilize quickly vs a traditional power supply I think you will find that in the first 12.5ms the traditional (slow) supply dips (or soars) less.
Looking a sims of each, the "fast" supply dips 2 volts in the first 12.5ms and the traditional (slow) supply dips only .27 volts in the first 12.5ms
(I admit the two supplies are different, one is a 464 volt supply and the other is 530 volts but I think my conclusion is valid)
Now if the current increase were sustained the traditional supply dips 4.5 volts and the fast supply dips only 2.75 volts for a current increase from 60ma to 70ma. but Class A amplifiers playing music don't do that.
As I said, we might be talking about two different things.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 05/30/17
No, I'm definitely not talking about a supply Dennis Fraker would use. I usually use choke input critical inductance filters. I'm too tired to model and show you what I'm talking about. Maybe later....
twystd
I re-model my critical inductance choke input supply and just by lowering the last cap to 50uf the voltage dip in the first 12.5ms is 2 volts (and of course the ripple is up) and the total drop is still 4.5 volts because the DCR didn't change.
I believe that's in keeping with what you said.
So I conclude it's all about storage. The bigger the last cap the more voltage is stored and the longer it takes for that voltage to drop due to a change is current draw. The supply has a lower impedance.
When the current draw increase is sustained long enough then the voltage eventually drops in accordance with the DCR of the rest of the supply.
The DCR of the supply vs. the impedance of the supply.
I conclude that designing for a fast settling time is designing for a high impedance supply.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
I tried a few PSUD runs, and the more capacitance there is on the south end of a northbound LC supply the smaller the drop, the smaller the rise and lower is the ripple.
For PP, Class A, I have yet to see where this sounds worse than a liddle capacitor, ripply supply.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
Set the current tap to the idle current draw of the circuit.There is no way to test for power supply impedance in PSUD2.
You can do a stepped response (stepped load) with the current tap.
That will tell you 2 things,
1. if the supply filter will ring.
2. the DCR of the PS, but it will not tell you the impedance.
If the supply is feeding a Class A circuit then the average current draw over time (the longest time being the length of the lowest frequency in the music) will be equal to the idle current draw (assuming linear tubes).
If the final cap is large enough then the voltage should be largely sustained through the cycle even at the lowest frequency of interest.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 05/28/17
Hi Tre, and thank you for your tips. It seems that i have been doing more or less what you said. In several of my amps the B+ voltages have been close to the Psud calculations. The next question is; After inputting the various data of the components that one plans to use or needs to get, is there an "ideal" graphical representation that you should be trying to obtain? Which leads to what are important data that one should understand after the simulation has concluded and is given in the tables. thanks again.
Those are broad questions.I assume the reason you are simming the supply is to predict the voltage and ripple?
Those are shown in the graph when you have V(C[last cap]) checked.
If your supply is cap input then you might also look at the charging currents to the first cap, I(C1).
The higher the first cap's value, the higher the charging current and the shorter the "on" time for each diode. This is also shown on the graph.
Also look at the diode "peak current" I(D1). Also shown on the graph.
Make sure the peak repetitive current rating (IFRM) of the rectifier tube is not exceeded.
With the step response you can see if the supply is ringing.
That is shown on the graph (with V(C[last cap]) checked) by the voltage "bouncing" up and down several times before settling down after the current "step".
A smooth "ski slope" (without the bouncing) shows a non-ringing behavior of the supply.
If you set the sim to run for 20 seconds after a reporting delay of 6 seconds and set the current step to increase the current by 10 or 20% after 8 seconds then you get a clean graph without things "running into" each other.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 05/28/17 05/28/17 05/28/17
A single L-C filter when handed a step-change in current draw is going to ring a bit. I think that the OP was looking for a realistic means to model the amplifier circuit. Not that the step change in current is not useful, but it has limitations. For example, when run w/o a turn on delay, the difference between a resistive load and a constant current one is about 100V on the initial peak.
I am fairly confident in saying that an L-C filter, when given enough capacitance will work quite well. I'd further the Class A bit, with the addition that a PP Class A is going to vary its draw from the PS hardly at all compared to an SE one. The SE being roughly constant when averaged over a complete cycle vs. nearly constant at any given part of it.
When adding a second L-C stage, the step change in current test can be quite useful. I recall a linestage PS I built( and then Sim'd ) that rang at about a half a Hz...nearly no ripple though...LOL
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
"I am fairly confident in saying that an L-C filter, when given enough capacitance will work quite well."
That's absolutely true. But then there are the designers who claim too much capacitance ruins the sound. Other designers prefer oil caps or other non-electrolytic types, often not available in larger values. Given all these variables and self-imposed limitations, there's no substitute for analyzing potential PS filters with SPICE.
--------------------------
Buy Chinese. Bury freedom.
"But then there are the designers who claim too much capacitance ruins the sound."
Presumably such designers are finding that they prefer the sound they get when they effectively introduce a tone control by means of limiting the low-frequency response of the amplifier, achieved by causing the power-supply impedance to become non-negligible at low frequencies. That is fine, if that is the sound colouration that they like. But there must surely be better ways of introducing a tone control for the amplifier than this.
I have nothing against tone controls, but I would think it is better to use a well-designed device with convenient adjustments available for personal taste, rather than a fixed low-frequency fall-off whose slope might, or might not, be to the individual listener's liking.
Chris
"there's no substitute for analyzing potential PS filters with SPICE."Agreed. I've never used PSUD2, but from what I've gleaned, it doesn't sound to be particularly appropriate for trying to optimise the performance of a power supply for an amplifier. As Tre says, the kind of time dependent load that a class A amplifier imposes on a power supply is not the kind that can be particularly well analysed if the only time-dependent thing one can model in the power supply software is the effect of a stepped load. LTSpice is really much more versatile for the kind of modelling that will realistically show the behaviour of the power supply when the amplifier is playing music. I don't see the point of trying to use PSUD2 for this kind of application.
Chris
Edits: 05/30/17
It is fairly easy and since i am usually using second hand transformers, i can run lots of different values of the different trans i have on hand and get close to the circuit specs.
When one ov those 'designers' can point to something sound as to why, I may just pay some attention. I suppose if the PS is playing with the rest of the SE tone machine, it may be possible...:)
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
.
Have Fun and Enjoy the Music
"Still Working the Problem"
My main concern was the proper representation of the power tubes and the driver tubes. But, it is good to know what other key bits one should be aware of. I am more concerned about ps design since in my last amp i somewhat over cooked a 5ar4. cheers, Dak
Did the supply have a critical inductance input choke?
If not then that 5ar4 could have been over cooked by excessive peak currents caused by too high a value first cap.
I've cooked rectifiers tubes and power transformers that way.
What is your audio circuit this time?
It shouldn't be hard to calculate the total current draw.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: