Gyraf Pultec low end loss

[ruffrecords]

Since the 5402 is not being driven as in the data sheet, the main determinant of the LF response is the primary inductance of the 5402. Anyone know what it is?

Cheers

Ian

 

[abbey road d enfer]

Since the 5402 is not being driven as in the data sheet, the main determinant of the LF response is the primary inductance of the 5402. Anyone know what it is?

Cheers

Ian

The 5402 is originally designed as a current xfmr for use in a Zero-Field topology. As such it has a low inductance; now I can't put a figure on it.

 

[ruffrecords]

The 5402 is originally designed as a current xfmr for use in a Zero-Field topology. As such it has a low inductance; now I can't put a figure on it.

Which would explain its poor LF performance in a conventional topology. Anyone tried to measure it?

Cheers

Ian

 

[gyraf]

The 5402 is originally designed as a current xfmr for use in a Zero-Field topology. As such it has a low inductance; now I can't put a figure on it.

No, it's not that it is made for anything like zero-field - that topology depends on keeping net transformer current cancelled by active means post-transformer like a virtual-earth mix bus. That is very different, and will only work for input transformers.

Lundahl never specifies inductance. I think I once measured the series-connected 5402 primaries to around 12H - but it was a primitive LC-meter, so I don't even know at what frequency that was. There was little doubt around here that it sounded right, so I never looked back...

Jakob E.

 

[abbey road d enfer]

No, it's not that it is made for anything like zero-field - that topology depends on keeping net transformer current cancelled by active means post-transformer like a virtual-earth mix bus.  That is very different, and will only work for input transformers. 

ZF can be applied to just any xfmr, but works better with low-inductance types. The parameters given for the 5402 show that its preferred application is not the typical medium-Z line level, even with the primaries in series..

Lundahl never specifies inductance. I think I once measured the series-connected 5402 primaries to around 12H

That is consistent with a typical source impedance of ca. 50 ohms when used with the two primaries in parallels.

 

[medway]

Have you noticed this transformer has a recommended drive impedance of 15 ohms?
Your soundcard may not be very capable of that.
At the output, even with the two primaries in series, that would be 60 ohms. The SRPP stage that drives it is probably about 10 times this.
It dpoes not mean it can't work, but definitely the LF response is somewhat sacrificed.

Thanks. Since this post I've read up quite a bit on how transformers work with impedance (admit my knowledge was quite thin before) and it all makes sense now.  It's probably a combo of the valves getting old (they are close to 10 years now inside this unit) and the DA.

Does coming into the unit unbalanced defeat the transformer? Looking at the schematics it seems to go around it but not sure. I think my soundcard has a lower output impedance on the unbalanced connection so might try that which should help at least.

 

[medway]

If your tube looses emission over time, low end can get weak.

I am aware of the design specs for the LL5402 (and have discussed this application with Per Lundall at several occasions) - but the way it behaves in this circuit is really nice. With jan6922's I usually get a -3dB point  between 22 and 28 Hz, which is good enough for most purposes...

Jakob E.

Seeing how mine as reduced over the years I'm sure when it was first built it was well into the very sub region and like you said perfectly fine (and even desirable for most applications).

What is a healthy impedance to drive it with? Is something sub 150 ok?

 

[jon83]

Seeing how mine as reduced over the years I'm sure when it was first built it was well into the very sub region and like you said perfectly fine (and even desirable for most applications).

What is a healthy impedance to drive it with? Is something sub 150 ok?

Did you ever figure this out? I'm having the exact same problem, curve is almost identical.

 

[abbey road d enfer]

Did you ever figure this out? I'm having the exact same problem, curve is almost identical.

If the info about the Lundahl xfmr is correct (12H), it is designed for low-ish source impedance, typically less than 800 ohms.
With Zs of 150 ohms, you should be fine.
Now there are a couple of things that may impact LF response.

• Core magnetized permanently; can be easily fixed by applying low frequency signal at max input level, with the unit off, and reducing progressively amplitude.
• DC current in one of the windings; check electrolytic caps.
• Mechanical damage to the core; very unlikely, and almost impossible to fix, but there should be visible signs

 

[jon83]

If the info about the Lundahl xfmr is correct (12H), it is designed for low-ish source impedance, typically less than 800 ohms.
With Zs of 150 ohms, you should be fine.
Now there are a couple of things that may impact LF response.

• Core magnetized permanently; can be easily fixed by applying low frequency signal at max input level, with the unit off, and reducing progressively amplitude.
• DC current in one of the windings; check electrolytic caps.
• Mechanical damage to the core; very unlikely, and almost impossible to fix, but there should be visible signs

I'm running out of a Prism, so pretty sure it's at 100ohms balanced, 50 unbalanced. Don't think the problem is there.

The curve is the same on both units I have, really strange. Starts dropping off around 200 - 300hz. If it's a component fault, it's the exact same problem on both units.

Just wondered if @medway had found the fault in his? Don't want to shell out on new tubes if it's something else.

 

[abbey road d enfer]

I'm running out of a Prism, so pretty sure it's at 100ohms balanced, 50 unbalanced. Don't think the problem is there.

Then you're left to troubleshooting #101. Follow signal ans see where it gets corrupted.

 

[jon83]

Then you're left to troubleshooting #101. Follow signal ans see where it gets corrupted.

Yeah, trying to delay the inevitable I think! No oscilloscope at the moment, was hoping to not have to fork out.

Thanks for the replies

 

[abbey road d enfer]

Yeah, trying to delay the inevitable I think! No oscilloscope at the moment, was hoping to not have to fork out.

Thanks for the replies

You don't necessarily need a scope. A scope is not particularly good at tracing frequency response problems. A soundcard with REW is more adequate.
Yoy have to know what you do, make sure that the levels you measure are not clipping the souncard input. If you have a soundcard with proper line inputs, that shoud do it.

 

[jon83]

You don't necessarily need a scope. A scope is not particularly good at tracing frequency response problems. A soundcard with REW is more adequate.
Yoy have to know what you do, make sure that the levels you measure are not clipping the souncard input. If you have a soundcard with proper line inputs, that shoud do it.

Really? I didn't know about that. Any chance you could explain a little? Which software would you recommend?

 

[jon83]

Is this the one?

https://www.roomeqwizard.com

 

[jon83]

You don't necessarily need a scope. A scope is not particularly good at tracing frequency response problems. A soundcard with REW is more adequate.
Yoy have to know what you do, make sure that the levels you measure are not clipping the souncard input. If you have a soundcard with proper line inputs, that shoud do it.

Ok, can't believe I didn't know about that. What a tool. How would you reccomend I use it to trace the signal?

 

[abbey road d enfer]

Ok, can't believe I didn't know about that. What a tool. How would you reccomend I use it to trace the signal?

You need to attach to your line input a very simple probe, i.e. a capacitor of about 1uF; with tube equipment it must be rated at about 250V minimum. Actually, before probing a point, you should check with a DMM what the DC voltage is, and rate the capacitor accordingly. If you're working on a Fender Twin, you're gonna need 500+V nominal voltage, but on line level stuff, 300V is generally adequate.
Another caveat is that when the probe touches a point with significant DC, it may damage the soundcard input if it's not protected. You should add a simple protection circuit as the one attached.
One end goes to the soundcard's input, the other you use to probe the signal as it goes from the input to the output, trough the various stages. You have to know that the moderate input impedance of the soundcard input may affect the circuit you're measuring.

This is just a basic jig; indeed there are more sophisticated arrangements, but this one should do the basic work.

 

[jon83]

You need to attach to your line input a very simple probe, i.e. a capacitor of about 1uF; with tube equipment it must be rated at about 250V minimum. Actually, before probing a point, you should check with a DMM what the DC voltage is, and rate the capacitor accordingly. If you're working on a Fender Twin, you're gonna need 500+V nominal voltage, but on line level stuff, 300V is generally adequate.
Another caveat is that when the probe touches a point with significant DC, it may damage the soundcard input if it's not protected. You should add a simple protection circuit as the one attached.
One end goes to the soundcard's input, the other you use to probe the signal as it goes from the input to the output, trough the various stages. You have to know that the moderate input impedance of the soundcard input may affect the circuit you're measuring.

This is just a basic jig; indeed there are more sophisticated arrangements, but this one should do the basic work.

Great, thanks! I'll give this a go.

 

[jokeramik]

Just a remark: I think the two output transformers are so close together, that they will interact. (Crosstalk)

(I suppose the two transformers above the mains filter are the output transformers.)

Also very near the mains input and the transformer.

These transformers should be as far apart as possible and mounted at 90 degrees to each other. So the interaction can be minimized.

Best regards!
jokeramik

 

[gyraf]

..the LL5402 are pretty well shielded, magnetic crosstalk is not a problem..