LL1660 P PZ Lundahl interstage ringing problem GM70pp

[adam2a3]

Hi everyone, I use an LL1660 P PZ gapless Lundahl interstage in the following configuration: Primary is parafeed driven (10uF cap) by a pair of RY50 triode connected in SRRP topology at approx 1KV upper anode voltage.
The secondary is configured as phase splitter driving the grids of a GM70 pp pair. I have noticed a severe ringing on the square wave, plus some resonance spots at approx 8 and 14 khz.
I tried loading the secondaries by a pair of resistors and concluded at a value of 4K7 in parallel with each secondary. This made things way better, though square looks like a sine over 16Khz.
A further experimentation by adding one capacitor in series with each resistor concluded at 3000pF in series with 4K7. This combination (dual zobel) provides more amplification and slightly better hf response compared to single resistor.
Any comments or ideas about a better approach to this problem?

[Bandersnatch]

hey-Hey!!!,
That is the second time I've heard of these ringing. Friend ov mine built PP 813 amps( triode-strapped ) with 6BL7 LTP driving 1660's and found ringing at 30 kHz at the interstage. He was less successful at taming the ring and leaving it sounding OK.
cheers,
Douglas

[revintage]

As Bandersnatch says you should leave it ringing. Don´t mind whats happening on the scope, it will sound best anyway. Loading the secondaries will make the amp be soundwise loose.

[LinuksGuru]

Originally Posted by adam2a3

I have noticed a severe ringing on the square wave, plus some resonance spots at approx 8 and 14 khz.

How big are these resonant points? If its 0.5 - 1 dB or less, it may be magnetic property of the core, and not resonant.

Square wave test is meaningless for audio transformers, it makes sense for high-frequency units used in switching power supplies.

[Shoog]

I disagree with the comments about leaving it ringing. I have found this sort of ringing to be very audible and quite fatiguing over time.
Lundahl is quite aware of this and usually suggest suitable zobel values for their transformers on their datasheets.

Shoog

[Bandersnatch]

Originally Posted by revintage

As Bandersnatch says you should leave it ringing.

I did *NOT* say leave it ringing. While the methods used to damp it left it sounding worse, I did not mean to imply that the amp was satisfactory with the ringing IT's....
cheers,
Douglas

[revintage]

OK, as Bandersnatch didn´t say you should leave it ringing. This will normally sound the best!

Shoog,
Lundahl usually suggest to leave the 1660 ringing. There is also no recommendations for a Zobel network for this transformer.

[adam2a3]

Thank you all for your remarks, I have still not evaluated any audible differences with and whithout resistors or zobels. I plan to use a relay for a/b comparison.
The sound in general is impressively good (one channel only, driving 4 Quad ESL-57 speakers in parallel-second channel is under construction). Bass guitar and bass drums sound significantly more real as compared with a GM70 Single ended amp.
At the two resonance frequencies (unloaded secondaries) the hight of mostly left side of square wave is getting almost double, as observed on the scope.
The scope is always conected at the output transformer secondary (at the dummy load) meaning that this transformer will not cure any previous stage anomalies.
Lundahl does suggest loading by resistors in this kind of circuit, but is not mentioning any values.
Another point for experimentation is to remove an electrolytic from LL1660 center tap to ground (that is bypassing neg bias pot wiper to ground and closing input signal circuit. Instead of using this electrolytic I am thinking about connecting the LL1660 secondary center tap directly to all GM70 filament ends (4 film capacitors in total). There is also a 10uF 2kV oil cap that goes from output transformer's center tap to ground . How about connecting it at the LL1660 center tap instead of going to the ground so it will close the output (anode) signal circuit directly?
Regards,Adam

[45]

Lundahl does suggest loading if the source impedance is substantially lower than specified.
See the (*) comment which is true for any configuration:

http://www.lundahl.se/pdfs/datash/1660.pdf

Cheers,
45

[Wavebourn]

You definitely need a transformer with better coupled primaries and secondaries for such a purpose. Load would help in case of excess stray capacitances, but in your case it would roll-off highs.