RF filter for a transformer input line amp

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Winston O'Boogie said:
What is the desk, is there a schematic anywhere?  What was there before for line in?  That sort of stuff etc.
If the line input was happy before, there's no reason it can't be now with a transformer there instead. 

How much butchery do ya wanna do?  ;D

It's a Soundcraft 200b, I never liked the dual transistor input with a cap inbetween and the DC servo that followed it. So I threw the input circuit out and put this new one in...
 
You might have output offset issues with that circuit. Presumably you will have a resistor in series with the 10K pot to limit gain to +20dB or whatever is appropriate for line in in which case you could just put a cap in series as well to AC couple and choose the cap value based on gain limit resistor. For +20dB max that would be ~100 ohms which would necessitate a cap of at least 100uF. Or put the cap on the output but then your pot might still be scratchy if it's the cheap variety.
 
living sounds said:
It's a Soundcraft 200b, I never liked the dual transistor input with a cap inbetween
Those transistors can significantly improve noise performance depending on the source Z. They were there for a reason.
 
living sounds said:
It's a Soundcraft 200b, I never liked the dual transistor input with a cap inbetween and the DC servo that followed it. So I threw the input circuit out and put this new one in...

What squarewave said really. 
Which IC is your new gain setting one according to the scheme attached, IC1 or IC2? 
No mic input at all now?
 

Attachments

  • Screenshot_2019-08-23-00-15-12-555.jpeg
    Screenshot_2019-08-23-00-15-12-555.jpeg
    227.1 KB · Views: 21
squarewave said:
You might have output offset issues with that circuit. Presumably you will have a resistor in series with the 10K pot to limit gain to +20dB or whatever is appropriate for line in in which case you could just put a cap in series as well to AC couple and choose the cap value based on gain limit resistor. For +20dB max that would be ~100 ohms which would necessitate a cap of at least 100uF. Or put the cap on the output but then your pot might still be scratchy if it's the cheap variety.

No problem with offset. I have a cap between the ground and the pot, but bridged it, because I didn't like the sound and it wasn't necessary.
 
Winston O'Boogie said:
What squarewave said really. 
Which IC is your new gain setting one according to the scheme attached, IC1 or IC2? 
No mic input at all now?

IC2, IC1 is gone. There really is no problem with noise, since I only use it as a line input with rarely more than 10db of gain (just for mixing).
 
I made more tests, and the problem doesn't seem to be transformer related. With the line switch off (and the transformer disconnected) I still get noise/oscillations down the line (turning up the high end of the EQs), unless C6 is present.
 
Here's a funny thing: If I send pink noise through the preamp, looking at the analyzer, the amplitude at the top of the spectrum actually increases while holding a 10n cap in the position of C6. Does it form a resonant circuit with the transformer somehow?

Edit: Yes, definitely a resonant circuit using a 100n cap, visible peak...
 
Get the channel all happy without the transformer first.
Not sure which way I would go right now, will think about it overnight.

One way, if noise isn't an issue, might be to have IC2 have the gain already in hand if it's not too much.  Sounds like it's 10dB if I read you correctly.  Then just use a pot as an input 'pre set' fader after the transformer, before the amp.
It's old school and not the best.  But you're not fiddling with IC2's gain.

(Edit:  I'd change the pot for a 1K log with a 100R at the bottom  in this case.)

Regardless, get IC2 behaving on its own, with EQ boosted or whatever causes problems with oscillation. 
 
some transformers will pass 1 M Hz and more, what is the definition of RF?

you could try some ferrite beads, they do good aginst all kinds of gremlins and will not hurt bandwidth.

measure your transformers out of circuit, inject into pri, look at unloaded sec response, add zobel to see what happens where.  you do not get something for nothing, zobel's work but they suck juice. could be an issue at a mic input where source drive ability is low.

best to suppress noise at the source, snip the antenna lead to the transmitter tower, wait til they sign off.

wind your own beads. those pre-packaged resistor type ferrite  inductors have terrible capacitance which will defeat purpose.
 

Attachments

  • ferrite beads.png
    ferrite beads.png
    10.8 KB · Views: 16
Thanks!

It seems RF coming through the transformer wasn't really the problem, I got instability even with the transformer disconnected. adding the series resistor and a small cap to ground directly before the op amp input seems to have cured the problem. Reducing the feedback resistor to 4.7k seems to work well, too.
 
living sounds said:
Thanks!

It seems RF coming through the transformer wasn't really the problem, I got instability even with the transformer disconnected. adding the series resistor and a small cap to ground directly before the op amp input seems to have cured the problem. Reducing the feedback resistor to 4.7k seems to work well, too.

NE5532 and the like are remarkable stable and should not need extensive tweaking to make them so. Often, simple engineering aspects like layout and proper decoupling at the chip are the culprits.

Cheers

ian
 
CJ said:
some transformers will pass 1 M Hz and more, what is the definition of RF?
um "radio frequency"....  Of course RF transformers pass RF.  AM radio band is nominally just below your 1MHz data point.
you could try some ferrite beads, they do good aginst all kinds of gremlins and will not hurt bandwidth.
does he actually have RF problem (typically rectification in active circuits that are not fast enough to keep up. ) Don't fix it if it ain't broke...

JR
measure your transformers out of circuit, inject into pri, look at unloaded sec response, add zobel to see what happens where.  you do not get something for nothing, zobel's work but they suck juice. could be an issue at a mic input where source drive ability is low.

best to suppress noise at the source, snip the antenna lead to the transmitter tower, wait til they sign off.

wind your own beads. those pre-packaged resistor type ferrite  inductors have terrible capacitance which will defeat purpose.
 
ruffrecords said:
NE5532 and the like are remarkable stable and should not need extensive tweaking to make them so. Often, simple engineering aspects like layout and proper decoupling at the chip are the culprits.

Cheers

ian


There is a standard 10uf electrolytic as well as a 100nf ceramic directly at both +V and -V pins of each op amp, going to a star ground on each channel PCB, all connected via low impedance cables going to a star ground at the center of the console frame.

It might be the improvised layout at the preamp input that is to blame, but the input filter proposed above seems to get rid of the problem so far.
 
living sounds said:
It might be the improvised layout at the preamp input that is to blame, but the input filter proposed above seems to get rid of the problem so far.

So you have in there a 4K7 and 680pF right now?  That seems quite brutal to me, both noise penalty wise and -3dB break point.
A series R  that's more towards a decade lower was where I was aiming. 


 
Winston O'Boogie said:
So you have in there a 4K7 and 680pF right now?  That seems quite brutal to me, both noise penalty wise and -3dB break point.
A series R  that's more towards a decade lower was where I was aiming.

No, forgot to upload the update, please see the attached file.

I have 1.1k in parallel (termination), 4.7k in series followed by 100p going to ground. And the feedback cap is now 4.7k.

BTW, if a low resistance feedback resistor is a problem for the op amp to drive, what about voltage followers?
 

Attachments

  • MyConsoleInput2.jpg
    MyConsoleInput2.jpg
    80.1 KB · Views: 15
OK got it.  Then go 470R series R and 1000pF shunt. 
The 5532 won't break a sweat driving that feedback load and following IC.  Doug Self shows AP distortion plots that are negligible at much worse loads than this.
 
Winston O'Boogie said:
OK got it.  Then go 470R series R and 1000pF shunt. 
The 5532 won't break a sweat driving that feedback load and following IC.  Doug Self shows AP distortion plots that are negligible at much worse loads than this.

Thanks, but why? Higher series resistance will decrease any effect on the transformer, won't it? Though I guess it may increase noise...

The Signetics / Phillips 5532 (I've got the ceramic variety rated for up to 125 ° C) get quite a lot warmer than the TI version (actually hot to the touch), but its distortion at the higer end measures lower in that circuit. That's in line with Samuel Groner's and Douglas Self's findings. Sounds better, too. There really is no chip like the vintage 5532/5534 availible today...
 
living sounds said:
Thanks, but why? Higher series resistance will decrease any effect on the transformer, won't it? Though I guess it may increase noise...

The Signetics / Phillips 5532 (I've got the ceramic variety rated for up to 125 ° C) get quite a lot warmer than the TI version (actually hot to the touch), but its distortion at the higer end measures lower in that circuit. That's in line with Samuel Groner's and Douglas Self's findings. Sounds better, too. There really is no chip like the vintage 5532/5534 availible today...

Lower series resistance will be better from a noise standpoint.  Take me out back and shoot me if I'm wrong, but I don't see it doing anything negative to the transformer tbh. 
The actual and reflected dcr and impedance on the secondary are all in parallel with your 1K1 loading resistor and then all that is in series with the series R of your input filter.  I would keep the whole lot under 1K myself. 
As I said at the start, you might be able to just up the value of the shunt C to 1800pF or 2200pF and add a smallish 200R or so series R.

I also think your original value of feedback resistor would be no sweat for your 5532 and, with your pot set at a value for usual gain needed, would be a better choice from a noise standpoint. 

But it's your desk dude, do what you need to or feel is best my friend 😊


 

Latest posts

Back
Top