I have a question related to ringing of output transformers. I have observed this behavior in two different transformers from Lundahl in two different designs, the LL1620 3.3K 60mA and LL9202 6.5K 50mA.
In both cases, the ringing was isolated to the output stage, no visible ringing on the previous stages. Mains transformer ringing is snubbed in both cases.
Here is a 1kHz and 10kHz square on the LL1620 primary.
And 1kHz and 10kHz square on the LL9202 from the primary.
Is it worth pursuing a Zobel network or similar to further damp the ringing? On the LL1620, looks to be near the audio band on the 10kHz square.
In both cases, the ringing was isolated to the output stage, no visible ringing on the previous stages. Mains transformer ringing is snubbed in both cases.
Here is a 1kHz and 10kHz square on the LL1620 primary.
And 1kHz and 10kHz square on the LL9202 from the primary.
Is it worth pursuing a Zobel network or similar to further damp the ringing? On the LL1620, looks to be near the audio band on the 10kHz square.
It doesn't matter what size of output impedance of the driver.
If you drive OPT with low impedance tube (for example 700R as 300B) or medium-high impedance (5k as 801a) the HF behaviour -and tend to ringing- also different.
If you drive OPT with low impedance tube (for example 700R as 300B) or medium-high impedance (5k as 801a) the HF behaviour -and tend to ringing- also different.
Last edited:
Is it beneficial to damp this ringing with a Zobel network on the OPT secondary, or let it ring at frequencies > 20kHz?
I would expect it to be beneficial if you want to apply negative feedback. Having a highly complex pole pair (that is, something that rings) in the loop usually complicates keeping it stable.
When you don't want to apply feedback, have no pets who can hear the ringing and don't believe in a certain controversial Japanese research paper from the 1990's about gamelan players who supposedly could subliminally hear above 26 kHz, then I doubt that the ringing does any harm apart from not looking pretty on a scope.
When you don't want to apply feedback, have no pets who can hear the ringing and don't believe in a certain controversial Japanese research paper from the 1990's about gamelan players who supposedly could subliminally hear above 26 kHz, then I doubt that the ringing does any harm apart from not looking pretty on a scope.
The 9202 10k picture actually okay, there are a sharp valley (rupture) at about 90-100kHz, it's over the audio band.
The 1620 -in this connection- almost useless, IMHO the HF bandwidth not more than 2xkHz, and may over it there is a rough rupture.
The Lundahl "universal" OPTs in my opinion designed -well- only one connection, the others are less usable.
The 1620 -in this connection- almost useless, IMHO the HF bandwidth not more than 2xkHz, and may over it there is a rough rupture.
The Lundahl "universal" OPTs in my opinion designed -well- only one connection, the others are less usable.
MarcelvdG - thank you, you have answered my question, much appreciated.
euro21 - the LL9202 is wired for 6.5K as measured above, perhaps the squares do not tell the whole story. The sound of the LL1620 is quite good, despite the measurements.
LL1620 FR in 6A5G SET amplifier, no NFB
LL9202 in 801A A2 SET amplifier, pentode input with 801A plate to cathode NFB
I have used in two projects now, the Lundahls have HF bandwidth limitations, I may look elsewhere in the future. I am replacing the LL9202 with Sowter UK iron in my 801A A2 amplifier.
euro21 - the LL9202 is wired for 6.5K as measured above, perhaps the squares do not tell the whole story. The sound of the LL1620 is quite good, despite the measurements.
LL1620 FR in 6A5G SET amplifier, no NFB
LL9202 in 801A A2 SET amplifier, pentode input with 801A plate to cathode NFB
I have used in two projects now, the Lundahls have HF bandwidth limitations, I may look elsewhere in the future. I am replacing the LL9202 with Sowter UK iron in my 801A A2 amplifier.
Last edited:
The ringing does not look severe at all. Any cure will likely be as bad or worse than the disease.
I'd consider the one from the 9202 to be totally acceptable. The other one I might try to do something about but I'm not sure what the best way to attack it would be.
When I have ringing like that, and no negative feedback, the first thing I do is turn on the scopes time cursors, and set them to measure the time of one ring cycle. The cursor also reports the ring frequency.
Example:
I measure the ring frequency to be 50kHz.
Then I think about the following:
There is no negative feedback in my amplifier.
My tweeters do not go to 50kHz.
My ears do not go to 50kHz.
The rise time of my square wave generator is 1 usec.
0.35/((1 x (e-6)) = 350kHz.
The bandwidth of my square wave is 350kHz.
The bandwidth of my CD player is 44.1/2 = 22.05kHz (much lower bandwidth than the square wave generator).
Nothing in that CD player is going to activate the 50kHz ringing (unless I have a dirty CD player with a 50kHz spur; and I can check for that with my scope.
A square wave coming from that amplifier that either is phase shifted at 50kHz, or that has amplitude error at 50kHz could cause the square wave to have 50kHz riding on the top and bottom flat sections.
The ringing is not due to 50kHz oscillation.
It is due to phase shift and amplitude response at very high frequencies (like 50kHz).
Although the 50kHz ringing does not look pretty, it does not affect the sound.
I connect a speaker and a CD player, put a CD in, and enjoy the music.
Example:
I measure the ring frequency to be 50kHz.
Then I think about the following:
There is no negative feedback in my amplifier.
My tweeters do not go to 50kHz.
My ears do not go to 50kHz.
The rise time of my square wave generator is 1 usec.
0.35/((1 x (e-6)) = 350kHz.
The bandwidth of my square wave is 350kHz.
The bandwidth of my CD player is 44.1/2 = 22.05kHz (much lower bandwidth than the square wave generator).
Nothing in that CD player is going to activate the 50kHz ringing (unless I have a dirty CD player with a 50kHz spur; and I can check for that with my scope.
A square wave coming from that amplifier that either is phase shifted at 50kHz, or that has amplitude error at 50kHz could cause the square wave to have 50kHz riding on the top and bottom flat sections.
The ringing is not due to 50kHz oscillation.
It is due to phase shift and amplitude response at very high frequencies (like 50kHz).
Although the 50kHz ringing does not look pretty, it does not affect the sound.
I connect a speaker and a CD player, put a CD in, and enjoy the music.
Last edited:
L0rdGwyn,
Are you driving those output transformers with an 801A triode?
I think the 801A has a plate resistance of about 3k.
Your frequency response curves look like the high frequency rolloff is at about 15kHz.
Are you using one of the recommended connection diagrams on the Lundahl data sheet?
Are you using the recommended driving impedance for that connection?
In my experience with various Lundahl interstage transformers, they only worked good when using both an exact recommended connection and the recommended driving impedance.
The Lundahl output transformers I have seen seem to use the same construction and winding techniques as the interstage transformers.
I would expect them to need to follow the same rules.
Are you driving those output transformers with an 801A triode?
I think the 801A has a plate resistance of about 3k.
Your frequency response curves look like the high frequency rolloff is at about 15kHz.
Are you using one of the recommended connection diagrams on the Lundahl data sheet?
Are you using the recommended driving impedance for that connection?
In my experience with various Lundahl interstage transformers, they only worked good when using both an exact recommended connection and the recommended driving impedance.
The Lundahl output transformers I have seen seem to use the same construction and winding techniques as the interstage transformers.
I would expect them to need to follow the same rules.
Last edited:
Thanks for the input everyone.
6A3sUMMER - thank you for explaining your method. The LL1620 is driven by the 6A5G, Rp 800ohms. Lundahl doesn't prescribe drive impedance in their OPT datasheets, only provides nominal primary impedance, in this case 3.3K.
The LL9202 is configured for 6.5K:8ohm, it is driven by an 801A at A2 bias with NFB taken from the plate to the cathode of the pentode gain stage. HF bandwidth improves with additional feedback with THD at 1W at 0.1%, but even then I seem to hit a point where transformer limitations kick in, never better than -1dB at 20kHz. The LL9202 will be replaced with a higher bandwidth transformer, this amplifier is being prototyped.
I'll measure the ringing frequency of the LL1620 more accurately on my scope and see if any intervention is needed, but I must say the amplifier sounds great as built, thought I would consult the forum to see if troubleshooting is worth the effort.
6A3sUMMER - thank you for explaining your method. The LL1620 is driven by the 6A5G, Rp 800ohms. Lundahl doesn't prescribe drive impedance in their OPT datasheets, only provides nominal primary impedance, in this case 3.3K.
The LL9202 is configured for 6.5K:8ohm, it is driven by an 801A at A2 bias with NFB taken from the plate to the cathode of the pentode gain stage. HF bandwidth improves with additional feedback with THD at 1W at 0.1%, but even then I seem to hit a point where transformer limitations kick in, never better than -1dB at 20kHz. The LL9202 will be replaced with a higher bandwidth transformer, this amplifier is being prototyped.
I'll measure the ringing frequency of the LL1620 more accurately on my scope and see if any intervention is needed, but I must say the amplifier sounds great as built, thought I would consult the forum to see if troubleshooting is worth the effort.
I wouldn't bother with such resonances when a transformer global nfb loop is not employed.
My experience as an audio transformer builder also says that materials type (core, dielectrics, wire) contribute to the most sound of the transformer. Frequency response much less, if it's not disastrous, if course.
My experience as an audio transformer builder also says that materials type (core, dielectrics, wire) contribute to the most sound of the transformer. Frequency response much less, if it's not disastrous, if course.
Last edited: