Actually, yes. I forgot about the 680pF cap until after I'd posted the scope traces. So, I disconnected it and rechecked things. That did sharpen up the trace a bit more, but still no sign of ringing, and not enough difference in the new scope trace to bother posting another image. The volume control pot was in circuit, but set to max volume, so it shouldn't have added any significant capacitance.
The 680pF cap had been put in, because I noticed that my cheap DVD player audio input source had some ultrasonic garbage (unfiltered sampling noise) added to the audio signal. The small cap cleaned it up.
Note, there were a couple of small changes made to the circuit when I added the tone control. I reduced the value of the volume control pot from 100k to 50k, and increased the input DC blocking cap from 0.22µF to a 3.3µF film cap. With the tone control disconnected and the 680pF cap removed, the input circuit was essentially the 3.3µF cap connected to the 50k pot.
I suspect the lack of ringing may be attributable to the possibly higher than normal transformer primary DC resistance, hence damping out any resonance. In the process of making a cheap transformer, they may have accidentally made a decent one.
valve frequency responce
-
Mike H
- Amstrad Tower of Power
- Posts: 20189
- Joined: Sat Oct 04, 2008 5:38 pm
- Location: The Fens
- Contact:
#32
Or, it's HF coupling is not actually very good, else ringing on the primary would come through. Next idea might be 'scope the primary Voltage. (?)
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
#33
I have been scoping the primary side all along. No ringing to be found. Transformer core appears to be made of wool.
I agree that HF coupling may not be overly spectacular. Yet, it's apparently good enough to give reasonable frequency response with modest NFB. (NFB keeps getting more and more modest as I try more experiments with this circuit.) Even without NFB, it had a -3 db top end of 15kHz which I don't think is too shabby (cheap OPT I keep reminding everyone).
Even though this remains on topic, and the OP seems to have fled, I'm thinking maybe I should start another thread discussing this little amp, as the circuit has some other 'non frequency response' characteristics that may be of interest.
I agree that HF coupling may not be overly spectacular. Yet, it's apparently good enough to give reasonable frequency response with modest NFB. (NFB keeps getting more and more modest as I try more experiments with this circuit.) Even without NFB, it had a -3 db top end of 15kHz which I don't think is too shabby (cheap OPT I keep reminding everyone).
Even though this remains on topic, and the OP seems to have fled, I'm thinking maybe I should start another thread discussing this little amp, as the circuit has some other 'non frequency response' characteristics that may be of interest.
-
Mike H
- Amstrad Tower of Power
- Posts: 20189
- Joined: Sat Oct 04, 2008 5:38 pm
- Location: The Fens
- Contact:
#34
Good lord. I would have expected the rise time to be sharper there. Something else slowing it down? That may be why no overshoot and ringing, rise time isn't fast enough to excite it. (?)
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
#35
Sorry, I didn't mean to say that the posted scope traces were primary side. Those are secondary side right at the 8 ohm resistor. I just meant to say that I've also been observing the primary side during experimentation and found no ringing.
However, your last comment got me re-examining things. Though the waveform rises much more sharply on the primary side, as would be expected, I noted it didn't look as fast as it should be. At first I thought it may be due to the breadboard layout and some long interconnecting leads. But then went digging around, and realized there was a 100pF cap from the pentode anode to ground. It was put in some time back when I thought I might be having some parasitic oscillation. It turned out not to be the case, and I thought I'd removed the cap. But it was hidden under the valve socket where I hadn't noticed it. That has now been removed, and the primary side rise time is much faster. Consequently, there is indeed some ringing visible. New scope traces are attached, one with 1kHz square wave, and one with 4kHz square wave. No NFB in either case. Upper trace is primary side, and lower trace is secondary side across 8 ohm resistive load. It's still quite heavily damped, but that may be due to fairly high primary winding DC resistance of 700 ohms.
However, your last comment got me re-examining things. Though the waveform rises much more sharply on the primary side, as would be expected, I noted it didn't look as fast as it should be. At first I thought it may be due to the breadboard layout and some long interconnecting leads. But then went digging around, and realized there was a 100pF cap from the pentode anode to ground. It was put in some time back when I thought I might be having some parasitic oscillation. It turned out not to be the case, and I thought I'd removed the cap. But it was hidden under the valve socket where I hadn't noticed it. That has now been removed, and the primary side rise time is much faster. Consequently, there is indeed some ringing visible. New scope traces are attached, one with 1kHz square wave, and one with 4kHz square wave. No NFB in either case. Upper trace is primary side, and lower trace is secondary side across 8 ohm resistive load. It's still quite heavily damped, but that may be due to fairly high primary winding DC resistance of 700 ohms.
- Attachments
-
- 4 kHz square wave response. No NFB
Upper trace: OPT primary
Lower trace: OPT secondary
-
- 1 kHz square wave response. No NFB
Upper trace: OPT primary
Lower trace: OPT secondary
-
Mike H
- Amstrad Tower of Power
- Posts: 20189
- Joined: Sat Oct 04, 2008 5:38 pm
- Location: The Fens
- Contact:
#37
A-ha that's more like it, still seems to be some early HF roll-off in there though somewhere.
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
#39
I think the early HF roll-off may have something to do with the way I'm deriving the signal to drive the lower triode. The output of the pentode is just as sharp as the input square wave. So, the slow-down is happening in the output stage.
The NFB does sharpen things up considerably, but I left it disconnected for these tests so as not to complicate things.
Nick,
I tried a 10 kHz square wave, but didn't have time to post another scope trace. It was tricky to estimate the oscillation period from one little wiggle, but it does appear to be in the range of 50 - 60 kHz. With that in mind, I decided to sweep a sine wave signal very slowly across that range to try again to see any resonance at either the primary or secondary, but nothing showed up.
In an earlier post, I mentioned that the waveform was visibly distorted between 30 kHz and 50 kHz. However, I didn't notice it this time, either with or without NFB. I believe this may be due to the way I had previously set the drive level to the lower triode. I'd been doing it by applying a sine wave signal to the amplifier input, and then adjusting the drive level pot to get the maximum output without noticeable distortion in the sine wave. Doing it that way, I realize that I was overdriving the bottom triode. This probably led to some instabilities that I erroneously blamed on parasitic oscillation in the pentode, and hence the small value capacitors that got sprinkled around the circuit. Later, I realized that the proper way to set the drive level is to monitor the signal at the common cathode resistor, and adjust the pot for a null at the fundamental frequency. (Some 2nd harmonic distortion remains visible at the cathode, as would be expected.) Adjusted this way, there is a minor reduction in output power for a given input signal, but not enough to worry about, and it can be compensated for, by increasing gain in the pentode stage. Also, I've upped the supply voltage, as I realized that it was nowhere close to the maximum allowable plate dissipation.
The NFB does sharpen things up considerably, but I left it disconnected for these tests so as not to complicate things.
Nick,
I tried a 10 kHz square wave, but didn't have time to post another scope trace. It was tricky to estimate the oscillation period from one little wiggle, but it does appear to be in the range of 50 - 60 kHz. With that in mind, I decided to sweep a sine wave signal very slowly across that range to try again to see any resonance at either the primary or secondary, but nothing showed up.
In an earlier post, I mentioned that the waveform was visibly distorted between 30 kHz and 50 kHz. However, I didn't notice it this time, either with or without NFB. I believe this may be due to the way I had previously set the drive level to the lower triode. I'd been doing it by applying a sine wave signal to the amplifier input, and then adjusting the drive level pot to get the maximum output without noticeable distortion in the sine wave. Doing it that way, I realize that I was overdriving the bottom triode. This probably led to some instabilities that I erroneously blamed on parasitic oscillation in the pentode, and hence the small value capacitors that got sprinkled around the circuit. Later, I realized that the proper way to set the drive level is to monitor the signal at the common cathode resistor, and adjust the pot for a null at the fundamental frequency. (Some 2nd harmonic distortion remains visible at the cathode, as would be expected.) Adjusted this way, there is a minor reduction in output power for a given input signal, but not enough to worry about, and it can be compensated for, by increasing gain in the pentode stage. Also, I've upped the supply voltage, as I realized that it was nowhere close to the maximum allowable plate dissipation.
-
Mike H
- Amstrad Tower of Power
- Posts: 20189
- Joined: Sat Oct 04, 2008 5:38 pm
- Location: The Fens
- Contact:
#40
Sorry only just noticed the preset arangement properly, not seen that before and can't think why it should be necessary, if the o/p stage is also the splitter then the opposite grid is normally just grounded?
Also noticed the 12k screen resistor, the decoupling 4.7uF ought to be a larger value?
Anyway I did think there's a loss of treble in there somewhere.
Also noticed the 12k screen resistor, the decoupling 4.7uF ought to be a larger value?
Anyway I did think there's a loss of treble in there somewhere.
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."
#41
With a larger value cathode resistor, then the grid of Triode-2 could be grounded, but with the low value, it doesn't ensure constant current, and so some additional help is needed to get the Triode-2 output to match Triode-1.
Initially, I just used a voltage divider between Triode-1 anode and ground. However, there's a positive feedback problem with that arrangement. As the upper anode goes more positive, this takes the grid of Triode-2 more positive and hence the bottom anode goes more negative, as required. However, due to the effect of the transformer, this causes a more positive voltage to be induced at the anode of Triode-1 making it go even more positive, and hence Triode-2 anode goes even more negative, and so on. This made the original circuit quite unstable. By taking the bottom leg of the voltage divider to the opposite anode and transformer terminal, the positive feedback effect largely cancels out, and the circuit is very stable (as long as the drive level pot isn't set ridiculously high).
The combination of 12k screen resistor and 4.7uF cap gives a corner frequency of 17Hz. I guess I could increase it a bit, but I don't think it will make any audible difference with the type of speakers I intend to use. On the other hand, I should probably bypass it with a smaller value film cap to ensure that the electrolytic isn't running into ESR problems at high frequencies. This started out to be a cheap amp, not a good amp. It happened to turn out better than I expected, but my goal is to keep the circuit as simple as possible, as befits a one-valve amp.
Initially, I just used a voltage divider between Triode-1 anode and ground. However, there's a positive feedback problem with that arrangement. As the upper anode goes more positive, this takes the grid of Triode-2 more positive and hence the bottom anode goes more negative, as required. However, due to the effect of the transformer, this causes a more positive voltage to be induced at the anode of Triode-1 making it go even more positive, and hence Triode-2 anode goes even more negative, and so on. This made the original circuit quite unstable. By taking the bottom leg of the voltage divider to the opposite anode and transformer terminal, the positive feedback effect largely cancels out, and the circuit is very stable (as long as the drive level pot isn't set ridiculously high).
The combination of 12k screen resistor and 4.7uF cap gives a corner frequency of 17Hz. I guess I could increase it a bit, but I don't think it will make any audible difference with the type of speakers I intend to use. On the other hand, I should probably bypass it with a smaller value film cap to ensure that the electrolytic isn't running into ESR problems at high frequencies. This started out to be a cheap amp, not a good amp. It happened to turn out better than I expected, but my goal is to keep the circuit as simple as possible, as befits a one-valve amp.
#42
Possible reason for bad HF response: construction not exactly to military standards:
Since there are a few off-topic aspects to this amplifier, I started a new thread here:
http://www.audio-talk.co.uk/phpBB2/viewtopic.php?t=4713
Since there are a few off-topic aspects to this amplifier, I started a new thread here:
http://www.audio-talk.co.uk/phpBB2/viewtopic.php?t=4713
-
Mike H
- Amstrad Tower of Power
- Posts: 20189
- Joined: Sat Oct 04, 2008 5:38 pm
- Location: The Fens
- Contact:
#43
Oh I get it, actually I HAVE seen that done for phase-splitters, but forgot.BobWeaver wrote:With a larger value cathode resistor, then the grid of Triode-2 could be grounded, but with the low value, it doesn't ensure constant current, and so some additional help is needed to get the Triode-2 output to match Triode-1.
We know another way to get around the unequal gain don't we chaps?
"No matter how fast light travels it finds that the darkness has always got there first, and is waiting for it."