• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Call it good enough?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have to use the "open in a new window" option to see them myself. Not sure why they don't show up in the post.

You have to be LOGGED IN, actually. With cookies turned on, for there to be the automagic opening of most graphics. Don't expect PDFs to preview, either. Ever. Its DIYAudio. (which I am definitely not complaining about!

Just saying,
GoatGuy ✓
 
I am logged in, and I can see other peoples images without having to open in a new window. I'm probably doing something wrong, LOL
 

Attachments

  • amp.jpg
    amp.jpg
    254.4 KB · Views: 153
That's pushing it a bit - single-tone tests are very limited, in any realistic signal intermodulation distortion is an issue, and 2nd and 3rd order IM products are as audible as 7th or 8th order (for the same amplitude), they all can fall at non-harmonically related frequencies.

Agree, to a point: it's complex enough before we start talking about the impact of material choices and where those IMD products end up in the Fletcher Munson curve. Then there's the issue of what each individual does and does not notice (and object to).

But I'd still always rather have x% of 2nd than x% of 7th.

I've been trying to mentally workout what we should be seeing from this CCT: an interesting race between the SE and PP bits.

OBTW: Lynn Olson did a bunch of work on 6SN7 distortion vs operating mode a while back which might be of use to Troy.
 
Last edited:
Agree, to a point: it's complex enough before we start talking about the impact of material choices and where those IMD products end up in the Fletcher Munson curve. Then there's the issue of what each individual does and does not notice (and object to).

But I'd still always rather have x% of 2nd than x% of 7th.

I've been trying to mentally workout what we should be seeing from this CCT: an interesting race between the SE and PP bits.

OBTW: Lynn Olson did a bunch of work on 6SN7 distortion vs operating mode a while back which might be of use to Troy.

Very good information in the link, Thanks! Lots to digest.
 
just did a little experiment, I bypassed the 830 ohm cathode resistors on the CCS'ed 6SN7's with 470uf Silmac 2 caps. 2nd and 3rd order harmonic distortion stayed about the same, maybe a couple hundredths of percent higher, but gain increased considerably, as to be expected due to the loss of negative feedback on the cathodes. I don't need the gain, so Ill remove the bypass caps. In a couple days I will try 8.2 volt zeners instead of the cathode resistors.
 
Last edited:
Nice chassis work, are the meters monitoring the OP valves cathode current, if so I presume you have sense resistors on the cathodes.


Making the bias adjustable and popping in a balance pot allows you to have a play around and set the amp up. Make sure to use fixed resistors to stop the bias from going to 0v and wire the wiper to one end of the pot in case the wiper lifts off, this will stop no bias being applied to the grids. I used 1M R's across the bias adjust pot to be 100% safe.


Have a look at this amp - 100w-monobloc3-2014 there is some very usefull info in the article, Partrick also goes into some detail about best class A using various OPT's and loads etc you may find interesting.


Lastly reducing the grid leak R's to 47k improved my amp, isn't that what they used on the GEC 400w amp?


Regards, Andy.
 
Nice chassis work, are the meters monitoring the OP valves cathode current, if so I presume you have sense resistors on the cathodes.


Making the bias adjustable and popping in a balance pot allows you to have a play around and set the amp up. Make sure to use fixed resistors to stop the bias from going to 0v and wire the wiper to one end of the pot in case the wiper lifts off, this will stop no bias being applied to the grids. I used 1M R's across the bias adjust pot to be 100% safe.


Have a look at this amp - 100w-monobloc3-2014 there is some very usefull info in the article, Partrick also goes into some detail about best class A using various OPT's and loads etc you may find interesting.


Lastly reducing the grid leak R's to 47k improved my amp, isn't that what they used on the GEC 400w amp?


Regards, Andy.

The meters have built in 2 ohm shunts, I hope that's going to work, if not can add a few ohms externally. I do have stop resistors in the bias PS so it wont set down to zero but I did not tie the wiper to one end , good idea! I have visited Patrick's site, a lot of deep information there. I keep going back and I learn a bit more each time. The article I have about the GEC amp mentions using 100K grid resistors when just using 4 output tubes. But, as I understand it, modern tubes may need lower values simply because they may not be made as well, and the grid can build up a charge. I had some Chinese 6550's that tended to "run away", in particular.
 
Last edited:
LED bias...subbed 4 red LED's for the 830 ohm cathode resistors for the 6SN7 driver tubes. Amplifier distortion stayed the same at lower power output, dropped a bit at mid to high output. She now has DC regulated filaments in the front end, I have to put my ear right up to an efficient little 4" speaker to hear any noise, here is a slight bit of hum but the tube hiss is just as loud. I think its about time to call it good and build the other one.
 

Attachments

  • IMG_1609.jpg
    IMG_1609.jpg
    1 MB · Views: 89
LED bias…subbed 4 red LED's for the 830 Ω cathode resistors for the 6SN7 driver tubes. Amplifier distortion stayed the same at lower power output, dropped a bit at mid to high output. She now has DC regulated filaments in the front end, I have to put my ear right up to an efficient little 4" speaker to hear any noise, here is a slight bit of hum but the tube hiss is just as loud. I think its about time to call it good and build the other one.

Sweet. LEDs as the “no-cheaper-or-better-but-definitely-au-courant” bias method (as opposed to say a 8 volt Zener, LOL) is fine. They definitely are more of a craftsman's Art though… you have to actually measure the VF of an LED (and use the same brand and model) to achieve the VF you're happy with.

Having been “in the business” of using semiconductor devices for most-every application (as well as starting off with a keen respect for 300 V B+ supplies… for valves), I have a now-somewhat-misguided “suspicion” against expecting run-to-run uniformity in semiconductor widgets, even from the same manufacturer, with the same model number.

In the Dark Ages (certainly after 1970 — when crystal dinosaur scales were used, but before 1990 — when manufacturing 'chops' got quite a bit tighter) those “data sheet specs” demonstrated huge manufacturing variability ranges. ± 35 to 60% wasn't at all uncommon. And were you to order “a bag” of parts, they'd vary all over the place.

However, of more recent experience, I've found manufacturing tolerances to be much tighter. I got — as an example — 250 J301 junction n-channel FETs, for a few pennies apiece.

It took me by surprise on setting up a mass-testing-while-drinking-beer rig, to find that almost all of the devices were within 3% of each other, operating-spec wise. There were a few outliers, but most were quite tightly near the center of the data sheet spec.

So, time's are a'changing.

GoatGuy ✓
 
Sweet. LEDs as the “no-cheaper-or-better-but-definitely-au-courant” bias method (as opposed to say a 8 volt Zener, LOL) is fine. They definitely are more of a craftsman's Art though… you have to actually measure the VF of an LED (and use the same brand and model) to achieve the VF you're happy with.

Having been “in the business” of using semiconductor devices for most-every application (as well as starting off with a keen respect for 300 V B+ supplies… for valves), I have a now-somewhat-misguided “suspicion” against expecting run-to-run uniformity in semiconductor widgets, even from the same manufacturer, with the same model number.

In the Dark Ages (certainly after 1970 — when crystal dinosaur scales were used, but before 1990 — when manufacturing 'chops' got quite a bit tighter) those “data sheet specs” demonstrated huge manufacturing variability ranges. ± 35 to 60% wasn't at all uncommon. And were you to order “a bag” of parts, they'd vary all over the place.

However, of more recent experience, I've found manufacturing tolerances to be much tighter. I got — as an example — 250 J301 junction n-channel FETs, for a few pennies apiece.

It took me by surprise on setting up a mass-testing-while-drinking-beer rig, to find that almost all of the devices were within 3% of each other, operating-spec wise. There were a few outliers, but most were quite tightly near the center of the data sheet spec.

So, time's are a'changing.

GoatGuy ✓

Both sides came in real close to 7.8 volts. The zeners didn't show today, and I had a bag of about 100 decent quality LEDS so I decided to give it a shot.
 
Would you call this good enough? These are the distortion numbers on my amp project, loosely based on the GEC 400 watt tube amp plans. I'm using 4 KT120's in parallel push pull, and instead of cathode followers driving the output tubes, I'm using paralleled 6SN7's with CCS plate loads. Output trans is Hammond 1650T. Measurements taken with Keithley 2015, at 1 Khz.

3 watts-.07% , 5 watts-.11%,
10 watts-.17%, 20 watts-.29%, 30 watts-.31%,
40 watts-.28%, 50 watts-.22%, 60 watts- .27%,
75 watts-.52% 90 watts-.61% and starting to clip

New distortion numbers with LED bias on driver tubes, and balance adjust pot on phase splitter (AC volts adjusted using Keithley 2015 ):
4 watts- .03%, 12 watts- .13%, 20 watts- .20%, 30 watts- .26%, 40 watts- .27%, 50 watts, .18%, 60 watts- .14%, 80 watts, .36%, 100 watts- .25%, just beginning to clip
It sounds GREAT! It has an immediacy that is quite surprising and pleasing.
 
Last edited:
Why not use the cathode follower?
I have always used The cathode follower

I am getting far less distortion this way, when actually driving the KT-120's. The orig GEC front end measures very well stand alone, but for some reason total distortion increases dramatically when its driving the output stage. I even reverted back to the GEC circuit earlier this week, as a sanity check, but 2nd and 3rd harmonic distortion was double or more. I'm wondering if the distortion of my modified circuit is such that it is being cancelled in the output stage. In any case, it measures and sounds good now, so I think I will leave it as is.
 
Last edited:
Congratulations on your successful project. A very good case can be made that only the distortion in the first Watt, or the first .01 Watt, matters. Performance near clipping gives bragging rights, but really only needs to be much better than a loudspeaker - not that difficult.


But performance within the working range, say -60 dBVU to 0 dBVU, actually matters much more. Not as sexy, but more important.


All good fortune,
Chris
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.