PSUD2 is not a full circuit simulator. It cuts corners. In this case the corner it has cut is that the constant current load still draws the current even when reverse biased. In real life this is unlikely to happen. You can ignore the warning, apart from recognising that it shows that real life is different from simulation.Max999 said:
Don't put too many CR sections in cascade. Newbies do this to get ultra low ripple (in simulation) but don't realise that in real life the ripple comes back via other means (e.g. poor grounding) and too many CR sections mean that the PSU has very high DC and subsonic output impedance.
I am confused about this one. So say we have a string of resistors of 1k, 18k, 18k, with appropriate capacitor size, hanging another 18k and cap for the first stage is going to give a high dc output impedance?
DC resistance is the sum of all those resistors (plus whatever effective resistance is inside the power supply itself).
We rarely think about this in tube guitar amp design, because there is lots of extra B+ available to waste, unlike, say, an FX pedal operating on 9 V DC.
-Gnobuddy
Yeah but it already has 37k in line between the rectifier and the last filter stage. Would adding another 18k, bringing up the total to 55k, change how the amplifying stage on the last filter cap operates? I think the cap would swamp out the 55k with its esr controlling what the stage sees.
Not at any reasonable AC frequency. But the quiescent (dc) operating point will change, because B+ will fall. Anode voltage will fall with it, maybe not enough to matter, but it will fall.
The thing is, all capacitors have infinite reactance at DC, no matter how big they are. So at DC, the capacitors are all open-circuits, and don't exist in the equivalent circuit. All you've got left is that string of resistors in series.
-Gnobuddy
The power supply is almost irrelevant because the valve will take 30 seconds to start working.
But that is how you get rid of motorboating, isolate the stages from each other's power supply.
It's all about the criterion for oscillation: if you have both unity gain, and sufficient phase-shift to have positive feedback, then you will have oscillations. This is true both for motorboating, and for high-frequency oscillation.
If you can push the additional phase shifts from additional RC power supply filters down to such a low frequency that you no longer have unity gain in the amp, then you won't have oscillation (motorboating).
My limited experience with motorboating mostly relates to my attempts to design and build my own cassette deck electronics in my teens. Cassette playback heads only spit out about one millivolt of signal maximum, so bringing that up to speaker level (or even line level) requires a lot of voltage gain, and it's easy to end up with instability or outright oscillation.
I could sometimes cure high-frequency oscillations with additional RC stages in the positive power supply rail. But I rarely managed to cure motorboating this way. Reducing the subsonic bass response of some stage in the amplification chain, however, often worked.
-Gnobuddy
I have seen that too. Same explanation, less capacitance moves up the frequency at which you get 180 total phase shift, and if this subsonic frequency moves up high enough, then the amplifier may still have enough gain to start oscillating (motorboating.)
So if we either increase the filter cap (move down the 180-deg frequency low enough for the amp to have less than unity gain), or reduce the infrasonic gain of the amplifier so it has less than unity gain even with the reduced filter cap, the motorboating will stop.
-Gnobuddy
The isolation has to work effectively at the motorboating frequency. Basically you have to ensure that the amplifier circuit has enough LF rolloff (but not too much phase shift) that the PSU has low output impedance (and low interstage coupling) at all frequencies where the amp has gain.Printer2 said:
But this is the case given one, two or five filter stages. It may be just me but I fail to see the dc resistance going back to the transformer as an important factor given the gain rolls off before the impedance of the PS node goes up..
For a preamp I do not understand why one would want a FAST power supply. It won't make your preamp FAST at all. In fact what you want is a very slow power supply with big caps at the end.
In the Music Reference RM-5 and Beveridge RM-1 I used regulators to deal with line variations coming through to the output. One must consider the low frequency "bounce" that is on the power line which often has a time constant of several seconds. A regulator is the best way to do this, but lots of parts.
One day I had a Marantz 7 on the bench and did my usual of watching the distortion trace (thats like a 1000x magnifier on the output). I ran the variac up and down a few volts to simulate a moving power line. It just sat still like it was regulated, but it wasn't. The trick is simple, make the power supply move SLOWER than the output capacitor can pass. So simple. Look at the schematic of that preamp sometime. The filtercaps go through many stages with rather large resistors between them with the phono section being the last stage.
In the Music Reference RM-5 and Beveridge RM-1 I used regulators to deal with line variations coming through to the output. One must consider the low frequency "bounce" that is on the power line which often has a time constant of several seconds. A regulator is the best way to do this, but lots of parts.
One day I had a Marantz 7 on the bench and did my usual of watching the distortion trace (thats like a 1000x magnifier on the output). I ran the variac up and down a few volts to simulate a moving power line. It just sat still like it was regulated, but it wasn't. The trick is simple, make the power supply move SLOWER than the output capacitor can pass. So simple. Look at the schematic of that preamp sometime. The filtercaps go through many stages with rather large resistors between them with the phono section being the last stage.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.