L0rdGwyn's DIY Audio
Feb 8, 2020 at 4:59 PM Post #91 of 6,723
IMAG1743.jpg
You are going to need a bigger, cozier chair if you have any hope of sitting there for long!
 
Feb 8, 2020 at 5:01 PM Post #92 of 6,723
You are going to need a bigger, cozier chair if you have any hope of sitting there for long!

You'd be surprised, it works quite well for me! Plus, I'm always shuffling around anyway, not sitting for long. It is my adjustable chair for my telescope, so I am used to it :)
 
Feb 8, 2020 at 7:41 PM Post #93 of 6,723
Hi LG,
Re your mentioning of various voltages coming from the transformer windings, I am under the impression that in the tube era a 10% variation was considered within specs. Is it really that important to have the exact voltage available, especially since the mains voltage may fluctuate a little as well?
 
Feb 8, 2020 at 7:58 PM Post #94 of 6,723
Hi LG,
Re your mentioning of various voltages coming from the transformer windings, I am under the impression that in the tube era a 10% variation was considered within specs. Is it really that important to have the exact voltage available, especially since the mains voltage may fluctuate a little as well?

Hi mordy - yes, +/- 10% is the goal. Closer to the specification is best since there will be mains variations. The issue is the 5.9V and 6.6V specification for the heater windings is at 115VAC. I am expecting 124VAC into the transformer from the wall, so those voltage ratings are low. In reality, I am seeing 6.3V and 7.1V, respectively. I am using dropping resistors for the two 5.9V (measured 6.3V) windings for the 4V/5V rectifiers and 4V driver tubes anyway. The voltages do not need to be exact with the resistors, but you can get pretty close. Without diving into the details (unless you want me to :)), I am trying to hit smaller targets on the voltages so that a larger variety of tubes can be used within their +/-10% specification.

I had not planned to use resistors on the 6.6V (measured 7.1V) windings for the power tube heaters. At 7.1V, I am about 13% above the 6.3V specification. Not horrible but not comfortable either, likely will do something to bring it down to a happy place.
 
Feb 9, 2020 at 12:05 PM Post #95 of 6,723
Continued working on the tube heater dropping resistors today.

I wired a pair of 4V driver tubes with the heaters in parallel, MOV MHL4. The heaters are center tapped and referenced to ground with a pair of 100ohm resistors, so this was a ton of fun with a thousand alligator clips. Thankfully, the voltage is right where I want it ~3.9V with 124VAC into the mains transformer.

IMAG1753.jpg

Next, I checked the voltage of the 4V rectifiers. First, a Marconi rebranded Mullard FW4/500 which draws 3A:

IMAG1765.jpg IMAG1764.jpg

Then a GEC U18/20, which draws 2.8A:

IMAG1770.jpg IMAG1761.jpg

As expected, their filament voltages are slightly off due to the difference in drawn current, but they are right where I want them to be, looks great.

Last thing I did was put together one of the CCS kits. I hooked it up to a 30VDC power supply and adjusted the trim pot until I got 6mA, the plate current for the driver tubes. This corresponded to 395ohm on the trim pot, so I'll likely get a 400ohm set resistor. Thanks for the idea, @A2029 :ksc75smile:

IMAG1777.jpg IMAG1775.jpg

So looks like my math was correct on the dropping resistors, now I can confidently drill holes in the chassis when it arrives. Supposed to be finished re-plating Tuesday, hopefully will get it by Friday and the real building will begin next weekend.
 
Feb 9, 2020 at 1:13 PM Post #96 of 6,723
Hi LG,
Re your mentioning of various voltages coming from the transformer windings, I am under the impression that in the tube era a 10% variation was considered within specs. Is it really that important to have the exact voltage available, especially since the mains voltage may fluctuate a little as well?

You have the pieces, but you need to put them together.

Yes, tube's can handle 10% shifts. Yes, wall voltage will vary. However you want to center your tube around an average wall voltage so that as your wall voltage changes and shifts, the voltage around the tube is only shifting +/-10% or less.

It's also important to note that just because tubes can deal with 10% variations doesn't mean that being 10% outside of the spec is good for hi-fi use. Ideally you want to keep the parameters as stable and ideal as possible.

If we were being super picky, what I would do is take your volt meter and monitor the wall voltage at different times of the day and on weekends. Figure out what your lowest and highest wall voltage was. Split the difference and run your amplifier off of a variac set to that average voltage. Set everything up such that you get as close to ideal voltages as humanly possible. That way your amp won't drift and wander as much in your home.

If you want an amp that you can take anywhere, just do the same thing but set it for 120v. It may or may not be ideal for your home, but it will work in darn near any wall socket in the US that wasn't built during ww2.
 
Last edited:
Feb 9, 2020 at 1:48 PM Post #97 of 6,723
Is it possible to hear a difference if the voltage changes somewhat more than the 10%?
From personal experience I have found that late at night or on weekends my equipment may sound better, possibly because of less electric interference from all kinds of devices and machinery.
 
Feb 9, 2020 at 3:50 PM Post #98 of 6,723
Is it possible to hear a difference if the voltage changes somewhat more than the 10%?
From personal experience I have found that late at night or on weekends my equipment may sound better, possibly because of less electric interference from all kinds of devices and machinery.

Well 10% difference would be obvious, but ideally we never get close to those kinds of differences.

But in general I would say it all depends on the circuit and what tubes you are using. I have seen some tube amps where the preamp tube is biased so close to the knees that a small change in voltage could let the tube slip into the knees and cause all sorts of havoc.

On the flip side I have seen amps where the tubes were biased with such headroom that it can handle larger variations without falling apart. You would still probably hear a change, but it just wouldn't be enough of a change to cry about.

But this is assuming good old fashion design. A tube with a plate resistor, cathode resistor, and cathode bypass capacitor.
 
Feb 9, 2020 at 3:54 PM Post #99 of 6,723
Continued working on the tube heater dropping resistors today.

I wired a pair of 4V driver tubes with the heaters in parallel, MOV MHL4. The heaters are center tapped and referenced to ground with a pair of 100ohm resistors, so this was a ton of fun with a thousand alligator clips. Thankfully, the voltage is right where I want it ~3.9V with 124VAC into the mains transformer.

IMAG1753.jpg

Next, I checked the voltage of the 4V rectifiers. First, a Marconi rebranded Mullard FW4/500 which draws 3A:

IMAG1765.jpg IMAG1764.jpg

Then a GEC U18/20, which draws 2.8A:

IMAG1770.jpg IMAG1761.jpg

As expected, their filament voltages are slightly off due to the difference in drawn current, but they are right where I want them to be, looks great.

Last thing I did was put together one of the CCS kits. I hooked it up to a 30VDC power supply and adjusted the trim pot until I got 6mA, the plate current for the driver tubes. This corresponded to 395ohm on the trim pot, so I'll likely get a 400ohm set resistor. Thanks for the idea, @A2029 :ksc75smile:

IMAG1777.jpg IMAG1775.jpg

So looks like my math was correct on the dropping resistors, now I can confidently drill holes in the chassis when it arrives. Supposed to be finished re-plating Tuesday, hopefully will get it by Friday and the real building will begin next weekend.
@A2029 rocks :)
 
Feb 11, 2020 at 12:52 PM Post #100 of 6,723
Continued working on the tube heater dropping resistors today.

I wired a pair of 4V driver tubes with the heaters in parallel, MOV MHL4. The heaters are center tapped and referenced to ground with a pair of 100ohm resistors, so this was a ton of fun with a thousand alligator clips. Thankfully, the voltage is right where I want it ~3.9V with 124VAC into the mains transformer.



Next, I checked the voltage of the 4V rectifiers. First, a Marconi rebranded Mullard FW4/500 which draws 3A:



Then a GEC U18/20, which draws 2.8A:



As expected, their filament voltages are slightly off due to the difference in drawn current, but they are right where I want them to be, looks great.

Last thing I did was put together one of the CCS kits. I hooked it up to a 30VDC power supply and adjusted the trim pot until I got 6mA, the plate current for the driver tubes. This corresponded to 395ohm on the trim pot, so I'll likely get a 400ohm set resistor. Thanks for the idea, @A2029 :ksc75smile:



So looks like my math was correct on the dropping resistors, now I can confidently drill holes in the chassis when it arrives. Supposed to be finished re-plating Tuesday, hopefully will get it by Friday and the real building will begin next weekend.

IXTP08N50D2 is an interesting choice for top device in the CCS cascode. I believe K&K used the IXTP01N100D in the past, but it's more expensive and not as hefty of a chip (but has lower capacitance values). What MOSFET is used as the bottom device on the other side of that heatsink?

Someday in the future I'll have to build a bunch of different CCS circuits and do an in depth shootout. I'll maybe build a blinding box to make it easy to switch and so that I can't see which CCS I'm listening to.
 
Feb 11, 2020 at 1:02 PM Post #101 of 6,723
IXTP08N50D2 is an interesting choice for top device in the CCS cascode. I believe K&K used the IXTP01N100D in the past, but it's more expensive and not as hefty of a chip (but has lower capacitance values). What MOSFET is used as the bottom device on the other side of that heatsink?

Hey Mischa - the top device is the chip on the opposite side of the heatsink, and it is the IXTP08N100D2. The IXTP08N50D2 in the photo is the bottom device. For some reason, Kevin has the bottom device marked Q1 on the board and the top device Q2.

Someday in the future I'll have to build a bunch of different CCS circuits and do an in depth shootout. I'll maybe build a blinding box to make it easy to switch and so that I can't see which CCS I'm listening to.

That would be very interesting, would be very curious to hear the results! :) maybe I would make a change based on your test.
 
Feb 11, 2020 at 3:04 PM Post #102 of 6,723
IXTP08N50D2 is an interesting choice for top device in the CCS cascode. I believe K&K used the IXTP01N100D in the past, but it's more expensive and not as hefty of a chip (but has lower capacitance values). What MOSFET is used as the bottom device on the other side of that heatsink?

Someday in the future I'll have to build a bunch of different CCS circuits and do an in depth shootout. I'll maybe build a blinding box to make it easy to switch and so that I can't see which CCS I'm listening to.

There is also the issue of thermals. The IXTP08N50D2 will only rise 2 degrees C per watt where as the IXTP01N100D will rise 5 degrees C per watt. Some tubes like the 5687 can create a few watts of dissipation across the CCS. Just 3 watts across the CCS is the difference between 6 degrees vs 15 degrees and may be the difference between bolting the mosfet to the top plate vs buying a dedicated heat sink.
 
Feb 11, 2020 at 8:43 PM Post #103 of 6,723
Hey Mischa - the top device is the chip on the opposite side of the heatsink, and it is the IXTP08N100D2. The IXTP08N50D2 in the photo is the bottom device. For some reason, Kevin has the bottom device marked Q1 on the board and the top device Q2.

Ah okay, that makes more sense having the IXTP08N100D2 at the top for more voltage handling. Slightly lower Crss capacitance as well than the IXTP08N50D2. IXTP08N100D2 has good Vgs as well, which helps to reduce the capacitances of the lower device slightly.

That would be very interesting, would be very curious to hear the results! :) maybe I would make a change based on your test.

That's if my ears are sensitive enough to reliably hear the minute changes :L3000:

There is also the issue of thermals. The IXTP08N50D2 will only rise 2 degrees C per watt where as the IXTP01N100D will rise 5 degrees C per watt. Some tubes like the 5687 can create a few watts of dissipation across the CCS. Just 3 watts across the CCS is the difference between 6 degrees vs 15 degrees and may be the difference between bolting the mosfet to the top plate vs buying a dedicated heat sink.

Agreed, the IXTP08 parts are better for higher currents/dissipation.
 
Feb 12, 2020 at 10:26 AM Post #104 of 6,723
Finished building the Linear Audio Autoranger last night, it is up and running. I was rushing and accidentally mounted two front panel LEDs too far into the chassis, so they aren't visible, but functionally it is good to go :ksc75smile: I'll fix the LEDs later this week.

Exterior:

IMAG1797-2.jpg

And interior:

IMAG1789-2.jpg IMAG1791-2.jpg

Why is this measurement device relevant to this project? As I mentioned before, the Autoranger allows one to take precision audio measurements at high signal output levels using an audio interface (I use the MOTU M4 DAC/ADC).

In the photos, I am feeding a 1kHz sine wave into the Glenn OTL from my MOTU M4. The Glenn OTL is connected to a switchable headphone dummy load I built, set to 300ohm. The output of the dummy load feeds the single-ended input of the Autoranger. You can see I am sending the Autoranger a 10V signal from the Glenn OTL. If I connected this 10V output directly to my MOTU M4 instead of the Autoranger, the ADC would be very unhappy. Instead, the Autoranger is providing 18dB of attenuation, so a very safe 1.28V signal enters the MOTU M4 and can be measured.

A 1kHz 10V signal from the Glenn OTL with this particular tube loadout corresponds to ~1% THD+N. Don't worry though, at that output level you wouldn't be able to hear the high levels of distortion since you would be deaf (at ~1V output, the Glenn OTL puts out around 0.07% THD+N with these tubes, Mullard ECC32 and sextet of Tung-Sol 6BX7GT).

IMAG1795-2.jpg

So, the utility of this setup is being able to test speaker amplifiers' high output voltages safely, and headphone amps if you want to crank it up and take measurements at unsafe listening volumes. Really cool device, should be lots of fun to play with for years to come :)

In other news, I received word that the chassis for this SET amp is FINALLY done. There were some...technical difficulties with the anodization, but I am happy to say the end result is good and I should have it this weekend to start building. I will have the house to myself, which means I will be living in my garage with the space heaters.
 
Last edited:
Feb 14, 2020 at 9:54 AM Post #105 of 6,723
At long last! The chassis has arrived. Time to check my planned layout, mark it up, and start drilling. My goal for the weekend is to get the holes drilled and parts mounted. After that, with just the wiring remaining, should come together pretty quickly.

IMAG1803-2.jpg
 

Users who are viewing this thread

Back
Top