LS28 power supply problem

[AudioIdiot]

Hi W9TR,
It is an R-core transformer,
And I expected that one bobbin contained the primary windings and the other the secondary windings.
But they both contain primary and secondary. So the theoretical low capacitance coupling between primary and secondary is not happening here.

As the primary is fried I have to unwind first the 4 secondary windings. So I am not going to do that.

I have ordered a custom build 120VA toroidal transformer. Changed some voltages, added a secondary winding for the +/- 12,6V that is used for the attenuator and input bias stage as they were being sourced by a single winding with two half wave rectifiers. Added an electrostatic screen between primary and secondary.
As the standard transformer looked like a 50VA type my guess is that the toroidal one will supply a cleaner voltage because of the headroom I now have.

I will mount the transformer on an aluminum plate and will put a magnetic screen around it.
I also found some rubber stand-offs with threaded mounting holes and will try to use these to further mechanically decouple the transformer from the amp.

As I have a full set of measurements with the original transformer I can compare if the already very low hum is further improved.

[W9TR]

That’s odd they would put the primary and secondary on the same bobbin on an r-core. Very strange. Defeats the primary purpose of an r-core. Anyway I look forward to your results!

[AudioIdiot]

So I promised to share my findings.

Audio Research LS28 before and after gain modification:

The measurements are all taken with the balanced I/O and unless otherwise noted the load impedance to the output is 100k using an Audio Precision System 2 Cascade.

Some static measurements:
Residual noise 20-22k
Before modification 18,77dB gain = -97 dBV L / -98 dBV R
After modification 12,95dB gain = 101 dBV L / 104 dBV R
This is the output noise with the volume at 103 and referes to an output voltage of 1V.

The 12V trigger has an open voltage of 12,45V and with a load of 7mA it dropped to 12V. This calculates to an output impedance of 63 Ohms.
The voltage seems to be protected by a PTC

The basic Schematic of the LS28. A very clever zero feedback balanced design. The two mirror imaged sections are the amplifiers for the hot and cold signal. The input stage is biased by a single current source that allows unbalanced to balanced conversion.

Channel mismatch is lower than 0,1 dB > 100k and lower than 0,005 dB from 20k to 22kHz
The modified gain has a slightly higher bandwidth with -1,75dB @ 200k versus -2,2 dB @ 200k for the original gain.


Remarks:
The output capacitor of 10µ/50V is under specified with 70V across the cap !!! It should at least be a 100V model. I think this is a liability and could cause quite some problems.

The 2A Slow blow fuse is to big for a 50VA transformer-based design. The total power dissipation sits around 45W. I would advise a for 115V mains : 0.8A 5x20mm IEC127 high I²t-value. For 230V mains: 0.4A 5x20mm IEC127 high I²t-value.

I also noticed that the voltage feeding the 180V regulator is 315V. That is quite a waste and unnecessary power dissipation. The transformer can be used from 100V to 250V AC so a large margin is not needed. I expect this comes from earlier tube rectifier-based designs where the voltage drop over the rectifier is much higher.

I can't add a PDF here or pictures that are not on the web, but I can tell you the unit measures absolutely fabulous. Very high bandwidth that is slightly influenced by the volume control position. Very low harmonic distortion up to 4V RMS into 100k Ohm with 2nd harmonics below -100dB. Excellent.

It is a pity that some choices are a bit strange. The fuse rating is absolutely wrong. The 50V output caps are dangerous. and I don't understand the ridiculous high voltage that is feeding the 180V regulator.

So far an update on the unit. Still waiting for the transformer.
Peter

[tdelahanty]

Originally Posted by W9TR

That’s odd they would put the primary and secondary on the same bobbin on an r-core. Very strange. Defeats the primary purpose of an r-core. Anyway I look forward to your results!

I'm not sure what an R-core transformer is, I've never heard that term. If both windings are on the same bobbin that is referred to as bifilar winding. This close coupling of primary and secondary increases efficiency of the transformer. I have known this type to be called an E-core, however I'm not the "last word" on transformer design???

[tdelahanty]

Originally Posted by AudioIdiot

So I promised to share my findings.

Audio Research LS28 before and after gain modification:

The measurements are all taken with the balanced I/O and unless otherwise noted the load impedance to the output is 100k using an Audio Precision System 2 Cascade.

Some static measurements:
Residual noise 20-22k
Before modification 18,77dB gain = -97 dBV L / -98 dBV R
After modification 12,95dB gain = 101 dBV L / 104 dBV R
This is the output noise with the volume at 103 and referes to an output voltage of 1V.

The 12V trigger has an open voltage of 12,45V and with a load of 7mA it dropped to 12V. This calculates to an output impedance of 63 Ohms.
The voltage seems to be protected by a PTC

The basic Schematic of the LS28. A very clever zero feedback balanced design. The two mirror imaged sections are the amplifiers for the hot and cold signal. The input stage is biased by a single current source that allows unbalanced to balanced conversion.

Channel mismatch is lower than 0,1 dB > 100k and lower than 0,005 dB from 20k to 22kHz
The modified gain has a slightly higher bandwidth with -1,75dB @ 200k versus -2,2 dB @ 200k for the original gain.


Remarks:
The output capacitor of 10µ/50V is under specified with 70V across the cap !!! It should at least be a 100V model. I think this is a liability and could cause quite some problems.

The 2A Slow blow fuse is to big for a 50VA transformer-based design. The total power dissipation sits around 45W. I would advise a for 115V mains : 0.8A 5x20mm IEC127 high I²t-value. For 230V mains: 0.4A 5x20mm IEC127 high I²t-value.

I also noticed that the voltage feeding the 180V regulator is 315V. That is quite a waste and unnecessary power dissipation. The transformer can be used from 100V to 250V AC so a large margin is not needed. I expect this comes from earlier tube rectifier-based designs where the voltage drop over the rectifier is much higher.

I can't add a PDF here or pictures that are not on the web, but I can tell you the unit measures absolutely fabulous. Very high bandwidth that is slightly influenced by the volume control position. Very low harmonic distortion up to 4V RMS into 100k Ohm with 2nd harmonics below -100dB. Excellent.

It is a pity that some choices are a bit strange. The fuse rating is absolutely wrong. The 50V output caps are dangerous. and I don't understand the ridiculous high voltage that is feeding the 180V regulator.

So far an update on the unit. Still waiting for the transformer.
Peter

I've owned older (70's- 80's) AR gear and found many of their choices seem to be poor engineering or compromises. Under-rated cap voltages, excessive input voltages to regulator stages, too high (excessive) tube bias etc. I'm not surprised by what you have mentioned. During my 33yr career in military electronics reliability has been of utmost importance.

[W9TR]

Originally Posted by tdelahanty

I'm not sure what an R-core transformer is, I've never heard that term. If both windings are on the same bobbin that is referred to as bifilar winding. This close coupling of primary and secondary increases efficiency of the transformer. I have known this type to be called an E-core, however I'm not the "last word" on transformer design???

R-cores are similar to a toroid core and are wound from a continuous strip of metal and are formed into a shape with two straight sides. The straight sides accommodate two bobbins that are wound on the core.

Instead of using a constant-width strip of metal, the R-core is wound from a strip of varying width, so that the finished core winds up with a circular cross section.

Usually the primary is wound on one bobbin and the secondaries on the other. This provides high electrostatic isolation and reduces inductive coupling between the primary and secondary.

BTW in bi-filar winding multiple wires are wound together as a ‘bundle’.

In a normal layup one wire is wound on the bobbin, then an insulating layer, then the next winding, etc.