As I work my way thru an upgrade of a tube amp I have been playing with (see this post) DC coupled long tail pair , I am now finalizing the amp before starting the cap and tube rolling process. In the stock form as shown on the schematic there was a fair amount of ringing (forgot to take a picture) so I went about tuning it some. First thing I did was to remove the 82pF feedback cap and the .22uF on the output and retest, resulted in large ringing, and since I would rather not have a cap in the feedback loop, I started playing with pF values on the output of the first stage, input to the 6SN7 stage to ground. I tried various values and the first 2 pics are 180pF and then 280pF, the 3rd pic is 220pF and the .22uF back on the output. All pics taken at 10KHz, 1W and the amp is pretty much flat to 30-40KHz depending on the cap used. The last pic is same as pic 3 but at 1KHz.
I know listening will be the ultimate test but I am trying to learn to correlate how the scope pattern looks versus how it sounds.
So, you gurus of amplifier design and tuning what do you think? I think the last pic looks the cleanest but again, I'm still learning.
Thank you, and Cheers!
180pF
280pF
220pF with .22uF-27R on output
1KHz
I know listening will be the ultimate test but I am trying to learn to correlate how the scope pattern looks versus how it sounds.
So, you gurus of amplifier design and tuning what do you think? I think the last pic looks the cleanest but again, I'm still learning.
Thank you, and Cheers!
180pF
280pF
220pF with .22uF-27R on output
1KHz
Second and third look good. I do not like fast overshoots (4). You've set the probe to 10:1 AND then calibrated it... right?
I'd leave the 0.22uF cap at the output. It's there to ensure stability with regard to various speakers and speaker cables that people may use.
I'd also connect the speaker cables, and then simulate the speaker load with a 6-ohm 50W resistor, and some added capacitance across it. Then, you'll be having a real-life environment, and the measurement will be the best you can get... that will work for your setup.
I'd leave the 0.22uF cap at the output. It's there to ensure stability with regard to various speakers and speaker cables that people may use.
I'd also connect the speaker cables, and then simulate the speaker load with a 6-ohm 50W resistor, and some added capacitance across it. Then, you'll be having a real-life environment, and the measurement will be the best you can get... that will work for your setup.
Last edited:
You need a series RC to ground, not just a capacitor to ground.
Try a 10k resistor to start, but vary the value as necessary.
This should eliminate the overshoot due to less phase shift.
But the rise time will be somewhat slower.
Try a 10k resistor to start, but vary the value as necessary.
This should eliminate the overshoot due to less phase shift.
But the rise time will be somewhat slower.
Last edited:
Awesome! Thanks guys for the quick replies, much appreciated.
Okay, my plan for the morning will be to pull the 220pF cap and measure as a baseline to see if the rise time improves, then start with a 10K in series with it and play with the combo values from there. I'm guessing the larger the resistor in series is, the greater capacity will need to be?
Yes it's a proper TEK 10x probe on my 475 scope, I will have to confirm calibration in the morning as I have a few TEK probes on the bench, and thanks for the reminder, I could be chasing an incorrect problem. 🙄 And yes I can put the dummy load at the end of the speaker cable, would I then measure at the end of the cable or still at the amp terminals?
This has been a great re-learning experience for me, I was right into this stuff back in the 80's but kind of lost a lot of knowledge due to brain being filled with office/work BS over the years and lack of hifi playing, raising child etc.. Now that work is slowing down, child has left home, I have a LOT more time to play hifi.
Cheers
Okay, my plan for the morning will be to pull the 220pF cap and measure as a baseline to see if the rise time improves, then start with a 10K in series with it and play with the combo values from there. I'm guessing the larger the resistor in series is, the greater capacity will need to be?
Yes it's a proper TEK 10x probe on my 475 scope, I will have to confirm calibration in the morning as I have a few TEK probes on the bench, and thanks for the reminder, I could be chasing an incorrect problem. 🙄 And yes I can put the dummy load at the end of the speaker cable, would I then measure at the end of the cable or still at the amp terminals?
This has been a great re-learning experience for me, I was right into this stuff back in the 80's but kind of lost a lot of knowledge due to brain being filled with office/work BS over the years and lack of hifi playing, raising child etc.. Now that work is slowing down, child has left home, I have a LOT more time to play hifi.
Cheers
You could just go direct into the BNC of your scope without a probe; that would tell you if it's the probe or not. I'm pretty sure frequencies like 1kHz wont need 50 Ohm cable / 50 Ohm termination.
Follow this procedure:
Start with a capacitor around 10nF with a series resistor, and adjust the resistor value until the overshoot is gone.
Then, keep that resistor value fixed, and reduce the size of the capacitor until you start getting overshoot.
Finally make the capacitor just large enough again so the peaking is gone.
This overshoot does not look like a probe cal problem.
I think I will make one other measurement. I'm going to open up the FB loop and see how the distortion measurement changes. It's currently about .2% THD @ 1W, 10W is 1%, triode wired.
cheers
cheers
I tend to like a little overshoot on my amps, with one or two cycles of well-damped ringing, it just sounds more "right" to my ears. YMMV, go with what sounds best to you.
Ok, finished plowing snow and am back on the amp. I opened up the feedback loop and removed all caps, and ensured the scope probe is calibrated, it was, and here are the pics. 1KHz, 10KHz and 100 Hz.
I'm going to close the feedback loop and start playing with R/C network as discussed yesterday. BTW distortion in open loop is 1W=.4% and 10W=3% compared to closed loop 1W=.25% and 13W=1%.
More later.
1KHz
10KHz
100Hz
I'm going to close the feedback loop and start playing with R/C network as discussed yesterday. BTW distortion in open loop is 1W=.4% and 10W=3% compared to closed loop 1W=.25% and 13W=1%.
More later.
1KHz
10KHz
100Hz
Okay I think I am getting really close, I'll let you guys tell me.
I put in a 25K pot in series with various cap values and adjusted the pot until I got a nice flat wave, what came up best is I think, 180pF in series with 6.6K. I also changed the output cap from .22uF to .1uF polycarbonate as the .22 rolled off too much and it was also a ceramic cap, the original feedback cap was a 56pF ceramic as well. Removing the feedback cap and changing the output cap from ceramic and getting the square wave looking good should go a long ways into getting rid of the tizz.
Am I close? 5Khz and 10KHz in the pics @ 1W.
Cheers
5KHz
10KHz
I put in a 25K pot in series with various cap values and adjusted the pot until I got a nice flat wave, what came up best is I think, 180pF in series with 6.6K. I also changed the output cap from .22uF to .1uF polycarbonate as the .22 rolled off too much and it was also a ceramic cap, the original feedback cap was a 56pF ceramic as well. Removing the feedback cap and changing the output cap from ceramic and getting the square wave looking good should go a long ways into getting rid of the tizz.
Am I close? 5Khz and 10KHz in the pics @ 1W.
Cheers
5KHz
10KHz
Thanks! No actually it's with a 15 Ohm load as my Big Reds are 16 Ohms. I should test it with an 8 Ohm load just to make sure it's still good.
I forgot to take pics of it bone stock so here is the other channel, I'm just getting ready to modify it now then will be able to listen tonight. I will have to do cap rolling tomorrow as it involves completely removing the cct brd and I'm almost out of steam.
Here's the stock waveforms, 5KHz and 10KHz, 1W.
Cheers
5KHz
10KHz
I forgot to take pics of it bone stock so here is the other channel, I'm just getting ready to modify it now then will be able to listen tonight. I will have to do cap rolling tomorrow as it involves completely removing the cct brd and I'm almost out of steam.
Here's the stock waveforms, 5KHz and 10KHz, 1W.
Cheers
5KHz
10KHz
Well, it's not as bad as stock but it's not perfect with an 8 Ohm load, but again, my speakers are 16 Ohm so I'm going to leave it setup for my system. It's easy to put in a switch if I want to make it adjustable.
Cheers
8 Ohms with mods as above
Cheers
8 Ohms with mods as above
Tube amps won't have near ideal square wave response, due to the output transformer.
And the response will change on a loudspeaker load as well.
And the response will change on a loudspeaker load as well.
Yeah.... output transformer will make things skewed... that's why I suggested initially doing your measurements with a speaker cable connected, and with a resistor and small capacitor in parallel., connected at the end of that speaker cable. You'll be tuning the square-wave response for your user-case.
"BTW distortion in open loop is 1W=.4% and 10W=3% compared to closed loop 1W=.25% and 13W=1%.".... so, open loop then?? ... Unless there's instability/oscillations, once you connected the intended speaker cable and speakers.
"BTW distortion in open loop is 1W=.4% and 10W=3% compared to closed loop 1W=.25% and 13W=1%.".... so, open loop then?? ... Unless there's instability/oscillations, once you connected the intended speaker cable and speakers.
I forgot to put the speaker cable in line, I'll have to do it tomorrow. I think I will put a switch in for open/closed loop, I'm listening tonight closed loop, and it's sounding WAY better, no tizz, better space around the instruments, and the bottom end is better, not sure why?? My go to reference is Jazz at the Pawnshop, just love it, I have 1 & 2, the 3 disc SACD set is on it's way from Greece, should be here in 3 weeks or so.
More later, and thanks everyone for helping me, it's a great learning experience.
cheers
More later, and thanks everyone for helping me, it's a great learning experience.
cheers
Well, the closed loop will definitely improve the damping factor and frequency response... so, keep that loop closed then.
You may want to check the frequency response as well... I use a good old fashion function generator (sine, triangle, square) as a source for testing, which I built probably 3 decades ago... try with open and then with closed loop.... see what happens when you open the loop. Just load the output when doing the tests.
You may want to check the frequency response as well... I use a good old fashion function generator (sine, triangle, square) as a source for testing, which I built probably 3 decades ago... try with open and then with closed loop.... see what happens when you open the loop. Just load the output when doing the tests.
Thanks! Did all that already. 😀 I use my HP339A distortion analyzer as a sine wave gen as it can measure dB, volts and has a number of other options for measuring amps. I also have a multi-wave generator, can't recall brand right now but has BNC out and puts out 20V into 50 Ohm, plenty for our uses. So closed loop freq response is flat from 10-25Khz and down 3 dB at 45Khz at 1W. Open loop it's flat 10-18Khz down 3dB about 25Khz using sine wave at 1W.
I had a massive listening session tonight, actually listening, not analyzing, damn near fell asleep listening to Kenny Burrrell Midnight Blue. Man some of those old recordings just sound terrific, so much better than a lot of modern recordings. Some of my faves are Lou Donaldson, Dr. Lonnie Smith, Jimmy Smith, they all seemed to hang out together and play on each others albums back in the 60's, pretty cool stuff. I have a lot of music on my server, FLAC files, about 800 CDs, and 5,000 albums...I'll be long dead before I get to hear them all. 😱🙄
I think my next project this week will be restoring my old Eico HF-81, a real classic that just sounds amazing stock. I'm reluctant to update it as I might destroy the magic, but I can't see how removing ceramic caps in the signal path would hurt.
Cheers
I had a massive listening session tonight, actually listening, not analyzing, damn near fell asleep listening to Kenny Burrrell Midnight Blue. Man some of those old recordings just sound terrific, so much better than a lot of modern recordings. Some of my faves are Lou Donaldson, Dr. Lonnie Smith, Jimmy Smith, they all seemed to hang out together and play on each others albums back in the 60's, pretty cool stuff. I have a lot of music on my server, FLAC files, about 800 CDs, and 5,000 albums...I'll be long dead before I get to hear them all. 😱🙄
I think my next project this week will be restoring my old Eico HF-81, a real classic that just sounds amazing stock. I'm reluctant to update it as I might destroy the magic, but I can't see how removing ceramic caps in the signal path would hurt.
Cheers