Added an output pad on EZ1290 to get Transformer Saturation but getting a-symmetrical clipping

[PatchDog]

Can anyone shed some light/knowledge on this? IS it to be expected? The a-symmetrical clipping is only occurring via the output transformer (Carnhill), not the amps, I know this because if I clip by raising gain and reducing the Trim Pot it's symmetrical. But engaging my output pad then hitting the gain hard till the TX drives only clips the lower half of the waveform, exactly the same occurs on both channels.

Could it be to do with the shunt .01uf - 1k5 resistor on the secondaries?

Voltage on primary pins reference from 0V is 23.3V & 22.6V if that may have something to do with it?

I was hoping to get some transformer saturation from this (which I am, it's just not symmetrical)

 

[KrIVIUM2323]

Hi,
Have you 'calibrated' your output stage as mentioned in Neve service manual?
The operation ask to load the ouput with 600r load, put an oscilloscope on the load, set the preamp for 26dbu output and use bias trimmer to observe symmetric clipping on oscilloscope.
If you have not then you know why yours behave like that.

The operation is described in page 20 of this document:
https://analogclassics.com/wp-content/uploads/2021/07/ez1290_build_guide.pdf

 

[PatchDog]

Hey man thanks for suggestion, yes I have done the biasing but admittedly only using the multimeter method (I have no scope) and setting the collector to ~22.66V,

I'd always been confused by the bias trimpot because it never changed the symmetry of clipping for me. However I've just realized I was always clipping the amp stage using the attenuator when trying, which looking at the schematic comes before the output bias stage. But now that I have recenty installed a pad on the output I'm able to clip the output stage of the 1290 without clipping my AD. So now maybe I can monitor that in my DAW and will see the change, I will try it.

 

[KrIVIUM2323]

Hi,
Well for the modules i've build i used the oscilloscope method as i think biasing through multimeter is not accurate enough ( i'm not saying it is not ok to do through dmm or this to be really important but let's say i have OCD ).
With my build biasing trimmer have an effect so maybe there is an issue in your build? I don't know.

i thought about the pad that could have an influence if the input impedance seen by the module is 'wrong' ( too low) but those output stage are theorically able to deliver a bit of juice...

Maybe someone with more insight and knowledge on this Neve like Ian could have an idea of what happen @ruffrecords ?

Edit: as you did edit to your last post, yes maybe your gain structure messed up what you seen before. If you can try to achieve the calibration procedure using a 600r bridged T pad on a breadboard if you have too much output: it will load the outputstage as it should and lower voltage to something acceptable by your AD ( 26dbu is high level).

Which kind of pad are you using atm? An Lpad or H pad? What values for resistors used?

 

[PatchDog]

Thank you friend much appreciate your help, it's much better now! It's not ideal but I'm using Cubase multiscope and test generator to dial the clip symmetry, but now that I have an output pad I can see the clip being adjusted as I trim the bias, awesome! It's not perfect but I have much better symmetry in the sweet spot of saturation.

If I push beyond the 'sweet spot' it still tends to go asymetrical, seems to be not much I can do about it it but this will work for my needs. Cheers

Sweet spot:


Not so sweet spot:


For the Pad I've been experimenting with different values, with help from Chat GPT, I think it's an H-Pad? 2 Series resistors and a shunt:

 

[KrIVIUM2323]

I'm glad you sorted the issue, but you did all the work.
The sweetspot look like on my oscillo, so it should be good. It's normal behavior to have unsymmetrical behavior when outside the good range ( before and after).

Ok about your pad i think it's a T pad or an L pad( and there is no such thing as an H pad, what i had in mind is called a balanced T pad, had a brain fart...).

Rather than using a 'learned idiot' ( chat gpt) about pad i would suggest something truthfull coming from a real intelligence owning a soul which doesn't spit out things in a dumb way ( ):

http://www.uneeda-audio.com/pads/

Worth a read imho, it can be of use one day or another.

 

[ruffrecords]

I use a 20dB pad consisting of a pair of 270 ohm resistors and a single 68 ohm which loads the driving source with close to 600 ohms.

The oscilloscope method is the only real way to set up the output stage. Other methods use a secondary indication which is not 100% reflective of the actual optimum bias.

You can buy a tiny hand held oscilloscope good enough for the job for around $20

Cheers

Ian

 

[PatchDog]

Rather than using a 'learned idiot' ( chat gpt) about pad i would suggest something truthfull coming from a real intelligence owning a soul which doesn't spit out things in a dumb way ( ):

http://www.uneeda-audio.com/pads/

Haha yeah for sure it is idiotic at times, but if you feed with enough data you get lucky sometimes lol, mostly not..... Cheers for the link, I did see this page but I couldn't figure out the math.

I use a 20dB pad consisting of a pair of 270 ohm resistors and a single 68 ohm which loads the driving source with close to 600 ohms.

The oscilloscope method is the only real way to set up the output stage. Other methods use a secondary indication which is not 100% reflective of the actual optimum bias.

You can buy a tiny hand held oscilloscope good enough for the job for around $20

Cheers

Ian

Thanks Ian, that is what I had heard before, I will defo get a cheap scope to do it properly.

Is your pad for the output? From the loading it appears so, this is what I was trying to get with Chat GPT (good loading for my Prism interface) but 20dB seems a bit much for me as wouldn't the the headroom be reduced? Since my Prism Line inputs only give me a switch between -10 and +4 (no gain control), so I run at +4, but -20dB means the level is low and have to raise the gain digitally (obviously don't want this) so I'm trying to find the sweetspot of saturation but have enough level that it is just below the +4 mark and doesn't clip.

For me -9dB roughly seemed correct (maybe more experimentation needed) and that was using 4.7k for series resistors and a 2.7k shunt, Chat GPT told me that was a 600ohm load but I'm not convinced that is correct.

My aim for the pads is to use the 1290 for line level stuff (I also have an input pad installed) so printing to and from my DAW mixes and drum stems to induce harmonics from the TX.