[L0rdGwyn]

By my calculation, a pair of 6528 in the Glenn gets down to about 12.25 ohms. However, I don't have any inefficient / low impedance cans...

How did you calculate this Ken? I can tell you how to measure it if you are interested in giving it a try.

I'm curious to see how my OTL drives the Caldera when my pair arrives, it also has a 15ohm output impedance. Never tried a planar with it.

 

[gibosi]

How did you calculate this Ken? I can tell you how to measure it if you are interested in giving it a try.

I'm curious to see how my OTL drives the Caldera when my pair arrives, it also has a 15ohm output impedance. Never tried a planar with it.

Ro = rp / AF+1

For a 6528

Ro is output resistance
rp = Plate resistance = 245 (from datasheet)
AF+1 = Amplification Factor +1 = 10 (from datasheet)

So Ro = 245 / 10 = 24.5 for one section.

And if the two sections are in series, as in the Glenn:

Ro is divided by 2. And thus, 24.5 / 2 = 12.25.

I can't remember where I found this, but it seems to get me pretty close.

 

[LobalWarming]

Ro = rp / AF+1

For a 6528

Ro is output resistance
rp = Plate resistance = 245 (from datasheet)
AF+1 = Amplification Factor +1 = 10 (from datasheet)

So Ro = 245 / 10 = 24.5 for one section.

And if the two sections are in series, as in the Glenn:

Ro is divided by 2. And thus, 24.5 / 2 = 12.25.

I can't remember where I found this, but it seems to get me pretty close.

Was it here: https://www.tubecad.com/2005/June/blog0048.htm

"As for output impedance, there isn't much debate. The cathode follower's output impedance equals:
Zo = rp / (mu + 1)
"

 

[gibosi]

Was it here: https://www.tubecad.com/2005/June/blog0048.htm

"As for output impedance, there isn't much debate. The cathode follower's output impedance equals:
Zo = rp / (mu + 1)
"

That article doesn't look familiar, but the calculation sure does.

 

[LobalWarming]

That article doesn't look familiar, but the calculation sure does.

Great! In life, tube rolling and stats, 1 out of 2 ain't bad.

 

[L0rdGwyn]

Ro = rp / AF+1

For a 6528

Ro is output resistance
rp = Plate resistance = 245 (from datasheet)
AF+1 = Amplification Factor +1 = 10 (from datasheet)

So Ro = 245 / 10 = 24.5 for one section.

And if the two sections are in series, as in the Glenn:

Ro is divided by 2. And thus, 24.5 / 2 = 12.25.

I can't remember where I found this, but it seems to get me pretty close.

Ahh okay, I'm sorry to say it is more complicated than it seems since this is an SRPP stage, the formula from the Tube CAD article is for a cathode follower (actually the cathode follower output impedance can be fairly accurately estimated by 1/gm). You will find different formulas around the web for the output impedance of SRPP, for identical triodes and the lower cathode resistor bypassed (like in GOTL), I came across this formula by a knowledgeable chap on diyAudio.

Zout = (Ra^2)/(2Ra+(mu+1)Rk)

Where Ra is the plate resistance and Rk is the upper triode cathode resistor, 250ohm in GOTL. For a 245ohm plate resistance, that would give an output impedance of around 20ohm but hard to say how accurate that is as the plate resistance is also dependent on the bias point and the GOTL will not bias a 6528 as they are referencing in the datasheet.

I measured the plate resistance of a 6528 on my tracer for around a 100V / 100mA bias point (ballpark of where it would be in the GOTL), got a plate resistance of 330ohm, which by the formula above would make for an output impedance of 34ohm, so that might be a little closer to the mark.

I actually have a LTSpice simulation of the GOTL circuit, calculations in LTSpice tend to be pretty accurate. Unfortunately I don't have a 6528 model so I can't say what the output impedance would be using it. I do have a 6336 model though, output impedance is around 65ohm in my simulation.

The absolute best way to find the output impedance though is to measure it! It is very easy to do, you just need an old headphone cable you don't mind cutting to access the wires, two different resistor values, two alligator clips, and a digital multimeter. If you or anyone else is interested in how that is done, I can describe the process.

 

[LobalWarming]

I'm interested.

Of course, the HF debate will be about whether it's best cutting up a custom silver cable or just an over-priced, but nicely braided, OFC cable.

 

[Monsterzero]

Ahh okay, I'm sorry to say it is more complicated than it seems since this is an SRPP stage, the formula from the Tube CAD article is for a cathode follower (actually the cathode follower output impedance can be fairly accurately estimated by 1/gm). You will find different formulas around the web for the output impedance of SRPP, for identical triodes and the lower cathode resistor bypassed (like in GOTL), I came across this formula by a knowledgeable chap on diyAudio.

Zout = (Ra^2)/(2Ra+(mu+1)Rk)

Where Ra is the plate resistance and Rk is the upper triode cathode resistor, 250ohm in GOTL. For a 245ohm plate resistance, that would give an output impedance of around 20ohm but hard to say how accurate that is as the plate resistance is also dependent on the bias point and the GOTL will not bias a 6528 as they are referencing in the datasheet.

I measured the plate resistance of a 6528 on my tracer for around a 100V / 100mA bias point (ballpark of where it would be in the GOTL), got a plate resistance of 330ohm, which by the formula above would make for an output impedance of 34ohm, so that might be a little closer to the mark.

I actually have a LTSpice simulation of the GOTL circuit, calculations in LTSpice tend to be pretty accurate. Unfortunately I don't have a 6528 model so I can't say what the output impedance would be using it. I do have a 6336 model though, output impedance is around 65ohm in my simulation.

The absolute best way to find the output impedance though is to measure it! It is very easy to do, you just need an old headphone cable you don't mind cutting to access the wires, two different resistor values, two alligator clips, and a digital multimeter. If you or anyone else is interested in how that is done, I can describe the process.

Just for the record, I knew all of that already.

 

[heliosphann]

Numbers hurt my head.

 

[L0rdGwyn]

Okay here is the gist of it.

1) Turn on the amplifier, listen with headphones, find a 1kHz test tone track on YouTube or a streaming service and turn up the volume to something resembling your normal listening volume, perhaps a bit louder. Once you've set the volume, don't change it for the rest of the measurement. Don't change test tone tracks or alter the volume in any other way!

2) Cut an old headphone cable (I am assuming a 1/4" TRS single-ended for this since we are talking GOTL), strip away part of the jacket so you can access the three wires. Separate the wires until you've identified the ground wire, left channel wire, and right channel wire. You can check which is which by using the continuity setting on the DMM.

3) You need two resistor values, let's say in the headphone impedance range, we'll use 300ohm and 32ohm for this example. You don't have to use these specific values, you just want to use resistors that are somewhat far apart in value so you can measure differences in voltage within the accuracy of your DMM.

4) Plug the stripped headphone cable into the warmed up amplifier. Take the 300ohm resistor (or whichever is the larger value of the two) and using two alligator clips, clip one end to the channel you are testing on the stripped headphone cable (left or right, doesn't matter) and attach the other end to the ground wire.

5) Set the DMM to AC voltage and attach the leads across the resistor - you are going to measure the AC voltage generated across the resistor as the amplifier amplifies the 1kHz test tone.

6) Press play on the 1kHz test tone and record the AC voltage generated. Remember, once you've measured the first voltage, don't touch the volume knob!

7) Now replace the 300ohm resistor with the 32ohm resistor, same channel and again attach the DMM across the resistor leads.

8) Press play again on the same 1kHz test tone and record the AC voltage generated.

9) Now you have four values: R1 (value of first resistor), R2 (value of second resistor), V1 (AC voltage measured across the first resistor), V2 (AC voltage measured across the second resistor).

Plug them into this formula, and you have the output impedance.

Zout = ((R1*R2(V1-V2))/((R1*V2)-(R2*V1))

Yay math!!!

So, just to illustrate that this does in fact work, I performed this exact process on my 6528 OTL just now. I have done this in the past, but using an oscilloscope instead of a DMM.

Here are my values:

R1 = 300ohm
R2 = 32ohm
V1 = 0.321V
V2 = 0.240V

Plugging into the formula...

Zout = ((300*32(0.321-0.239))/((300*0.239-32*0.321))
Zout = 12.8ohms

Using a different set of 6528 tubes in the past and using an oscilloscope, I got around 15ohms, so it's a little off but pretty close. Very well could be the output impedance has changed slightly using a different set of tubes.

 

[L0rdGwyn]

Here is my headphone dummy load box, TRS jack is on the back, knob on the front changes the load from 32ohm / 80ohm / 120ohm / 300ohm. BNC jacks on the front so I can connect it directly to my scope. I was a psycho for making this thing, I would never do this nowadays lol but I'm glad I did, I use it all the time.

 

[L0rdGwyn]

Numbers hurt my head.

But ya gotta use those numbers to make the good amplifier sounds.

 

[bcowen]

@bcowen took note of that, and will forever haunt you if when you order the next amp (after Telemachus)!

He's said that so many times I think he's just trying to get himself to believe it. But I'll probably haunt him anyway just 'cause it'll be fun.

 

[gibosi]

Plugging into the formula...

Zout = ((300*32(0.321-0.239))/((300*0.239-32*0.321))
Zout = 12.8ohms

Using a different set of 6528 tubes in the past and using an oscilloscope, I got around 15ohms, so it's a little off but pretty close. Very well could be the output impedance has changed slightly using a different set of tubes.

The formula I found actually works pretty good for your amp, 12.25ohms. Will have to scrounge a couple resisters and an old headphone cable to see what the value is for Glenn's SRPP. When I finally get around to it, I will post my results. But I'm slow so don't hold your breath. lol

 

[L0rdGwyn]

The formula I found actually works pretty good for your amp, 12.25ohms. Will have to scrounge a couple resisters and an old headphone cable to see what the value is for Glenn's SRPP. When I finally get around to it, I will post my results. But I'm slow so don't hold your breath. lol

Yes it does, that's because my amplifier is a parallel cathode follower output, so use the formula to calculate the output impedance of one section, then divide by two as the sections are in parallel.

As I mentioned, the output impedance of a cathode follower can also be estimated by 1/gm. So for 6528, 1/37,000umhos = 1/0.037mhos = 27ohms. Now divide by two with the sections paralleled, you get 13.5ohms, again pretty close to the real world measurement.

But this is only for a cathode follower, the calculation is more complex for SRPP. I think it is best measured for accuracy.