[skyskraper]

do they do tube matching at ae also?

 

[AtomBoy]

AE does do tube matching, but apparantly not on the tubes we're using. It doesn't show up as an option on the order form. They're cheap enough that I'm ordering 4 of each for spares and swapping.

About the power supply- I'm planning on upgrading from the wall wart eventually- I've got an Ebay find on the way. The cheap option from Digi-Key will get me started and be good enough for the amp I'll give to Dad for Christmas!

 

[drewd]

Quote:

Originally Posted by skyskraper with a well tuned discrete buffer is it possible that the offset would be low enuff that the caps could be gotten rid of entirely?

Unfortunately, no, because the plate voltage is around 12 volts and that voltage sits on the input to the buffer, so it's on the output, too. Since the amplifier operates on a single voltage power supply, the plate voltage serves sort of the same purpose as the virtual ground circuit in many transistor amps (although it's there for some other reasons, too). Pete's original article touches on this a little bit, but not with much detail.

If you're thinking in terms of biasing an opamp and wondering why that voltage doesn't appear on the output of the buffer, the reason is because of the way that an opamp works. For example, the PIMETA or PPA cascode pulls the output of the opamp (and thus the input of the buffer) toward V- (in terms of the virtual ground). The opamp then produces whatever voltage is necessary to offset the cascode, causing the voltage on the output of the opamp (and thus the input of the buffer) to be zero (when referenced to the virtual ground). Thus, no DC offset to overcome.

The hybrid amp is a single ended triode with no feedback, so whatever voltage sits on the plate (the output of the tube) is not nulled, so it also sits on the input to the buffer. That voltage has to be there for a couple of reasons: to provide a relatively high AC impedance on the plate for high gain and low distortion and to simulate a dual power supply for the buffer. Remember, what we call ground on something like a PIMETA is really just V+/2 - it's just a reference. But in a tube amplifier like the hybrid, ground is really zero volts. Since the buffer needs some voltage to swing around, the plate voltage serves that purpose nicely.

Does that make sense?

-Drew

 

[Finch&Music]

Quote:

Originally Posted by drewd I thought I'd also speak up about output cap selection. I've got two hybrid amps running now, one built with budget parts (Panasonic caps and pot) and the other with semi-pricy parts (Elna Silmic-II and Wima caps, ALPS Blue Velvet pot). There is a difference and it's noticeable. A I have to say that if it's noticeable with my ears, then it's probably a big difference. The second amp, with the expensive parts sounds better, with quite a bit more detail and sense of space. That being said, the budget amp sounds pretty darned good. After poking around with amp number three, I believe that the significant difference lies with the output coupling caps, which shouldn't be too surprising. The Panasonic FC and poly caps just don't perform as well as the Elna Silmic-II and Wima poly caps. Amp number three with Panasonic caps on the power side and the Elna/Wima combo in the audio path sounds just as good to me as the full Elna/Wima amp. That's with 12FK6 tubes and a STEPS running at 30V. Grid bias is at 13.8V. My off the cuff feeling is that if your aim is to get into a tube amp on the cheap and you want to go with the budget parts, you'll do just fine here. If you want to spend a few dollars extra and go for the upgraded output caps, you'll get the most bang for the buck there. If you go all out and outfit the entire amp with them, I don't know if you'll hear the difference - but I'll be the first to point out that after spending 10 years duking it out with US Navy sonar systems, my hearing isn't what it used to be! -Drew

Drew, a very interesting and enlightning post. I am glad you post it in this stadium. Most of us are still in the process of choosing and/or gathering components, so you left some choises.

I think you are quite right about the crucial point about the output couplingcap and as you already said it is not surprising at all.

However you are the only one yet to have listen in reality to its quality(-differences), but I keep in mind that much expierence on output cap quality is also applicable on the millet amp(-board).

I have again some questions/suggestions:

1. In the output stage there is one resistor and one combined cap functioning in the signalpath. Is it possible that also the resistor quality is of influence? Could you try already some different resistors of more than middle of the road quality?

2. In other tube amp configurations you see also combined output cap's. In the millet amp there is the combination of one big electrolytic with one small 0.220 polyprop. The last one is a low value filmcap. I have seen combinations where they use one very substantial filmcap parallel to one even more substantial electrolytic (bipolar or non-polar), bypassed by some small filmcaps. Do you expect any benefit of that sort of combinations?

3. If you expect benefits of other combinations is it possible to look again to the board layout, because I have observered that there are very good filmcaps (in a small box) with substantial values, but they are bigger than now can be put to the board in the outputposition. For example Evox/Rifa or Wima have beautiful polyester or polyprop box-caps at lagere values, but they have a leadspace of 27,5 mm or sometimes even a little greater (also their fysical sizes are greater). May be a little late, but another look at the board might be benefical to prospective builders.

Greetings

PM: By the way (question of a non-expert): is the millet amp a output transformerless amp (OTL) in the strict sense of the theory? Or have OTL amps also no outputcaps?

 

[dsavitsk]

Quote:

Originally Posted by drewd After poking around with amp number three, I believe that the significant difference lies with the output coupling caps, which shouldn't be too surprising. The Panasonic FC and poly caps just don't perform as well as the Elna Silmic-II and Wima poly caps. Amp number three with Panasonic caps on the power side and the Elna/Wima combo in the audio path sounds just as good to me as the full Elna/Wima amp. That's with 12FK6 tubes and a STEPS running at 30V. Grid bias is at 13.8V.

fwiw, I exchanged the 100uF elna's for 470uF Nichicon Muse KZ's (from Percy) and it made a pretty big difference. Whether the difference is the size or the brand, or both is anyone's guess, however. The point, however, is that Drew is right that the output caps matter.

 

[dsavitsk]

Quote:

Originally Posted by Finch&Music 1. In the output stage there is one resistor and one combined cap functioning in the signalpath. Is it possible that also the resistor quality is of influence? Could you try already some different resistors of more than middle of the road quality?

The resistor quality probably matters, but it probably matters a lot less than the cap selection.

 

[drewd]

Quote:

Originally Posted by dsavitsk fwiw, I exchanged the 100uF elna's for 470uF Nichicon Muse KZ's (from Percy) and it made a pretty big difference. Whether the difference is the size or the brand, or both is anyone's guess, however. The point, however, is that Drew is right that the output caps matter.

The difference is almost certainly in the size. The output caps, in combination with the output resistor and the headphone's impedance, will set the low frequency corner, that is, the -3dB point. It's a first order high pass filter. That means that the rolloff is going to start at ten times the corner frequency. For a 300 ohm headphone, the -3dB point is around 33Hz, but the rolloff starts at ten times that, or 333Hz. For a set of Grados (at 32 ohms), the -3dB point is close to 310Hz, with the rolloff starting at 3KHz - uh oh!

The 470uF cap moves the cutoff substantially - to 7Hz at 300 ohms and about 66Hz at 32 ohms. For low impedance cans, it wouldn't be out of line to use a 680uF cap, I suppose. I've got 470uF caps in mine for my DT880s.

To figure the cutoff frequency on your own, the formula is just 1/RC, where R is the impedance of your headphones and C is the capacitance of the output capacitors.

 

[amb]

Quote:

Originally Posted by drewd To figure the cutoff frequency on your own, the formula is just 1/RC, where R is the impedance of your headphones and C is the capacitance of the output capacitors.

drewd, you forgot something in the denominator... the formula should be:

f = 1 / (2 * pi * R * C)

For a 100uF output cap, -3dB point is 5.3Hz for 300 ohm load or 49.7Hz for 32 ohm load. For 470uF, it's 1.1Hz and 10.6Hz, respectively.

 

[drewd]

Quote:

Originally Posted by amb drewd, you forgot something in the denominator... the formula should be: f = 1 / (2 * pi * R * C) For a 100uF output cap, -3dB point is 5.3Hz for 300 ohm load or 49.7Hz for 32 ohm load. For 470uF, it's 1.1Hz and 10.6Hz, respectively.

Doh! I knew that I was stretching back too many years for my signals and transforms class!

Thanks!

-Drew

 

[aeroes]

Will a TREAD sufficient enough to supply the Millet?

 

[jamont]

Quote:

Originally Posted by Voodoochile As for the PSU, while a wall-wart is okay, a decent linear makes a marked improvement. Jamont mentioned this earlier also. I found an open-frame linear (Sola or Power-One) 28v PSU for about $15. These retail for closer to $70, and they show up in surplus and on ebay fairly often.

An opamp based amp, like PIMETA or similar, will have a much better PSRR than a discrete amp like this one, so putting some extra bux into a good PSU makes sense. I found an audible improvement when I replaced the unregulated wall-wart with a good regulated supply. A cheap way to do this would be to get the CUI wall-wart and add a TREAD to regulate it, leaving out the rectifier.

 

[skyskraper]

thanks drew, that did make sense. i kinda worked out what was going on from his article but i didnt realise all 12v were present

 

[Jaypetermen]

Quote:

Originally Posted by blip ... I guess I could always use the millet as a pre-amp for another gain stage... Like maybe for my pimeta or some kind of simple voltage gain stage.

Can the Millet-Hybrid amp be use as a pre-amp as is or does it need to be modified.

For example, could you use it pre-M3?

 

[Jaypetermen]

Quote:

Originally Posted by drewd That's with 12FK6 tubes and a STEPS running at 30V. Grid bias is at 13.8V.

I am about to start building a STEPs & M3. I currently have the AD8610AR op-amps (which are rated up to 27VDC) and I am also thinking about building a Millit-Hybrid, since I could not resist the group buy.

I would like to configure the STEPs to be able to power both amps (one at a time of course) without changing the voltage. Would 27VDC be a good setpoint for both of the stock amps?

 

[skyskraper]

Quote:

as far as making the amp work as a preamp, you should not have to do anything to it. The amp should do a great job driving a high impedance load like a power amp. Output cap selection becomes pretty easy - anything over 47uF will work just fine. Pete briefly touches on that in the AudioXpress article available on his web site.

with the right transformer selection you should be able to power both of them at the same time using the steps.

m3 < 500 ma. hybrid < 400 ma. the lm317 in the steps can push out 1.5amp (on paper) so you would have plenty of room to jiggle.

and maybe for safety's sake, setting the voltage at like 24 or something that isnt bordering on the max for the m3 op amp would be a good idea. fried chips =