Amek AT8 Hybrid Line Driver Questions

[limo]

This is my BC3 circuit diagram on page5, but your schematic the R19 connection in your circuit diagram is different.
R19 and U1D+12 should be connected together, not to GND.

 

[GreyboxAudio]

This is my BC3 circuit diagram on page5, but your schematic the R19 connection in your circuit diagram is different.
R19 and U1D+12 should be connected together, not to GND.

Thanks for sharing! It adds to the confusion tho. Because if I connect the circuit this way I get +39dB of gain and a low pass filter at 12kHz

 

[GreyboxAudio]

For reference: I've stitched together several see-through pictures of the PCB and will try to re-trace everything from scratch.
Update: it looks like everything is connected identically to the schematic.

 

[mhelin]

I don´t know if I posted this already: https://mae.tcrcloud.de/index.php/s/rq8Kxt2A27Rc2c4?path=/Amek/Amek BC3

They oddly specified LL1539 to be used as an input transformer as it is a balanced driver output transformer really (must be an option as the schematic doesn't make sense otherwise)...

 

[jensenmann]

When I was reworking a BC3 years ago I was really astonished to find LL1539 at the inputs. Indeed it was an option for broadcasters. I don´t remember whom I asked but I´ve been told that Rupert Neve specified these to be used in this spot. Thanks to the very low DCR it´s a good choice regarding noise.

 

[Matt Syson]

AMEK used a LL1539 as an input transformer (option) in several broadcast orientated desks as it can handle LINE level signals and was still reasonable for Mic level. A 1:1 transformer doesn't really care where it is located in a circuit. it is worthwhile to note that you should be careful not to accidentally magnetise the core of any transformer as it will become more distorted and possibly a touch microphonic as I vaguely remembrer a discussion between AMEZK and Lundahl about it (33 years ago). Lundahl Transformers are good candidates for negative resistance LF distortion improving circuits as used by other desk manufacturers. AMEK just used them as straight isolation when the customer needed galvanic isolation.

 

[sahib]

They oddly specified LL1539 to be used as an input transformer as it is a balanced driver output transformer really (must be an option as the schematic doesn't make sense otherwise)...

I have the test report and the internal memo for that. I just have to look through which folder it is in.

 

[GreyboxAudio]

For reference: I've stitched together several see-through pictures of the PCB and will try to re-trace everything from scratch.
Update: it looks like everything is connected identically to the schematic.

I've modified one of my MZ-15-RN channels so that I can easily swap the Insert send between the original AT8 and my own PCB. Even with all resistances matched the original AT8 still behaves differently and I'm starting to wonder if the compensation caps of the two NE5534 did serve a bigger purpose

 

[GreyboxAudio]

I'm slightly losing my mind about this. I have set up the buss output on a breadboard so that I can easily swap the AT8 boards and eliminate any influnces from the overall Mozart Channelstrip. I'm sending a 1Vpp 1kHz Sine wave into it and am measuring both outputs. With the original AT8 everything is nice and symmetrical and with my rebuilt AT8 it's not. I have measured and compared resistances between all pins and have even added a trimmer to match the original untrimmed resistor at Pin 11. And even though my AT8 rebuild only uses 1% resistors instead of laser trimmed resistors, this should not lead to a >500mV difference in output signals.

If anyone has any ideas or insights into this circuit my sanity will probably thank you

Currently my last resort is to breadboard an AT8 with NE5532/NE5534 for comparison. If that still measures differently then it's probably the ghost of Rupert Neve haunting my circuits or AMEK used Quantum Tunneling in the output stage...

 

[GreyboxAudio]

Okay, I'm completely lost now. The protoboard version I built today with an NE5532 and two NE5534 performs to spec. If I remove the compensation caps of the NE5534 it still performs to spec. If I replace the NE5534 with OPA1677 it still performs to spec. And when I replace the NE5532 with an LM4562 it still performs to spec.
So the compensation caps of the NE5534 don't matter. The Op-Amps don't really matter. And yet boards I had manufactured at JLCPCB don't perform to spec and even my SPICE simulation does not perform to spec
What in little shelford is going on here?!

 

[limo]

My son is studying semiconductor design and image sensors at university.
I talked to him a little this morning. In his opinion, if the prototype version was working correctly, the lines on the board were probably too thin.
It's no wonder that a 1% error in resistor can make a huge difference depending on board design and layout.

He told me so and showed me some photos.
This is an NEC floppy drive IC that I happened to have recently cracked.



Printing the resistor pads is the very technology used to manufacture potentiometers.
Resistors and pattern lines could be printed from an early stage and were cost-effective, so it seems to have been used when producing hybrid ICs by gathering transistors and ICs.

Toshiba power amplifier IC "TH9013P" was also manufactured using a similar method.
http://www.easyaudiokit.com/bekkan2019/retro-amp/amp.html

 

[GreyboxAudio]

How do you mean "the lines on the board were probably too thin."?

 

[amplexus]

How do you mean "the lines on the board were probably too thin."?

Possibly traces to thin or spacing to thin- if the resistance/capacitance tolerances are that tight a slight variation due to trace width or stray capacitance might be causing issues?