¿ªÔÆÌåÓý

Date

Re: Q: Power Supply Design user group?

 

This page has now moved to

On Mon, Oct 17, 2022 at 01:49 PM, Frank Wiedmann wrote:


For books and papers about SMPS, you might want to take a look at


Re: Can't see the noise of a white noise generator #NOISE

 

¿ªÔÆÌåÓý


Le 27/07/2023 ¨¤ 09:42, John Woodgate a ¨¦crit?:

A bit off-beam, that. The reverse-biased B-E junction is a fairly good noise generator, but I expect that reverse breakdown is not included in the model.

I don't think it is included. I found it the hard way when I simulated a full-wave precision rectifier. It works fine in simulation but does not in practice, because the b-e junction goes into zener when the opamp output goes to the negative rail.


Re: Difference between finding DC point before AC, and pure DC simulation

 

I have also observed this. The reason is probably that any possible inaccuracies in the DC solution of a .tran analysis will usually disappear during the first few timesteps. For a .ac analysis, on the other hand, there is no such self-correction and depending on the circuit, the results might be very sensitive to its DC bias point. So I would guess that the accuracy settings for the DC solution of a .ac analysis are more strict than for a .tran analysis. This matches with your observation that the .tran analysis often finds the DC solution faster.


Re: Can't see the noise of a white noise generator #NOISE

 

¿ªÔÆÌåÓý

A bit off-beam, that. The reverse-biased B-E junction is a fairly good noise generator, but I expect that reverse breakdown is not included in the model.

On 2023-07-27 02:46, Andy I wrote:

Hyu wondered why a transient simulation doesn't show any random noise.

The reason is simple:? SPICE is noise-free in .TRAN analysis.? It always was.? SPICE is deterministic.

You can simulate noise in SPICE by using a .NOISE analysis.

You can also "observe" noise in a .TRAN analysis in LTspice, by adding a B-source (behavioral source) with a white() argument.? This will add a random (but not Gaussian) noise signal to your circuit, but you get to decide how much noise to add, so it will not necessarily behave like the real components (resistors, diodes, or transistors) you're trying to simulate.

? ?Bnnn Nxxx Nyyy V=2*noise(time*2e6)

will generate white-ish noise with a bandwidth of about 1 MHz and a peak amplitude of +/- 1 volt.? But carefully note that the simulation settings (Maximum Timestep and Plotwinsize) may affect how accurately the simulation follows the signal generated by the white() function.

Andy


Re: Can't see the noise of a white noise generator #NOISE

 

On Wed, Jul 26, 2023 at 02:46 PM, Andy I wrote:

SPICE is deterministic
Is LTspice deterministic? For large simulations that have difficulty converging?

-marcel


Re: Reflected power in RF power supplies

 

Min,

Don't think about lumped vs. continuous circuits as?an either-or.? The continuous view is always correct; the lumped one may be close enough but it does not replace the continuous one.

That's true for transmission lines too, e.g., coax.? A real transmission line has continuously distributed L, R, and C.? Regardless of its length, we could instead represent it as lumps of cascaded LRC sections, and it's good enough if the frequency is low enough.? There are reflections on long transmission lines even if you model it as a series of LC or LRC lumps.

For these RF signal generators, it is all the same to them whether they drive (say) a 150 ohm resistor directly, or a mismatched load on the end of a very long cable.? In either case, the impedance seen is X, the SWR is Y, the reflection coefficient is so much, etc.

At the RF device itself, it can't tell whether the thing it's driving has a forward-moving wave and a reflected wave.? That doesn't matter.? We can talk about things like VSWR even if there are no standing waves.? It's just a measurement tool.? Similarly, we can talk about "reflected power" even if there isn't any transmission line..? Anyway, those are abstractions.? What really exists, are voltages and currents, as a function of time and location along the line..? The electrons don't see forward or reflected waves, they just respond to the E-M fields acting on them, instant by instant.

At the output of the RF device, it sees a load impedance.? You can express it in different ways.? A useful one is reflection coefficient.? Even a 150 ohm resistor, connected to a "50 ohm" circuit by soldering it directly to the output pins (hence a lumped circuit model), has something we can measure and call "reflection coefficient".??We can also measure "reflected power", which takes into account both the magnitude of the reflection coefficient, and the output power from the RF device.? And as it happens, the reflected power is a useful metric for telling us how much mismatch the RF device can tolerate.

It might help to think of "reflected power" this way.? It means the same conditions as a long piece of transmission line with a mismatched load, having the same input impedance.

But we are waaay off-topic for this group -- even though the problem of modeling transmission lines (lumped vs. continuous) is important in SPICE and LTspice.

Andy


Re: Reflected power in RF power supplies

 

Thank you very much for all the kind replies.

I have uploaded a photo into the album?reflected_power_of_13.56mhz_generator where the key specs of two Advanced Energy RF generators are listed. It says 'the reflected power limit' is 600 W. My confusion was what is the definition of 'reflected power' in lumped circuits since I feel like that the generator can still be modeled using lumped circuit elements. Andy's explantions are excellent!?


Re: Can't see the noise of a white noise generator #NOISE

 

Hyu,

We should also note here that most transistor SPICE models do not include base-emitter reverse breakdown, which is something your circuit depends on.

LTspice has a BJT parameter for B-E reverse breakdown voltage.? The parameter is "BVbe".? I don't remember if that was unique to LTspice or if all SPICEs had it.? However, many (if not most) BJT models omit that parameter, for the simple reason that almost nobody uses transistors with the base-emitter junction in reverse breakdown.? It is usually something to be avoided.

You'll note that the emitter of Q1 in your simulation is at +9V.? It is not in breakdown in your simulation.? Reverse breakdown is needed to exploit the high noise zener-like breakdown.

Checking standard.bjt, none of the models there have BVbe.? You can add it.? If I remember correctly, group member Bordodynov has uploaded BJT models that include BVbe, when questions such as yours came up in the past.

And there is another hitch.? Even if the model includes reverse breakdown, I don't know if the simulation includes the noise that you get in practice with real transistors in breakdown.? Because it is something most designers avoid, it might be something that was never added to the internal BJT model equations in SPICE.? It's a useful technique to use in practice for generating noise, but rarely ever simulated.? (This refers to the .NOISE analysis that does show transistor noise.)

Another thing to note is that the typical B-E breakdown voltage is on the order of 6 or 7V, if I remember correctly.? So the 9V supply voltage is too close to the point where it doesn't break down.? I suggest using a larger supply voltage, 12V or more.? It might work fine with a 9V battery -- but it seems a little close for comfort.

Andy


Re: Can't see the noise of a white noise generator #NOISE

 

Hyu wondered why a transient simulation doesn't show any random noise.

The reason is simple:? SPICE is noise-free in .TRAN analysis.? It always was.? SPICE is deterministic.

You can simulate noise in SPICE by using a .NOISE analysis.

You can also "observe" noise in a .TRAN analysis in LTspice, by adding a B-source (behavioral source) with a white() argument.? This will add a random (but not Gaussian) noise signal to your circuit, but you get to decide how much noise to add, so it will not necessarily behave like the real components (resistors, diodes, or transistors) you're trying to simulate.

? ?Bnnn Nxxx Nyyy V=2*noise(time*2e6)

will generate white-ish noise with a bandwidth of about 1 MHz and a peak amplitude of +/- 1 volt.? But carefully note that the simulation settings (Maximum Timestep and Plotwinsize) may affect how accurately the simulation follows the signal generated by the white() function.

Andy


Can't see the noise of a white noise generator #NOISE

 

Hello everyone,

I uploaded a file called "White noise generator.asc".? The circuit it's from a book.

I want to check the level of the noise generated but can't see any.

What am I doing wrong?

Thanks in advance
Hyu


Re: Square wave into a bridge rectifier (by member "FlightRisk")

 

Thank you Andy, that makes sense about the nodes and capacitor ground. I would probably have not found that myself. I did try it with real diodes and went back to the ideal diode just to simplify things. I want to work with real diodes designed to handle that current now that it is working. I appreciate the help.

Fred


Re: Reflected power in RF power supplies

 
Edited

Min wrote, "What does it mean when it shows 'reflected power' 100W? Can someone kindly explain it?"

Not without seeing it.

One guess is that 100W is the maximum allowed reflected power that the device can tolerate, being 'sent' back into it from the load that it's driving.

There is a fair amount of disagreement and argumentation about the terms.? Expressions like "reflected power" and "VSWR" (SWR) essentially measure the amount of an impedance "mismatch" compared to a reference impedance.? If you define your reference impedance to be 50 ohms, and if the actual impedance of a device is something other than 50 ohms, then you can also "measure" (calculate) that impedance difference from 50 ohms and write it as a SWR number, or a "reflection coefficient" number, or a?"reflected power" number if you also know what the driving power is.? In the case of an RF power source, probably the "reflected power" is the most important to know, because it causes damage to the source if it is exceeded.

All these measurements do not require any length of transmission line.? You can "measure" or calculate SWR as a number without seeing any standing waves, because there is a unique correspondence between reflection coefficient and SWR, which does not depend on any length of transmission line.? If you can measure the load's impedance, you can calculate its reflection coefficient and its SWR, relative to (say) 50 ohms.? Knowing that, and knowing the driven power, you can calculate the reflected power too.

To everyone who insists that SWR (VSWR) does not exist unless you have at least half a wavelength or so of t-line, my answer is: "Get over it."? The thing that drives that line doesn't know, or care, how long it is!? All it sees is an impedance at the end it's driving.? Given that impedance at the driven end, behavior is 100% identical whether the line is 1 mm long or 100 km long.? VSWR is a convenient measuring stick, not a literal procedure.

Andy


Re: Reflected power in RF power supplies

 

I see 13.56 MHz & think industrial RF heating.

The load impedance probably varies with what is being heated. RF power amps prefer a resistive load. When the load becomes complex/reactive (not purely resistive), the power delivered to the load impedance is not as desired and there can be power reflected back.

To simulate that (basis of this group), one would need to be able to measure that or know from someone else's measurement what a given load's characteristics are.

Or, I'm off base, and the load is actually a resistor and the power dissipated in the resistor is what performs the heating of whatever object or process the RF is used for...and the process has no effect on the VSWR or ISWR on the transmission line between the PA and the load.



On Wednesday, July 26, 2023, <minsun@...> wrote:
Try to understand what 'reflected power' means in a 13.56 MHz RF power supply. The wavelength for 13.56 MHz is more than 20 m. Therefore, I think a 13.56 MHz power generator still falls into the lumped element circuit domain. However,? definitions like 'reflected power' and 'VSWR' appear in the power generator's datasheet. I get confused about the physical meanings of 'reflected power'. What does it mean when it shows 'reflected power' 100W? Can someone kindly explain it?

Thanks a lot,
Min



--
Murray


Re: Reflected power in RF power supplies

 

¿ªÔÆÌåÓý

I think you need to upload the data sheet as a PDF to Files >Temp, and then tell us you did that. We need to read it in detail to see what those specifications are about. Does the manufacturer expect that the power supply will be connected to a remote load by coaxial cable?

On 2023-07-26 15:22, minsun@... wrote:

Try to understand what 'reflected power' means in a 13.56 MHz RF power supply. The wavelength for 13.56 MHz is more than 20 m. Therefore, I think a 13.56 MHz power generator still falls into the lumped element circuit domain. However,? definitions like 'reflected power' and 'VSWR' appear in the power generator's datasheet. I get confused about the physical meanings of 'reflected power'. What does it mean when it shows 'reflected power' 100W? Can someone kindly explain it?

Thanks a lot,
Min


Re: Reflected power in RF power supplies

 

¿ªÔÆÌåÓý

I suggest you direct your question the manufacturer Gambetti.

Le 26/07/2023 ¨¤ 16:22, minsun@... a ¨¦crit?:

Try to understand what 'reflected power' means in a 13.56 MHz RF power supply. The wavelength for 13.56 MHz is more than 20 m. Therefore, I think a 13.56 MHz power generator still falls into the lumped element circuit domain. However,? definitions like 'reflected power' and 'VSWR' appear in the power generator's datasheet. I get confused about the physical meanings of 'reflected power'. What does it mean when it shows 'reflected power' 100W? Can someone kindly explain it?

Thanks a lot,
Min


Reflected power in RF power supplies

 

Try to understand what 'reflected power' means in a 13.56 MHz RF power supply. The wavelength for 13.56 MHz is more than 20 m. Therefore, I think a 13.56 MHz power generator still falls into the lumped element circuit domain. However,? definitions like 'reflected power' and 'VSWR' appear in the power generator's datasheet. I get confused about the physical meanings of 'reflected power'. What does it mean when it shows 'reflected power' 100W? Can someone kindly explain it?

Thanks a lot,
Min


Re: Regarding basic simulation of ACST

 

Hello Andy & Tony,

Thanks for your help. Yes 200 ohm was not able to trigger ACST (mV output), increasing that to 1K in high side switching mode made circuit working though simulation time is quite long. This is auto switching in Q-1 & Q-3.?

Regards,
jagdish


Re: Square wave into a bridge rectifier (by member "FlightRisk")

 

Fred,

Because the file you uploaded mentions "inrush current"? --

Note that the inrush current (and the time to charge the capacitor) depend on what diodes you use for D1-D4.? With the default diode "D", the peak inrush current was almost 13 Amps.? But it might be only 6 Amps or less depending on the diode you pick (and in that case, it takes twice as long?for the capacitor to charge).

Andy


Locked Re: New Simulator Written by Mike Engelhardt -- Released today

 

Remember to discuss QSPICE in the QSPICE group, not the LTspice group.

Andy


Locked Re: New Simulator Written by Mike Engelhardt -- Released today

 

FYI I just tried downloading the released version of QSPICE.? I have Avast anti-virus.? It gave me the following fail message:
"Threat secured.? We've moved xbx0.0 to your quarantine because it was infected with Win32:InjectorX-Gen[TrJ]"


------ Original Message ------
From "Mike Engelhardt" <engel@...>
Date 7/25/2023 10:13:35 AM
Subject [LTspice] New Simulator Written by Mike Engelhardt -- Released today

Dear Group,
?
Today is the day.? My new simulator, formerly called S¡¤P¡¤Q¡¤R, released today as QSPICE.
?
It's been three and a half years of development.? It started out as a mission to get SPICE right: completely rearchitecting the timestep control, eliminating singularities and device IV curve discontinuities, advancing countless aspects of the numerical methods over Berkeley SPICE.? Basically, QSPICE is what I would have written 25 years ago when I wrote LTspice had I known then what I know now.
?
But, with affiliation with IC mfgs, it became more than "just" a superior SPICE.? QSPICE is a mixed-mode simulator that allows massive amounts of digital logic to be presented to your SPICE simulation.? C++(or Verilog) is compiled to native Intel object code.? Odds are, the logic evaluates faster in simulation than hardware.
?
Physicist Maurizio Di Paolo Emilio, Editor in Chief of Power Electronics News and EEWeb, and EE Times correspondent has been grilling me for the last couple months and has a detailed article:
?
? ?
?
The formal press release has hit the wire almost two hours ago:
?
? ?
?
QSPICE is currently in unrestricted beta, meaning anyone can download it without waiting from
?
--Mike Engelhardt
Author of QSPICE