Building Started...

[rlyach]

jimiclow,

Thanks for the info. I ordered a couple of knobs as well. I was not aware of the size difference. 

[Paul Joppa]

It is unclear to me what exactly is being plotted, so I can't comment on that.

However, I will say that a major virtue of switched attenuators is that they have excellent tracking. Normal pots are quite poor; linear ones claiming 10% and audio-taper usually being worse. Premium pots specifically intended for audio rarely spec better than 5%. But switched attenuators almost always use 1% resistors.

The disadvantage of switched attenuators is cost. Reliable switches are expensive, and switches with many positions are likewise expensive; the combination of both  is expensive indeed. Goldpoint products seem costly relative to potentiometers, but among available switched attenuators their products are remarkably affordable.

In our experience, even cheap switch contacts have no sonic problems - even the crappiest switch sounds good as long as it continues to work at all. This does not appear to be the case for the sliding contacts of potentiometers, though there is little available data beyond anecdotal evidence of the form "brand A sounds better than brand B."

None of the above addresses the effects of the resistive element. There, you get into subjective assessments that don't correlate well with things that are measured and/or quoted in specifications. A search of the web will provide a variety of opinions ...  :^)

[rlyach]

Paul,

Thanks for weighing in. The graph shows channel to channel matching over the entire resistance range of the kit POT and an RS Alps POT. I set the pot and measured the resistance of each channel and then moved on to the next measurement point. The % error is the difference between the channels as calculated against the average of the channels. The db error is calculated assuming a 2V signal and looking at the Vout error of the pot. At very low values of resistance (high volume) the error is small in the db scale. At low volume levels a large % mismatch will lead to a large db difference between the left and right channels. I hope I explained the graph adequately.

[Doc B.]

None of the above addresses the effects of the resistive element. There, you get into subjective assessments that don't correlate well with things that are measured and/or quoted in specifications. A search of the web will provide a variety of opinions ...  :^)

Yes. The Goldpoint populated with Dale resistors is fairly widely regarded to sound better than the miniV with surface mount resistors. I imagine the resistance value tolerances are similar. And I concur based upon our recent adventures designing the BeeQuiet attenuator, the resistor type seems to have a lot more influence upon the sound than the switch contact material. Of course then there is the added twist of which form of attenuator is chosen, series, ladder, shunt, etc. Like anything else in audio you can analyze this stuff into the ground and still never have a clue about what sounds good until you actually start cutting, trying, and listening. And of course like anything else on the internet there is an excess of opinion on the subject and a startling percentage of those opinions posted are from insane people. Which I see as all the more reason to just buy some parts and try stuff myself.

[Paul Joppa]

Hey Randy-

Perhaps it's the residual of Valentine's dinner  :^)  (yes, homemade wine was involved...), but this is still not clear to me.

Assuming R1 is the series part of the pot resistance, and R2 is the shunt to ground, then if I understand you correctly:

"% error" = 2*(R2A-R2B)/(R2A+R2B)

However, the absolute resistance is not the same as the attenuation of the pot.

I still don't know what exactly is the "Vout error" that goes into the dB calculation.

Vout/Vin=R2/(R1+R2)

and there is no consideration of R1 in the above "% error" expression. I would suggest that:

dB error = 20*log10(R2A/(R1A+R2A)) - 20*log10(R2B/(R1B+R2B))

but I can't tell whether that is what you used.

I know this is a picky and technical question, but I sense that you are a technically inclined person, so I am pursuing it - let me know if I am getting overbearing; it's something I am unfortunately prone to do!

best,
-Paul J.

[rlyach]

Paul,

The percent error calculation I used is exactly the same as what you showed. The db error that I calculated is the db difference between the channels with respect to the intended db attenuation. I simply took the intended attenuation and multiplied by the percent error. The way I did it may not be the correct way to look at the db error. Your formula shows the db difference between channel A and channel B, which seems simpler and more straight forward. Here is another plot of db mismatch plotted against the average db attenuation rather than resistance. It shows that the BH supplied POT is better than the RS pot. Unfortunately I did not take more high resistance readings on the RS pot so I can't see its performance at very low volumes. I will repeat this study once I get my Goldpoint attenuator, and I will make sure I add enough data to fill the entire range.

Thanks for the help. I am not an audio engineer so I am still making sense of all this. Most of my experience is with integrated circuit design and manufacturing. I don't mind in the least that you are pushing me. I love learning and find amplifier circuits fascinating. Thanks for the help.

[rlyach]

Now I get to be embarrassed. After looking at my data I realized that audio pots are NOT symmetrical.... I confused R1 and R2. Here is the comparison again using the proper POT orientation. Now it is very clear that the POT shipped with my kit is better than the POT that I purchased at Radio Shack. Also, the standard POT only has a max mismatch error of .65db, and only at very low levels. This has been very educational.

Update: The graph has been updated with new and complete data.

[rlyach]

Well... I am waiting for my attenuator and switch to come in so I have had to stop. Here is my progress so far. This kit is really fun to build. The manual instructions are very easy to follow. I love the high resolution pictures. You can zoom way in to look for more detail.

[Doc B.]

Looks good!

[jimiclow]

Randy, are you sure you don't want to go the 45 route? You can still put a 2A3 in a 45 amp.

[rlyach]

Jimiclow,

At the risk of displaying my ignorance and starting another long discussion, I will say that I chose to wire for the 2A3 because of power. I will be driving a pair of Orca  speakers and wanted the extra 2 Watts or so. I did look at the bias point of the 45 setup and it was a little confusing to me. I measured the DC resistance of the plate choke red wire to the black wire. It read ~260 ohms. Based on my calculations, that puts the bias point at 57V on the cathode and 35mA of plate current. Assuming the same ~4K load from the output transformer I calculate a max of 2.5 Watts at 4.9% distortion. I have already purchased a pair of JJ 2A3-40's (match) just for this amp, so I don't think I will switch to 45's.

Do the 45's really sound better. From a purely design perspective (from an admitted novice), the specs seem the same or slightly lower at a lower power output. Also, If I wire for the 45 and put 2A3's in the amp, the bias current will be only ~35mA and the usable range of the plate voltage will drop by about 50V. It will also reduce the output power as compared to the 2A3 wiring.

Now I am waiting for the someone with more experience to set me straight. 

[Paul Birkeland]

Assuming the same ~4K load from the output transformer I calculate a max of 2.5 Watts at 4.9% distortion.

Now compare the sound pressure levels available from your Orcas with 2.5 watts vs. 3.5 watts.

[rlyach]

That would be an interesting test, but Blumenstein Audio does not publish any sound pressure vs power data. They conservatively spec the Orca sensitivity at 89dB at 1Watt and 1Meter. Just for clarity, I calculated the 2A3 maximum power output at 4.4 Watts as compared to the calculated 2.5 Watts for the 45s. These calculations were based on the published plate curves for the respective tubes. Also, this number uses the full swing of the tube which includes 0V bias, and that is not a good place to operate, so it is better to compare the output power using the specs published for the two different tubes used in the Stereomour, which are 1.75 Watts vs 3.5 Watts. Those are the numbers that drove my decision.

[Paul Birkeland]

You can calculate the difference in SPL without knowing the efficiency of the speaker. 

[Paul Joppa]

...and it's close to 3dB, which is audible but - shall we say? - not huge. I'd go with the 2A3 myself for Orcas.

Yes, by the way, the available power is in theory a bit greater than what we quote, or what is quoted in the spec sheets. Chalk it up to the transformer losses, about 10% (0.5dB).

Also, the RCA standard (250v, 60mA, 2500 ohm load) is 3.5 watts and I don't want to quote a higher number, people will just think I'm exaggerating. But in fact, at 300v/50mA into 4000 ohms, the theoretical efficiency is around 20% greater - very close to your 4.4 watts in fact.

For what it's worth, there are two "standards" for power. One is the power at clipping, defined as a barely visible flat top to the sine wave output, the other usually as 5% THD. Neither is all that precise, since with no feedback the linearity of the individual tube has a large effect.