What does a Neve 1073 and a Behringer ADA-8000 Have in common?
Not a lot. Unless of course you are like me and decide to transplant some Neve soul into an ADA-8000. I have this Behringer ADA-8000 with a burned out power supply anyway … time to experiment!
You are looking at the main section of the Behringer ADA-8000. The red arrow points to the shorting/burned out capacitor that destroyed the transformer. After ordering a new transformer and getting everything to work, I started to tinker.
In the picture you can see the power supply and the SPDIF to ADAT I/O chips (the Alesis Semiconductor AL1402 ADAT optical decoder and AL1401A ADAT optical encoder – no. 1 and 2 are on left). The clock circuitry is also nearby. One chip is for multiplexing 4 stereo SPDIF channels into a single ADAT out, and the other operates in reverse, taking a single ADAT Channel and splitting it into 4 stereo SPDIF outs. These are then connected to the actual D/A and A/D converters.
First I thought if I change the opamp on the input section I could get better performance. Wrong. The old ADA-8000 used a TL074 opamp. The new machine is much better and has a very low noise and high end opamps which are very hard to beat. At least I tried – this is my input section with a replaced opamp (the bottom right one), the bottom left one is the original. What we have here is the input opamp of 2 sections inputing into an Alesis Semiconductor AL1101 A/D converter (IC17) , it sends a stereo SPDIF signal to the control section (seen above):
So the first step really was to find the schematic, After some searching I found it here: http://www.gyraf.dk/schematics/Behringer_ada8000_analouge.PDF Thank you Gyraf Audio for another great job well done!
looking at the input section of the schematic:
I had a Sowter 9145 transformer lying around – which is equivalent to a Neve P/N 10468 and a spare NEVE BA284 board. The first thought was to bypass the input section completely and transplant the whole BA283 preamp section instead of the existing one. However exmining the existing circuit I noticed that the input section is an ultra low noise class A amplifier – it is a design that has been used in SSL consoles, monolithic preamp chips (such as the SSM2019) and a very good description of a similar preamp by DIY wiz Rod Eliott is here. This seems quite decent, so i decide to modify the input section a bit and add just transformer into the soup. Here is the modified schematic as I planned it:
The idea is to feed the fantom power via the center tap thus reducting the noise coming from the +48 supply. The also improved the balance in phantom situations. The 6k resistors are needed to load the trasformers properly, although in initial test I decided (read – lazy) to leave the original ones in place to “see if loading is sufficient”. So now I had the design, and I was ready to tackle the task of modifying the actual hardware! After much dismantling, I managed to seperate the front from the back. The front and back are seperated by cables that carry the preamplified signal to the converters. All the mods are going to run on the front end, so we might as well get to know it.
This is the back side of the front panel. After measuring all the components – I determined which ones needed to be soldered and changed:
Here u can see the following changes: temporarily removed the XLR to allow access to the components. The caps were removed and the 6.81 resistors changed to send the 48V to the center tap of the transformer. Now the transformer is tacked on:
This is the other side of the same board. This is a fully balanced configuration and thus we ground the input center tap. Noise rejection should be very high. Tests conducted after construction revealed noise reduction by at least 20db. The output leads from the transformer go to the vacated caps inputs. The center tap is connected to the +48v input. This configuration works surprisingly well for the non-phantom power situation too, as there is a signal path to ground. OK now the reassembly:
For this project I didnt concern myself with aesthetics and chose to use electrical tape to secure the transformer to the board. The front (class A preamp) and rear board are reconnected using the connectors. Here is a view of the whole project before closing the lid and firing it up:
Ok now we need to test this thing! The nice folks at Guitar Center sold me a quality signal splitter cable that I then hooked up to the input sections. Here it is all hooked up and ready to compare the two sections:
Hey it works! The levels are a little off because I didn’t properly load the transformer to get the correct gain, also slightly worse noise figures and frequency responses. My ear cant hear 10Hz or 49.99KHz, and neither can the ADA-8000. However, pushing up the levels all the way up to 70db gain there is a 20db drop in broadband noise. A contributing factor is the fact that transformers have a much lower operation band and common mode (hum) rejection is extremely high. Still impressive and useful.
Ok now is the time to actually send in the white noise and compare the signals:
Hmmmmm… Looks the same! Does that mean that they sound the same? Not in a million years! They may have an identical frequency response but look at this null test. (A null test is combining the signals in a way that is suppose to cancel them out so you can detect the difference. In this case it was done with one signal phase inverted):
Now we see the picture! The difference between them is called phase shift. Transformers add an invisible/inaudible component to music called phase shift – for each frequency phase is shifted ever so slightly in a different direction. This is inaudible by the human ear alone. However: When we mix multiple channels together, the unexpected phase shifts mean a broader mix and less phase cancellations. An overall easier mix by far, hence: Transformer input channels are preferred by studios for better noise rejection, hum rejection and introducing musical phase shifting into the music.
I hope to get my hands soon on a real 1073 and compare the two!