tube rectifier for larger amp? (bass amp)

[R.G.]

It would be interesting to see a simulation of a full-wave CT rectifier HT circuit where the DC secondary winding resistances were mismatched, as normally occurs in practice. Of the PT's I've measured, the DCR mismatch has been below 10%.

Hmmm. Interesting concept. I'll try that when I get some time to play with the simulator.

The effect of that PT DCR imbalance would be worse when using semiconductor diodes, although I haven't come across anyone taking measurements of the ripple at the fundamental frequency, and whether some padding resistance in one diode arm can be used to null out that fundamental.

Yeah, but I suspect that the effect is complicated. Higher resistance in one leg, to the extent that the windings themselves were balanced in terms of the number of turns, would reduce the peak voltage the cap charged to that half-cycle. But because of the oddities of the cap charging to a peak which is determined by the voltage and the currents through the resistances, I suspect that the charging pulse width would expand on the high-resistance side, so the difference in cap peak voltage might not be as much as the difference in resistance. Just a theory.

I'll check when I get a chance.

[R.G.]

I did some simulation. It's, well, complicated.

Complicated enough that I thought I was doing something wrong with the simulator; of course, that is still possible.

I set it up with identical but out of phase 300Vac sources driving 100R resistors through 1N4007 diodes into a 47uF cap loaded by 1K to 2K ohms. Ripple was in the 20-50V p-p range depending on loading. So far, so good.

Then I ran the Fourier analysis on the ripple. I expected the balanced case to show nearly all 120Hz (I'm in the USA, so my simulator runs on 60Hz, not 50z ) and almost no 60 Hz. That was not the case. I got less AC mains component than 2x mains, but not hugely. The 2x mains was about 18V (at whatever loading I happened to be doing when I did the Fourier) but the 1x mains was about 11V. This inconsistency seemed to persist through all the simulations.

I tried imbalancing the winding resistances by 100R to 20R, and that did increase the 1x mains some. The problem is, the high 1x mains component seemed too high on all runs. This odd result largely obscured any conclusions to draw from the change in resistance.

I also tried ideal diodes instead of 1N4007 models. Same result, so it wasn't an oddity about the diodes. I also tried a main voltage source of 300Vac and a voltage-controlled-voltage-source mirror of it for the two source. Same result.

I tried removing the cap and running a Fourier on the full-wave-rectified. 2x mains obligingly leapt to 180V or so, and 1x mains plummeted to much lower than it was when the cap was in circuit. This part matches what I'd expect, and is a check on the matching of the voltage sources and ideal diode models. But it leaves me with the question about why adding the cap to matched resistances and matched diodes added in a 1x mains component.

At this point, it looks like there are competing mechanisms, but I can't tell exactly what they are.

[trobbins]

That does sound counter-intuitive, and needs a bit of thought.

RG, just to clarify any fourier related quirks:
- the fourier period was an integer number of 60Hz cycles?
- the fourier period was over a steady state portion of the simulation (ie. not over any start-up conditions which would indeed include a fundamental component due to charge up pulses).
- a fourier spectrum of the capacitor current pulses contains only 120Hz components?

[gui_tarzan]

I've heard (rightly or wrongly, I don't know) that guitarists love the tube rectifier for the "sag" and diodes don't give that same sag leading to a tighter tone. If that is actually the case, since I play mostly bass I would think that diodes would be preferable in a bass amp. You know, "I like tight bottom ends".

[soundmasterg]

I've heard (rightly or wrongly, I don't know) that guitarists love the tube rectifier for the "sag" and diodes don't give that same sag leading to a tighter tone. If that is actually the case, since I play mostly bass I would think that diodes would be preferable in a bass amp. You know, "I like tight bottom ends".

Tube rectifiers are of course diodes also, but yes you would probably prefer solid state diodes and higher capacitance in the first couple stages too for bass use.

Greg

[R.G.]

- the fourier period was an integer number of 60Hz cycles?

I didn't dig that deeply. I'll have to go look.

- the fourier period was over a steady state portion of the simulation (ie. not over any start-up conditions which would indeed include a fundamental component due to charge up pulses).

I believe so yes, but again, I'll go look.

- a fourier spectrum of the capacitor current pulses contains only 120Hz components?

Didn't try that, but will look.

[Dave H]

Then I ran the Fourier analysis on the ripple. I expected the balanced case to show nearly all 120Hz (I'm in the USA, so my simulator runs on 60Hz, not 50z ) and almost no 60 Hz. That was not the case. I got less AC mains component than 2x mains, but not hugely. The 2x mains was about 18V (at whatever loading I happened to be doing when I did the Fourier) but the 1x mains was about 11V. This inconsistency seemed to persist through all the simulations.

It didn't get that result when I ran it in LTSpice (at 50Hz mains ) Balanced (2x100R) there wasn't any 50Hz, unbalanced (100R,110R) the 50Hz was about 20dB down on the 100Hz or about 7V 100Hz and 0.65V 50Hz when I switched to linear y axis. I don't really know how to run the Fourier analysis so I'm not sure how accurate it is and there's a nasty slope on the baseline to compensate for.

[R.G.]

Then it was probably a quirk about how I ran the Fourier analysis. Your results are much more like I'd expect.

[Dave H]

Quirky it is. Look at the slope on the baseline. The second 'linear' one looks better.

[trobbins]

I don't think there is anything amiss with the noise floor - the lin vs dB difference appears correct. The noise floor response in fourier results is often dependant on the detailed setup of the fourier calculation process, with the main influence being the window type used (eg. Hanning). You can see that difference very markedly when using a software oscilloscope style soundcard input to a program like Room EQ Wizard V5, or Visual Analyser - both great programs for looking at real time spectrums, whereas programs like TrueRTA just use a default window for their spectrum analysis.

[gui_tarzan]

Tube rectifiers are of course diodes also, but yes you would probably prefer solid state diodes and higher capacitance in the first couple stages too for bass use.

Greg

Of course you are correct, I should have said "silicon" diodes. Seems to me it is good enough for about half of the vintage Fender amps and cheaper than a tube. However, is there a real current/noise difference between the two?

[R.G.]

There may or may not be a noise difference between them. We're still hacking on that issue.

There definitely is a current difference between them. Tubes of all kinds are limited in how much current they can boil off the cathode without damaging the cathode. Too much current can shorten tube rectifier life dramatically. This is the reason that all tube rectifiers have a specification for the maximum value of the first filter cap. This cap's value largely determines the size of the current pulses that go through tube rectifiers. Too big a cap, the rectifier gets more or less slowly damaged.

Solid state diodes also have the equivalent issue, the maximum repetitive peak current they can withstand. However, the repetitive peak currents for solid state diodes is hugely larger than that for tube rectifiers, even relative to their average current values. So solid state diodes have no reasonable limit to the size of the first filter cap. This is especially true in a standard guitar amp where the resistances of the high voltage windings limit the pulse currents themselves.

You can use bigger caps with solid state diodes, and that makes for a stiffer B+ supply. That can change the effective "feel" of the amp, anecdotally most in the bass regions. The bigger filtering may also make a generally lower hum/ripple floor, and may also cut down on interstage coupling, so making for a quieter, less quirky amp. These last are subtle and layout/wiring dependent as well.

Silicon diodes of the "general rectifier" category can also have issues with suddenly snapping off when they turn off, and this emitting a blat of RF noise that gets picked up in the rest of the amp and appearing like a bit of buzzy hum. The cure for this is to fast, soft recovery silicon diodes, or to use R-C snubbers across general purpose rectifiers. No hill for a climber.