Ready to go SMPS for 15W tube amps!

[Cliff Schecht]

There are magnetics in that output stage design, just not your typical output transformer. They are doing some kind of impedance conversion using some switches (FET's?) and probably custom magnetics.

It's just a neat idea at this point, for that price it doesn't appeal to many in the consumer market. He could make a much cheaper version but pride and greed can get in the way of making money sometimes..

[cxx]

I read a paper that looked at replacing the power supply in a bassman with a high quality lab supply with zero ripple. The result was a very sterile poor sounding guitar amp. The conclusion was that it was the lack of modulation that made it sound bad. I can't put my hand on the paper now, but it was one of the few academic papers on guitar amps that I've seen.

Found what I was thinking of: http://www.ampbooks.com/home/classic-ci ... AB-ripple/

I think that the supplies on the crate v series were replaced by switching ones when manufacturing moved overseas. The consensus was that it ruined what was a reasonably nice cheap tube amp. I had several of the VC series which I rode into the ground and used for parts. Those were nice.

[LeftyStrat]

I have seen that paper and this is definitely a concern.

I actually think I'm going to build a couple of Rockets, one with this supply and one with a traditional supply, in order to compare the two.

[Milkmansound]

Wow it's a lot smaller an I expected it to be! Might fit into a 1U rack?

Yep will fit in a 1u, and probably is less than a pound. The weight savings alone is enough for me to want to make it work.

Interesting. Might be just the ticket for a 1U tube reverb

[LeftyStrat]

Found what I was thinking of: http://www.ampbooks.com/home/classic-ci ... AB-ripple/

I just had an interesting thought about this. 60 Hz and 120 Hz don't match up with any guitar frequencies. Low A on a guitar is 110 Hz. If you were tune a guitar so that A = 120 Hz, (or 100 Hz in places where 50 Hz is used), it should probably sound like crap, at least at all multiples, since there should be some phase cancellation. Or at least different volumes for A, depending on the phase of the note in with relation to the grid cycle.

I'm also wondering if "swirl" might be attributed lowered filtering causing the beating of the different frequencies reinforcing and canceling each other.

What would really be interesting is injecting a variable frequency ripple on to a well regulated supply. Might there be better frequencies than 100 or 120 Hz?

[Cliff Schecht]

Lefty you are thinking about CMR (common-mode rejection) which is an amplifiers ability to reject signals common to both the plus and minus inputs. Although we are running our PI's in a single-ended to differential type fashion, any 60 Hz or 120 Hz noise is common to both sides of the PI and is rejected by the differential action. If the noise isn't the exact same on both tubes (either on the plates or grids) then you don't get perfect cancellation (same if the triodes aren't matched, resistors aren't matched, etc as these mismatches all reduce CMRR).

And we don't look at the wall AC phase compared to our signals phase because by the time our amplifiers actually see any voltage, it's already been rectified and filtered. Push-pull stages don't require as much filtering because the push-pull action of the amplifier cancels any noise common to both plates/grids of the output stage (same with the PI). Since our signals are "differential" within the PI, the noise gets subtracted out while the signal gets amplified.

[LeftyStrat]

I'm thinking more about that ripple on preamp tubes. Especially those biased to clip one side of a waveform more than the other. The amount of flattening of the waveform would vary with the ripple. And I really know nothing, just kind of thinking out loud.

[LeftyStrat]

http://www.ampbooks.com/home/classic-ci ... AB-ripple/

Just want to know why this is the case.

[diagrammatiks]

It's how the ripple is fixed in amplitude.

There's still ripple on a regulated supply unless it's very very well filtered.

the difference is that the ripple voltage and the high voltage are no longer modulated completely by current draw demands downstream of the reservoir.

the node and ripple percentage are fixed at whatever regulated output the supply puts out.

its also specific to class A/B amps as well.

in a single ended amp or class A the power supply reacts very differently.

[Cliff Schecht]

I'm not at all satisfied with the conclusions made in that paper. No FFT's of any waveforms and no explanation as to the difference in sound between 50 Hz and 60 Hz sources. They jump to the conclusion seemingly right at the point where they should start digging further. It would not be hard to artificially introduce this ripple into a switching supply and characterize how much is needed to get "that sound".

I'm not convinced that "the" sound of a 5F6A is caused just by the ripple supply. It seems like while they talk about the transformer/rectifier impedance they focus primarily on the voltage sag in their testing and neglect to even measure the current draw from the amp on either traditional or experimental supplies.

I think the reason the SMPS took the nuts off of their amp is because their supply was too perfect in the way it delivered current to the load. This is where the internal impedances of a transformer/rectifier really determine how an amp is going to react when asked to deliver a full-power transient. Looking at the 5F6A voltage waveform shows that bit of wiggling as it tries to get back to a steady state point but they don't show what the current is doing. It's like we're only getting half of the story..

Am I missing something or are these results somewhat inconclusive and unscientific?

[diagrammatiks]

I would have definitely liked to see a follow up study.

I think that saying that changing the power supply significantly changes the sound of the amp is a pretty obvious conclusion.

Increasing the filtering a lot also has significant changes.

I know that a lot of hifi amps use regulated high voltage supplies so I dunno.

best buy one of those supplies and a/b it.

[cxx]

The web page did seem a little subjective in its assertions. But I can't read German and didn't translate the original paper linked to on this page. Does anyone read German?

What I got out of it was that they implemented a no ripple supply that had the same sag characteristics the original supply and it sounded bad. Of course the sounded bad part is certainly subjective. I'm sure some people would like the sound (probably jazz players).

I don't find their assertions off the wall though. Modulation of sounds seems to be fundamental to perception. The whole basis of what notes sound good together and 10 million different stomp boxes seem to be based on it. Can you capture the differences in a FFT? Probably. You certainly can capture the excessive effect of ghost notes that way. I'm not so sure that you will be able to distinguish between the desirable and undesirable amounts of modulation just by looking in the frequency domain.

Switching supplys are different than the one in this paper. Correct me if I'm wrong, but I understood the switching supply to work at much higher frequencies to reduce the magnetic component costs. It does this by chopping a rudimentary DC at this high frequency and regulating. I don't think any 60/120 Hz gets through.

Sounds like we'll get some closer to home info from those who are building with these supplies. It will be interesting to see.

[LeftyStrat]

This is Google's translation of their conclusions:

The sonic results

When plugging the guitar into the amplifier was found, however, that the guitar introduced hum compared to that in the original AC heating unit interspersed hum is so strong that no serious subjective improvement of Ripple behavior is responsible.

The sonic results with the new, hum-free anode supply was, however, Surprisingly, very disappointing.

Despite the almost complete agreement of the curves (with the exception of the 100-Hz Hum), the amplifier is no longer recognizable sound, the sound has completely changed.

The amplifier loses the "lightness", the "Touch Sensitivity", the "liveliness" and "sharp" "Bizzeligen" overtones. The playing is not as much fun, it is as if one more must strike and the sound comes a little later, it's a little tired, it's a strain.

To "side effects" be excluded, such as changes in the negative Bias has been built an adapter that allows you only supply between the anode the original power supply and the low noise power supply can switch back and forth. For this purpose, the Anode supply output of the "new" power supply to an octal plug set, then instead of The GZ34 in their version of the amplifier circuit board is inserted. Thus, a change possible between the two power supplies in minutes, so that the previous sound impression is still fresh in your mind so that you can compare the sound impressions better.

The effective connection with the "new" power supply eliminating the burden attributable to heating the GZ34 and the attributable burden anode supply side leads only to a very low Increase of the negative grid bias of the output tube of the order 0.5 V, which after We estimate did not sound critical, the change is about 1% lower than the normal fluctuations in line voltage.

This result is surprising, since the waveforms of the anode supply for Strain, except for the omission of the 100 Hz hum, same. It was investigated how a change of the internal resistance of the sand Affect sound impression by varying the bias of the triode was 6080. Here, a very strong influence of the sound impression noted.

If the grid bias of 6080 has reduced so much that the supply voltage of the Amplifier under load remains virtually constant, then the amp sounds "lax" and "cheap" about the same as a 200 €-transistor amplifier electronics from the wholesale market. In all settings of the bias and thus loses the amount of voltage drop in "Bassman" its characteristic penetrating, he sounds less aggressive softer. From the "WILDCAT" is a well-behaved house cat.

The conclusions

Currently we can not yet explain the observed effect conclusively. The sensitivity of the Sound impression to (seemingly) smallest changes on the power supply is surprisingly and was not anticipated by us.

The most likely explanation is that the 100 Hz modulation of Gitarrentons with greater Modulation (from the use of the limit) not as a nuisance, but as a sound-forming element must be understood. Presumably this modulation gives only the sounds of their Bassman Penetration and "Bizzeligkeit", it may possibly be a very fast "tremolo" . understand

It follows then the question that then it really a different sound would be between the U.S. and Europe resulting images, as the Line frequencies of 60 Hz and 50 Hz are different. This results in a frequency of the modulation of 120 or 100 Hz

It gives the following reference to the frequencies of the tones:
G 98 Hz
Gis 103.8 Hz
A 110 Hz
B 116.54 Hz
H 123.47 Hz

The question here is whether an audible effect has yet to be clarified. It would be entirely possible be that one for a "totally authentic" sound impression of the amp is best for a Converter fed at 60 Hz.

Furthermore, it could play a role, that the dynamics and the mean values ​​of the anode supply by the time-delayed effective influence of the internal resistance of the Triodenstufe supplying power supply module at another time conditions as those with whom Our tests were conducted, compared with the original unit of differences could result.

The idea of low-noise power supply has initially proven to be misguided. It serves its actual purpose, the avoidance of hum, not because this essentially by Coupling in the guitar itself is created. At the same time the sound, probably due the absence of no al disorder to be considered 100 Hz modulation of the output signal at affect larger modulations and / or highly misunderstood by even effects.

The results obtained are nonetheless valuable and bring us further. For they show with Insistence, the importance of the details of the power of the sound formation.

It is the engineer's still the thinker once difficult to accept this result
The familiar world view must be expanded. One also sees the importance of self-critical and comparative work, especially in audio and music area. One must in the course of the Work process oriented and over again on a good reference.

[es345]

So an interesting test would be to use a "clean" powersupply, to inject a generated comparable hum via G2 modulation and then to compare the sound via A/B comparison. In a second step one could change the frequeny of the injection.

I need more time ....

[LeftyStrat]

So I started looking into building something with this, but the biggest problem is how to mount it. If I mount it outside the chassis, the HT will be exposed. But putting it inside the chassis takes up a bit of real estate.