SSL-2 audio interface

[Tubetec]

I bit the bullet this morning and emailed SSL about it ,
I gave them my best appraisal of the issue at hand ,
finger crossed it lands on the desk of someone helpfull in their technical department ,
It looks clear cut to me that theres an underlying issue with the grounding strategy , it allows an unwanted feedback path exist and at some point down the line the situation worsens and manifests itself as full-on 'motorboating' .

Awww man , I hope this doesnt turn into another 'Dave vs Goliath' scenario(story of my life) and we find a satisfactory conclusion .

 

[Tubetec]

Oh well ,
didnt realise its a bank holiday in the UK , so just an automaton responding for now ,
Hope your all getting the same nice weather as were having here in Ireland

 

[Tubetec]

A few interesting developments ,

Nicads as we know take half a dozen or so cycles from new to condition the cells , the cell packs I got had evidently sat a long time on the shelf , they work ok after a few charge cycles ,but I still feel the internal resistance is a bit high , I get around 3500mah into them on a 1A charge , on discharge at 1A I only get around 2500ma back , from a 4000mah cell , the supplier has agreed to replace them .

I made a permanent connection from the headphone out 0v to chassis , the osscilation is gone , once the lid is back on the box and its all sealed up any trace of the instabillity is also gone , even at full gain on hi-z input (open) and full volume on the headphone output .
There is a trace of residual switching noise , the magnitude of which varies depending on the quality of the power source , as the nicads have had a chance to form this noise drops lower and lower .
I also replaced the wire link bridging out the blown regulator with a small inductor , that gets the internal noise down another handfull of db . As it is now the hi-z input is very usable , Im hoping the fresh nicads will give a small further improvement and along with better wiring and maybe some extra passive filtering should mean the internal switching harmonics are almost entirely masked by circuit noise .

SSL might be dealing with a backlog of emails after the long weekend , so no response yet ,
Hopefully my thread here hasnt caused an avalanche of inquiries from thousands of SSL2 owners ,
Either way Im pretty close now to solving the issue off my own bat , still be nice to get a chance to discuss the issue with someone in SSL though , that might never happen ,being realistic about it .

I'll post a pic of the noise floor of the unit later for comparative purposes with HP200CD's results .

I had been thinking of investing in the SSL SiX mixer , kinda hangs in the balance now depending on what happens next .

 

[Tubetec]

Nothing appart from an automated email responce from SSL yet

 

[Cappy]

I decided to make a bid on an SSL-2 usb interface secondhand ,
The shock-jock wannbe Covid lockdown sensation fantasy didnt work out in the end,
theres a pile of these units out there for sale for silly money ,

noise ,distortion and bandwidth tests show very good performance ,

I was able to pick up the unit with the intact VST plugin liscence for 50% full rip ,
The Model 2+ incorporates a few extra features like a second heaphone cue and unbalanced in/out on phono ,
Model 2 and 2+ share the same pcb so the base medel provides for some possible hacks like adding an extra set of inputs ,

The 4K button is a farce , it would be better off reconfigured as a channel mute switch ,

 

[Cappy]

What makes the 4k button a farce?

 

[Tubetec]

Well it does the same as many modern condenser mics by applying a broad mid/high bump , ok it might work with certain source/mic combinations ,
I'd much prefer it as a channel mute switch instead .

 

[Tubetec]

I got a reply from an SSL tech , so thats good news at least .

 

[Tubetec]

Ive been doing a few further experiments with battery power via usb , with some curious results I cant really explain .
So intitially I connected the interface with a USBC -USBC cable with an older style USB A adapter at the computer end , that works fine , interface is picked up , drivers install correctly and all the sample rates are available to choose from in the Asio panel .
Then I tried powering the unit via battery , no connection to the computer , that also works fine .
Then I hacked the USBC cable , spliced in the battery on the positive and negative+sheild ,positive wire from the pc is dissconnected ,USB Green/white connected through , the unit powers up as usual ,but its not picked up at all by the pc .
Then I made up another cable , USB A at the pc end , USB C at the interface end , USB data (green and white) connected as usual , 5v positive from the PC dissconnected , 5v negative and sheild connected to battery minus and on through to the computer , the device is seen by the PC but the connection reverts to USB 1 so I can only choose 44.1 or 48 khz , its the same four wires but somehow Windows picks up the device differently and I get that windows message saying this device can perform faster in another port .
Im at a bit of a loss to try and explain why this happens , maybe someone here knows more about the USB system can explain .

 

[Tubetec]

The guy who sold me the nicads exchanged them with fresh from the factory cell packs
Noticeable increase in capacity , better voltage stabillity and an even lower amount of regulator switching noise in the interface output , The 4000 mAh packs will power the interface all day long ,

Its become clear the moment the unit touches the computer usb bus noise in the analog signal path jumps up ,even with an external battery powering the interface .
This particular laptop doesnt have a battery , so the mains adapter adds its own dirt into the mix ,

It seems to me the power and USB data signal down the same few feet of cable causes unwanted interactions , typical USB cabling has extremely thin power conductors ,so voltage drops and capacitive coupling do you no favours at all .

As I said earlier , its the analog signal path Im interested in here , If I can clean up the noise floor by removing the digital functions of this device ,Im willing to consider that option .

The contact I made over at SSL was helpful ,
I asked for a partial schematic of the PSU section , I told him that I didnt need the details of their proprietary audio circuits . Anyway he kindly sent me what I wanted , explaining that the details of the rest had to remain under wraps .
I wouldnt mind an opinion on it ,but I wont post it up openly without express permission to do so .

 

[Tubetec]

The regulator/switch U36 ,which I blew up at the very start I had left in place on the pcb , it turned out it had a partial short from input to output , when I applied the two nicad packs the equalisation current from the two batteries burnt a direct path through ,

Anyway I finally extracted the offending component with the scalpel ......
Now I set the unit up with the nicad pack delivering 5.4v into the audio supply , where originally there was only 4.5v , downregulated from the USB rail .
Im using the USB 10,000mAh powerbank with a 'power only' 2 wire and screen USB charge cable to energise the USB part of the interface , It has much larger conductors than a standard USB data cable.

Swapping over to mains powered computer laptop for the USB power ,
Ive got REW calibrated and it correctly shows the vast amount of the noise floor is in the region of -130dbv , there is a bump at the upper end of the range centered on 100Khz ,
around 15db higher noise in that area ,192khz sample rate .
Now we share a 0V connection with the computer some little squrrley whirley sounds are induced when I come in contact with the metal front panel of the interface , but we can work around that .

One thing thats very impressive about the noise floor of the interface is when you apply or remove the phantom power the audio is muted for around 3 seconds , in REW its visible , noise floor plummets to nearly -200dbv for a few seconds while the 48v circuits stabilise , then crossfade from zero ,so no clicks or pops .

Theres a 100uf/25v on the pcb , directly after the blown regulator in the audio 5v line , its one of those small SMD electrolytics ,it does a very poor job for itself .
I placed a small foil cap across the SMD cap and added a quality 100uf out in the battery connector , thats again changed the spectrum of the noise floor ,appreciably attenuating hi frequency mush .
I'll check again with REW later ,

I'm fitting a 4 pin XLR but USB power will still be available in the usual way ,

 

[Tubetec]

Thats the result of HP100's first test at the top,
Look at the difference below ,same conditions , hi-z input ,max gain , 192khz sample rate ,
Ive been able to reduce the noise spikes by more than 40db .
Ive made my reference level 0.775v rms in my measurements so with the gain faded down I get noise of around -130dbv right across the audio range , there is however a bump at the upper end of the scale which means the averaged noise level doesnt change all that much .

I added a 1000uF electrolytic after the battery , with an axial inductor of 0.6 ohms in the positive lead to the pcb and another 1000uF suplementing the 100uF cap on the pcb . As the capacitors have charged up and formed backround noise levels have changed further for the better since I took the measurment last night , I still have some very tiny instabillity in ch2 , the magnitude of which is now also greatly reduced .
I'll probably try adding an extra inductor into the negative battery lead next , the Nicad has around 5.4v on load so can easily afford to drop a few hundred mV in the extra series resistance ,
Note in the third picture , mic/line input at minimum gain , theres still a big bump in the noise floor of the unit in the 80-100 khz region , better inductances might tame that some more .

 

[Tubetec]

I allowed things another day of settling in time ,
The seething mess that was the noise floor is now back to random white noise , no trace of whine ,fizzle or hum now .Numerically the noise measurement doesnt show much difference ,but the noise floor is now as innocuious as it gets to the ear , a world away from what it was to begin with .
I'll have a look again in REW tomorrow to see whats happened to the noise hump in the 80-100khz region , I'll also post a wav file for comparative purposes for other owners of the SSL2/2+.

 

[Tubetec]

Below is the diagram of what Im using to power the audio 5V rail in the SSL2 ,
making the 0V connection between the headphone and line out sockets took away the noise spikes in the output , the extra inductances and capacitances reduce the backround fizzle and fuzz to a very large degree ,
But I still appear to have a noise bump higher up , something like another reduction of 20db in the 80-100 khz region should leave me with an exceptionally quiet supply .

The weird thing about all this is nicads have noise in the region of -200dbv , so any noise that does exist comes from within the circuits themselves , or maybe its picked up externally by the unscreened battery and its associated cables , to that end I got a small metal cash box to house the batteries , inductors and capacitances .

The diagram below is probably what I'll try next , adding extra inductances in the 0V line .
Hopefully the fully balanced approach should get me where I want to be as far as noise rejection .

 

[capacitor]

I know you haven't had a ton of responses, so as a novice I just wanted to drop a quick word of thanks. I'm finding this thread very interesting and instructive & I wish I had something to contribute. Please carry on!

 

[Tubetec]

Muchly appreciated Capacitor ,
The thread has a good read count ,
Im a little suprised I didnt get more responces ,
I know its been a rather long and winding , in some ways its as much to keep a record for myself .

In any case Im really happy with the results Im getting from the batteries replacing USB power , theres definately a lot of room for improvement in many audio interfaces , the results I got speak for themselves ,
While the spikes werent that noticable in either mic or line mode , they were still there to some degree , Hi-z mode wasnt of use to me the way it was originally ,

The nice thing about the mods is there non destructive to the pcb ,ok you need to remove the preregulator switch ,U36 but thats easily done with a razor blade , no real chance to do damage ,
Then theres the bridging of a few pins near the input sockets , even with standard USB power the proper grounding produces very much better results .Dual rail standalone battery power is the icing on the cake for me and it surely pushes this device up a notch or two in terms of quality of both the analog and digital output .

 

[ruffrecords]

I notice from the tear down pictures that all the pots are dual ones. I can understand that for the stereo controls (monitor, monitor mix and phones) but not really required for the individual gain controls. I guess it is probably just as cheap for them to buy all the same type considering how many they buy or is there some more subtle reason?

Cheers

ian

 

[Tubetec]

Maybe using a two track pot with both sections in parrallel might help iron out any tollerances better ?
There is some small inconsistancies towards the very upper end of the gain pot travel , but usually you wont end up running the device flat out ,so probably not much of an issue .
I didnt find schematics for the audio circuits in the SSL2 and SSL themselves were only able to release a few details of the power supply path to me .

 

[ccaudle]

Im a little suprised I didnt get more responces

My guess is that most people don't have one of these, so don't have anything specific to contribute, and the people that do have one are willing to let you do all the potentially device killing experimentation before they decide to play along at home.

Does the headphone output have any RFI protection? Your description of getting better performance when tying the headphone out shell connection back to device ground made me wonder whether they left it floating because the typical case is a fully floating output (i.e. just headphones), and the worse performance you see is only seen when another piece of non-floating equipment is connected.

 

[Tubetec]

Here I matched exactly the conditions in HP100CD's test in REW ,
Noise is reduced by around 13db ,
Theres a complete lack of any kind of noise spurs or spikes even though its connected to a laptop running off the mains . This is the measured noise with the input channels faded down,with the audio running off nicads and the USB side of the circuit bus powered.